2 - CDTA

intimidatedunadvisedΒιοτεχνολογία

19 Φεβ 2013 (πριν από 4 χρόνια και 3 μήνες)

486 εμφανίσεις

Technologies Avancées - Numéro 15 - Janvier 2003
1
TECHNOLOGIES AVANCEES est une revue semestrielle éditée par le Centre de
Développement des Technologies Avancées et traitant des thèmes suivants :
￿￿Architecture des Systèmes
￿￿Génie-Logiciel
￿￿Intelligence Artificielle et Systèmes Experts
￿￿Système d'Information
￿￿Automatique, Robotique et Productique
￿￿Image et Parole
￿￿Télécommunications
TECHNOLOGIES
AVANCEES
Directeur de la Publication
Bessalah Hamid, Directeur du CDTA
Comité International de Lecture
H. BAUDRAND,ENSEEIHT - France
M. BELLANGER,CNAM - France
M. CHERIET ETS Montréal - Canada
P. COIFFET,Université Paris XI - France
B. COURTOIS,C.N.R.S - France
P. EXERTIER,CERGA - France
N. GABOUZ,U.D.T.S - Algérie
J.P. GUILLOIS,ENST - France
H. HACENE,U.S.T.H.B - Algérie
R. HOUACINE,U.S.T.O - Algérie
P. LEPRINCE,Université de Paris XI - France
I.N. MIHAILESCU,Université de Bucarest - Roumanie
F. NEKKA,Université de Montréal - Canada
A. PRUSKI,Université de Metz - France
S. SAHLI,Université de Constantine - Algérie
M. SELLAMI,Université de Annaba - Algérie
C.C. TSCHERNING,Université de Copenhague - Danemark
A. TOUZI,SEES/MS Adrar - Algérie
Secrétariat de Rédaction :
H. HAMOU
Z. SIHAOUI
CENTRE DE DEVELOPPEMENT DES TECHNOLOGIES AVANCEES
CITE DU 20 AOUT 1956, BP N°17, BABA-HASSEN, ALGER, ALGERIE
Tél.: (213) 21.35.10.18 / 21.35.10.40 / 21.35.10.75 - Fax: (213) 21.35.10.39
E-mail: revue@cdta.dz
￿￿Microélectronique
￿￿Traitement de surface
￿￿Télédétection
￿￿Technologie et Application des Lasers
￿￿Physique et Application des Plasmas
￿￿Biotechnologie
2
Technologies Avancées - Numéro 15 - Janvier 2003
5
11
17
22
29
35
S o m m a i r e
recherche et développement
The effect of disorder on the optical absorption edge of
SIO/TEO2 thin film
B. YAGOUBI, C.A. HOGARTH
Review of biosorption applications to nuclear waste
treatment
H. BENAISSA
Effect of the gauges dimensions on the piezoresistive
pressure sensor response
Z. DIBI, A. BOUKABACHE and P. PONS
Supervising of teleoperated movements whith a
neurofuzzy approch
Z. AHMED FOITIH
Validation des mesures gravimétriques par la méthode de
collocation
S.A. BEN AHMED DAHO, S. KAHLOUCHE
Analysis method of wavel et combinations
A. HAJRAOUI and A. HAJRAOUI
4
Technologies Avancées - Numéro 15 - Janvier 2003
5
THE EFFECT OF DISORDER ON THE OPTICAL
ABSORPTION EDGE OF SIO/TEO THIN FILM
2
*B. YAGOUBI - **C.A.HOGARTH
*Physics Department - University of Mostaganem
BP 227, Mostaganem, Algeria.
**Department of physics, Brunel University
Uxbridge, Middlesex UB8 3PH, Great Britain.
Abstract
The optical absorption edge of vacuum co-
evaporated SiO/TeO in thin film form, is
2
studied. A general equation based on the
absorption being due to indirect transitions in k-
space is found to be compatible with our
experimental results. We show, in particular,
how the optical energy gap depend on the
compositions of the two oxides. The optical gap
systematic change is observed only in certain
range of the compositions. Whereas it is not
respected in other compositions. This could be
due to the change of the optical energy gap of
SiO from 2.25eV to 2.73eV at rapid and slow
evaporations respectively.
The optical absorption edge at low energies of
nearly all amorphous semiconductors as well
as oxides is characterised by an absorption
coefficient which rises exponentially with
photon energy. The absorption edge of
amorphous SiO/TeO thin film is, however,
2
about as sharp as the direct edge in their
corresponding crystalline materials. This
makes a strong support as evidence against the
interpretation that the localised states in the
gap, are responsible for the Urbach tail.
Introduction
It is well known, in microelectronics, that thin
films of thickness less than 1 can be used as
capacitors, protective coatings, temperature
and pressure sensors, thermistors and
insulating layers.
Recent l y, t he el ect ri cal and opt i cal
characteristics in a number of thin oxide films
and amorphous materials, (see Kocka et al [1],
Nebel et al [2]), have been studied. Much
experimental data has been gathered, but the
electrical and optical behaviour itself in these
materials, is not yet completely understood.
Their characteristics depend mostly on the
preparation technique used and their
behaviour
may differ from that of the corresponding bulk
material. In the present work, thin films of mixed
SiO/TeO were prepared by vacuum co-
2
evaporation. Most workers agreed that films
prepared by this technique present a high state
of disorder. Many workers have investigated the
electrical and the optical properties of SiO, TeO
2
and SiO mixed with other oxides(Ilyas and
Hogarth [4], Bidadi and Hogarth [5], Al-ani et al
[5]). In this paper we discuss the disorder effect
on the optical absorption edge in SiO/TeO thin
2
film.
1. Experimental techniques
The mixed oxide films of SiO and TeO of
2
thickness of order 300nm, for all experiments,
were prepared by thermal evaporation in
-6
vacuum better than 10 torr. Inside the
evaporation system, a substrate where the
material to be deposited, can be protected by a
shield during the initial source heating to adjust
the deposition rate, so that, the film can be grown
at a preadjusted deposition rate.
Tantalum boat was used as a source heater for
the evaporation of both SiO and TeO. The
2
substrate for thickness measurements was
masked by a thin wire in close contact to its
surface. After evaporation of the film the wire
was removed and a thin film of aluminium was
deposited on the substrate. The thickness of the
oxide film was measured using a Sloan
instrument angstrometer (model M-100).
Accur at e cal cul at i on f or t hi ckness
measurements may be found elsewhere
(Márquez et al [6]). The optical absorption
measurements were made using a Perking-
Elmer Lambda 3UV/VIS spectrophotometer
equipped with a source and a reference beam.
The effects of the substrate on the UV
characteristic was minimised by placing another
cleaned uncoated substrate in the path of the
reference beam.
Technologies Avancées - Numéro 15 - Janvier 2003
6
a
b
Figure 2. Electron micrograph(a) and diffraction(b) patterns of 56%SiO/44%TeO
2
thin
film after annealing
Figure 1. Electron micrograph(a) and diffraction(b) patterns of 56%SiO/44%TeO
2
thin film at room temperature.
a
b
Technologies Avancées - Numéro 15 - Janvier 2003
7
The structure of layers of thickness of order
50nm was deposited on a carbon-coated mica
substrate held at 90°C and it was investigated in
an electron microscope (JEO model JEM7) by
electron diffraction.
2. Results and discussions
The electron diffraction pattern at room
temperature, shown in figure 1b, indicates that
the structure of SiO/TeO thin film is amorphous,
2
while at temperature higher than 500K figure 2,
the structure tends to become crystalline as
indicated by the appearance of the sharp rings.
The electron micrograph shown in figure 2a
shows some kind of small islands but without
sharp edges which means that they are not
metallic crystallites.
The electron micrograph shown in figure 2b for
the same sample shows, however, a random
distribution of small metallic islands about the
same shape and with sharp edges indicating
that these islands could be chains of metallic
crystallites. Their growth effect, usually, causes
a decrease in the d.c conductivity as the
temperature is raised.
It is, therefore, more convenient to use the Davis
and Mott [7] model to study the SiO/TeO optical
2
absorption at room temperature. These authors
neglected all transitions in which both the initial
and the final states are localised. This
assumption is based on the argument that in the
same region of space there is a little chance of
finding a localised state derived from both the
valence and conduction bands. In this case they
found the following formula for the optical
transition.
2
￿h￿￿￿(h￿￿- E ) (1)
opt
where is the frequency of the incident radiation
and E is defined as the optical energy gap and
opt
corresponds to the smallest energy separating
the localised states of one propagating band (eg.
valence band) tail and the extended states of the
other (eg. conduction band ).
Figure 3 shows the absorption spectra near the
fundamental edge for a series of co-evaporated
SiO/TeO samples about the same thickness'
2
but different proportions. It is seen that the
shape of the absorption edge is not abrupt as in
crystalline materials and clearly different from
those expected for a non-metallic crystal. It is
also observed that the position of the
absorption edge moves towards shorter
wavelengths with the increase of SiO content in
the composition. As the content of silicon oxide
is further increased (87 % SiO), the position of
the absorption edge seems to return back to
shorter wavelengths, that is the lower knee of
c u r v e ( d ) m a y b e l o c a t e d
300
350
400
450
500
550
600
0,0
0,2
0,4
0,6
0,8
1,0
￿
(nm)
Figure 3. Optical transmittance versus wavelength for various
compositions of SiO/TeO
2
amorphous thin film (300nm thick).
Optical
transmittance
Technologies Avancées - Numéro 15 - Janvier 2003
8
Technologies Avancées - Numéro 15 - Janvier 2003
0,0
0,5
1,0
1,5
2,0
2,5
3,0
3,5
4,0
4,5
0
100
200
300
400
500
600
700
800
900
h
￿
(eV)
(
￿
h
￿
)
1/2
(cm
-1/2
eV
1/2
)
Figure 4. (
￿
h
￿
)
1/2
versus photon energy for various compositions of SiO/TeO
2
amorphous thin film (300nm thick).
0
1
2
3
4
5
3
4
5
h
￿
(
eV)
56%SiO
30%SiO
87%SiO
Log
(
￿
)
Figure 5. Log(
￿
) versus photon energy for various compositions of
SiO/TeO
2
amorphous thin film
9
Technologies Avancées - Numéro 15 - Janvier 2003
somewhere between the transition knees of
curves (a) and (b) corresponding to 79% SiO
and 56% SiO respectively. The two knees
observed in curve (d) of the optical transmittance
could, probably, be due to the deposition rate
variation during the evaporation procedure.
The optical absorption coefficient calculated
from the transmittance curves, is given by￿= (2.303/d)Log(I/I ) (2)
0 t
where I and I are the intensities of the incident
0 t
and transmitted light respectively and d is the
film thickness.
The high absorption region may be analysed by
1/2
plotting (h) against h in accordance with
equation (1). These results are shown in figure 4
and are of linear form in the high absorption
regi on. The val ue E obtai ned from
opt
extrapolation of the linear regions are 1.51eV,
1.87eV, 2.37eV and 2.04eV for 30% SiO, 56%
SiO, 79% SiO and 87% SiO respectively. We
notice that the optical energy gap increases as
the SiO content increases up to 79% SiO. In
order to increase the SiO content little further in
the structure, we have to increase the SiO
deposition rate. The optical gap value obtained
for 87% SiO which is less than that of 79% SiO
could be, therefore, due to the change of SiO
optical energy gap from 2.25eV to 2.73eV at
rapid and slow evaporation rates respectively
[5]. This is due to the fact that at rapid
evaporation, the atoms to be deposited on the
corning glass substrate would not have enough
time to arrange themselves in a crystalline form,
instead the material becomes more disordered
leading to a deeper localised states tail in the
forbidden band, hence a smaller optical gap.
The density of the localised states in the
propagating band tails is of a particular concern
mainly because of its contribution to the optical
transitions and to the conductivity at low
temperatures (thermally-activated hopping
conduction). It has been suggested that these
localised states are due to the lack of long-range
order (atomic disorder). Most workers believe
that the role of these localised states may be
seen at low energies corresponding to low
4
absorption (10 cm). It has also been suggested
that the Urbach tail could be interpreted as
arising from transitions between these localised
states [8].
In fact, the optical absorption edge at low
ener gi es of near l y al l amor phous
semiconductors is characterised by an
absorption coefficient which rises exponentially
with photon energy. The absorption edge of
amorphous SiO/TeO thin films shown in figure 5
2
is, however, not exponential form. These results
are not similar to those obtained for amorphous
silicon [9] and amorphous germanium [10]. In
these materials, it is seen that the absorption
edge, is about as sharp as the direct edge in their
corresponding crystalline materials. The results
found in amorphous SiO/TeO, do not obey the
2
following relation: ￿ = exp(h/E) (3)
t
known as Urbach rule and where E is supposed
t
to be the width of the localised states band. The
results obtained on SiO/TeO may indicate either
2
the material is not homogenous, which is not the
case, or that the assumption based on the fact
that Urbach tail arises from the localised states,
can not be satisfied. Following the arguments of
Dow and Red field [11], it is rather more
reasonable to say that the Urbach rule arises
from an electric field broadening of an exciton
because the exciton states do exist in both
amorphous and crystalline materials.
Conclusion

The electron diffraction pattern at room
temperature, indicates that the structure of
SiO/TeO thin film is amorphous and thus, the
2
Davis and Mott [7] model could be used to study
the SiO/TeO optical absorption edge at room
2
temperature. We have seen that the optical
energy gap increases as the SiO content
increases up to certain proportion. For further
increase of the latter, the energy gap seems to
shift back to lower energies. This is, probably,
due to the change of SiO optical energy gap from
2.25eV to 2.73eV at rapid and slow evaporation
rates respectively. The main reason for this is
that at fast evaporation rate, the atoms would not
have enough time to arrange themselves in a
crystalline structure. The material becomes,
therefore, more disordered leading to a deeper
localised states tail in the band gap, hence a
smaller optical gap. The optical absorption edge
at low energies in amorphous SiO/TeO thin film
2
could arise from an electric field broadening of
an exciton.
Acknowledgements
One of us (B.Y) would like to acknowledge the
financial support from algerian government.
10
Technologies Avancées - Numéro 15 - Janvier 2003
REFERENCES
[1] J. Kocka, O. Klima, E. Sipek, C. E. Nebel, G.
H. Bauer and M. Hoheisel.,Phys. Rev., B, 45,
6593, 1992.
[2] C.E. Nebel, R. A. Street, N. M. Johnson and J.
Kocka., Phys. Rev., B, 46, (1992) 6789.
[3] M. Ilyas and C.A. Hogarth., J.Mater. Sci.
Letters., 2, 535, 1983.
[4] H. Bidadi and C.A. Hogarth., Thin Solid
Films., 27, 319, 1975.
[5] S.K.J. Al-ani, K.I. Arshak and C.A. Hogarth,
J.Materials Science.,19, 1737, 1984.
[6] E. Márquez, J. M. González-Leal, R.
Jiménez-Garay, S. R. Lukic and D. M. Petrovic.,
J. Phys., D:Appl. Phys., 30, 690, 1997.
[7] E.A. Davis and N. F. Mott, Phil. Mag.. 22 903,
1970.
[8] J. Tauc, R. Grigorovici and A. Vancu,
Phys. Stat. Sol. 15, 627, 1966.
[9] J.E. Fischer and T. M. Donovan, J. Non-cryst.
Solids, ,202, 1972.
[10] T.M. Donovan, W. E. Spicer and J. M.
Benett, Phys. Rev. Lett., 22, 1058, 1969.
[11] J.D. Dow and D. Redfield, Phys. Rev. B1)
3358, 1970.
11
Technologies Avancées - Numéro 15 - Janvier 2003
REVIEW OF BIOSORPTION APPLICATIONS
TO NUCLEAR WASTE TREATMENT
H. BENAISSA
Laboratoire de Matériaux Sorbants et Traitement des Eaux, Département de Chimie
Faculté des Sciences, Université de Tlemcen, B.P. 119, 13000 Tlemcen, Algérie.
Tél./Fax. (213) 0 43 20 03 30
Abstract :
Application of biosorption technology to the
treatment of wastes water produced from the
nuclear processes has been given significantly
attention recently by the research community. A
brief update on the developments and
applications of biosorption in the area of nuclear
waste treatment is presented.
Key Words: nuclear waste treatment -
biosorption radionuclide ions - material of
natural origin.
Introduction:
Certains processing operations for nuclear
materials generate waste waters in large
quantities which contain a variety of
radionuclide ions some of which are toxic and
some of which are valuable [18]. These liquids
cannot be release into streams or sewers
without causing environmental damage.
Treatment of such waters to remove the
radionuclide ions is required [18], [13].
Uranium is one of the most serious
contamination concerns because of its
radioactivity and heavy - metal toxicity [9]. The
biological leaching process of the uranium ore,
produces vast quantities of complex, dilute
uranium solution with typical uranium
concentration of 200-500 mg U / L, low pH
(usually in the range of pH 1 to 2), and a variety
of other ions such as heavy metals and sulfates
[34]. The increasing demand for uranium has
prompted a search for cheaper and more
efficient methods for its recovery [22]. Thorium
is a nuclear fuel element and is often present in
considerable quantities in uranium ores. The
interest in sequestering thorium from waste
solutions originates from the need to protect the
environment from this radioactive element and
its decay products, and from the potential
commercial value of the element as a nuclear
fuel for breeder reactors [23]. Because effluents
of nuclear processes often contain radium-226
activities two orders of magnitude above the
standard of 10 pCi / L, some form of radium-226
removal is required [26]. Neptunium, an
activation product of uranium-238, can occur in
wastewaters from nuclear facilities. A discharge
-5
limit of 3.10 µC / L has been proposed for Np.
i
There is concer within the nuclear industry that
this requirement could be difficult to meet for Np
using existing technologies [20]. Malfunctions
and leaks of nuclear reactor cooling fluids
containing highly radioactive elements of
different kinds further demonstrate the urgent
need for efficient and element selective
materials of this sort [13]. Existing processes for
treating such waste waters suffer from many
disadvantages. Currently available resins and
activated charcoals are or become ineffective
when the metal concentration is below certain
levels or when the trace metals are
accompanied by major quantities of common
salts, and when the removal of carrier-free
radioelements is attempted [13]. There is,
therefore, a need for a simple inexpensive but
efficient means for the removal of radioactive
contaminates such as radium, polonium,
strontium, uranium, plutonium and thorium
residues from waste waters [36].
Cer t ai n speci es of mi cr oor gani sms:
filamentous fungi, yeasts, algae, bacteria, have
been shown to selectively retain radionuclides
ions from an aqueous environment [22], [26],
[36], [21], [27]. This ability is exhibited in dead
microorganisms as well as live ones.
Biosorption is the sequestering of metal ions by
solid materials of natural origin. This uptake is
often considerable near those of commercial
ion exchange resins and frequently selective,
and occurs via a variety of mechanisms [21].
The biosorption of radionuclides from dilute
aqueous solution has
12
Technologies Avancées - Numéro 15 - Janvier 2003
been proposed as an alternative process for the
treatment of dilute radioactive wastes from the
nuclear industry economically [27].
The present paper provides a brief update on the
developments and applications of biosorption in
the area of nuclear waste treatment from
literature. We have restricted our coverage to
examples of radionuclide ions biosorption by
inactive biomass.
1. Phenomenon of biosorption
One of the processus that has been shown to be
selective in sequestering radionuclides from
complex dilute solutions is biosorption [34].
Biosorption is the sequestering of metal ions by
solid materials of natural origin. This uptake is
often considerable and frequently selective [21]. A
number of different types of inactive microbial
biomass have recently been identified as
adsorbents that present competitively high
radionuclide uptake capacity and selectivity which
depends strongly, both on the biomass type and
the pH [22], [26].
An important consideration for the practical
utilization of biosorption phenomenon for the
uptake of radionuclides from dilute solutions is the
amount of radionuclide that can be accumulated
by the biosorbent mass [22]. Equilibrium
biosorption isotherms have been used to quantify
the radionuclide uptake capacity of various types
of biomass which were also compared to others
types of adsorbents [27]. The amounts of
radionuclides which biosorbents can accumulate
vary from a few micrograms per gram of
biosorbents to several percent of the dry cell
weight [22]. Table 1 presents several examples
where subst ant i al amount s of cert ai n
radionuclides have been observed to be
accumulate.
As example, Rhizopus arrhizus biomass,
produced as a by product of industrial
fermentation, has been shown to possess very
high uptake capacity for use as a biosorbent of
radioactive ions [22], [26]. These fungal cells have
uptake capacity of 15% of biomass ( dry weight )
has been defined as an economic threshold for
practical applications of biosorption when
compared with alternative methods such as
traditional adsorption, ion exchange, chemical
precipitation, solvent extraction and reverse
osmosis [9], [3]. Table 2 compares uranium and
thorium capacities of several materials at
equilibrium concentration of 5mg/Liter. This
concentration is specified as a maximum
permissible uranium and thorium in surface
waters by certain developed countries [22].
Table 2: Uranium and Thorium Biosorption
Uptake Capacity (mg/g) at pH 4 [22].
The biomass of Rhizopus arrhizus at pH 4
exhibited the highest uranium and thorium
biosorptive uptake capacity. For example, this
exceeds the capacity by 2.5 times of a common
anionic exchange resin (IRA-400) used by
uranium production companies for selective
separation of uranium from other ions in solution
[22]. Table 3 showns a brief listing of radium
uptake capacities of various adsorbents.
Radioactive Ion
Rhizopus arrhizus
Ionex IRA-400
Activated Carbon
F-400
Uranium
80
31
15
Thorium
140
3
20
Radioactive
Ions
Uranium
Thorium
Microorganism
Rhizopus
arrhizus
Saccharomyces
cerevisiae
Pseudomonas
aeruginosa
Penicillium
chrysogenum
Rhizopus
arrhizus
Metal Uptake
-1
(g metal g
cells dry wt)
0.18
0.15
0.15
0.08
0.17
Reference
[Tsezo and
Volesky,
82b]
[Strandberg
et al., 81]
[Strandberg
et al., 81]
[Jilek et al.,
75]
[Tsezos et
Volesky,
82a]

Adsorbents
Natural zeolite
Manganese
zeolite
Zirconium
salts
Bio-Rex ion
exchange
resin
Activated
Carbon
Municipal
wastewater
activated
return sludge A
Municipal
wastewater
activated
return sludge B
Radium
Uptake
Capacity
(pC/ g)
i
2800
2100
2750
2900
3500
40.000
75.000
References
[Arnold and
Crowse, 65]
[Arnold and
Crowse, 65]
[Arnold and
Crowse, 65]
[Tsezos and
Volesky,81]
[Tsezos and
Keller, 83]
[Tsezos and
Keller, 83]
[Tsezos and
Keller, 83]
Table 1: Levels of Radionuclides Accumulation
by Microorganisms.
Table 3: Examples of Radium Adsorption /
Biosorption Uptake Capacity.
13
Technologies Avancées - Numéro 15 - Janvier 2003
The biological origin adsorbents have a
superior radium uptake capacity compared to
conventional adsorbents. Resting cells of
several species of microorganisms removed
neptunium (Np) from an aqueous solution.
Concentrations of up to 15 mg Np per g cells
(dry weight) at pH 4, were obtained [20].
It is evident from these examples that biological
origin adsorbents have emerged as a novel
class of adsorbents of superior radionuclide
uptake capacity compared to conventional
adsorbent. It is therefore reasonable to propose
that these adsorbents may form the basis for a
new technology for radionuclide sequestering
[26]. It should be noted that the values and
comparisons reported in the literature for
radionuclides loading capacity only have a
relative meaning because of different testing
conditions (e.g. temperature, Ph and
wastewaters composition) and methods [9].
2. Mechanism of biosorption
The understanding of the biosorption
mechanism is very important for the successful
manipulation of the biosorption process, for a
better interpretation of biosorption results, and
for the efficient design of technical applications
of the phenomenon [24]. Biosorption occurs via
a variety of mechanisms including active
transport, ion exchange or complexation, and
adsorption or inorganic precipitation. While the
first mechanism is associated with living cells,
the later mechanisms have been reported for
living and nonliving microorganisms [23]. The
cell wall of any microorganism contain many
potential sites for the uptake of ions and it is
unlikely that any one type of molecule or
functional group is responsible for the
biosorption of the radionuclides [2], [4].
As example, the examination of the mechanism
of uranium and thorium by inactive cells of
Rhizopus arrhizus has been examined, using
electron microscopy, energy dispersive X-ray
analysis, electron paramagnetic resonance and
infrared spectroscopy. It has revealead that the
cell wall of the fungal biomass is responsible for
the observed metal uptake and the chitin, a
structural acetelated aminopolysaccharide of
the cell wall, is the most important cell wall
component with respect to the observed
radionuclide biosorptive uptake [23], [33]. In
order to improve the understanding of
mechanism of uanium uptake by inactive cells
of Rhizopus arrhizus, the uptake of uranium by
chitin along with the use of spectral
investigation techniques was investigated. The
chitin nitrogen and a free radical associated
with the chitin macromolecule are responsible
for the biosorption of uranium as well as of other
cations by the cell wall chitin [23], [25]. The
uranium biosorption includes at least three
process, including uranium coordination with
the nitrogen of chitin, the adsorption of
additional uranium in the cell wall chitin
structure, and the precipitation of uranyl
hydroxide within the cell wall matrix [24]. The
understanding that has developed on the
mechanism of uranium has also allowed the
elucidation of the mechanism by which other
ions present in solution, compete with uranium,
for uptake by the biomass. Such co-ions reduce
the uranium uptake of the biomass by
competing for the nitrogen coordination site on
chitin [27] [23, 24] [25]. Uranium biosorption
mechanisms by biosorbents of various
biological origins may be different. Process
such as complexation, ion exchange,
coordination, adsorption, chelation and
microprecipitation may be synergistically or
independently involved in the metal biosorption
[9], [37].
The mechanism of thorium sequestering by
inactive cells of Rhizopus arrhizus was
investigated following the same method as for
the uranium biosorption mechanism. The
thorium sequestering mechanism appeared
somewhat different from that of uranium [23].
Two processes participate in the uptake. The
first process coordinates thorium to the nitrogen
of the chitin cell wall network, where it is
retained. The second process involves
adsorption of thorium by the external sections of
the fungal cell wall. The two processes, unlike
the proposed hypothesis for uranium
biosorption, appear unrelated. Thorium
adsorption is the outer cell wall section
dominates the overall observed thorium
biosorptive uptake and, consequently, controls
the process.
A generalised mechanism of radionuclide ions
biosorption cannot be advanced at present. The
available evidence suggests that the detailed
mechanism of biosorption is radionuclide ion
and microorganism specific [30].
14
Technologies Avancées - Numéro 15 - Janvier 2003
3. Applications
Commerci al uti l i zati on of bi osorpti on
phenomenon has been rare. There appear to
be no instances for technically sophisticated
processes based on removal radionuclide ions
from wastewaters by biosorption phenomenon
in use on a large scale [18].
Attemps to use biomass of fungal for
radionuclide removal are reported in U.S.
Patent N°. 4 320 093 [36]. The authors have
reported that uranium or thorium cations are
removed from aqueous suspensions or solution
by treatment of the aqueous material with the
biomass derived from fermentation of a fungal
microorganism of the genus Rhizopus, e.g.
Rhizopus arrhizus. The process can be utilized
to treat aqueous tailings from uranium ore
extraction processe, to reduce the radioactive
content of the tailings prior to disposal. In U.S
Patent N°. 4 021 368, Nemec et al., 77 disloses
a process for retention of metal ions such as
uranium from solution, using biomass of
mycelia of microorganisms of fibrous fungi ,
st i ff ened by addi ng t o t he bi omass
pol ymeri zabl e component s, aff ect i ng
polymerization and mechanically granulating
the produit. A patent was issued to Heide et al.,
75 for the use of a matrix of green algae to
extract uranium from seawater. A proprietary
technical biological origin adsorbent has been
developed by [2]. Inactive cells of Rhizopus
arrhizus have been immobilized, into the form of
particles of desirable size, for the sequestering
of radionuclide ions from dilute aqueous
solution. The immobilized biomass particles are
porous and are members of a new generation of
biological origin adsorbents. Batch and
continuous laboratory scale pilot plant
experiments were carried out. The results have
shown that the immobilized microbial biomass
can successfully recover all of the uranium from
dilute ( less than 300 mg U / L ). The biomass
maintained its biosorption capacity of about 50
mg U / L over 12 examined successive
biosorption - elution cycles with no apparent
indication of failure.
REFERENCES
[1] W.D. Arnold and D.J Crowse, "Industrial
Engineering Chemical Process Design
Development”, Vol. 4, pp. 3, 1965.
[2] T.G Beveridge and S.F Koval, "Binding of
Metals to Cell Envelopes of Escherichia Coli K-
12", Applied Environmental Microbiology, Vol.
42, pp. 325-335, 1981.
[3] J.A. Brierley, G.W. Goyak and C.L. Brierley,
"Considerations for Commercial Use of Natural
Pr oduct s f or Met al s Recover y", I n:
Immobilization of Ions by Biosorption, H.,
Eccles, and S., Hunt (Eds.), Ellis Horwood, Ltd.,
Chichester, U.K, 1986.
[4] R.H. Crist, K. Oberhauser, N. Shauk and M.
Ngkuyen, "Nature of Binding Between Metallic
Ions and Algae Cell Walls", Environmental
Science Technology, Vol. 15, pp. 1212-1217,
1981.
[5] N. Friis and P. Myers-Keith, "Biosorption of
Urani um and Lead by Streptomyces
longwoodensis”, Biotechnol. Bioeng., Vol. 26,
pp 21-28, 1996.
[6] E. Guibal, I. Saucedo, J. Roussy and P. Le
Cloirec, "Uptake of Uranyl Ions by New Sorbing
Polymers: Discussion of Adsorption Isotherms
and pH Effect", Reactive Polymers, Vol. 23, pp.
147-156, 1994.
[7] E. Guibal, C. Roulph and P. Le Cloirec,
"Infrared Spectroscopic Study of Uranyl
Biosorption by Fungal Biomass and Materials of
Biological Origin", Sci. and Technol., Vol. 29,
pp.2496-2503, 1995.
[8] E.Z. Heide, M. Pashke, K. Wagener, and
Wald M., "Aus Kultivierbaren Mutanten von
Einzelligen Grunalgen Bestehende Matrix
Sowie Verfahren zur Urangewinnung Mittels
Dieser Matrix", Germany Patent, N°: 2 345
430, 1975.
[9] M.Z.C. Hu, J.M. Norman, B.D. Faison B.D
and M.E. Reeves, "Biosorption of Uranium by
Pseudomonas aeruginosa Stain CSU:
Characterization and Comparison Studies",
Biotechnol. Bioeng., Vol. 51, pp. 237-247,
1996.
[10] R.H. Jilek, K. Prochazka, K. Stamberg,
Katzer J. and Nemec P., "Some properties and
Development of Cultivated Biosorbent", Rudy,
Vol. 23, pp. 282-286, 1975.
[11] J.J. Lenhart, L..A. Figueroa, B.D.
Honeyman and D. Kaneko, "Modeling the
Adsorption of U(VI) onto Animal Chitin Using
Coupl ed Mass Transfer and Surface
Complexation", Colloids and surfaces, Vol. 120,
pp. 243-254, 1997.
15
Technologies Avancées - Numéro 15 - Janvier 2003
[12] Z. Michael, C. Hu and M. Reeves,
"Biosorption of Uranium by Pseudomonas
Aeruginosa Strain CSU Immobilized in a Novel
matrix", Biotechnol. Prog., Vol. 13, pp. 60-70, 1997.
[13] R.A.A. Muzzarelli, M. Wecks and O. Flippin
"Removal of Trace Metal Ions from Industrial
Waters , Nuclear Effluents and Drinking Waters,
with the Aid of Cross-Linked N-Carboxymethyl
Chitosan", Carbohydrate Polymers, Vol. 11, pp.
293-306, 1989.
[14] P. Nemec, H. Prochazka, K. Stamberg, J.
Katzer, J. Jilek and P. Hulak, “Process of
Treating Mycelia of Fungi for Retention of
Metals", U.S. Patent N°: 4 021 368, 1977.
[15] E. Piron, M. Accominotti and A. Domard,
"Interaction Between Chitosan and Uranyl Ions.
Role of Physical and Physicochemical
Parameters on the kinetics of sorption",
Langmuir, Vol. 13, pp. 1653-1658, 1997a.
[16] E. Piron and A. Domard, "Interaction
Between Chitosan and Uranyl Ions. Part 1. Role
of physicochemical parameters", Int. J. Bio.
Macromol., Vol. 21, 327-335, 1997b.
[17] I. Saucedo, E. Guibal, CH. Roulph and P.
Le Cloirec, "Sorption of Uranyl Ions by a
Modified Chitosan : Kinetic and Equilibrium
Studies", Environmental Technology, Vol. 13,
pp. 1101-1115, 1992.
[18] S.E. Shumate and G.W. Strandberg,
"Accumulation of Metals by Microbial Cells", In :
Comprehensive Biotechnology : The Principles
of Biotechnology Engineering Considerations,
Eds. Moo-Young, M., C.L., Cooney, and A.E.,
Humphrey, Pergamon Press, pp. 235-247,
1985.
[19] G.W. Strandberg, S.E. Shumate and J.R.
Parrott, "Microbial Cells as Biosorbents for
Heavy Metals: Accumulation of Uranium by
Saccharomyces cerevisiae and Pseudomonas
aer ugi nosa", Appl i ed Envi r onment al
Microbiology, Vol. 41, pp. 237-245, 1981.
[20] G.W. Strandberg and W.D. Arnold,
"Microbial Accumulation of Neptunium ",
Journal of Industrial Microbiology, Vol. 3, pp.
329-331, 1988.
[21] M.E. Treen Sears, B. Volesky and R.J.
Neufeld, "Ion Exchange / Complexation of the
Uranyl Ion by Rhi zopus Bi osorbent",
Biotechnology and Bioengineering, Vol. 26, pp.
1323-1329, 1984.
[22] M. Tsezos and B. Volesky, "Biosorption of
Uranium and Thorium", Biotechnology and
Bioengineering, Vol. 23, pp. 583-604, 1981.
[23] M. Tsezos and B. Volesky, "The Mechanism
of Thorium Biosorption by Rhizopus Arrhizus",
Biotechnology and Bioengieering, Vol. 24, pp.
955-969, 1982a.
[24] M. Tsezos and B. Volesky, "The Mechanism
of Uranium Biosorption by Rhyzopus Arrhizus",
Biotechnology and Bioengineering, Vol. 24, pp.
385-401, 1982b.
[25] M. Tsezos, "The Role of Chitin in Uranium
Adsorption by R.Arrhizus", Biotechnology and
Bioengineering, Vol. 25, pp. 2025-2040, 1983.
[26] M. Tsezos and D.M. Keller, "Adsorption of
Radium - 226 by Biological Origin Adsorbents",
Biotechnology and Bioengineering, Vol. 25, pp.
201-215, 1983.
[27] M. Tsezos, "The Selective Extraction of
Metals from Solution by Microorganisms. A
Brief Overview", Canadian Metallurgical
Quarterly, Vol. 24, pp. 141-144, 1985.
[28] M. Tsezos, M.H.I. Baird, and L.W. Shemilt,
"Adsorptive Treatment with Microbial Biomass
of Radium - 226 Containing Wastewaters",
Biochemical Engineering Journal, Vol. 32, pp.
B25-B38, 1986.
[29] M. Tsezos, M.H.I. Baird and L.W. Shemilt,
"The Kinetics of Radium Biosorption", The
Chemical Engineering Journal, Vol. 33, pp.
B35-B41, 1986.
[30] M. Tsezos, "Adsorption by Microbial
Biomass as a Process for Removal of Ions from
Process or Waste Solutions", In: Immobilization
of Ions by Biosorption, Eds. H., Eccles, S. Hunt,
Ellis Horwood, Chichester, U.K., pp. 201, 1996.
[31] M. Tsezos, S.H. Noh and M.H.I. Baird,
Canadian Patent & Development Ltd., Patent
file 265 861, 1987.
[32] M. Tsezos, "The Performance of a New
Bi ol ogi cal Adsor bent Model l i ng and
Experimental Results", Second International
Biohydrometallurgy Symposium, University of
Warwick, Warwick, England, July (STL
Publications, Surrey, U.K.), 1987.
16
Technologies Avancées - Numéro 15 - Janvier 2003
33] M. Tsezos and S. Mattar, "A Further Insight
Into the Mechanism of Biosorption of Metals by
Examining Chitin EPR Spectra", Talanta, Vol.
33, pp. 225-232, 1986.
[34] M. Tsezos, R.G.L. Mc Gready and J.P. Bell,
"The Continuous Recovery of Uranium from
Bi ol ogi cal l y Leached Sol uti ons Usi ng
Immobilized Biomass”, Biotechnologie and
Bioengineering, Vol. 34, pp. 10-17, 1989.
[35] M.E. Treen Sears, B. Volesky and R.J.
Neufeld, "Ion Exchange / Complexation of the
Uranyl Ion by Rhi zopus Bi osorbent",
Biotechnology and Bioengineering, Vol. 26, pp.
1323-1329, 1984.
36] B. Volesky and M. Tsezos, "Separation of
Uranium by Biosorption", U.S. Patent , N°: 4 .
320 . 093, 1982 .
[37] B. Volesky, "Biosorbents for Metal
Recovery", Trends Biotechnology, Vol. 5, pp.
96-101, 1987.
17
Technologies Avancées - Numéro 15 - Janvier 2003
EFFECT OF THE GAUGES DIMENSIONS ON THE
PIEZORESISTIVE PRESSURE SENSOR RESPONSE
Z. DIBI*, A. BOUKABACHE** AND P. PONS***
*Laboratoire d'Electronique Avancée, Université de Batna,
Rue Chahid Boukhlouf El Hadi, Batna 05000 Algérie
e-mail : zohirdibi@yahoo.fr
**Département d'Electronique, Université de Constantine,
Route Aïn-El-Bey, Constantine 25000, Algérie
e-mail : aliboukabache@yahoo.fr
***Laas-Cnrs, 7 Av Roche, 31400 Toulouse, France
e-mail : ppons@laas.fr
Abstract
In this paper we present a new study, which can
quantify the sensitivity loss of piezoresistive
pressure sensors due to the gauges
dimensions. The work presents a result of
simulation by taking real dimensions of the
gauges, usually considered as being punctual.
The work we present deals with the modelling of
a structure realised by microelectronic
techniques and chemical etching, containing
four piezoresistors of the P type connected in
Wheatstone bridge implanted on the square
silicon membrane. We modelled the defects of
non-punctuality on the electrical response of
the two types of gauges (normal and tangential)
according to their positions on the membrane
and we examined the effect of this default on the
sensitivity of the sensor. When taking into
account the real dimensions of the gauges, we
found an over-estimation of the sensitivity
approximately of 10% for each 100µm length.
This effect induces not only an important
sensitivity loss, but can also provide a plausible
explanation to the higher values of offset
voltage if the gauges are not mismatched.
Key words: Pressure, Sensor, Gauge,
Piezoresistivity, Membrane, Punctuality.
Introduction
The silicon pressure sensors, based on
piezoresistive effect, gives high values of gauge
factor, compared to those given by metallic
materials [5]. This advantage, added to the fact
that the microelectronic use intensively this
material,
explains the intensification of research on this
subject. Nevertheless, in practice, there exist
several limitations such as thermal behaviour
[1], lack of parallelism of the membrane [2] and
the finite gauge dimensions. Based on one, two
or four piezoresistors implanted on the surface
of the silicon membrane, their electrical
response is directly related to the tensor's
stress. In the literature we considered each
gauge as punctual as far as the surface of the
piezoresistor is too small compared to
membrane surface. In practice, this assumption
is not accurate. This paper presents an
opposite proof and gives a simulation result of
the defect of the piezoresistor dimensions on
the electrical response and on the sensor
sensitivity.
1. Basic Equations
The piezoresistive pressure sensor is based on
the phenomenon of the piezoresistivity in strain
gauges, which are implanted in a silicon
membrane figure 1. In order to reduce the
temperature sensitivity and eliminate the
voltage supply variations dependence, the
gauges are mounted in Wheatstone bridge
figure 2. The four piezoresistors are placed on
the edges of the membrane. The voltage supply
is applied between pins 3 and 1; the response is
collected between pins 4 and 2.
Under the effect of the pressure, the membrane
is deformed and stresses take place on its
surface. The piezoresistors convert the stress
to an electrical signal.
Figure 1. Implantation of strain gauges in
silicon membrane
Figure 2. Schematic representation of
experimental structure of the implantation of
four gauges mounted in Wheatstone bridge.
1.1.The membrane deflexion
The relation that governs the membrane
th
deflexion w is given by a 4 order partial
derivative equation [ 4].
(1)
￿￿: Constant relating the anisotropy of
mechanical properties of silicon;
D: rigidity coefficient of the membrane flexion;
P: applied pressure.
The two parameters and D are calculated using
the equations from [4]
(2)
(3)
18
Technologies Avancées - Numéro 15 - Janvier 2003
D
P
=
4
y
y)w(x,
4
+
2
y
2
x
y)w(x,
4
Si
2á+
4
x
y)w(x,
4
￿
￿
￿￿
￿
￿
￿
￿ ￿
￿
Si
= +
2G(1-
2
E
y

)
D =
E
y
h
3
12 1
2
( )￿ ￿
h is the membrane thickness; E G and í
y,
represent respectively the Young, Coulomb
modules, and the Poisson's coefficient of
silicon. Due to the anisotropy of this material,
the numerical values of the mechanical factors
depend on the crystallographic plan in which
the membrane is fabricated and the orientation
of its sides.
In order to obtain the solution of the system
described by the differential equation and the
limit conditions, Galerkin method has been
used [4].
A membrane deflexion subject to a pressure P
is given by the equation:
(4)
Where i and j are even numbers; a is the length
of the membrane side. The constants depend
only on the anisotropic factor in the case of
square membrane.
w(0,0) corresponds to the deflexion in middle
with an amplitude calculate by:
(5)
k is a constant. In order to normalise the
expression of the deflexion with respect to the
membrane centre, we let:
Which allows rewriting the equation (1) as:
(6)
1.2. Stress
The equations of the two plane elements of the
constraints matrix normalised with respect to the
term, is given by:
(7)
(8)
Figures 3 and 4 shows the three-dimensional
variations of these constraints.
j
a
y2
n
ji,

i
a
x2
ij
k
*
2
2
a
y2
1
2
2
a
2x
-1 w(0,0)=y)w(x,
￿￿
￿
￿￿
￿
￿ ￿￿
￿
￿￿
￿
￿
￿￿
￿
￿
￿￿
￿
￿￿
￿
￿￿
￿
￿
￿
￿￿
￿
￿
￿￿
￿
￿￿
￿
￿￿
￿
k
4
a
D 16
P
= w(0,0)
w(0,0)
y)w(x,
= v)u,(
N
wand
a
2y
= v,
a
2x
=u
P(l/h)
2
k *
v
2

v)u,(
N
w
2
+
u
2

v)u,(
N
w
2
2
3
= v)u,(
1 ￿￿
￿
￿￿
￿
￿
￿
￿
￿
￿￿
k *
u
2

v)u,(
N
w
2
+
v
2

v)u,(
N
w
2
2
3
= v)u,(
2 ￿￿
￿
￿￿
￿
￿
￿
￿
￿
￿￿
￿
Si
k
ij
Si
20
Technologies Avancées - Numéro 15 - Janvier 2003
This piezoresistor can be placed so that its
principal sides are perpendicular or parallel to
the side of the membrane. To optimise the
sensitivity of the sensor, we generally use four
piezoresistors, two parallels and two
perpendiculars to the edges, mounted in a
Wheatstone bridge. Figure 6 shows the result
obtained by simulation concerning the loss of
Wheatstone bridge sensitivity according to the
position gauge on the membrane. It gives high
values of the loss if the gauge is placed far away
from the medium extremity of the membrane.
As an example, if we place a gauge at 10µm
(practical value) an approximately 10% loss is
obtained.
Figure 6. Sensitivity loss of the sensor
according to the position of the gauges
considered as punctual.
3.Simulation of the nonpunctuality
defect
3.1. Introduction
In this part, we present a simulation which takes
into consideration the effect of (100µm)
piezoresistor length and (20µm) width, which
was regarded before as being negligible with
Figure 5. Relative variations of the electrical
response according to the gauge position on
the membrane (u=2x/l and v=2y/l).
respect to the dimensions of the square
membrane (L=2 mm). The method used is
based on the calculation of the average value of
the stress applied to the gauges by the use of a
Simpson numerical method of integration by
means a MATLAB software.
3.2 Electrical response
The two kind of electrical response losses
between a punctual gauge and a 20µm width at
the same position are representing respectively
on figure 7-1 and 7-3. On figure 8-1 and 8-3 we
are represented the losses between a punctual
gauge and a 100µm length. We notice that the
error made on normal resistor RN is larger than
that made on RP. It is also observed that the
more the piezoresistors are moved to the
membrane centre the more the loss increases.
3.3 Sensitivity of the sensor
The Wheatstone bridge carries out a very good
compensation by lowering the error. By taking
into account the real dimensions of the gauges,
it is shown that the fact of considering them as
being punctual induces a sensitivity error over-
estimated of about 10% if we take 100µm
gauges length and 2% if we take a 20µm width
figure 7-2 and 8-2. Because of the symmetry of
the square membrane, we can then assume
that we have approximately 10% loss for each
100µm. This simulation also showed the origin
of the disparity of the behaviour between the
two types of gauges RN and RP because they
have not the same response when we take real
gauges dimensions. It also confirms the
exactness of the assumption that the parallel
gauges are punctual contrary to the
perpendicular ones [2].
Figure 7. Losses between a punctual gauge
and that of width 20 µm
1. Electrical response of the normal gauge
2. Sensitivity of the Wheatstone bridge
3. Response of the parallel gauge
21
Technologies Avancées - Numéro 15 - Janvier 2003
Figure 8 Losses between a punctual gauge
and that of 100 µm length
1. Electrical response of the normal gauge RN
2. Sensitivity of the Wheatstone bridge
3. Electrical response of the parallel gauge RPConclusion
The electrical response of a parallel gauge RP
varies rather little in the medium of the edges
contrary to that perpendicular gauge RN which
varies in a more marked way.
Real dimensions of the gauges influence the
response of the sensor with regard to those,
which are placed parallel to the edges, the
reduction in sensitivity is less important than
those which perpendicular to the edges. This
simulation has found the optimum structure of
the sensor. Positioning the gauges far from the
extremity of the membrane increase the loss of
the electrical response and sensitivity of the
bridge. The choice of a pressure sensor
containing four piezoresistors connected in
Wheatstone bridge, compensates the loss of
the electrical response of a single gauge by
dividing in all the case this loss by two.
The fact of taking into account real dimensions
of the gauges revealed, when regarding them
as punctual, an over-estimation error on the
sensitivity of the pressure sensor of
approximately 10% for each 100µm length. It
has been shown also that the defect of
punctuality is felt more for larger dimensions of
piezoresistors, which leads us to divide each
normal piezoresistors into two elements.
REFERENCES
[1] A. Boukabache, G. blasquez, P. Pons, Z.
Dibi, “Characterisation and modelling of the
mismatch of TCRs and their effects on the drift
of the offset voltage of piezoresistive pressure
sensor”, Sensors and Actuators 84, 292-296,
September 2000.
[2] Z. Dibi, A. boukabache, P. Pons, “Effect of
the silicon membrane flatness defect on the
piezoresistive pressure sensor response”, The
t h
7 IEEE International Conference on
Electronics, Circuit and Systems ICECS'2K
Kaslik, Lebanon; 853-857, December 2000.
[3] A. Götz, F. Campabadal, C. Cané,
“Improvement of pressure sensor performance
and process robustness through reinforcement
of the membrane edges”, Sensors and
Actuators A 67 (138-141), 1988.
[4] S. P. Timoshenko, S. Woinowsky-Krieger,
“Theory of plates and shells”, Mc Graw-Hill,
1982.
[5] Yoshitaka Matsuoka et All, “Characteristic
analysis of a pressure sensor using the Silicon
piezoresistance effect for high-pressure
measurements”, J. Micromech. Microeng. 5.
25-31, 1995.
22
Technologies Avancées - Numéro 15 - Janvier 2003
SUPERVISION OF ROBOT TELEOPERATED
MOVEMENTS WITH A NEUROFUZZY APPROACH
Z. AHMED-FOITIH*, S. GALERNE**, W. NOUIBAT*,
N. BERRACHED*, N. QUETIN**
* Laboratoire de Recherche en Robotique - Département d'Electronique
Faculté de Génie Electrique - Université des Sciences et de la Technologie d'Oran
B.P. 1505 Oran el M'naouer Oran - Algeria
Tél.: (041) 42 29 81 - Fax: (041) 42 29 81- E-mail: foitih@mail.univ-usto.dz
** CEMIF-Systèmes complexes - Université d'Evry Val d'Essonne
40,rue du Pelvoux, Courcouronnes 91 020 Evry Cedex - France
Abstract :
Robot task performing requieres human
intervention at all task levels (planning,
execution and supervision levels) with adapted
assisting means already implemented. This
paper is concerned with the assistance of to
logical fault diagnosis of teleoperated tasks with
respect to the a priori defined plan. Results on
Neurofuzzy prediction of the goal chosen by a
human operator along a grasping task are
presented, with the associated evolution
perspectives.
Keywords
Teleoperation, Manual control, NeuroFuzzy
expert systems, Fuzzy logic, neural networks,
Supervision, Robotics, Man-machine interface.
Introduction:
Telerobotics are concerned with remote control
of robotized devices by one or several human
operators. The tasks carried out with heavy
constraints on cost and security, are complex,
non repetitive and mostly unpredictable. In
addition the environment is not well-known and
not very well mastered. These characteristics
require the use of many resources (mobile
basis, robots, transducers…) and different
control modes (automatic, manual or mixed).
The associated complexity imposes on one
hand a hierarchisation of the activities
(planning, execution, supervision) and on the
other hand a human intervention at these
different levels. Moreover the number of
information processed by the operators
requires an assistance to decision making [7].
More or less detailed plans are elaborated off-
line and at the supervision level, diagnosis of
the correct development of interventions has to
be effected. There will be a physical diagnosis
that concerns the state of used resources and a
logical diagnosis that concerns the task
evolution with respect to planning. Logical
diagnosis brings especially the interpretation of
teleoperated moves (manually controlled
moves) into play.
Beyond t he supervi si on aspect, t he
interpretation capacity of movements controlled
by the operator may moreover simplify the man-
machine dialogue at the execution levels hence
offering the operator an automatic control mode
coherent with the task in progress. The
competition between automatic and manual
control modes [4] could be modified. Moreover
the logical diagnosis implementation allows
keeping simple information of the real mission
development and to draw up a mission report.
The interpretation of teleoperated movements
has to be managed in real time. In other words,
a sufficient reactivity with respect to the process
in progress, numerous variable that could be
heterogeneous and imprecise (transducer
information or appraisal on tasks expressed in
natural language) should exist. In addition, the
presence of human operators in the control loop
needs adapted tolerance. Fuzzy logic perfectly
answers these requirements and moreover
avoids complex process and physical system
modeling [1].
The work presented in this paper deals with the
interpretation of manual controlled moves in
grasping object situation and, mainly the
estimation of the operator aimed goal in a
partially known environment that may contain
obstacles. This robot move recognition is
23
Technologies Avancées - Numéro 15 - Janvier 2003
carried out by the supervision unit from
instantaneous data during either the starting
phase, or simple translation phase, or the
grasping final phase. The robot move chosen
by the operator can depend on unknown factors
by the supervision unit such as parallax
problem, vision field limitation, particular
configuration, comfort and work load of the
human operator, robot motion capability.
The first criterion is established using fuzzy
logic Indeed fuzzy logic avoids complex
modeling of theman-machine system, and is
well adapted to heterogeneous process and
inaccurate data (data delivered by sensors or
human expertise on tasks and expressed in
natural language (“ large ”, “ medium ”, “ small”).
The first part of the paper proposes an
improvement of the neurofuzzy supervision
expert by parameter tuning (membership
functions, rules) and by adding new decision
variables (acceleration, historical account).
Three references robot moves are used.One is
computed with a trapezoidal speed profile, and
the other ones are manually controlled (direct,
and with by passing paths).
In the second part, well-known obstacles which
cannot be grasped by the robots (wall, tables...)
are introduced. Planning methods such as
potential field or visibility graph methods are
used. The latter mode naturally offering several
feasible paths segmented by sub-objectives
which are processed by the neurofuzzy
supervision expert.
The first supervision unit and the new one are
tested on several environments including
different obstacles and the results are
compared.
Furthermore, the supervision unit is improved in
order to take into account other advanced
telemanipulation characteristics. For instance,
particular grasping configurations (for tools,
objects with handle...) are easily introduced
through approach points. In the same way, the
vision field of the operator depending on the
camera positions may be introduced through
the visibility graph method.
1. The grasping problem
Most of automatic trajectory generation
programs for robots use a trapezoidal speed
profile with a starting phase at constant
acceleration, followed by a constant speed
transfer phase and a final constant deceleration
phase. In the same way, a manual controlled
trajectory of the effector of a manipulator arm
may be cut out in three phases: starting phase,
transfer phase and final approach [2].
Thus an external observer might search
parameter combination related to the robot
effector and the objects (distance, speed,
angle, etc.) within each phase. The starting
phase can only be detected by the supervisor
with the acceleration parameter. Transfer
phase corresponds to a large speed on a
trajectory with a stable direction.
Coiffet [3] defines an approach point before the
grasping point (these points are in fact
reference frames with position and orientation
parameters). The existence of such points
leads to the fact that distance parameter is a
pertinent parameter. Small, medium and large
distances between robot effector and the object
supposed to be grasped, are defined according
to the trajectory phases: in "medium" distance
area, the observer will search for deceleration
or small value of speed Figure 1.
Figure 1 Characteristic areas in a grasping
task
In "small" distance area, the observer will
search a final approach phase where the robot
effector moves slowly and softly at the grasping
point. In this latter phase, the operator may
have to modify the initial trajectory according to
grasping constraints: special part of the object
to grasp, viewing problem...Figure 1.
In addition to the distance parameter,
interesting parameters used for trajectory
recognition are the angle between the
instantaneous speed vector and the segment
joining the effector and the object Figure 2, as
well as the instantaneous speed on trajectory.
Figure 2 Definition of parameters ￿i, di and V.
24
Technologies Avancées - Numéro 15 - Janvier 2003
The combination of these three parameters
allows to define some rules. At first, an initially
engaged trajectory will be detected because of
a large value of speed; then, with respect to
each phase, the aimed object might be recognized.
Rules are set up from specific situations which
depend on these chosen parameters.
The transfer phase is ideally characterized by
an important speed in the direction of the aimed
object. Thus the object is far and seen under a
small angle. In the same way a specific situation
may be brought out for the final approach
phase. At this moment the robot effector is close
to the aimed object and the approach speed is
necessarily slow. On the other hand, approach
movements may involve large angular values.
Beside these specific situations, ambiguous
situations must be separated. For instance,
when the robot effector is in movement nearby
an object, the speed value will inform on the
operator's purposes. A slowing down seems to
precede a grasping, while a large speed value
indicates a fast passing through towards a
farther target.
These rules translated in common language are
easily useable in a process like a neurofuzzy
expert. Thus for each parameter combination
corresponds a situation, and an instantaneous
prediction value for a grasping task.
2. Neurofuzzy prediction
The proposed neurofuzzy expert prediction
module is based on two neural network that
fulfill three functions as in the fuzzy expert figure
3 which are fully explained in the paper by
Quetin [8] :
Figure 3 Neurofuzzy prediction module.
The first net has four layers :
- The first layer represents the inputs in which
each neuron is an expert input variable.
- The second layer achieves the fuzzyfication of
the input variables.
- The third layer represents the fuzzy rules, in
which the number of neurones is equal to that of
the rules.
- The fourth layer corresponds to the outputs of
the fuzzy sets.
The second net is a multi-layer neural net that
process of the outputs deffuzzyfication.
The main variables needed for the grasping
task prediction are the distance d between
effector and the object which varies 0 and 375
cm, the angle varying between 0 and 180°, the
magnitude of the effector speed varying
between 0 and 85 cm/s.
The membership functions chosen for the
fuzzyfication are of triangular shape and are
labelled as "small", "medium" and "large"
Figure 4. Quantitative definition of the different
parameters is experimental and may depend on
each manipulation operator.
Figure 4 Generation of the prediction.
Membership functions of the output variables
(fuzzy predictions for each object) are also
triangular.
The defuzzyfication of these variables is carried
out with the center of gravity method.
During the learning phase we have compared
the neurofuzzy net's output with the one
generated by the fuzzy expert.
This operation is based on the backpropagation
method. For the defuzzyfication we have used
four hidden layers, random initialization of the
weights and a decreasing gradiant step for each
n iteration.
3. experiment description and first results
The observer uses a simplified model of the
environment from which the object aimed by the
human operator is to be recognized. First,
objects are supposed to be distributed in a two
dimensional space and do not obstruct each
other. All are visible by the human operator.The
experiment consists in a simulation on a two
dimensional screen from manually controlled
trajectory obtained with the help of a joystick.
Thus these trajectories are noisy ones because
of natural shaking in manual control.
Two goals are represented as disks situated at
different distances. At the beginning, they are
seen under an angle of about 30°. Three
trajectories have been chosen in order to test
the neurofuzzy expert criterion: two manual
controlled trajectories and an automatic one
with a trapezoidal speed profile Figure 5
- Trajectory n°1: is straight but noisy in the
direction of object n°1. It represents the
standard test with the automatic one, the
prediction criterion should be at a high level for
object n°1 from the beginning to the end of the
trajectory. Nevertheless, for experimental
reasons trajectory n°1 is not perfect, it includes
a slowing down at half way, and a gap at the
end.
- Trajectory n°2: for which the operator is
undecided at the beginning is aimed to show the
evolution of predictions for each object.
Figure 5 Experimentals trajectories
Distance, angle and speed parameters are
computed at each instant. A parabolic
approximation method on 7 points of these
parameters is used in order to obtain a
sufficiently precise estimate. For each
trajectory, instantaneous prediction graphs are
presented Figure 6 and 7. The analysis of these
graphs indicates one, four areas with prediction
variations.
Results are discussed for each area, from the
characteristics of experimental trajectories.
Figure 6 Predictions for trajectory n°1
Trajectory n°1
This trajectory aimed the object n°1. Initially,
prediction 1 exceeds 50%, with only a low level
due to insufficient speed. Prediction 2 is also
high because the angles 1 and 2 have almost
the same value. These results, comparable for
both objects, thus less significant, show that
available information is not enough to predict
which object is aimed by the operator. Then,
prediction 1 goes up at about 75%, as if object 1
was aimed; 2 being large, prediction 2 is
undeterminate. Afterwards prediction 1 falls,
that corresponds to a speed decrease. This
problem has been already discussed. On the
contrary, at the end of the trajectory the
distance parameter becomes the most
important one, prediction 1 goes high at 80%.
This area fits well with the final approach area
defined at the beginning of this paper.
Trajectory n°2
This trajectory already presented as undecided
at the beginning, leads to both prediction 1 and
2 at 50%. This is comparable with the precedent
result of trajectory n°1, grasping of object 1
being as possible as the one of object 2;
afterwards there is no more prediction (large
angles).
Figure 7 Predictions for trajectory n°2
25
Technologies Avancées - Numéro 15 - Janvier 2003
Then prediction 1 comes back; the values of 1
and d1 recalling the aim of object 1; but no
slowing down is being detected. The grasping
prediction falls at no prediction, and finally goes
up at 50% to become stable at the final
approach. During all this approach, we can
notice that the angle is not stable. This is due to
the object proximity, while the operator has
been misled with something like parallax. This
area fits exactly with the skirting of object 1, that
may be necessary because of special grasping
configuration or because of well adapted vision
angle. Here, angles over 90° may appear
without disturbing the prediction.
First experimental results are satisfying; taking
into account that the dynamic aspect with the
acceleration or the historical evolution of the
criterion would improve such results.
4. Improvement of the neurofuzzy expert criterion
The acceleration can be added to the criterion in
order to bring out starting and final phases in a
grasping task. Unfortunately acceleration
computing generates a very noisy signal which
jams the criterion. Its use is difficult and needs
adjustments and heavy filtering which
introduces delay between real signal and
prediction delivered by the computed
neurofuzzy expert criterion. Nevertheless,
results applied on trajectory n°1 show a little
improvement of prediction criteria with a
noticeable reacting delay at the very beginning
of the trajectory Figure 8.
This criterion applied to trajectory n°2
shows a widening and growing of signal peaks
which improve the prediction value and here
gives the observer an earlier and higher
prediction for object n°1 at x=296 with 80%.
On the other hand, knowledge from precedent
values of the criterion are not used. With a
memorized history it is possible to obtain a
smoother criterion. This method is equivalent to
a filtering process which induces a delay for the
prediction.
Figure 8 Predictions for trajectory n°1 with
acceleration
Figure 9 Predictions for trajectory n°2 with
acceleration
Figure 10 Predictions for trajectory n°1 with
history
Figure 11 Predictions for trajectory n°2 with
history
Including the acceleration variable and history
in the criterion, globally corresponds to the
initial criterion but with better correspondence
to the trajectories as shown on figures 10 and
11. In particular, trajectory n°2 appears to be
more stable and in the last part of the trajectory
does not go lower than 50% for 0% in the two
precedent criteria. Thus, the neurofuzzy expert
criterion used later on in this paper will be this
latter one. At present, the automatic observer
thus realized does not take into account any
problem due to the environment, such as
obstacles or viewing problem.
26
Technologies Avancées - Numéro 15 - Janvier 2003
This criterion has been implemented and tested
on the same test-bench including two
obstacles; the manual controlled trajectory is
skirting the obstacle in the direction of object
n°1 as shown on figure 12. The previous
criterion is compared with the new one Figures
13 and 14. Figure 15 shows the shortest
trajectories computed with the visibility graphs
method with intermediate points that appears
like circles.
Figure 14 Predictions with knowledge of the
positions of obstacles
The new neurofuzzy expert criterion uses
information from only one computed trajectory,
the shortest one, which explain prediction 1 is
not improved while prediction 2 uses
intermediate points situated at x=200. This
paradoxical result brings out the necessity to
deal with more than one computed trajectory.
Further development will have to implement this
feature easily feasible with the visibility graphs
method.
Figure 15 Obstacles and computed
trajectories
27
Technologies Avancées - Numéro 15 - Janvier 2003
However knowledge of the environment can
improve the performance of the neurofuzzy
expert criterion.
5. Robot moves interpretation using
knowledge about the environment
The most interesting knowledge about the
environment to be used to interpret the robot
moves, is the presence of known obstacles may
be interesting. Effectively, in the presence of
obstacles, the human operator will manage to
avoid contact and will pass around the
obstacles probably with the shortest way. So it
seems reasonable to consider path planning
algorithms able to calculate the shortest
traj ectory. Two methods are usual l y
implemented: potential fields and visibility
graphs methods. The first method seems to
generate soft trajectories which are close to
manual controlled trajectories. Nevertheless,
the visibility graphs method providing
segment ed t r aj ect or i es t hat i ncl ude
intermediate goals is directly compatible with
neurofuzzy expert criterion already developed.
Intermediate goals are computed as significant
virtual objects of the aimed object. Moreover,
afterwards for supervision necessity or in order
to elaborate a mission report, the intermediate
goals help to synthesize the robot moves with a
very few number of points.
Figure 12 Manual controlled trajectory
Figure 13 Predictions with no knowledge
28
Technologies Avancées - Numéro 15 - Janvier 2003
Figure 16 Obstacles and shadow areas with
computed trajectories
The human operator has to manage with many
problems such as limitation of the field of vision.
Effectively, one can think that a non visible area
will not be passed through by the robot
especially if there is no camera aboard. So
taking into account all known obstacles,
shadow areas can be calculated and computed
with the visibility graph method as new
obstacles. An experiment with two fixed
cameras shows new computed paths for
objects 1 and 2 Figure 16. Now, at the
beginning both trajectories are identical and
predictions are similar; but from the last
common intermediate point trajectories and
predictions are diverging Figure 17. This latter
criterion gives the observer a high level
prediction from the begining of the trajectory
unlike precedent criteria.
Figure 17 Predictions with knowledge of the
positions of obstacles and cameras
Conclusion and prospects
Such study using neurofuzzy expert and
visibility graphs methods shows, that it is
possible to predict which object the operator is
going to grasp in order to suggest him an
automatic process or also for supervision
consideration and elaboration of a mission
report. Other improvements are to consider,
like grasping constraints on objects or
capability to perform other kind of tasks
(drilling, cutting, welding etc.).
REFERENCES
[1] B. Bouchon Meunier, "Les raisonnements
approximatifs et incertains, Lettre du club CRIN
"Logique floue", Paris, Mai 92.
[2] F. Chavand, S. Galerne, N. Quetin
Assistance to an operator in teleoperated
robotics. 8th Europ. Conf. On Human Decision
Making And Manual control. Copenhague,
1989.
[3] P. Coiffet, La Robotique. D. Hermès. Paris,
1986.
[4] Groupe "Coopération Homme-machine" ,
GDR Automatique du CNRS, Lille, Sept. 93.
[5] C.C Lee, “Fuzzy logic in control system,
IEEE Transactions on systems mans and
cybernetics”. Vol.20, n°2,; Part I&II, 1989.
[6] E.H. Mamdani, “Applications of fuzzy
algorithms for simple dynamic plant”, Proc.
IEEE. 121 (12), P 1585/88, 1974.
[7] N. Quetin, "Autonomy and feedback levels
in man-machine systems: the case of space
telerobotics”. 10th europ. Conf. On human
decision making and manual control. P149-
158, Liège, Belgique, 1992.
[8] N. Quetin, Z. Ahmed-Foitih, S.Galerne,
"Logical fault diagnosis in advanced
t el eoperat i on", I FAC DARS, Vi enne,
september 1995.
[9] Z. Ahmed-Foitih, N. Quetin, S. Galerne,W.
Nouibat et N. Berrached, "Interprétation des
mouvements téléopérés par une approche
neuro-floue pour le diagnostique logique en
robotique d'intervention", 3ème conférence
maghrébine d'automatisme, d'électrotechnique
et d'électronique industrielle, COMAEI'98,
Bejaia, Algérie. pp. 168-171, Décembre 98.
29
Technologies Avancées - Numéro 15 - Janvier 2003
VALIDATION DES MESURES GRAVIMETRIQUES
PAR LA METHODE DE COLLOCATION
S. A. BEN AHMED DAHO, S. KAHLOUCHE
Centre National des Techniques Spatiales
B.P. 13 Arzew - 31200 - ALGERIE
Abstract
A val i dat i on met hod of t he gr avi t y
measurements was applied on a zone in the
North of Algeria. Using the Least Squares
Collocation (LSC) method and Gravsoft
software of the Copenhagen University, the
gravi ty data provi ded by the Bureau
Gravimétrique International (B.G.I.), consisting
of 2206 gravity measurements covering the
region, are analysed and validated.
The validation was applied to predict free air
gravity anomalies reduced from the effect of the
spherical harmonic coefficient set OSU91A. The
detected gross errors represent 4.7% of the land
data. The results obtained show the non-
homogeneity of the data used and their
insufficient accuracy.
Résumé
Une méthode de validation des mesures
gravimétriques a été appliquée sur une zone
couvrant l'Algérie du Nord. Par utilisation de la
collocation par moindres carrés et du logiciel
Gravsoft de l'Université de Copenhague, les
données de gravité fournies par le Bureau
Gravimétrique International et composées de
2206 mesures couvrant la région sont analysées
et validées.
La validation a été appliquée aux anomalies à
l'air libre réduites de l'effet du modèle
géopotentiel de référence OSU91A. Les erreurs
grossières détectées représentent 4.7 % des
données terrestres, les résultats obtenus
montrent l'inhomogéneïté des mesures utilisées
et l'insuffisance de leur précision.
Mots clés : collocation par moindres carrés,
anomalies de gravité, modèle géopotentiel,
fonctions de covariance empirique et
analytique, potentiel perturbateur, géoïde.
Introduction
Les valeurs de la gravité jouent un rôle
important dans la modélisation du champ de
pesanteur terrestre, lequel est utilisé d'une
façon permanente dans le calcul précis des
orbites des satellites. Elles contiennent
également des informations sur la distribution
des masses à l'intérieur de la Terre
(applications géophysiques), et dans le cas de
mesures répétées au cours du temps, sur les
variations temporelles de la Terre (applications
géologiques). En outre, diverses applications
en océanographie et en navigation sont basées
sur les données du champ de pesanteur. La
gravité étant une quantité fondamentale en
physique de la Terre, les mesures effectuées
doivent être analysées et validées pour qu'elles
puissent être exploitées.
La validation des données est une procédure
extrêmement stricte qui garantit la qualité et
l'i ntégri té de l a banque de données
gr avi mét r i ques. El l e est appl i quée
systématiquement, avant intégration dans la
banque, à tout ensemble de données. Son
principe consiste à faire une comparaison entre
les valeurs observées et prédites, estimées par
une technique puissante.
L'objet de ce travail est de présenter une
mét hode de val i dat i on des mesures
gravimétriques dans des petites zones au Nord
de l'Algérie. La zone choisie pour notre
application s'étale de 34° à 38° N en Latitude et
de -3° à 9° E en Longitude, et contient 2206
mesures.
Dans ce contexte, la validation a été appliquée
aux anomalies réduites de l'effet du modèle de
référence OSU91A, en utilisant la collocation
par moindres carrés. Les résultats obtenus, le
taux d'erreur, permettent de se prononcer sur la
qualité des mesures effectuées.
Dans le cas de la collocation par moindres
carrés (collocation sans paramètres), les
formules (3) et (5) se réduisent respectivement
aux formules de prédictions suivantes :
(6)
(7)
2. Description des données
2.1. Anomalies de gravité
Les anomalies de gravité utilisées dans le cadre
de ce travail ont été fournies par le Bureau
Gravimétrique International (B.G.I.). Ces
mesures dont la précision à priori est de 5 mgals
ont été rattachées au Système de Référence
Géodésique GRS67 par le B.G.I. La répartition
géographique des mesures couvrant l'ensemble
du territoire national est représentée sur la figure
1.
Figure 1 : Répartition géographique des 12472
mesures de gravité fournies par le B.G.I.
2.2. Sélection des mesures
La collocation est en général appliquée
localement, et on traite alors des quantités
résiduelles (observations et signal) par rapport à
une représentation en harmoniques sphériques
du potentiel disponible jusqu'à un degré et ordre
donnés.
La zone choisie pour notre application s'étale de
[ 34° à 38° N ] en latitude et de [ - 3° à 9° E ] en
longitude et contient 2206 points. Ce choix est
conditionné par la densité relativement élevée
des points et par la forte variation du terrain dans
cette région.
30
Technologies Avancées - Numéro 15 - Janvier 2003
1. Méthode de collocation
La collocation peut être considérée comme une
extension de la méthode des moindres carrés.
Dans cette méthode, on estimera, en plus du
vecteur des paramètres, une variable aléatoire
modélisable appelée signal qui exprime la
différence entre la réalité et le modèle qui s'y
adapte.
En général, le modèle mathématique de la
collocation est donné par la relation suivante:
(1)
où la quantité A.X représente le modèle linéarisé
qui est une fonction du vecteur des paramètres X
de p composantes, la matrice A(q x p) exprime
l'effet du vecteur des paramètres dans
l'observation, la fonction S représente le signal
considéré comme une fonction irrégulière
oscillant autour d'un plan de référence et
supposée être toujours une quantité aléatoire
modélisable, la quantité aléatoire n de
dimension (q x 1) caractérise l'erreur
d'observation produite lors de la mesure et L est
le vecteur des observations.
Tout le problème réside dans la détermination de
la courbe (AX+S), ce qui revient à ajuster le
modèle et à estimer la valeur du signal S.
Les estimations linéaires optimales des vecteurs
des paramètres X et du signal S de l'équation (1)
sont données respectivement par (Moritz, 80) :
(2)
(3)
où est le vecteur de covariance entre le signal
S et le vecteur des observations L,
est la somme de la matrice de covariance des
observations et de la matrice de variance -
covariance du bruit associé.
Les matrices des variances - covariances
Et décrivant les précisions des estimations
et sont données par :
(4)
(5)
L A X S n ￿ ￿ ￿.
￿ ￿
$
.....X A C A A C L
T T
￿
￿
￿
￿1
1
1
$
..( )S = C
sl
C L AX
￿
￿
1
E
XX
￿
￿ ￿
(..)A C A
T 1 1
E
avec H = A
SS
￿ ￿ ￿
￿
￿ ￿ ￿
C C C I H C
A C A A C
ss sl ls
T T
..( ).
(..)..
1
1 1 1
C
sl
C C C
LL nn
￿ ￿
E
XX
E
SS
$
X
$
S
$
..S C C l
sl ll
￿
￿1
E C C C C
SS ss sl ll ls
￿ ￿
￿
..
1

31
Technologies Avancées - Numéro 15 - Janvier 2003
Pour des contraintes purement numériques, la
zone d'étude a été divisée en 12 zones
rectangulaires disjointes.
Enf i n, une correct i on at mosphéri que
recommandée par l'Association Internationale
de Géodésie a été appliquée pour éliminer
l'influence des masses atmosphériques. Les
anomalies à l'air libre de la zone d'étude ainsi
que leur distribution sont représentées sur la
figure 2.
Figure 2: Carte des anomalies à l'air libre
observées de la zone 34° - 38° N et -3° - 9° E.
Contours tous les 10 mgals (Min. = - 82.59,
Max. = 165.37, Moy. = 29.60).
2.3. Modèle géopotentiel
L'étude des perturbations orbitales des satellites
artificiels, sous l'influence de l'attraction
gravitationnelle terrestre, a permis d'établir des
modèles géopotentiels représentatifs du
potentiel gravitationnel de la Terre. Ces modèles
se présentent sous la forme de développements
en harmoniques sphériques; une classe
particulière de polynômes et fonctions qui
permettent une modélisation appropriée de la
forme de notre planète.
L'expr essi on du dével oppement en
harmoniques sphériques du potentiel Terrestre
W (somme du potentiel gravitationnel V et du
potentiel centrifuge ) est donnée par [9]:
(8)
où :
( ) sont les coordonnées sphériques (rayon
vecteur, colatitude et longitude) du point de
calcul, GM est la constante gravitationnelle, a
est le demi-grand axe de l'ellipsoïde de
référence; sont les fonctions associées de
Legendre normalisées,
et sont les coefficients normalisés du
potentiel gravitationnel et représente le
potentiel centrifuge,, étant la
￿
￿￿
vitesse angulaire de rotation de la Terre.
Le potentiel normal généré par l'ellipsoïde de
référence est représenté par l'expression [9]:
(9)
où M' est la masse de l'ellipsoïde de référence.
De même, l'expr essi on du pot ent i el
perturbateur; différence entre le potentiel
terrestre et le potentiel normal, est donnée par la
formule suivante [9]:
où et représentent les différences
entre les coefficients normalisés du potentiel
terrestre et les coefficients normalisés du
potentiel normal, et où n'ont été considérés que
les termes de degré ; du fait qu'on suppose
généralement que la masse de l'ellipsoïde de
référence est égale à la masse de la Terre, et que
l'origine du repère du développement est
confondue avec le centre de gravité de la Terre.
Par ailleurs, si l'on considère un développement
en harmoniques sphériques du potentiel
perturbateur jusqu'au degré n = et en
posant =0, on aura :
(10)
Cette expression permet de dériver les formules
de toutes les quantités gravimétriques qui sont
en relation avec le potentiel perturbateur T.
• Hauteur du géoïde (Formule de Bruns)
(11)
où est la pesanteur normale donnée par la
formule de Somigliana.
• Anomalie de gravité à l'air libre
(12)
Les anomalies réduites sont obtenues en
retirant les anomalies de gravité du modèle
géopotentiel calculées par (12) à partir des
anomalies de gravité observées :
(13)
￿ ￿
W
GM
r
a
r
C F S G
e
n
nm nm nm nm
m
n
n
￿
￿￿
￿
￿￿
￿
￿
￿￿
￿￿
￿￿
..+ ￿
00
F P Cos
nm nm
￿ ( ￿ ￿ ) cos m
G P Cos
nm nm
￿ ( ￿ ￿ ) sin m
r,,￿ ￿
nm
P
nm
C
nm
S
)(Sin r
2
1
222
￿￿￿￿
)Sin(P
r
a
J1
r
GM
U
1n
n2
n2
n2
'
￿￿
￿
￿￿
￿
￿
￿￿
￿
￿
￿￿
￿
￿￿
￿
￿￿
￿
￿￿
￿
￿ ￿

r
MG
.GS+.FC
r
a
r
GM
T
n
0m
nmnmnmnm
2n
n
￿
￿￿￿￿￿
￿
￿￿
￿
￿
￿￿
￿
￿￿
￿
nm
C￿
nm
S￿
n ￿ 2
N
max
￿M
￿ ￿ .GS+.FC
r
a
r
GM
UWT
n
0m
nmnmnmnm
N
2n
n
max
￿￿
￿￿
￿￿
￿￿
￿
￿￿
￿
￿￿￿
￿ ￿ .GS+.FC
r
a
r
GMT
N
n
0m
nmnmnmnm
N
2n
n
max
￿￿
￿￿
￿￿
￿￿
￿
￿￿
￿
￿
￿
￿
￿
￿ ￿
￿ ￿ .GS+.FC
r
a
1n
r
GM
g
n
0m
nmnmnmnm
N
2n
n
2
m
max
￿￿
￿￿
￿￿￿￿
￿
￿￿
￿
￿￿￿
￿g
red
￿g
m
￿g
obs
￿ ￿ ￿g g
red m
= g
obs
￿
32
Technologies Avancées - Numéro 15 - Janvier 2003
Ces anomalies reflètent les caractéristiques
locales de la zone en question puisque les
contributions des grandes longueurs d'onde
sont éliminées.
Le problème de choix du modèle géopotentiel
qui ajuste d'une façon optimale le champ de
gravité en Algérie n'a pas été résolu. Dans le
cadre de ce travail, le modèle OSU91A [10],
développé en harmoniques sphériques jusqu'au
degré et ordre 360, a été utilisé. La figure 3
montre la carte des anomalies à l'air libre moins
la contribution du modèle global OSU91A.
Figure 3 : Carte des anomalies observées
moins la contribution de OSU91A. Contours
tous les 10 mgals (Min. = -70.99,
Max. = 125.56, Moy. = 1.45).
3. Fonction de Covariance
La fonction de covariance exprime la
dépendance statistique des quantités liées au
potentiel perturbateur. On suppose que la
différence entre le modèle terre (ellipsoïde) et la
réalité (géoïde) obéit à des lois statistiques.
L'approximation optimale au sens des moindres
carrés du potentiel terrestre est obtenue quand
la fonction de covariance empirique est utilisée
dans l'estimation des différentes quantités
gravimétriques du champ de pesanteur
Terrestre.
3.1. Covariance empirique
La fonction de covariance empirique de
l'anomalie réduite de chaque zone a été calculée
avec le programme EMPCOV [15] en utilisant la
formule suivante:
(14)
La somme est effectuée pour toutes les
combinaisons des points dont la distance
sphérique est comprise entre
et. N es t l e nombr e de
combinaisons, et mi nut es d'ar cs,
représente la dimension de l'intervalle
d'échantillonnage. Le choix de cette valeur
dépend du pas de la grille utilisée.
3.2. Ajustement de la fonction de covariance
locale
L'application de la méthode de collocation
nécessite la connaissance d'un modèle
analytique de covariance. La fonction de
covariance locale d u p o t e n t i e l
perturbateur utilisée est donnée par [11, 12]:
(15)
avec :
où : représente la variance des erreurs des
coeff i ci ent s harmoni ques du modèl e
géopotentiel,
N : degré maximum du modèle de référence,￿ : coefficient multiplicatif empirique que l'on
estimera,
A : facteur d'échelle de prédiction,
: rayon de convergence [c'est le rayon de la
sphère de Bjerhammar],
r et r' : rayons vecteurs des points P et Q.
: rayon moyen de la terre.
B : entier positif.
Cependant, l'utilisation du modèle précédent
comme modèle de fonction de covariance
locale exige l'estimation de trois paramètres : le
rayon de la sphère de Bjerhammar ( ) et deux
facteurs d'échelle ￿, A. Le problème consiste
donc à ajuster les valeurs de ces paramètres
afin de se rapprocher le plus possible des
valeurs empiriques. Une procédure itérative
par moindres carrés est utilisée pour
l'estimation de ces derniers, en imposant à B
d'être un entier, ce qui conduit dans ce cas à
des expressions finies pour la somme des
séries en question.
Les résultats de l'ajustement des fonctions de
covariance empirique sur le modèle de
Tscherning décrit ci dessus ont été obtenus par
le programme COVFIT8 [5].
C
N
g g
ss i j
( ).￿ ￿￿ ￿￿￿
￿
1
Q
i
et Q
j
￿
ij
(/￿ ￿ ￿￿ 2 )
( +/2)￿ ￿￿
￿￿ ￿ 7 5.
K( )￿
K K( ) ( ) ( )￿ ￿ ￿￿ ￿K
G L
K
et K
G
E
l
E
l
l
l
N
L
B
i N
i
i
GM
R
R
rr
P Cos
A
i i i B
R
rr
P Cos
( ).
'
( )
( )
( )( )( )'
( )
￿ ￿ ￿ ￿
￿ ￿
￿
￿￿
￿
￿￿
￿
￿￿
￿
￿￿
￿
￿
￿ ￿ ￿
￿￿
￿
￿￿
￿
￿
￿
￿ ￿
￿￿
￿
￿
￿
2
2
1
2
2
1
1
1 2
￿
l
R
B
R
B
R
E
L'observation est considérée comme erronée si [15]:
> (18)
où k est une constante, généralement prise
égale à 3 et représente la variance des observations.
Les valeurs prédites des anomalies de gravité
de l'ensemble B ont été estimées à partir des
données de l'ensemble A, puis comparées aux
observations de l'ensemble B. Cependant, si la
différence entre et est supérieure
à 20 mgals (seuil fixé au préalable en fonction
de la précision et de la densité des données
utilisées), alors, cette observation est rejetée.
De même pour les données de l'ensemble A à
partir de l'ensemble B.
La procédure est répétée sans les observations
erronées et sur tous les points des ensembles A
et B. Alors, si la même observation est rejetée,
on peut affirmer qu'elle est entachée d'erreur et
par conséquent, el l e sera él i mi née
automatiquement.
Le taux d'erreur détecté en utilisant cette
méthode est de l'ordre de 4.7%. Les
statistiques des 2100 mesures validées sur
2206 mesures brutes, sont regroupées dans la
table 1.
Les écarts entre les observations et les
prédictions sont maximums, jusqu'au 20 mgals,
dans les régions dépourvues de données, et
d'envi ron 1 mgal l e l ong des l evés
gravimétriques.
33
Technologies Avancées - Numéro 15 - Janvier 2003
4. Validation des mesures gravimétriques
La procédure de validation a été appliquée en
utilisant la collocation par moindres carrés. Les
données réduites de chacune des 12 zones ont
été divisées en deux ensembles disjoints A et
B, à condition qu'ils aient la même distribution.
En outre, un échantillonnage avec un pas de 5'
(~ 10 km) sur les données contenues dans
cette zone a été effectué (il ne s'agit pas de
valeurs moyennes). A partir de ces anomalies
réduites, deux fonctions de covariance
empiriques ont été calculées séparément. Ces
valeurs empiriques ont été utilisées pour
estimer les valeurs des paramètres du modèle
de la fonction de covariance locale (modèle de
Tscherning ).
Soit l'anomalie prédite par collocation
(LSC) à partir d'un ensemble de valeurs.
Cette valeur est donnée par [9] :
(16)
où est le vecteur de covariance entre les
observations et les prédictions , C
est la somme de la matrice de covariance des
quantités et de la matrice de variance -
covariance du bruit associé.
L'écart type (r.m.s) des différences
est donné par:
(17)
où est la variance des anomalies de gravité.
￿g
pred
￿g
red
￿ ￿
￿
g C C g
pred g red
￿
￿
..
1
C
g￿
￿g
red
￿g
pred
￿g
red
￿ ￿g g
red pred
￿
￿
2
0
1
( )..￿ ￿
￿ ￿
g g C C C C
red pred g g
T
￿ ￿ ￿
￿
C
0
￿ ￿g g
red pred
￿
k g g
red pred g
.( ( ) )￿ ￿
2 2
1
2
￿ ￿
￿
￿ ￿
￿
￿g
2
red
g￿
￿g
pred
Table 1 : Statistiques (en mgal) des données validées.
Anomalies Moyenne Ecart type Minimum Maximum
28.07 29.67 -82.59 136.20
0.24 22.84 -67.75 123.26
Prédictions ( ) 0.43 20.89 -54.67 116.39
Différences ( - ) -0.18 6.65 -19.94 19.9
￿g
obs
red
g￿
pre
g￿
red
g￿
pre
g￿
34
Technologies Avancées - Numéro 15 - Janvier 2003
Conclusion
La méthode utilisée dans le cadre de ce travail
n'est pas récente dans son principe, mais elle
est d'un intérêt majeur pour la validation des
mesures gravimétriques ; type de données
utilisé généralement dans la modélisation du
champ de pesanteur local.
Le taux d'erreur détecté en utilisant cette
technique reste élevé par rapport au nombre de
mesures utilisées. Les résultats obtenus sur la
zone test, considérée comme la plus dense,
ont montré de graves lacunes dans la
couverture gravimétrique du pays qu'il
conviendrait de combler aussi bien pour
satisfaire les besoins de la géodésie que ceux
de la géophysique et de la géologie.
En perspective, et afin de procéder à une
détermination précise du champ de pesanteur
local à partir des mesures gravimétriques
validées, il est nécessaire de disposer d'une
couverture gravimétrique dense et uniforme sur
tout le territoire national.
Remerciements
Les auteurs tiennent à remercier le Prof. C. C.
Tscherning (secrétaire général de l'A.I.G.) et le
Prof. R. Forsberg de l'uni versi té de
Copenhague pour leur avoir fourni le logiciel
GRAVSOFT, et le Prof G. Balmino, directeur du
B.G.I. et Secrétaire Général de l'U.G.G.I., pour
avoir mis à leur disposition les données
gravimétriques sur l'Algérie.
REFERENCES
[1] B.G.I., Bulletin d'information du Bureau
Gravimétrique International, N° 72, 1993.
[2] S. Benahmed Daho, "Détermination du
géoïde gravimétrique en Algérie par la méthode
de collocation", Thèse de magister en
techniques spatiales - CNTS, 1996.
[3] S. Benahmed Daho, S. Kahlouche, “The
gravimetric geoid in Algeria : first results”.
Scientific Assembly of the International
Association of Geodesy, Rio de Janeiro - Brazil
- 3- 9 September 1997.
[4] S. Kahlouche, I. Kariche, S. Benahmed
Daho, “Comparison between altimetric and
gravi met ri c geoi d i n t he sout h-west
mediterranean basin”, Scientific assembly of
the international Association of Geodesy, Rio
de Janero Brasil 3-9 September 1997.
[5] P. Knudsen, “Estimation and Modelling of the
Local Empirical Covariance Function using
gravity and satellite altimeter data”, Bulletin
Géodésique, vol. 61 1987.
[6] G. Kraiger, “Influence of the curvature
parameter on Least-squares prediction”,
Manuscripta geodaetica, 1988.
[7] E. Min., “A comparison of Stokes numerical
integration and collocation, and a new
combination technique”, Bulletin géodésique
pp. 223 - 232, 1995.
[8] H. Moritz, “Covariance Function in Least
Squares Collocation”, Report N° 240, Ohio
State University, 1976.
[9] H. Moritz , “Advanced Physical Geodesy”, H.
Wi chmann-Abazcus Press, Karl sruhe-
Tundridge Wells, 1980.
[10] R. H. Rapp, F. Sanso, “Determination of the
geoid: present and future”, International
Association of Geodesy symposia, N° 106,
1995.
[11] C.C Tscherning, “A FORTRAN IV program
for the determination of the anomalous potential
using stepwise least squares collocation”,
Reports of the Department of Geodetic
Science, N° 212, Ohio State University, 1974.
[12] C.C Tscherning, “Closed Covariance
Expressions for Gravity Anomalies, Geoid
Undulations, and Deflections of the Vertical
Implied by Anomaly Degree-Variance Models”,
Reports of the Department of Geodetic
Science, N° 208, Ohio State University, 1974.
[13] C.C, Tscherning, Rapp R.H. , Goad C., A
“comparison of methods for computing
gravimetric quantities from High Degree
Spherical Harmonic Expansions”, Manuscripta
geodaetica, Vol. 8, pp. 249-272, 1983.
[14] C.C. Tscherning, Geoid “Modelling using
collocation in Scandinavia and Greenland”,
Marine Geodesy, vol. 9, N° 1, 1985.
[15] C.C. Tscherning, “Geoid determination by
Least Squares Collocation using GRAVSOFT”,
Lecture notes for the international school for the
determination and use of the geoid, Milano,
October 1994.
35
Technologies Avancées - Numéro 15 - Janvier 2003
Abstract:
In this paper, an analysis method of wavelets
combinations is described. It is a new tool for
extracting the useful information from them. It
differs from the more well known signal
analysis. The waveform of each wavelets
combination is cutted to pieces. Then, the
wavelet hidden in each piece is identified. This
analysis method of wavelets combinations is
depended on their detection and on the hidden
wavelet identification. The details of this
analysis are exposed in the main text of this
paper. Experiment results are presented to
demonstrate the performance of this new
analysis method.
Résumé:
Dans cet article, une méthode d'analyse de
combinaisons d'ondelettes est décrite. C'est un
nouveau outil très intéressant pour l'extraction
de l'information utile de celles-ci. Ce traitement
est différent à de nombreuses analyses
connues du signal. Le chronogramme de
chaque combinaison d'ondelettes est découpé
en morceaux. L'ondelette supposée cachée
dans chacun de ces morceaux est identifiée.
Cette méthode d'analyse de combinaisons
d'ondelettes est dépendante de leur détection
et de l'identification de leurs ondelettes. Les
détails de cette analyse sont exposés dans cet
article. Des résultats expérimentaux sont
présentés pour montrer la performance de cette
nouvelle méthode d'analyse.
Key words : absolute error- absolute average-
amplification- attenuation - detection- gaussian
noise- identification- filter operator- lifetime-
random signal- signal analysis- signal
processing- signal to noise ratio- transient
signal- wavelet- wavelets combination-
wavelets family- wavelets mother.
Introduction :
Each wavelets combination [4, 1, 2] may be a
translation of an useful information. This
information depends on the number of wavelets
and of their families in each combination.
Because these wavelets may be from the same
family or from the lot of families.
ANALYSIS METHOD OF WAVELET COMBINATIONS
Amina HAJRAOUI and Abderahmane HAJRAOUI
GEEA Laboratory, Abdelmalek Essaadi University
Sciences Faculty, BP.2025, Tetouan, Morocco
So, each wavelets combination may be as
follow:
Where is a wavelets mother of the
following family . But, , , and are
wavelet coefficients of amplification, of
translation, and of dilatation respectively. The
greatest quantity of information, which may be
traduced by these wavelets combinations, take
us to remark the importance to search a method
to analyse them. In fact, this method is exposed
bellow. Its performance is studied in analysis of
a random signal. Therefore the preparation
description of wavelets combinations is
presented in the second section. Then, the
indicated analysis method is explained in the
fourth section. The experiment results and
discussion are introduced at the end.
1. Preparation of wavelets combinations
1.1 Choice of the useful wavelets:
Attention is given to choosing proper wavelets
for any applications. The first efficient way is to
record the real phenomenon, like medical,
sound or image signals. The record of synthetic
signals may be used also. From these signals,
the useful wavelets are extracted. This
operation provides new libraries of standard
wavelets with the required properties. The
information at different resolution scales is
provided by these useful wavelets. The cost of
translating a real phenomenon to a combination
of useful wavelets can be measured by how
many wavelets must be superposed to obtain a
desired degree of approximation and this cost
can be minimized by fast search through the
constructed library of standard wavelets. The
great number of these wavelets help everyone
to construct any combination of wavelets. Each
combination may have wavelets from any
families. To study the proposed analysis
method, we have chosen wavelets, which
everyone has the following properties:
)(ti￿￿
￿ ￿i￿￿
i￿
ia
ib
36
Technologies Avancées - Numéro 15 - Janvier 2003
a.wavelet is causal =0 if t<0
b.wavelet is from the space:
c.wavelet has a null average:
d.wavelet has a finite lifetime equal to
e.wavelet has a maximal magnitude equal to
1Volt
1.2 Co n s t r u c t i o n o f wa v e l e t s
combinations:
The wavelets combination is a transient signal.
Its construction is a random operation.
Because, the number of wavelets in each
combination, and the appearance of this
combination in the studied signal are random. In
addition, the choice of a wavelet and of its
amplification, are random too. So, each used
wavelets combination is expressed as follow:
with
Where J is chosen less or equal to 9. is a
random integer. But is an indicator which is

defined as follow:
2. Random signal of wavelets combinations
It is a function of the indicated combinations [5,
3, 6] and of a noise. It constructed to study the
performance of the proposed analysis method.
It exposed as follow:
Where B(t) is a gaussian noise. But, , and
are the random lifetime of each wavelets
combi nat i on and t he cor r espondi ng
appearance time respectively. This respect
the following condition :
￿
￿
￿
pdtti
0
2
)(￿￿
￿
￿
￿￿￿￿
J
m
TJuiV tmTttC uiJ
0
)1( )().(.)( ￿￿￿
],........,,[ 21 JJV ￿￿￿￿
wheretBttStS anc........).........(..)..(....).....( ￿￿￿
￿￿
￿￿
￿
￿￿￿￿
￿￿￿￿￿
￿￿
aanuan
uanananV
an
TttTJtif
TJtttifttC
ttS
J
p)1(................0
)1(.....)....(
)(
aanna Ttt ￿￿￿ )1(
￿
￿
￿
￿￿￿
￿￿ ￿
otherwise
TJtif
t
u
TJ u
..................0
)1(0...........1
)()1(
)(ti￿￿
JV
ant
)(
2
RL
0)(
0
￿
￿
￿
dtti￿￿
aT
uT
i￿
ant
3. Anal ysi s met hod of wavel et s
combinations
3.1 Combination pieces production
After amplifying the random signal , all
wavelets combinations are detected by mean of
the absolute average [4]. Each waveform of the
detected wavelets combination is cutted to
pieces. Each one of these pieces has the same
lifetime as a wavelet. If the last piece has not the
required lifetime, it will be ignored. Because, it is
known that a wavelet is hidden in each indicated
piece. This described operation depends on the
appearance and on the disappearance times
and of each wavelets combination. The
number of wavelets in each combination is
always a random integer.
3.2 Identification
Being weak or powerful, each hidden wavelet
must be i denti fi ed. So, al l wavel ets
combinations are identified too and are
analysed. Because the meaning of each
wavelet is known. The indicated identification is
realized by mean of the absolute error [4]. This
error , between samples of a combination
piece and of a known wavelet, must be
minimum.
If and are samples of and
of its filtered signal respectively, this error is
defined below. Any low pass filter may be used.
But its cut-off frequency must depend on the
greatest frequency of the used wavelets
combinations. is the corresponding filter
operator [4]:
Where ,N, and are sampling frequency,
wavelet lifetime number, and appearance time
of a wavelet respectively. Max() is a local
amplification rate of the hidden wavelet .
4. Results and discussion
After the preparation of the wavelets
combinations in the conditions which are
defined in section 2, these combinations are
detected in the random studied signal .
..)().()]([.)(
0
￿
￿
￿￿￿
N
m
eaieaicaii mTtMaxmTtSFTtae i￿￿
￿.,....)]([.,.)]([.)( eaicaicai TtSFTtSFTSuptMax ￿￿
)(kSc
eT
(.)i￿￿
)]([ kSFT c
[.]FT
(.)iae
￿.)].([......,.eaic TNtSFT ￿
(.)cS
(.)cS
)(tSc
(.)i￿￿
39
Technologies Avancées - Numéro 15 - Janvier 2003
Recommandations aux Auteurs
Les articles de Technologies Avancées se répartissent en deux rubriques: recherche-développement et synthèse. Ils peuvent
être rédigés en Français ou en Anglais. Les articles de recherche-développement des Technologies doivent porter soit sur des
travaux ayant une certaine originalité et apportant une contribution novatrice, aidant au développement des Technologies
Avancées, soit sur des réalisations et études concrètes qui n'introduisent pas forcément des idées nouvelles, mais qui
présentent un intérêt primordial dans la maîtrise de la technologie et des concepts scientifiques contemporains. Les articles de
synthèse ont pour but de faire le point sur l'état de l'art dans un domaine ayant trait aux Technologies Avancées. Cette rubrique
est destinée à fournir une vue d'ensemble compréhensible pour un lecteur non-spécialiste du domaine.
Critères de Publication:
Les critères communs aux deux rubriques sont: l'apport scientifique, technique ou technologique, la clarté de l'exposé et la
qualité du texte. Pour les articles de recherche-développement, sont exigés: l'intérêt théorique ou de développement et la rigueur
de la présentation, l'importance pratique des réalisations pouvant, soit produire un effet économique, soit permettre la maîtrise
d'une technologie de pointe. Les articles de synthèse seront évalués à partir de leur facilité de lecture et de leur présentation qui
doit favoriser l'approfondissement ultérieur du domaine. Pour les auteurs principaux de chaque article qui paraît dans la revue il
n'existe pas de frais de participation.
Présentation:
Le texte sera présenté en double colonne dactylographié, à double interligne sur une page avec les dimensions adéquates, en
utilisant une numérotation arborescente: 1, 1.1, 1.1.1, il est cependant demandé de ne pas dépasser trois niveaux.
1. En première page, doivent être mentionnés: le titre (police de caractère 14), les noms des auteurs, leurs adresses
professionnelles, leurs numéros de téléphone et de télécopie, ainsi qu'un résumé de 200 mots en anglais suivi de 5 à 6 mots clés.
2. Illustrations: Chaque figure ou groupe de figures doit pouvoir occuper une ou deux colonnes du texte. Les dimensions finales
et maximales d'une illustration sur une colonne sont de 8,9 x 24 cm, et de 18,4 x 24 cm pour une illustration sur deux colonnes.
Les figures doivent être référencées et numérotées dans le texte. Il est nécessaire de joindre à l'article un tirage séparé de
chaque figure, légende table, … leurs dimensions doivent être telles qu'après réduction, ils ne soient pas inférieurs à ¼ de leur
dimension originale.
Les photographies: doivent être soumises en trois (03) exemplaires. Les épreuves doivent être de première qualité, sur
papier glacé avec contraste bien marqué. Nos recommandations la consultation de l'illustrating science: Standard for Publication
(publié en 1988 par le Concil of Biology Editors, INC Bethesda, MD 20814, USA) pour la préparation des dessins, des graphiques
sur ordinateur, des textes prêts à la photographie, qui y est clairement expliquée et illustrée, de même que le procédé de quadrichromie.
Bibliographie: toute référence doit être indiquée dans le texte par des numéros entre crochets, et ce dans l'ordre
alphabétique du nom du premier auteur. Les références seront rassemblées dans la bibliographie par ordre alphabétique de la
façon suivante :
Auteur 1, Auteur 2, … (année) titre - revue - numéro - page début, page fin (article ou communication)
Auteur 1, Auteur 2, … (année) titre, éditeur (pour un livre).
Modalités de publication:
1. Après examen par le comité de rédaction et le comité international de lecture de 2 de ces membres, les textes nécessitant une
révision partielle seront renvoyés aux auteurs pour une nouvelle rédaction.
2. Après acceptation définitive d'un manuscrit, les auteurs recevront, avant publication, des épreuves d'impression qu'ils doivent
vérifier dans les délais indiqués. Aucune modification de texte ne pourra être apportée à ce stade où seules les erreurs pourront
être rectifiées.
3. Les auteurs recevront, sur leur demande et contre-remboursement, le nombre de tirés à part souhaité.
4. Les textes proposés pour publication doivent être adressés en cinq exemplaires, sous plis recommandé avec accusé de
réception à:
Monsieur le Directeur de la Publication "Technologies Avancées "
Centre de Développement des Technologies Avancées
Cité du 20 Août 1956, BP 17 Baba Hassen, Alger, Algérie