106402 M.Tech. (Instrumentation)

muscleblouseΤεχνίτη Νοημοσύνη και Ρομποτική

19 Οκτ 2013 (πριν από 3 χρόνια και 10 μήνες)

112 εμφανίσεις

Presentation by:

Ameer
Iqbal


Roll No.
-

10
EE
64
R
02

M.Tech
. (Instrumentation)

Contents


Introduction


Components


Sensors


Pattern recognition


Wireless electronic nose


Advantages & limitations


Applications


Future & conclusion


2

Biological Nose




Detection and identification of odour


Quantifying smells are useful in gas chromatography


Human nose is very sensitive


Subject to fatigue, inconsistencies, adaptation etc.


Smelling toxic gases may involve risk




Fig. Conduction route diagram of animal olfactory system




3

Electronic Nose


Instrument intended to mimic the human sense of
smell


Combines human sensitivity & instrument’s objective


Consists of:


Sample handling system


Sensing system


Pattern recognition system

4

Fig. Schematic diagram of an electronic nose

Electronic Nose


Correspondence of electronic nose with biological nose

Biological nose

Electronic nose

Lungs

Pump

Mucus, Hair, Membrane

Inlet Sampling System

Olfactory cells

Sensors

Olfactory vesicle

Data pre
-
processing module

Olfactory centre

Pattern recognition module

Nerve Impulses

Electrical signal

5

Components


Sample handling system


Generates the headspace of sample to be analyzed


Exposes the odorant to the sensors


Sensing system


Array of different sensors


Each sensor has different sensitivity to different gases


Produces a pattern characteristic of the odour

Fig.
Response of sensor array to different pure chemicals

6

Sensing system


Quantity & complexity of the data collected can make
analysis of data in an automated system difficult.


Using array of sensors, each sensor designed to
respond to a specific chemical


Number of unique sensors must be at least as great as
the number of chemicals being monitored


Difficult to build highly selective chemical sensors


Expensive also


7

Sensing system


Use of
Artificial neural networks (ANN)


ANN combined with a sensor array


Number of detectable chemicals is greater than that of
sensors


Less selective sensors can be used


Less expensive too


Electronic noses incorporating ANNs have been
demonstrated in various applications.


8

Electronic nose sensors


Conductivity Sensors

(a) Metal Oxide Sensor


Oxides of tin, zinc, titanium etc. doped with
platinum


Active material


Doped material deposited between two metal
contacts over a resistive heating element


Operating temp.:
200
°
C
-
400
°
C


As VOC passes over the active material,
resistance changes


Resistance changes in proportion to the
concentration of the VOC.

9

Conductivity Sensors

(b) Polymer Sensor


Active material is a conducting polymer


e.g.
Polypyroles

,
thiophenes

,
indoles

etc


When exposed, chemicals forms bond with polymers


Bonding may be ionic or covalent


Transfer of electrons along polymer chain is affected, i.e.
conductivity changes


Operate at ambient temperature, no heater required

10

Conductivity Sensors

Metal Oxide Sensor

Polymer Sensor

Susceptible to poisoning by sulphur
compounds present in the odorant
mixture

Difficult and time consuming to electro

polymerize the active material

Wide availability

Susceptible to humidity & can mask the
responses of VOC

Relatively low cost, hence widely used

Electronic interface is simple, suitable
for portable instruments

11

Fig. Comparison between metal oxide & polymer sensors

Piezoelectric Sensors

(a)
Quartz crystal microbalance (QCM)


Consists of a resonating disk with metal
electrodes on each side connected to lead wire


Resonates at a characteristic frequency (10
-
30
MHz) when excited with an oscillating signal


Polymer coating serves as sensing material


Gas adsorbed at the surface of the polymer
increases the mass, reduces resonance frequency


Reduction is inversely proportional to mass
adsorbed by the polymer


12

Piezoelectric Sensors

(b)
Surface acoustic
-
wave (SAW)


An ac signal is applied across the input metal
transducer


Fingers of this gas sensor creates an acoustic wave
that "surfs" the piezoelectric substrate


When the wave reaches the metal fingers of the
output transducer, ac voltage is recreated


Voltage is shifted in phase as a result of the
distance travelled.


Phase shift depends on the mass & the absorption
properties of the sensing polymer layer


SAW devices are less sensitive than QCMs


13

Piezoelectric Sensors


Limitations:


More complex electronics are needed by these sensors
than conductivity ones


Resonant frequencies can drift due to the active
membrane ageing


Requires frequency detectors

14

MOSFET Sensors


Gate is covered by noble metal catalyst, e.g. platinum,
palladium, or iridium


Charge applied to the gate leads to current flow from
source to drain


VOCs sweeping over the catalyst forms products that
alter the sensor's gate charge


Channel conductivity varies

15

MOSFET Sensors


Advantage:


Can be made with IC fabrication processes, batch to
batch variation is minimized


Disadvantage :


Reaction products should penetrate the catalytic metal
layer in order to influence the charge


Hermetic seal for the chip’s electrical connections in
harsh environments

16

Optical Sensors


Glass fibre coated on its sides & ends with a
thin active material containing fluorescent
dyes


Pulse of light from an external source
propagates along the fibre


VOCs can alter the polarity of the dyes


Dyes responds by shifting fluorescent
spectrum of the light


Simple fabrication
-

Fluorescent dyes can easily
be coupled

17

Signal processing & pattern recognition


Main sequential steps:


Pre
-
processing


Feature extraction


Classification and


Decision making


Data base of the expected odorant should be compiled

18

Signal processing & pattern recognition


Pre
-
processing


Compensates for sensor drift


Compress the transient response of the sensor array


Reduces sample to sample variations


Feature extraction


Reduce the dimensionality of the measurement space


Can be more readily inspected visually


Extract information relevant for pattern recognition


Performed with linear transformations e.g. PCA & LDA


Nonlinear transforms, e.g.
Sammon

nonlinear maps and
Kohonen

self organizing maps


19

Signal processing & pattern recognition


Classification


Bayesian classifiers, Artificial Neural Network(ANN) etc
are used


Trained to identify the patterns that are representative of
each odour


Identify the odorant by comparing it with trained ones


Decision Making


Used for application specific knowledge


Can determine that given sample does not belong to any
one in database



20

Wireless Electronic Nose


Developed in 2010


Can perform remote multipoint odour monitoring


Signal from isolated locations can be combined and
processed at a database server


Data measured are delivered via
ZigBee

wireless
network

Fig.
ZigBee

node of wireless electronic nose network.

21

Wireless Electronic Nose

(a)Electronic Hardware


MCU acquires gas sensor data through ADC interface &
sends the data to
ZigBee

wireless network


Real time clock in MCU stamps date and time of the
transmitted data

Fig. Block diagram of wireless electronics nose.

22

Wireless Electronic Nose


Sensors were designed to be particularly sensitive to
different gases


Temp. & humidity sensors for environmental conditions.

Name of

sensors

Compound

to be detected

TGS
3870

Carbon monoxide

TGS
4161

Carbon dioxide

TGS
825

Hydrogen sulfide

TGS
826

Ammonia

KE
-
25

Oxygen

SHT
15

Temp. sensor

SHT
15

Humidity sensor

Table
-

Sensors used for the developed electronic nose

23

Wireless Electronic Nose

(b)Gas Flow System


Two solenoid valves control the flow of reference air &
air sample


Reference air
-

air filtered by activated carbon (valve
1
)


Valve
1

open
-

valve 1 closed
-
valve
2

open
-

valve
2

closed

Fig. Gas Flow System

24

Wireless Electronic Nose

(c) Principal Component Analysis (PCA)


Data from ADC is stored in
2
-
D array versus sampling
time


Data for each sensor are subtracted by their mean values


Covariance matrix of the subtracted data is computed


Eigenvectors &
Eigenvalues

of the covariance matrix are
then calculated


Then principal components are chosen and featuring
vectors are formed

25

(d)
ZigBee

Technology


ZigBee

is famous for its low cost, low power consumption
& miniaturization


Tree topology, with benefits of star & mesh, was used


Mostly operates in sleep mode, low power consumption


End nodes acquire e
-
nose data and send them to the
router nodes


Router nodes combines its own data & send to base
nodes


Data was sent to database server


Wireless Electronic Nose


Fig. Tree Topology

26

Wireless Electronic Nose

Fig. Normalized data set from
wireless electronic nose.

Fig. PCA plot between PC
1 &

PC
2


27

Electronic Nose


Advantages


Detection of poisonous gas is possible


Can be done in real time for long periods


Cheaper than Trained human sniffers


Individuals vary, e
-
nose don’t


Limitations


Time delay between successive tests


Insensitivity to some species


According to application, e
-
nose has to be changed

28

Applications


Environmental control (air quality, gas emission levels
of factories, chemical plant monitoring etc.)


Medical applications (urine, skin, breathe odour
analysis, ulcer monitoring etc)


Food industry (coffee, fermentation process,
identification of bacteria etc.)


Defence and security industries (detecting land mines)


Pharmaceutics, chemical industry (odour, quality
control of pharmaceutical compounds etc.)


Semiconductor industrial process

29

Future Work


Research is being done on IC E
-
Noses


Miniaturizing current Technology


Improvement in sensitivity for lower levels of organisms
or smaller samples


Minimizing cost


30

Conclusion


Humans

are

not

well

suited

for

repetitive

tasks
.

Electronic

nose

has

the

potential

to

become

standard

tool

for

smelling
.

Researches

are

still

going

on

to

make

electronic

nose

much

more

compact

than

the

present

one

and

to

make

e
-
nose

ICs
.

31

References

[
1
]

T
.
Pogfay
,

N
.
Watthanawisuth
,

W
.
Pimpao
,

A
.
Wisitsoraat
,

S
.

Mongpraneet
,

T
.
Lomas

&

M
.
Sangworasil
:

“Development

of

Wireless

Electronic

Nose

for

Environment

Quality

Classification”,

International

conference

0
n

Electrical

Engineering/Electronics,

Computer,

Telecommunications

and

Information

technology,

19
-
21

May,

2010

[
2
]

S
.
H
.

Saeed
,

Z
.

Abbas
,

B
.

Gopal
:

“Experimental

use

of

electronic

nose

for

analysis

of

volatile

organic

compound”,

Multimedia,

Signal

Processing

and

Communication

Technologies,

2009
.

IMPACT

'
09
,

14
-
16

March

2009

[
3
]
Nagle

H

T,

Gutierrez
-
Osuna

R,

Schiffman
:

“The

how

and

why

of

electronic

nose”,

IEEE

Spectrum,

Sep

1998

[
4
]

Lars

J
.

Kangas
,

Lars

H
.

Liden
,

Sherif

Hashem
,

Richard

T
.

Kouzes
:

“Electronic

noses

&

their

applications”,

IEEE

Technical


Applications

Conference

and

Workshops,

1995

[
5
]

http
:
//en
.
wikipedia
.
org/wiki/Electronic_nose

32

Thank you

33