Neuro-Fuzzy Based Controller for Landing Phase of an Aircraft

sciencediscussionAI and Robotics

Oct 20, 2013 (3 years and 7 months ago)

63 views

Neuro
-
Fuzzy Based Controller for Landing Phase of an Aircraft


C. Mygana Krishna Moorthy

78, Kangayam Cross Road,

Tirupur


641 604

Tel: 91
-
421
-
2421925

E
-
mail: cmkmoorthy@rediffmail.com


The controller design for the landing phase of an aircraft, aims at g
uiding the
aircraft through a predetermined slope of 3 deg and then from a pre
-
selected altitude
reduce the descent, so as to flare the aircraft and make a smooth touch down. In addition
to guiding the aircraft through 3 deg slope, the velocity of the airc
raft should also be
decreased proportionately, so that it lands with appropriate speed. The landing phase also
comprises lateral control in which the aircraft is aligned with the centerline of the
runway. This work deals with exploring the possibility of u
sing Neural Network and
Fuzzy
-
logic based controller to cater the demands of auto
-
landing. Advantages of
integrating Neural Network and Fuzzy
-
logic system is also analyzed. In case of Neural
Network based controller, reinforcement learning strategy is adop
ted. Reinforcement
learning refers to learning method framed by temporal difference and evolutionary
programming. Reinforcement learning based Neural Network controller is designed for
the landing problem. Fuzzy system is composed by deriving the knowledge

from experts
and framing suitable rules. Gaussian, trapezoidal, and triangular membership functions
are employed. Fuzzy system is easy to design and implement, because of use of linguistic
variables, whereas Neural Network has the power of learning from e
xperience, which
requires training the controller. Hence a Neuro
-
Fuzzy system capable of achieving best of
both systems behavior and performance is tried out in this work.