Neural Network Controller for Two-Degree- Freedom Helicopter Control System

runmidgeAI and Robotics

Oct 20, 2013 (3 years and 8 months ago)

59 views

STUDENT

PHOTO

Department
of
Electric and Electronic Engineering

Faculty of Engineering

Universiti Putra Malaysia, 43400 Serdang, Selangor, Malaysia.

E
-
Mail:

Supervisor:


Department of
Electric and Electronic Engineering

Faculty of Engineering

Universiti

Putra Malaysia,
43400
UPM Serdang,

Selangor, Malaysia.

Tel: +603


8946 4449

E
-
Mail
:
ribhan@eng.upm.edu.my

Neural Network
Controller for Two
-
Degree
-
Freedom
Helicopter Control System


Name:

Matric No.:

Project Summary




For

emergency

services,

one

of

the

fittest

candidates

is

two
-
degree
-
freedom

(
2
-
DOF)

helicopter
.



Helicopter

is

inherently

complicated

unstable

and

nonlinear

dynamic

system

under

a

significant

influence

of

disturbances

and

parameter

perturbations
.

The

system

has

to

be

stabilized

using

a

feedback

controller
.



Automatic

control

features

of

2
-
DOF

helicopter

are

usually

approximated

by

linear

quadratic

regulator

(LQR)

controller,

which

is

not

efficient

due

to

its

linear

feature
.




A

control

system

that

adapts

the

nonlinear

dynamics

of

various

flight

regimes

as

they

occur

has

the

potential

to

achieve

superior

performance
.

Towards

this

aim

artificial

intelligence

e
.
g
.
,

Neural

Network

(NN)

based

system

is

on
-
demand
.


This

Research

presents

the

nonlinear

control

of

2
-
DOF

helicopter

using

an

efficient

artificial

NN

controller
.

It

is

developed

using

back

propagation,

feed

forward

NN

model

to

control

helicopter’s

motors

and

consequently

pitch

and

yaw

angles

using

MATLAB

software
.



The

practical

implementation

into

the

existing

2
-
DOF

Helicopter

gives

precise

results

for

the

changes

in

the

pitch

and

yaw

angles
(~

33
%

pitch

and

69
%

yaw

improvement

compared

to

LQR)


followed

by

increased

stability

and

safety

in

controlling

the

helicopter
.



This

may

be

useful

in

controlling

helicopters

precisely

whenever

complicated

maneuvers

needed

to

conduct

by

pilots
.



Performance Results

Quanser

2
-
DOF Helicopter (
Left
) and
Dynamics of
2
-
DOF
Helicopter (Right)

Project
Development Flow

Conclusions


The

proposed

controller

exhibits

good

tracking

performance

for

an

unstable

helicopter

system
.



Increased

stability

and

safety

in

controlling

the

helicopter

remotely

for

emergency

/

rescue

services
.


Useful

in

controlling

helicopters

precisely

whenever

complicated

maneuvers

needed

to

conduct

by

pilots
.


SIMULINK
Block Diagram
for
The Closed
-
loop
Helicopter Controller
S
ystem

Simple
System Block Diagram

Pitch & Yaw Response for controller comparison














Metric

Value of Pitch Response for controller

LQR

LQR+I

Proposed NN

Rise Time

1.85 sec

1.21 sec

0.5 sec

Overshoot

76.51%

53.2%

20.5%

Settling Time

8.05 sec

4.07 sec

5.0 sec

Metric

Value of Yaw Response for controller

LQR

LQR+I

Proposed NN

Rise Time

2.35 sec

1.64 sec

0.7 sec

Overshoot

86.51%

71.4%

12.5%

Settling Time

20.05 sec

4.05 sec

2.0 sec


Left
-
Up: NN controller Pitch
Response


Left
-
Down: NN controller Yaw
Response


Right
-
Up: GUI for NN test &
validation of data


Right
-
Down: NN Best
-
fit plot


Just
-
Down: NN Performance /
Efficiency