Beruflich Dokumente
Kultur Dokumente
NO
Compare
PID with Stop
FLC
Ye
s
Flowchart for FLC design suing optimal
PID
Measure
Redesign Design FLC e(t), ie(t)
NO
Compare
FLC
with PID
Stop
YES
Design of FLC for Air heating process
Design of conventional PID controller
Conduct the open loop studies
Conduct closed loop studies
Repeat PID in discrete form
Obtain the variation of error & integral
error
Fuzzification
Rule base
Decision making logic
Defuzzification
Step1: Design of PID controller
Create a model of AHS
Give the step i/p and get the open loop
response
G(s)=Kpe-Ds/(1+sζ )
Obtain the optimal values of Kp, ζ and
D using Ziegler Nichol’s tuning
e(t) = c(t) – r(t)
ie(t) = en-1 (t) + en(t)
t t
e(t) ie(t)t
18
-0.2
t t
m(t)
t
Variation of manipulated
variable
Step2: Fuzzification
e(t) = 1-(-0.2)/2
= 0.6
Ie(t) = 52-18/2
= 17
MV = 40-0/2
= 20
ERROR INTEGRAL MANIPULATED
ERROR VARIABLE
e(t)=-0.2 to 1 ie(t)=18 to 52 Mv=0 to 40
e(t)
VS S M L VL
RULE
ie(t) BASE
VS S M L VL
m(t)
Step3: DECESION MAKING
LOGIC
EXAMPLE
If e(t)=0.4, if ie(t)=0.2 them m(t)
is VL
If e(t)=0.5, if ie(t)=0.9 them m(t)
is VL
MIN-MAX criteria
MIN{e(t)=0.4, ie(t)=0.2}=0.2
MIN{e(t)=0.5, ie(t)=0.9}=0.5
Max{0.2, 0.5} = 0.5
Therefore MV=0.5
Step4: DEFUZZIFICATION
Centreof Area method – provides
smooth control action
μ=Σμ(wj)(wj)/Σμ(wj)
j=1 to n where n denotes no.of
linguistic variables
wj denotes the support value
μ(wj)
denotes the value of
membership grade
Drawbacks of conventional
FLC
Tedious to find the universe of
discourse ‘y’ for every support ‘x’
y=[x-x1/x2-x1][y2-y1]+y1 = y
x
Fortuning – change the width of
membership function with the top
held constant