Artificial Intelligence in Control Engineering exercise

Lecturer: Dr.Pham Viet Cuong

September 12nd, 2018

Group 9:
Nguyen Chinh Thuy 1513372
Nguyen Tan Phu 1512489
Le Van Hoang Phuong 1512579
Do Tieu Thien 1513172
Nguyen Tan Sy 1512872
Nguyen Van Qui 1512702

1 Problem
Given a single-variable function

y(x) = 4x4 − 5x3 + exp−2x − sin(x) − 3cos(x)

subject to x ∈ [0, 5], use the Genetic Algorithm to find the global maximum of the function y(x).

2 Configuration
To solve the problem with the Genetic Algorithm, we use a configuration that is shown in the table 1.

Table 1: Configuration of the Genetic Algorithm

Parameter Value
Number of binrary bits 13
Number of workers 100
Number of generations 200
Probability of mating 0.95
Probability of mutation 0.01
Range of x [0, 5]
Number of discrete points of x 200

1
 3 Implementation
3.1 Python code
1 #−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−
2 # Libraries
3 #−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−
4 im po rt numpy a s np
5 im po rt m a t p l o t l i b . p y p l o t a s p l t
6 im po rt tqdm
7 from IPython imp ort d i s p l a y
8 im po rt time
9
10 #−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−
11 # C l a s s o f G e n e t i c Algorithm
12 #−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−
13 c l a s s GeneticAlgorithm ( o b j e c t ) :
14 ”””
15 [ Arguments ]
16 f u n c : F u n c t i o n t o f i n d t h e maximum v a l u e s
17
18 n b i t s : Number o f b i n r a r y b i t s
19
20 n w o r k e r s : Number o f w o r k e r s
21
22 n g e n : Number o f g e n e r a t i o n s
23
24 m a t i n g p r o b : P r o b a b i l i t y o f mating
25
26 m u t a t i o n p r o b : P r o b a b i l i t y o f mutation
27
28 x r a n g e : Range o f x
29
30 x n p o i n t s : Number o f d i s c r e t e p o i n t s o f x
31 ”””
32 def init ( s e l f , func , n b i t s , n w o r k e r s , n gen ,
33 mating prob , m u t a t i o n p r o b , x r a n g e , x n p o i n t s ) :
34 super ( GeneticAlgorithm , s e l f ) . init ()
35 # P a ra m e t e r s o f t h e G e n e t i c Algorithm
36 s e l f . func = func
37 self . n bits = n bits
38 s e l f . n workers = n workers
39 s e l f . n gen = n gen
40 s e l f . mating prob = mating prob
41 s e l f . mutation prob = mutation prob
42 s e l f . x range = x range
43 s e l f . x n points = x n points
44
45 # Inialize variables
46 s e l f . w o r k e r s = np . random . r a n d i n t ( 0 , 2 , s i z e =( s e l f . n w o r k e r s , self . n bits ) )
47 s e l f . b e s t w o r k e r = None
48
49 # Compute t h e o b j e c t i v e
50 d e f c o m p u t e o b j e c t i v e ( s e l f , pred , e s p=1e −8) :
51 r e t u r n pred + e s p − np . min ( pred )
52
53 # Convert b i n a r y t o d e c i m a l
54 d e f d ec od e ( s e l f , w o r k e r s ) :
55 delta = ( s e l f . x range [ 1 ] − s e l f . x range [ 0 ] )
56 norm range = w o r k e r s . dot ( 2 ∗∗ np . a r a n g e ( s e l f . n b i t s ) [ : : − 1 ] ) / ( 2 ∗ ∗ s e l f . n b i t s −1)
57 r e t u r n norm range ∗ d e l t a + s e l f . x r a n g e [ 0 ]
58
59 # Natural s e l e c t i o n
60 def s e l e c t ( s e l f , objective ) :
61 i d x = np . random . c h o i c e ( np . a r a n g e ( s e l f . n w o r k e r s ) , s i z e= s e l f . n w o r k e r s , p=o b j e c t i v e /
o b j e c t i v e . sum ( ) )
62 return s e l f . workers [ idx ]
63
64 # Mating
65 d e f mate ( s e l f , p a r e n t , w o r k e r s ) :
66 i f np . random . rand ( ) < s e l f . m a t i n g p r o b :
67 i d x = np . random . r a n d i n t ( 0 , s e l f . n w o r k e r s , s i z e =1)
68 c r o s s p o i n t s = np . random . r a n d i n t ( 0 , 2 , s i z e= s e l f . n b i t s ) . a s t y p e ( np . b o o l )
69 p a r e n t [ c r o s s p o i n t s ] = w o r k e r s [ idx , c r o s s p o i n t s ]
70 return parent
71

2
72 # Mutation
73 d e f mutate ( s e l f , c h i l d ) :
74 f o r i in range ( s e l f . n b i t s ) :
75 i f np . random . rand ( ) < s e l f . m u t a t i o n p r o b :
76 c h i l d [ i ] = 1 i f c h i l d [ i ]==0 e l s e 0
77 return child
78
79 # Perform a s t e p o f e v o l u t i o n
80 def step ( s e l f ) :
81 r e a l v a l s = s e l f . d ec od e ( s e l f . w o r k e r s )
82 o u t v a l s = func ( r e a l v a l s )
83 objective = s e l f . compute objective ( out vals )
84 s e l f . b e s t w o r k e r = s e l f . w o r k e r s [ np . argmax ( o b j e c t i v e ) , : ]
85 s e l f . b e s t o b j e c t i v e = s e l f . f u n c ( s e l f . de co de ( s e l f . b e s t w o r k e r ) )
86
87 s e l f . workers = s e l f . s e l e c t ( o b j e c t i v e )
88 w o r k e r s c o p y = s e l f . w o r k e r s . copy ( )
89 f o r parent in s e l f . workers :
90 c h i l d = s e l f . mate ( p a r e n t , w o r k e r s c o p y )
91 c h i l d = s e l f . mutate ( c h i l d )
92 parent [ : ] = c h i l d
93
94 return real vals , out vals
95
96 # Perform t h e whole e v o l u t i o n
97 def evolve ( s e l f ) :
98 f i g = p l t . f i g u r e ( f i g s i z e =(25 , 8 ) )
99 ax1 = f i g . a d d s u b p l o t ( 1 , 2 , 1 ) ; ax1 . s e t t i t l e ( ” Workers e v o l u t i o n ” )
100 x = np . l i n s p a c e ( ∗ s e l f . x r a n g e , s e l f . x n p o i n t s )
101 ax1 . p l o t ( x , f u n c ( x ) )
102 ax2 = f i g . a d d s u b p l o t ( 1 , 2 , 2 ) ; ax2 . s e t t i t l e ( ” O b j e c t i v e e v o l u t i o n ” )
103
104 best objectives = [ ]
105 f o r i i n tqdm . tqdm ( r a n g e ( s e l f . n g e n ) ) :
106 # Perform a s t e p o f e v o l u t i o n
107 real vals , out vals = s e l f . step ()
108
109 # Plot the workers e v o l u t i o n
110 i f ’ sca ’ in l o c a l s () :
111 s c a . remove ( )
112 s c a = ax1 . s c a t t e r ( r e a l v a l s , o u t v a l s , s= s e l f . x n p o i n t s , lw =0 , c= ’ r e d ’ , a l p h a
=0.5)
113
114 # Plot the o b j e c t i v e e v o l u t i o n
115 b e s t o b j e c t i v e s . append ( s e l f . b e s t o b j e c t i v e )
116 ax2 . p l o t ( b e s t o b j e c t i v e s , ”b” )
117
118 d i s p l a y . c l e a r o u t p u t ( w a i t=True )
119 display . display ( plt . gcf () )
120 time . s l e e p ( 0 . 0 1 )
121
122
123 #−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−
124 # Main e x e c u t i o n
125 #−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−
126 # D e f i n e a f u n c t i o n , which we want t o f i n d i t s maximum
127 f u n c = lambda x : 4∗ x ∗∗4 − 5∗ x ∗∗3 + np . exp (−2∗x ) − np . s i n ( x ) − 3∗ np . c o s ( x )
128
129 # I n s t a n t i a t e a G e n e t i c Algorithm
130 GA = G e n e t i c A l g o r i t h m ( f u n c=func , n b i t s =13 , n w o r k e r s =100 , n g e n =200 ,
131 m a t i n g p r o b = 0 . 9 5 , m u t a t i o n p r o b = 0 . 0 1 , x r a n g e = [ 0 , 5 ] , x n p o i n t s =200)
132
133 # Evolution
134 GA. e v o l v e ( )
135 b e s t w o r k e r = GA. b e s t w o r k e r
136 x o p t = GA. d ec od e ( b e s t w o r k e r )
137 y opt = func ( x opt )
138 p r i n t ( ” \ nOptimal r e s u l t : ” )
139 p r i n t ( ” ( x , y )max : (%.4 f , %.4 f ) ” % ( x o p t , y o p t ) )

3
3.2 Result

(a) Workers evolution

(b) Objective evolution

Figure 1: Optimal result with(x, y(x))max: (5.0000, 1875.1080)