一个基于Matlab的人工免疫算法

frogfish

1.   %Immune Algorithm based on the immune network model for function f(x1,x2) optimum
   
2. %copy right SCUT Guangxing Tan 2005.02.18
   
3. clear all;
   
4. 
   
5. %Parameters
   
6. Size=120;   
   
7. G=200;     
   
8. CodeL=15;
   
9. 
   
10. E=round(rand(Size,2*CodeL));    %Initial Code
    
11. 
    
12. %Main Program
    
13. for k=1:1:G
    
14. time(k)=k;
    
15. 
    
16. for s=1:1:Size
    
17. m=E(s,:);
    
18. y1=0;y2=0;
    
19. 
    
20. %Uncoding
    
21. m1=m(1:1:CodeL);
    
22. for i=1:1:CodeL
    
23.    y1=y1+m1(i)*2^(i-1);
    
24. end
    
25. x1=10.24*y1/65535.0-5.12;
    
26. m2=m(CodeL+1:1:2*CodeL);
    
27. for i=1:1:CodeL
    
28.    y2=y2+m2(i)*2^(i-1);
    
29. end
    
30. x2=10.24*y2/65535.0-5.12;
    
31. %f(X1,X2)=(a/(b+(x1*x1+x2*x2)))*(a/(b+(x1*x1+x2*x2)))+(x1*x1+x2*x2)*(x1*x1+x2*x2)
    
32. %here -5.12=<x1<-2,  -5.12=<x2<=5.12, a=3.0,  b=0.05;
    
33. 
    
34. % f(X1,X2)=90+(a/(b+(x1*x1+x2*x2)))*(a/(b+(x1*x1+x2*x2)))+(x1*x1+x2*x2)*(x1*x1+x2*x2)
    
35. % here -2=<x1<=5.12,  -5.12=<x2<=5.12
    
36. 
    
37. % f(X)=(-x+3*M_PI-0.5)(1+sinx*sinx)     if 1.5M_PI<x<=2M_PI;
    
38. 
    
39.    if  x1<-2
    
40.     F(s)=(3.0/(0.05+(x1*x1+x2*x2)))*(3.0/(0.05+(x1*x1+x2*x2)))+(x1*x1+x2*x2)*(x1*x1+x2*x2);
    
41.      else
    
42.     F(s)=90+(3.0/(0.05+(x1*x1+x2*x2)))*(3.0/(0.05+(x1*x1+x2*x2)))+(x1*x1+x2*x2)*(x1*x1+x2*x2);
    
43.    end
    
44. end
    
45. 
    
46. sumf=0;
    
47.   for s=1:1:Size;
    
48.    sumf = sumf+F(s);
    
49. end
    
50.   favg=sumf./Size;
    
51.   
    
52. %initiate density  
    
53.    x=F./sumf;
    
54. 
    
55. % This function updates the concentrations of the population antibodies
    
56. % according to the algorithm update rule:
    
57. %UpdateConcentrations
    
58. %Hamming distance
    
59. for i=1:1:Size
    
60.    for j=1:1:Size  
    
61.         H(i,j)=0;
    
62.       for Q=1:1:20
    
63.            H(i,j)=H(i,j)+E(i,Q).*(1-E(j,Q))+E(j,Q).*(1-E(i,Q));
    
64.        end
    
65.         affinity(i,j)=F(i)*H(i,j);
    
66.     end
    
67. end
    
68. % 亲密度计算完毕
    
69. %**************以上是人工免疫的，以下是遗传算法的交叉、突变代码*****************
    
70. %Euclider distance           
    
71. %for i=1:1:Size;
    
72.    %for j=1:1:Size;  
    
73.        %H(i,j)=0;
    
74.             %H(i,j)=sqrt((x1(i)-x1(j)).^2+(x2(i)-x2(j)).^2);
    
75.             %affinity(i,j)=F(i)./(H(i,j)+1);
    
76.             %end
    
77.             %end   
    
78. 
    
79. % This function normalizes the values of the concentrations of the
    
80. % population. If the concentrations are perceived as percentages of
    
81. % the total concentration of the system then their sum must be 1.0.
    
82. % This function recalculates the concentrations so that this rule
    
83. % is followed.
    
84. c =0.8;
    
85. k1 =0.1;
    
86. k2 =0.3;
    
87. %x = popv(:,fitness);
    
88. for i = 1:1:Size
    
89.       xsum = 0.0;   
    
90.   for j = 1:1:Size
    
91.     xsum= abs(xsum+(F(j).*affinity(i,j)-k1.*F(j).*affinity(j,i)-k2/(c*Size)).*F(i));
    
92.        %m(j,i)=popv(i,fitness).*affinity(i,j);
    
93.        %m(i,j)=popv(j,fitness).*affinity(j,i);
    
94.        %n(i,j)= m(j,i)-k1.*m(i,j)-k2/(c*popsize);
    
95.        %sum= sum+n(i,j).*popv(j,fitness);   
    
96.    end  
    
97.    x=x+x.*c.*xsum;
    
98. end
    
99.    
    
100.    Fitness=x./sum(x);
     
101. 
     
102. Ji=1./Fitness;
     
103. %****** Step 1 : Evaluate BestJ ******
     
104. 
     
105. 
     
106. fi=Fitness;                          %Fitness Function
     
107. [Oderfi,Indexfi]=sort(fi);     %Arranging fi small to bigger
     
108. Bestfi=Oderfi(Size);           %Let Bestfi=max(fi)
     
109. BestS=E(Indexfi(Size),:);      %Let BestS=E(m), m is the Indexfi belong to max(fi)
     
110. bfi(k)=Bestfi;
     
111. 
     
112. %****** Step 2 : Select and Reproduct Operation******
     
113.    fi_sum=sum(fi);
     
114.    fi_Size=(Oderfi/fi_sum)*Size;
     
115.    
     
116.    fi_S=floor(fi_Size);        %Selecting Bigger fi value
     
117.    
     
118.    kk=1;
     
119.    for i=1:1:Size
     
120.       for j=1:1:fi_S(i)        %Select and Reproduce
     
121.        TempE(kk,:)=E(Indexfi(i),:);  
     
122.          kk=kk+1;              %kk is used to reproduce
     
123.       end
     
124.    end
     
125.    
     
126. %************ Step 3 : Crossover Operation ************
     
127. pc=0.80;
     
128. n=ceil(20*rand);
     
129. for i=1:2:(Size-1)
     
130.     temp=rand;
     
131.     if pc>temp                  %Crossover Condition
     
132.     for j=n:1:20
     
133.         TempE(i,j)=E(i+1,j);
     
134.         TempE(i+1,j)=E(i,j);
     
135.     end
     
136.     end
     
137. end
     
138. TempE(Size,:)=BestS;
     
139. E=TempE;
     
140.    
     
141. %************ Step 4: Mutation Operation **************
     
142. %pm=0.001;
     
143. %pm=0.001-[1:1:Size]*(0.001)/Size; %Bigger fi, smaller Pm
     
144. %pm=0.0;    %No mutation
     
145. pm=0.1;     %Big mutation
     
146. 
     
147.    for i=1:1:Size
     
148.       for j=1:1:2*CodeL
     
149.          temp=rand;
     
150.          if pm>temp                %Mutation Condition (varify incident)
     
151.             if TempE(i,j)==0
     
152.                TempE(i,j)=1;
     
153.             else
     
154.                TempE(i,j)=0;
     
155.             end
     
156.         end
     
157.       end
     
158.    end
     
159.    
     
160. %Guarantee TempPop(30,:) is the code belong to the best individual(max(fi))
     
161. TempE(Size,:)=BestS;
     
162. E=TempE;
     
163. M(k)=F(Size);
     
164. end   % Is special for if k
     
165. 
     
166. Max_Value=F(Size)
     
167. BestS
     
168. x1
     
169. x2
     
170. figure(1);
     
171. plot(time,M);
     
172. xlabel('Times');ylabel('Best F');
     
173. figure(2);
     
174. plot(time,bfi);
     
175. xlabel('times');ylabel('Best Fitness');

复制代码

yanzi621

:lol不错的

wangahu2008

能用吗??发表者

chg555

在其他地方看到过,好象还缺函数..

oobb1314

确实牛逼！！！！！！！！！！！！！顶