SHOW:
|
|
- or go back to the newest paste.
| 1 | function [x_opt,y_opt,W]=PSO(epsilon,Nmax,xmin,xmax) | |
| 2 | ||
| 3 | W=0; | |
| 4 | N=2; %wymiar problemu, tu dla dwóch zmiennych | |
| 5 | C=20; | |
| 6 | w=0.8; % wartosc powinna malec w czasie dzialania algorytmu | |
| 7 | c1=1; | |
| 8 | c2=1; | |
| 9 | ||
| 10 | R.x=zeros(N,C); | |
| 11 | R.v=zeros(N,C); | |
| 12 | R.y=zeros(1,C); | |
| 13 | ||
| 14 | for i=1:C | |
| 15 | R.x(:,i)=(xmax-xmin).*rand(N,1)+xmin; | |
| 16 | R.v(:,i)=0.02*(xmax-xmin).*rand(N,1)-0.01*(xmax-xmin); | |
| 17 | [R.y(i),W] = fit_fun(R.x(:,i),W); | |
| 18 | end | |
| 19 | ||
| 20 | x_opt_i=R.x; | |
| 21 | y_opt_i=R.y; | |
| 22 | [y_opt,numer]=min(R.y); | |
| 23 | x_opt=R.x(:,numer); %najlepsze rozwiązanie to cząstka o numerze ... | |
| 24 | ||
| 25 | while true | |
| 26 | for i=1:C | |
| 27 | R.v(:,i)=w*R.v(:,i)+c1*rand*(x_opt-R.x(:,i))+... | |
| 28 | c2*rand*(x_opt_i(:,i)-R.x(:,i)); | |
| 29 | R.x(:,i)=R.x(:,i)+R.v(:,i); | |
| 30 | [R.y(i),W]=fit_fun(R.x(:,i),W); | |
| 31 | if R.y(i)<y_opt | |
| 32 | y_opt=R.y(i); | |
| 33 | x_opt=R.x(:,i); | |
| 34 | if y_opt<epsilon | |
| 35 | return; | |
| 36 | end | |
| 37 | end | |
| 38 | if R.y(i) <y_opt_i(i) | |
| 39 | y_opt_i(i)=R.y(i); | |
| 40 | x_opt_i(:,i)=R.x(:,i); | |
| 41 | end | |
| 42 | if W>Nmax | |
| 43 | return; | |
| 44 | end | |
| 45 | end | |
| 46 | end | |
| 47 | ||
| 48 | ||
| 49 | function [y,w]=fit_fun(x,w) | |
| 50 | ||
| 51 | w=w+1; | |
| 52 | %y=x(1)^2+x(2)^2; | |
| 53 | m1=400; | |
| 54 | m2=50; | |
| 55 | k1=x(1); | |
| 56 | k2=2e5; | |
| 57 | b=x(2); | |
| 58 | g=9.81; | |
| 59 | x2_0=-(m1+m2)*g/k2; | |
| 60 | x1_0=-m1*g/k1+x2_0; | |
| 61 | T=5; | |
| 62 | dt=0.01; | |
| 63 | A=0.06; | |
| 64 | f=2*pi*10; | |
| 65 | ||
| 66 | assignin('base','m1',m1);
| |
| 67 | assignin('base','m2',m2);
| |
| 68 | assignin('base','k1',k1);
| |
| 69 | assignin('base','k2',k2);
| |
| 70 | assignin('base','b',b);
| |
| 71 | assignin('base','g',g);
| |
| 72 | assignin('base','x1_0',x1_0);
| |
| 73 | assignin('base','x2_0',x2_0);
| |
| 74 | assignin('base','T',T);
| |
| 75 | assignin('base','dt',dt);
| |
| 76 | assignin('base','a',A);
| |
| 77 | assignin('base','f',f);
| |
| 78 | sim('model.mdl');
| |
| 79 | plot(t,x1); | |
| 80 | - | y=0; % obliczenie wartosci funkcji celu na podstawie x1 ZROBIĆ !!!!! |
| 80 | + | figure; |
| 81 | ||
| 82 | [l,m]=butter(12,0.16,'low'); | |
| 83 | x1_f=filter(l,m,x1); | |
| 84 | plot(t,x1_f); | |
| 85 | ||
| 86 | - | %wykres sygnału filtrowanego |
| 86 | + | y=max(abs(x1_f)); |
| 87 | y=y+1e6*p; | |
| 88 | p=max(0,-k1)^2+max(0,k1-2e4)^2+ ... | |
| 89 | max(0,b)^2+max(0,b-3e3)^2+ ... | |
| 90 | - | figure |
| 90 | + | |
| 91 | plot(t,x1); | |
| 92 | - | y=mean(max(x1_f)+min(x1_f)); |
| 92 | + | |
| 93 | ||
| 94 | ||
| 95 | ||
| 96 | error('To by było na tyle'); |
