Particle swarm optimization Algorithm (PSO) for optimal solution

Explanation of particle swarm optimization algorithm (PSO): http://blog.csdn.net/myarrow/article/details/51507671 (various references on the Internet)

Using PSO to find the function extremum.

function [f]= Fun (x, y)%UNTITLED2 shows a    Summary of this function here = (x-a) ^2 + (y-b) ^2; end

Constructs a two-tuple function, obviously with a minimum, in (50, 50).

Use five of particles to find it. The function output value corresponding to each point is considered a particle's fitness.

1 clear All;2 CLC;3%Initialize4n =5;5P = rand (5,2) * -;6p = cell (5,1);7  fori =1: N8P (I,1) = {[P (I,1), P (I,2)]};9 EndTenv = rand (5,2) *0.5; OneVmax =0.5; APbest =p; -Gbest = cell (1); -Fit_v = Zeros (5,1); theKey =1; -  fori =1: N -Fit_v (I,1) = Fun (P{i} (1), P{i} (2)); -     if(Fit_v (I,1) < Fit_v (Key,1) ) +Key =i; - End + End AGbest (1) = Pbest (Key,1); at%Loop -t =1; -  while(T < -) -      fori =1: N -          forj =1:2 -V (i, j) = V (i, J) +2*rand () * (Pbest{i, T} (j)-p{i} (j)) +2*rand () * (Gbest{t} (J)-P{i} (j)); in             if(ABS (V (i, j)) >Vmax) -V (i, j) = ABS (V (i, J))/V (i, j) *Vmax; to End +P{i} (j) = P{i} (j) +V (i, j); - End theFit_v_temp = Fun (P{i} (1), P{i} (2)); *         if(Fit_v_temp <fit_v (i, T)) $Fit_v (i, t+1) =fit_v_temp;Panax NotoginsengPbest (i, t+1) =p (i); -         Else theFit_v (i, t+1) =fit_v (i, t); +Pbest (i, t+1) =pbest (i, t); A End the End +t = t +1; -Key =1; $      fori =1: N $         if(Fit_v (i, T) <fit_v (key, T)) -Key =i; - End the End -Gbest (t) =pbest (key, t);WuyiEnd

Operation result: [50.0030796073836,50.0004541255463]

Figure 1 gbest trajectory (red dot indicates last convergent position)

Fig. 2 The fitness curve of each particle

Can be found in 100 times around the basic convergence.

Particle swarm optimization Algorithm (PSO) for optimal solution