matlab遗传算法解决TSP旅行者问题

function res = gademo() % ELEC 6604 Dr. G. Pang % Example on the MATLAB implementation of the Genetic Algorithm % Feb 2017 % based on paper2 (EV charging, 3 players) % Problem description: % Three players: % player 1, demand 6 kW/hr, 5 time intervals % player 2, demand 6 kW/hr, 6 time intervals % player 3, demand 3 kW/hr, 3 time intervals % max power each interval is 3.2 kWh, each vehicle per interval has a max limit of 1.6 kWh % $20 $45 $30 $x $y $z % EV1 a1 b1 c1 d1 e1 % EV2 a2 b2 c2 d2 e2 f2 % EV3 a3 b3 c3 clear all,close all % Free parameters are d1, a2, b2, c2, d2, e2, b3, c3 % Part 1: Initialization % The randsize is the number of random cases generated % Each case will be checked to see if it can satisfy the contraints to be a parent. % If so, it will be added to parentpop. % The desired number of initial parent is popsize, which is the population size. randsize=200; popsize=20; EP=[20 45 30 50 60 70]; parentpop=[ ]; childpop = [ ]; bestTparent=0; checksize = [ 0 0 ]; % Part 2: To generate only the "popsize" number of valid parents to enter into the Main Loop while (checksize(1) < popsize ) pop = rand(randsize,8)*1.6; % 8 free parameters, final 14 parameters for (i=1:randsize) tmp=create(pop(i,:)); if (tmp ~= 999) parentpop = [parentpop ; tmp]; end end checksize=size(parentpop); end % End of the while loop % To generate the parents and sort the best popsize to enter into the main loop results=cost(parentpop,EP); [A,index]=sort(cost(parentpop,EP),1,'ascend'); newparentsort=parentpop(index,:); parentpop=newparentsort([1:popsize],:); % To test crossover functions using two individual parents parent1 = parentpop(3,:); parent2 = parentpop(4,:); [child1,child2]=crossover1(parent1,parent2); [child3,child4]=crossover2(parent1,parent2); % To test mutation function [child5]=mutation( parentpop(1,:)); %-- Part 3: Canonical Main Loop---------------------------------------------------- NumGeneration = 200; for igen = 1:NumGeneration %---- Part 3A, to generate the children by GA as new generation --------- childpop=[ ]; childpopsize=[ 0 0 ]; % parentpop has popsize number of parents % To generate the popsize number of children while (childpopsize(1) < popsize ) % To generate the random index for crossover and mutation ind=randi(popsize,[1 2]) ; parent1 = parentpop(ind(1),:); parent2 = parentpop(ind(2),:); [child1,child2]=crossover1(parent1,parent2); [child3,child4]=crossover2(parent1,parent2); [child5]=mutation( parent1); if (child1 ~= 999) childpop = [childpop ; child1]; end if (child2 ~= 999) childpop = [childpop ; child2]; end if (child3 ~= 999) childpop = [childpop ; child3]; end if (child4 ~= 999) childpop = [childpop ; child4]; end if (child5 ~= 999) childpop = [childpop ; child5]; end childpopsize=size(childpop); end % End of the while loop % childpop has over popsize number of children % Elite: parentpop and childpop are added together before sorting for the best popsize to continue % truncation selection to generate parents for the next generation totalpop=[parentpop ; childpop ]; [A,index]=sort(cost(totalpop,EP),1,'ascend'); newtotalsort=totalpop(index,:); parentpop=newtotalsort([1:popsize],:); % The best parent is saved for display plot_opti(igen)=A(1); %--------- Part 3B, for comparison, generate popsize parents randomly, and save the best ---- Tparentpop = [ ]; checksize = [ 0 0 ]; while (checksize(1) < popsize ) pop = rand(randsize,8)*1.6; % 8 free parameters, final 14 parameters for (i=1:randsize) tmp=create(pop(i,:)); if (tmp ~= 999) Tparentpop = [Tparentpop ; tmp]; end end checksize=size(Tparentpop); end % End of the while loop % Just pick up the top "popsize" number of valid, random cases Tparentpop = Tparentpop(1:popsize,:); size(Tparentpop) % To sort for the best, and save the best result if it is better than before [A,index]=sort(cost(Tparentpop,EP),1,'ascend'); if (igen == 1) plot_rand(igen) = A(1); % The best rand case is saved for display bestTparent = Tparentpop(index(1),:); else if ( A(1) < plot_rand(igen-1) ) plot_rand(igen) = A(1); % The best rand case is saved for display bestTparent = Tparentpop(index(1),:); else plot_rand(igen) = plot_rand(igen-1); end end end % END of the Main FOR loop % Part 4: Results [plot_opti ; plot_rand] parentpop(1,:) bestTparent figure,plot(1:igen,plot_opti,'b',1:igen,plot_rand,'r') title('GA algorithm for EV charging problem: blue is the optimized case') end