gga_fis.m

书籍代码:遗传演算法原理与应用_活用MATLAB(Source Code)

% Use Chang's GA to find the best PID gain varaibles
% cd d:\ga_chang
% gga_fis.m
% gena_fis.m is defined to evaluate fitness function

%===========================================
global Kp Ki Kd wn zeta Emin main_loop 
global tippa MIN_offset MUL_factor
%===========================================
% Define options
Emin=20000;  wn=1; zeta=0.1; main_loop=0;
MIN_offset=20000; MUL_factor=1;
tippa=readfis('pidfuzz');
wn=1;zeta=0.1;
%        Kp     Ki    Kd
range=[   1      1     1
        600     10   150]; 
obj_fcn='gena_fis';
bit_n = 40;		% Bit number for each input variable
gen_no =15;	% Number of generations
popuSize =22;		% Population size
xover_rate = 0.97;	% Crossover rate (BETTER)
mutate_rate = 0.04;	% Mutation rate (BETTER)
elite=1;
IC=[];
%idmotora
%************************************************************
% call GA written by Chang
tic
[popu, popu_real, fcn_value, upper, average, lower, ...
    BEST_popu, popuSize, gen_no, para] = GA_genetic(obj_fcn,...
    range, IC, elite, gen_no, popuSize);
tc=toc/60;
fprintf('==> Computation time is (%.2f) minutes.\n',tc);            

% Simulate the system and calculate the ERROR.
fprintf('\n            <<FINAL RESULTS>>  \n'); date
load gena_data; main_loop=1;
figure(3);
xWhitebg;
x = (1:gen_no)';
plot(x, upper, 'o', x, average, 'x', x, lower, '*');
hold on;
plot(x, [upper average lower]);
hold off;
legend('Best', 'Average', 'Poorest');
xlabel('Generations'); ylabel('Fitness');
figure(2);subplot(212);
[t,x,Y]=sim('pidb_fuzzy',[0 20]); y=Y(:,2);in=Y(:,3);
plot(t,[in y]);grid
title('I/O responses, After GA search, gga-fis.m file used')
xlabel(['Time(s),Kp=' num2str(Kp) ',Ki=' num2str(Ki) ',Kd=' num2str(Kd)]);