Genetic algorithm gives incorrect output

Hi,

I have a problem with a piece of code running a genetic algorithm. The algorithm optimizes input parameters to a model which has only time as an output. However when I put the parameters, that came out of the genetic algorithm optimization, into my model, the model gives out NaN. The optimum that the genetic algorithm is finding is not a viable solution for the model.

I hope someone can help me out, I checked everythin but I really can't seem to find the mistake.

Kind regards,

Wouter

%% Main file for simulation
clear all
close all
clc


%% Course options

% Distances = [2000 1000 17000;2000 3000 15000;10000 1000 9000;10000 3000 7000];
% Coursepar.distance = Distances(i,:);
% Coursepar.gradient = [0 atan(6/100) 0];

Coursepar.distance = [18000 2000];
Coursepar.gradient = [0 atan(6/100)];
%% Selecting Riders
Riderpar.Names = {'TT_specialist','TT_specialist','TT_specialist','TT_specialist','Climber','Climber','Climber','Climber'};

%% Strategy parameters
Optimpar = [450,450,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,0]/5;
% Optimpar = [450,450,450,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,0]/5;


Stratpar.min_finish = 4;
Stratpar.pref = 1;
Stratpar.sui_allow = 0;


%% Environment parameters

Environment.rho = 1.225;
Environment.g   = 9.81;
Environment.Cr  = 0.0027;

%% Running simulation

tic
fun = @(OptimPar)multisegrun(Riderpar,Coursepar,Stratpar,Environment,Optimpar);
[x,t,exitflag] = ga(fun,19,[],[],[],[],[70 70 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0],[160 160 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 4],[],[1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19])
toc

Answer 1

[Alan Weiss]

I suggest that you give an initial population consisting of at least one feasible point, meaning a point that gives a non-NaN value of the fitness function. Set this population in the InitialPopulationMatrix option. If you cannot locate any such point, well, move on to the next suggestion.

What might be even better is to debug your multisegrun file so that it doesn't ever return NaN. This is more work, but fixes the underlying cause of the issue.

Alan Weiss

MATLAB mathematical toolbox documentation