MATLAB Answers

is it possible to use cell array to define Lb and Ub of genetic algorithm in Optimization Toolbox

4 views (last 30 days)
Nourhan Elsayed
Nourhan Elsayed on 30 May 2020
Commented: Ameer Hamza on 30 May 2020
i am creating a code to minimize a function using genetic algorithm where i have 2 decision variables x15 and x16.
x15 is a 2x6 vector, x16 is 1x6 vector.
next code is my file for variables defininsion
function Variables(x15, x16)
% Declaration of global variables
global x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 x14 x15 x16 x17 x18
%% Input Data
x1 = 6; % j
x2 = 2; % m
x3 = 1; % i
x4= 2880; %P_G "6L34DF"
x5 = [0, 0, 1204, 2147, 1761, 5100];% P_Prop
x6 = [250, 600, 1500, 450, 450, 450]; % P_EL
x7= [2847, 1533, 197, 2267, 1478, 438]; % T
x8= 2919.786477; % u
x9= -6490.928826; % v
x10= 10952.80516; % w
x11 = 4400; % 8V31DF P_SMCR
x12 = 1128.256228;% a
x13 = -1940.085409; % b
x14 = 7938.533808; % c
%% Decision Variables
x15 = zeros(x2,x1);
x16 = zeros(x3, x1);
%% Dependent Variables
x17 = x12 .* x15.^2 + x13 .* x15+ x14; % SFOC_ME
x18 = x8 .* x16.^2 + x9 .* x16 + x10 ; % SFOC_G
end
and this is my File for ObjectiveFunction (Fittness Function)
function y = ObjectiveFunction(x15, x16)
% Declaration of global variables
global y x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 x14 x15 x16 x17 x18
%% Objective Function
y = 2*(((x11' .* x15) .* x17) * x7') +((x4'.* x16) .* x18) * x7'; % TFC
end
and i have inserted Lower and Ubber Bound into the optimization toolbox as follows
lb = {{zeros(2,6)}, {zeros(1,6)}};
ub = {{ones(2,6)}, {ones(1,6)}};
nvar = 2
in addition, i refere to the next constraints file
function [c,ceq] = NonLinearConstraints
%% Declaration of global variables
global x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 x14 x15 x16 x17 x18
%%
for J = 1:x1
c = zeros(1,2*x1)
c(J) = x5(J) - x11 * x15(J)
c(J+x1) = x6(J) - x4 * x16(J)
end
ceq = [];
end
i have got this message
Error running optimization.
GA requires the following inputs to be of data type double: 'lb','ub'.
my question is, if this data type is not accepted by the ga, what is the code i should use to define lb and ub for a vector decision varialbe not a scaler variable as usual?
my second question is, does my code allow the algorithm to see the decison variables among other variables or not?
Thank you in advance :)

  0 Comments

Sign in to comment.

Answers (1)

Ameer Hamza
Ameer Hamza on 30 May 2020
Edited: Ameer Hamza on 30 May 2020
Define lb and ub as simple vectors. For example
nvars = 18;
lb = zeros(1, nvars);
ub = ones(1, nvars)
Also you need to pass nvars=18, not nvars=2 to the ga().

  2 Comments

Nourhan Elsayed
Nourhan Elsayed on 30 May 2020
thank you Ameer for your answer
if nvars = 18, does it mean that the algorithm is going to change all the 18 variables. i think yes. but i dont want the algorithm to change all the 18 variables, just x15 and x16.
Ameer Hamza
Ameer Hamza on 30 May 2020
Ok. In that case, use nvars=2 and just define lb and ub to be [0 0] and [1 1].
Can you show how you are calling ga() with the function in your question? It will help in understanding if there is an issue in writing the commands.

Sign in to comment.


Translated by