Using different solvers in Optimization Problem

Hi,

I am using problem based optimization problem and i am getting same optimized values for ga, patternsearch and particle swarm .but time taken is different.

Please advice what could be the mistake

Answer 1

[Walter Roberson]

If you are getting exactly the same results, then there are a few possibilities:

• you might have a mistake in your equation implementation that causes it to return the same values all of the time;
• you might have a mistake in how you specify the solver, causing it to always be set the same;
• you might have specified options such as integer constraints that are leading Problem Based Solver to override your solver request because the other solvers cannot handle the options / request

If you are getting the same results to within round-off error but not bit-for-bit identical:

• you might be using a simple linear equation or quadratic equation; all of the functions should converge for those kinds of function
• the result you are getting might be right at one of the constraint boundaries, so all of the solvers might end up at much the same solution

[NN]

% Minimise the cost of power 
prob.ObjectiveSense = 'minimize';
prob.Objective =dt *Cost'*PESSV;


showproblem(prob)

options = optimoptions('ga',prob.optimoptions,'Display','iter');
%options = optimoptions('patternsearch',prob.optimoptions,'Display','iter');
%options = optimoptions('ga',prob.optimoptions,'Display','iter');
%options = optimoptions(aprticleswarm',prob.optimoptions,'Display','iter');
[values,~,exitflag] = solve(prob,'Options',options);
fval = evaluate(prob.Objective,values);

[NN]

Kindly review this code and plz let me know if there is a correction.

[NN]

I have specified my constraints as below:

% Satisfy power load with power from PV, grid and ESS
prob.Constraints.loadBalance = Ppv + windV+ PgridV + PESSV == Pload;

prob.Constraints.grid = PgridV <=3e6;
prob.Constraints.solar = Ppv <=1e6;

[Walter Roberson]

Can you post your complete code for testing ?

[NN]

function [Pgrid,Pess,Eess] = battSolarOptimizetrial3(N,dt,Ppv,Pwind,Pload,Einit,Cost,FinalWeight,batteryMinMax)


% Minimize the cost of power from the grid 
 
prob = optimproblem;

% Decision variables
PgridV = optimvar('PgridV',N);
PessV = optimvar('PesssV',N,'LowerBound',batteryMinMax.Pmin,'UpperBound',batteryMinMax.Pmax);
EessV = optimvar('EesssV',N,'LowerBound',batteryMinMax.Emin,'UpperBound',batteryMinMax.Emax);

% Maximise the profit from the usage of battery and PV
prob.ObjectiveSense = 'minimize';
prob.Objective =dt *Cost'*PessV;


% Satisfy power load with power from PV, grid and battery
prob.Constraints.loadBalance = Ppv + Pwind+ PgridV + PessV == Pload;

prob.Constraints.grid = PgridV <=3e6;
prob.Constraints.solar = Ppv <=1e6;


showproblem(prob)
options = optimoptions('ga',prob.optimoptions,'Display','iter');

[values,~,exitflag] = solve(prob,'Options',options);
fval = evaluate(prob.Objective,values);


tb1=struct2table(values)
% Parse optmization results
if exitflag <= 0
    Pgrid = zeros(N,1);
    Pess = zeros(N,1);
    Eess = zeros(N,1);
    
else
    Pgrid = values.PgridV;
    Pess = values.PessV;
    Eess = values.EessV;
    
end

[NN]

• you might have a mistake in how you specify the solver, causing it to always be set the same;--- I just changed the solver name in the solve command below:

options = optimoptions('ga',prob.optimoptions,'Display','iter');

• you might have specified options such as integer constraints that are leading Problem Based Solver to override your solver request because the other solvers cannot handle the options / request

How do i correct it without modifying the constraints?what is the possible solution ?

[NN]

Please let me know what it means by moving the comment

[Walter Roberson]

You posted content as an Answer, but the content was not a solution to your question and was instead a reply to what I had posted. I used moderation facilities to move your content to be a Comment rather than an Answer

[NN]

Ok Now understood I was actually asking what is the solution Please let me know if there are any solution after reviewing the code.

[Walter Roberson]

I haven't had time to test it yet.

[NN]

Ok, please check if you are free .

When i used patternsearch, i got a warning like below:

You have passed PATTERNSEARCH options to LINPROG. LINPROG will use the common options and ignore the PATTERNSEARCH options that do not apply.

To avoid this warning, convert the PATTERNSEARCH options using OPTIMOPTIONS.

[NN]

Hi,

Can you please chekc on the above problem and suggest a solution please