竞争粒子群（competitive swarm optimizer，CSO）优化算法学习笔记

竞争粒子群（competitive swarm optimizer，CSO）优化算法学习笔记

~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~

先导知识

粒子群优化算法：
https://blog.csdn.net/chensanwa/article/details/79284698?spm=1001.2014.3001.5506

参考文献：
R. Cheng and Y. Jin, A competitive swarm optimizer for large scale optimization, IEEE Transactions on Cybernetics, 2014, 45(2): 191-204.

竞争粒子群与粒子群的不同

竞争粒子群与粒子群优化算法不同，竞争粒子群中的每个粒子的最佳位置和全局最佳位置不再由pbest和gbest更新，而是由一种竞争机制更新。

每次竞争后，将根据胜者的信息更新败者的信息。

算法思想

一个种群P(t)有m个粒子，t为迭代次数。
每个粒子位置为：Xi(t) = (xi,1(t),xi,2(t), . . . ,xi,n(t))
速度为：Vi(t) = (vi,1(t),vi,2(t) ,…,vi,n(t))
种群被分成m/2对(假设m为偶数)，每对间两个粒子进行竞争。
拥有更好适应值得为胜者，直接进入下一代种群，败者则根据胜者信息更新自己的位置和速度。更新完毕后传入下一代种群

更新公式为：
V(l,k) (t+1)=R1 (k,t) V(l,k) (t)+R2 (k,t) (X(w,k) (t)-X(l,k) (t))
+ φR3k,t(Xkt -Xl,kt)（1）
X(l,k) (t+1)=X(l,k) (t))+V(l,k) (t+1) （2）
l表示败者，w表示胜者，R()为随机产生的向量，X ̅k (.)模式是全局粒子平均位置，φ为控制X ̅k (.)影响的参数。
公式（1）的第一部分与PSO中的惯性部分类似，第二部分为败者从胜者处学习，第三部分则是全局学习，可以与PSO中的三部分对照看。

由上可以得出伪代码

文章图片均来自：R. Cheng and Y. Jin, A competitive swarm optimizer for large scale optimization, IEEE Transactions on Cybernetics, 2014, 45(2): 191-204.

示例代码

代码来自PlatEMO工具箱，下载地址：https://github.com/BIMK/PlatEMO

classdef CSO < ALGORITHM
% <large/none> <constrained/none>
% Competitive swarm optimizer
% phi — 0.1 — Social factor

%------------------------------- Reference --------------------------------
% R. Cheng and Y. Jin, A competitive swarm optimizer for large scale
% optimization, IEEE Transactions on Cybernetics, 2014, 45(2): 191-204.
%------------------------------- Copyright --------------------------------
% Copyright © 2021 BIMK Group. You are free to use the PlatEMO for
% research purposes. All publications which use this platform or any code
% in the platform should acknowledge the use of “PlatEMO” and reference “Ye
% Tian, Ran Cheng, Xingyi Zhang, and Yaochu Jin, PlatEMO: A MATLAB platform
% for evolutionary multi-objective optimization [educational forum], IEEE
% Computational Intelligence Magazine, 2017, 12(4): 73-87”.
%--------------------------------------------------------------------------

methods
    function main(Algorithm,Problem)
        %% Parameter setting
        phi = Algorithm.ParameterSet(0.1);
        
        %% Generate random population
        Population = Problem.Initialization();
        
        %% Optimization
        while Algorithm.NotTerminated(Population)
            % Determine the losers and winners
            rank    = randperm(Problem.N);
            loser   = rank(1:end/2);
            winner  = rank(end/2+1:end);
            replace = FitnessSingle(Population(loser)) < FitnessSingle(Population(winner));
            temp            = loser(replace);
            loser(replace)  = winner(replace);
            winner(replace) = temp;
            % Update the losers by learning from the winners
            LoserDec  = Population(loser).decs;
            WinnerDec = Population(winner).decs;
            LoserVel  = Population(loser).adds(zeros(size(LoserDec)));
            R1  = repmat(rand(Problem.N/2,1),1,Problem.D);
            R2  = repmat(rand(Problem.N/2,1),1,Problem.D);
            R3  = repmat(rand(Problem.N/2,1),1,Problem.D);
            LoserVel = R1.*LoserVel + R2.*(WinnerDec-LoserDec) + phi.*R3.*(repmat(mean(Population.decs,1),Problem.N/2,1)-LoserDec);
            LoserDec = LoserDec + LoserVel;
            Population(loser) = SOLUTION(LoserDec,LoserVel);
        end
    end
end

end