基于Matlab怎么实现野狗优化算法

%====欢迎关注公众号：电力系统与算法之美====
 
function DOA()
 
%% ====参数设置====
 
popsize=20;    % 种群规模
Iteration=1000;     % 迭代次数
lb = -10;     % 各维度的下限
ub = 10;  % 各维度的上限
dim = 2;  % 优化变量的个数
 
P= 0.5;  % Hunting or Scavenger  rate. 
Q= 0.7;  % Group attack or persecution?
beta1= -2 + 4* rand();  % -2 < beta < 2     
beta2= -1 + 2* rand();  % -1 < beta2 < 1    
naIni= 2; % minimum number of dingoes that will attack
naEnd= popsize /naIni; % maximum number of dingoes that will attack
na= round(naIni + (naEnd-naIni) * rand()); % number of dingoes that will attack
 
%% ====初始化种群位置=====
Positions=lb + (ub - lb).*rand(popsize, dim);
for i=1:size(Positions,1)
    Fitness(i)=sum(Positions(i,:).^2); % get fitness
end
[best_score, minIdx]= min(Fitness);  % the min fitness value vMin and the position minIdx
best_x= Positions(minIdx,:);  % the best vector
[worst_score, ~]= max(Fitness); % the max fitness value vMax and the position maxIdx
curve=zeros(1,Iteration);
 
%% Section 2.2.4 Dingoes'survival rates
 
for i=1:size(Fitness,2)
    survival(i)= (worst_score-Fitness(i))/(worst_score - best_score);
end
 
 
%% =====开始循环===========
for t=1:Iteration
    for r=1:popsize
        if rand() < P  % Hunting
            sumatory=0;
 
            c=1;
            vAttack=[];
            while(c<=na)
                idx =round( 1+ (popsize-1) * rand());
 
                band= 0;
                for i=1:size(vAttack, 2)
                    if idx== vAttack(i)
                        band=1;
                        break;
                    end
 
                end
 
                if ~band
                    vAttack(c) = idx;
                    c=c+1;
                end
            end
 
            for j=1:size(vAttack,2)
                sumatory= sumatory + Positions(vAttack(j),:)- Positions(r,:);
            end
            sumatory=sumatory/na;
 
            if rand() < Q  % group attack
                v(r,:)=  beta1 * sumatory-best_x; % Strategy 1: Eq.2
            else  %  Persecution
                r1= round(1+ (popsize-1)* rand()); %
                v(r,:)= best_x + beta1*(exp(beta2))*((Positions(r1,:)-Positions(r,:))); % 
            end
        else % Scavenger
            r1= round(1+ (popsize-1)* rand());
            if rand() < 0.5
                val= 0;
            else
                val=1;
            end
 
            v(r,:)=   (exp(beta2)* Positions(r1,:)-((-1)^val)*Positions(r,:))/2; % 
        end
        if survival(r) <= 0.3  % Section 2.2.4, Algorithm 3 - Survival procedure
            band=1;
            while band
                r1= round(1+ (popsize-1)* rand());
                r2= round(1+ (popsize-1)* rand());
                if r1 ~= r2
                    band=0;
                end
            end
            if rand() < 0.5
                val= 0;
            else
                val=1;
            end
            v(r,:)=   best_x + (Positions(r1,:)-((-1)^val)*Positions(r2,:))/2;  % Section 2.2.4, Strategy 4: Eq.6
        end
        % Return back the search agents that go beyond the boundaries of the search space .
        Flag4ub=v(r,:)>ub;
        Flag4lb=v(r,:)<lb;
        v(r,:)=(v(r,:).*(~(Flag4ub+Flag4lb)))+ub.*Flag4ub+lb.*Flag4lb;
        % Evaluate new solutions
        Fnew= sum(v(r,:).^2);
        % Update if the solution improves
        if Fnew <= Fitness(r)
            Positions(r,:)= v(r,:);
            Fitness(r)= Fnew;
        end
        if Fnew <= best_score
            best_x= v(r,:);
            best_score= Fnew;
        end
    end
    curve(t)= best_score;
    [worst_score, ~]= max(Fitness);
    for i=1:size(Fitness,2)
        survival(i)= (worst_score-Fitness(i))/(worst_score - best_score);
    end
 
end
 
 
%======结束优化===============
 
%% 进化曲线
figure
semilogy(curve,'Color','r','LineWidth',2)
grid on
title('收敛曲线')
xlabel('迭代次数');
ylabel('最佳适应度');
axis tight
legend('DOA')
 
 
display(['最优解: ', num2str(best_x)]);
display(['最小值: ', num2str(best_score)]);
 
end