求大佬帮帮我改改代码，把遗传算法的部分改为粒子群算法能实现的就行

Question

close all;
monte_num=200;  % 蒙特卡洛仿真次数
%% 不同P_on
%clear all;
clc;
n=10:5:40;        % 用户数
n_max=max(n);
m=10;           % 子信道数

pfa=0.01;      % 虚警概率
Ts=5*10^-3;      % 感知时间1ms
Tr=195*10^-3;  % 传输时间     待定
data=1;%0.5;                             %  信道，能量损耗影响因子
Qfa=0.05;                                % 系统虚警概率要求
pdoh=0;             
%monte_num=10;  % 蒙特卡洛仿真次数
match_method=2;  
compare=3;       % 比较的方法个数
Pd_req=0.95; %0.95;
conspdreq=1;     %考虑各信道检测概率需求法
kc=60;     % 种群数目,暂定必须是偶数,且能被4整除

    Gain_sum=zeros(compare,length(n));
    Overhead_sum=zeros(compare,length(n));
    profit_sum=zeros(compare,length(n));
    Pd_min_sum=zeros(compare,length(n));
    Pd_ave_sum=zeros(compare,length(n));
    Pf_sum=zeros(compare,length(n));
    num_co_sum=zeros(compare,length(n));

for k=1:monte_num  
    k
    seq=0;
    R_cr=(rand(1,n_max)*0.99+0.01)*10^7;       %　各节点传输速率  0.1Mbps -10M bps   %  zeros(1,n_max); %
    R_ch=(rand(1,m)*0.99+0.01)*10^7;           %　各信道传输速率  0.1Mbps -10M bps
    
    p_d_matrix=rand(m,n_max)*0.7+0.3;        %  各用户各信道检测概率矩阵  均值0.8
   % P_on=rand(1,m)*0.6+0.4;  %0.7;      % 信道空闲概率
    
       
    % 遗传算法
    P_on=ones(1,m)*0.5;
    [ fit_best_finalvalue, fit_best_co_num, Pd_final_ave, Pf_final  ] = PM_GAMP_22( n, m, P_on, pfa, Ts,Tr,data,Qfa,R_cr,R_ch,p_d_matrix,60);
    
    seq=seq+1;
    %Gain_sum(seq,:)=Gain_sum(seq,:)+Gain_final;
    %Overhead_sum(seq,:)=Overhead_sum(seq,:)+Overhead_final;
    profit_sum(seq,:)=profit_sum(seq,:)+fit_best_finalvalue;
    %Pd_min_sum(seq,:)=Pd_min_sum(seq,:)+Pd_final_min;
    Pd_ave_sum(seq,:)=Pd_ave_sum(seq,:)+Pd_final_ave;
    Pf_sum(seq,:)=Pf_sum(seq,:)+Pf_final;
    num_co_sum(seq,:)=num_co_sum(seq,:)+fit_best_co_num;
    
    P_on=ones(1,m)*0.7;
    [ fit_best_finalvalue, fit_best_co_num, Pd_final_ave, Pf_final  ] = PM_GAMP_22( n, m, P_on, pfa, Ts,Tr,data,Qfa,R_cr,R_ch,p_d_matrix,60);
    
    seq=seq+1;
    %Gain_sum(seq,:)=Gain_sum(seq,:)+Gain_final;
    %Overhead_sum(seq,:)=Overhead_sum(seq,:)+Overhead_final;
    profit_sum(seq,:)=profit_sum(seq,:)+fit_best_finalvalue;
    %Pd_min_sum(seq,:)=Pd_min_sum(seq,:)+Pd_final_min;
    Pd_ave_sum(seq,:)=Pd_ave_sum(seq,:)+Pd_final_ave;
    Pf_sum(seq,:)=Pf_sum(seq,:)+Pf_final;
    num_co_sum(seq,:)=num_co_sum(seq,:)+fit_best_co_num;
    
    P_on=ones(1,m)*0.9;
    [ fit_best_finalvalue, fit_best_co_num, Pd_final_ave, Pf_final  ] = PM_GAMP_22( n, m, P_on, pfa, Ts,Tr,data,Qfa,R_cr,R_ch,p_d_matrix,60);
    
    seq=seq+1;
    %Gain_sum(seq,:)=Gain_sum(seq,:)+Gain_final;
    %Overhead_sum(seq,:)=Overhead_sum(seq,:)+Overhead_final;
    profit_sum(seq,:)=profit_sum(seq,:)+fit_best_finalvalue;
    %Pd_min_sum(seq,:)=Pd_min_sum(seq,:)+Pd_final_min;
    Pd_ave_sum(seq,:)=Pd_ave_sum(seq,:)+Pd_final_ave;
    Pf_sum(seq,:)=Pf_sum(seq,:)+Pf_final;
    num_co_sum(seq,:)=num_co_sum(seq,:)+fit_best_co_num;
    
end;

    Gain_monte=Gain_sum/monte_num;
    Overhead_monte=Overhead_sum/monte_num;
    profit_monte=profit_sum/monte_num;
    % Pd_min_monte=Pd_min_sum/monte_num;
    Pd_ave_monte=Pd_ave_sum/monte_num;
    Pf_monte=Pf_sum/monte_num;
    num_co_monte=num_co_sum/monte_num;

figure;
grid on;hold on; 
plot(n,profit_monte(1,:)','-sk');    % plot(n,profit_monte,'-s')  找到错误了，plot画不同曲线的时候，是按照每列一条曲线画的，而我是按照每行一条曲线画的，当compare~=length(n)时，由于前面有n限制，因此是按照行话，当compare=length(n)，就按照默认的列画了
plot(n,profit_monte(2,:)','-dk'); 
plot(n,profit_monte(3,:)','-*k'); 
xlabel('认知用户数');ylabel('总系统效益(bps)');
legend('\it{P_{on}}=0.5','\it{P_{on}}=0.7','\it{P_{on}}=0.9','location','southeast');

figure;
grid on;hold on; 
plot(n,num_co_monte(1,:)','-sk');   
plot(n,num_co_monte(2,:)','-dk'); 
plot(n,num_co_monte(3,:)','-*k');     % 重新仿的话按照顺序3,4就行了
xlabel('认知用户数');ylabel('参与合作感知的认知用户数');
legend('\it{P_{on}}=0.5','\it{P_{on}}=0.7','\it{P_{on}}=0.9','location','southeast');