【布局优化】基于Memetic算法求解集群无线传感器网络中节能覆盖控制优化问题含Matlab源码...

1 简介

Memetic 算法是一种结合遗传算法和局部搜索策略的新型智能算法。通过与局部优化策略的结合，局部调整进化后产生的新个体，强化了算法的局部搜索能力。Memetic 算法区别于遗传算法的关键是局部搜索，局部搜索策略的效率及可靠性决定Memetic 算法的求解速度和质量。本文基于模拟退火 Memetic 算法求解桥梁传感器优化布置问题。模拟退火算法( SA) 是模拟物理退火过程中的加温、等 温、冷却等过程，基于固体物质退火过程的优化算法，是一种迭代自适应启发式概率性搜索算法，利用Metropolis 准则控制温度下降过程。

2 部分代码

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% PARAMETERS %%%%%%%%%%%%%%%%%%%%%%%%%%%%

function [coverage_rec,avg_packets_to_bs,avg_packets_to_ch,DEAD,S,last_round,CLUSTERHS,avg_ch]=LEACH(sense_node,...

                                              rmax, p, best_gene, node_x, node_y, sink_x, sink_y,...

                                              packet_bit, local_search_for_alive_neighbor_flag) 

%avg_ch: average of culser heads when all nodes run out

global target_x target_y grid_range_x grid_range_y span sense_range;

if local_search_for_alive_neighbor_flag==1

    gene_of_each_round=best_gene;

else

    gene_of_each_round=ones(1,length(best_gene));

end

Eo=0.25;

%Eelec=Etx=Erx

ETX=50*0.000000001;

ERX=50*0.000000001;

%Transmit Amplifier types

Emp=0.1*0.000000001;

%Data Aggregation Energy

EDA=5*0.000000001;

flag=0;                                                         

%Data Aggregation Energy

%Computation of do

for i=1:1:sense_node

        S(i).xd=node_x(i);

        XR(i)=S(i).xd;

        S(i).yd=node_y(i);

        YR(i)=S(i).yd;

        S(i).G=0;

        S(i).sons=0;

        %initially there are no cluster heads only nodes

        S(i).type='N';

        S(i).E=Eo;    

end

S(sense_node+1).xd=sink_x;

S(sense_node+1).yd=sink_y;

dist_node=zeros(sense_node+1,sense_node+1); % calculate the distance between two nodessense_node+1 indicates of sink

for i=1:(sense_node+1)

    for j=1:(sense_node+1)

        dist_node(i,j)=dist(S(i).xd,S(i).yd,S(j).xd,S(j).yd);

    end

end

coverage_rec=zeros(1,rmax+1);

countCHs=0;

%counter for CHs per round

rcountCHs=0;

cluster=1;

countCHs;

rcountCHs=rcountCHs+countCHs;

flag_first_dead=0;

last_round=0;

global_count=0;

%%%%%%%%%%%%%%%%%%%%%%%%% END OF PARAMETERS %%%%%%%%%%%%%%%%%%%%%%%%

%%%%%%%%%%%%%%%%%%%%%%%%% START OF ROUNDS %%%%%%%%%%%%%%%%%%%%%%%%%%

for r=1:1:rmax

  %Operation for epoch

  r

    if mod(r, round(1/p) )==0

       for i=1:sense_node  % update G, in order to determine if the epoch is done

         S(i).G=0;

         S(i).cl=0;

       end

    end

    for y=1:sense_node

       S(y).sons=0; % record the number of nodes belonging to each head

    end

% hold off;

%Number of dead nodes

dead=0;

%Number of dead Advanced Nodes

dead_a=0;

%Number of dead Normal Nodes

dead_n=0;

%counter for bit transmitted to Bases Station and to Cluster Heads

packets_TO_BS=0;

packets_TO_CH=0;

%counter for bit transmitted to Bases Station and to Cluster Heads 

%per round

PACKETS_TO_CH(r+1)=0;

PACKETS_TO_BS(r+1)=0;

for i=1:1:sense_node

    %checking if there is a dead node

    if (S(i).E<=0)

%         disp('dead=');

        dead=dead+1;

        if local_search_for_alive_neighbor_flag==1  && gene_of_each_round(i)==1

            global_count=global_count+1;

%             clf;

            gene_of_each_round_old=gene_of_each_round; % record the old topology

            gene_of_each_round(i)=0;  % update the gene and re-evaluate if the node is dead                                

            gene_of_each_round=local_search_for_alive_neighbor(i,gene_of_each_round,S,dist_node);

%             plot_topology(gene_of_each_round_old,gene_of_each_round,sense_range,node_x,node_y,target_x,target_y);

%             pause;         

        end 

            gene_of_each_round(i)=0;  %double check 

            S(i).type='D';          

    end

end

% recheck 

if sum(gene_of_each_round)==0

    for b=1:sense_node

        if S(b).E>0

            gene_of_each_round(b)=1;

        end

    end

end

for j=1:sense_node

    if S(j).E>0 && gene_of_each_round(j)==1

        S(j).type='N';

    end

end

% after check, evaluate the coverage

 [coverage_rec(1,r+1),temp]=fit_foreach(gene_of_each_round);

STATISTICS(r+1).DEAD=dead;

DEAD(r+1)=dead;

%When the first node dies

if (dead==1)

    if(flag_first_dead==0)

        first_dead=r

        flag_first_dead=1;

    end

end

countCHs=0; 

cluster=1; %counter index

alive_node=0;

for i=1:1:sense_node

   if(S(i).E>0)

   temp_rand=rand;    

   alive_node=alive_node+1;

   if ( (S(i).G)<=0  && gene_of_each_round(i)==1) % possibly act as a head if G<=0

       %Election of Cluster Heads    

       if(temp_rand<= (p/(1-p*mod(r,round(1/p)))))

            countCHs=countCHs+1;

            packets_TO_BS=packets_TO_BS+1;  

            S(i).type='C';

            S(i).G=round(1/p)-1;   

            % G will not be 0 if it has acted as a head. G can be used as a index to determine if act as a head

            C(cluster).xd=S(i).xd;

            C(cluster).yd=S(i).yd;

            distance=dist_node(i,sense_node+1);% distance between the node and sink

            C(cluster).distance=distance;

            C(cluster).id=i;

            X(cluster)=S(i).xd;

            Y(cluster)=S(i).yd;

            cluster=cluster+1;

        end     

    end

   end 

end

CLUSTERHS(r+1)=(cluster-1);

if (alive_node==0 && flag==0 )||(r==rmax && flag==0 ) % record the last_round if the max round is achieved or all nodes are dead

    last_round=r;

    flag=1;

end

packets_TO_CH=0;

%Election of Associated Cluster Head for Normal Nodes

for i=1:1:sense_node

   if ( S(i).type=='N' && S(i).E>0 && gene_of_each_round(i)==1)

     if(cluster-1>=1) % if there are more than one head

       min_dis=dist_node(i,sense_node+1);

       min_dis_cluster=1;

       for c=1:1:cluster-1

           temp=min(min_dis,sqrt( (S(i).xd-C(c).xd)^2 + (S(i).yd-C(c).yd)^2 ) );

           if ( temp<min_dis )

               min_dis=temp;

               min_dis_cluster=c;

           end

       end     

          S(i).E=S(i).E- ( ETX*(packet_bit) + Emp*packet_bit*( min_dis * min_dis)); 

        if(min_dis_cluster~=1)

          S(C(min_dis_cluster).id).E = S(C(min_dis_cluster).id).E- ( (ERX + EDA)*packet_bit ); 

          S(C(min_dis_cluster).id).sons=S(C(min_dis_cluster).id).sons+1;

          packets_TO_CH=packets_TO_CH+1;

        end

        S(i).min_dis=min_dis;

        S(i).min_dis_cluster=min_dis_cluster;      

     else

        min_dis=dist_node(i,sense_node+1);

        min_dis_cluster=1;

        S(i).E=S(i).E- ( ETX*(packet_bit) + Emp*packet_bit*( min_dis * min_dis)); 

        packets_TO_BS=packets_TO_BS+1; 

     end

 end

end

      PACKETS_TO_CH(r+1)=packets_TO_CH; 

      PACKETS_TO_BS(r+1)=packets_TO_BS;     

         % the number of packets to be sent to BS from heads

for h=1:sense_node %% calculate the energy consumption of the transmission between head and bs

    if S(h).type=='C' && S(h).E>0

       dis=dist_node(h,sense_node+1);% distance between the node and sink

       S(h).E=S(h).E- ( EDA*(packet_bit) + ETX*(packet_bit) + Emp*packet_bit*( dis * dis)); 

    end

end

% hold on;

countCHs;

rcountCHs=rcountCHs+countCHs;

for i=1:sense_node %% reset the energy index to 0 if it's minus 

    if S(i).E<0

        S(i).E=0;

    end

end

% if r==1 || r==500 || r==1500 || r==3000 || r==4500 || r==4900

% % if r==rmax

%     energy_analysis(100,100,S,sense_node);  

% %     print(figure(10),'-dmeta',[int2str(r)]);  

% 

%     pause;

% end

end

% energy_analysis(grid_range_x*span,grid_range_y*span,S,sense_node);

% plot_topology(1,sense_range,gene_of_each_round,node_x,node_y,target_x,target_y);

gene_of_each_round

fprintf('\nglobal_count=%d',global_count);

figure(2);

if last_round~=0

    avg_ch=sum(CLUSTERHS)/last_round;

    avg_packets_to_bs=sum(PACKETS_TO_BS)/last_round;

    avg_packets_to_ch=sum(PACKETS_TO_CH)/last_round;

else

    avg_ch=sum(CLUSTERHS)/rmax;

end

plot(sense_node-DEAD(2:(rmax+1)),'b');     %  the dead nodes will appear from 2nd round 

figure(3);

target_num=length(target_x(1,:))*length(target_y(:,1));

plot(coverage_rec(2:(rmax+1))/target_num,'r')

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%   STATISTICS    %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%                                                                                     %

%  DEAD  : a rmax x 1 array of number of dead nodes/round 

%  DEAD_A : a rmax x 1 array of number of dead Advanced nodes/round

%  DEAD_N : a rmax x 1 array of number of dead Normal nodes/round

%  CLUSTERHS : a rmax x 1 array of number of Cluster Heads/round

%  PACKETS_TO_BS : a rmax x 1 array of number packets send to Base Station/round

%  PACKETS_TO_CH : a rmax x 1 array of number of packets send to ClusterHeads/round

%  first_dead: the round where the first node died                   

%                                                                                     %

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

end

3 仿真结果

4 参考文献

[1]高亮, 刘文君, and 周驰. "用Memetic算法求解有时间约束的TSP问题." 华中科技大学学报：自然科学版 36.7(2008):4.

博主简介：擅长智能优化算法、神经网络预测、信号处理、元胞自动机、图像处理、路径规划、无人机等多种领域的Matlab仿真，相关matlab代码问题可私信交流。

部分理论引用网络文献，若有侵权联系博主删除。