基于遗传算法/贪心算法的移栽路径优化研究(附全部MATLAB代码)

工业EAs

已于 2023-10-24 23:52:39 修改

阅读量185

点赞数 1

文章标签：贪心算法 matlab 人工智能 1024程序员节

于 2023-10-09 20:55:20 首次发布

本文链接：https://blog.csdn.net/qq_17633137/article/details/133715589

版权

本文四种算法源码下载地址:【免费】遗传算法/贪心算法/固定次序法移植路径规划MATLAB代码资源-CSDN文库

补苗移栽和稀植移栽是温室多叶移栽的两种主要作业方式。补苗移栽是将穴盘中的空穴或劣质幼苗替代为健康幼苗以保证幼苗出厂的一致性，能减少因缺苗导致的大田移栽的漏种；稀植移栽是将密集穴盘内的幼苗移栽到稀疏穴盘内，使幼苗有更多的养分和更大的生长空间。移栽机的末端执行器在执行移栽任务时，大部分时间穿梭于供苗穴盘和目标穴盘之间，在不改变机器结构和成本的情况下，可通过移栽路径规划缩短移动距离来提高移栽效率。本文的代码实现了类似于TSP旅行商问题的移植算法。

代码：

rng default
rng (1)
%% order
sizes = 500/8; % 格子Y轴长度
size2 = 250/4; % X轴长度
end_base =[30 30]; % 左边偏移
end_width = 4; % X轴格子个数
end_long = 8; % Y轴格子个数
end_num = end_width*end_long;
end_emptynum = 12; % 缺苗个数
end_v = ones(1,end_num); end_v(randi(end_num,1,end_emptynum))=0; end_Mat = reshape(end_v,end_long,end_width); % 矩阵
%end_Mat=[1 1 1 1;1 0 1 1;1 0 1 1;1 1 1 1] 指定的话直接覆盖矩阵就行。
end_pos = [];
noneed_pos=[];
for i = 1:end_long
    for j = 1:end_width
        if end_Mat(i,j)==0
            pos = end_base+[(j-1+0.5)*size2 (end_long-i+0.5)*sizes ];
            end_pos=[end_pos;pos]; % 所有缺苗位置
        end
        if end_Mat(i,j)==1
            pos = end_base+[(j-1+0.5)*size2 (end_long-i+0.5)*sizes ];
            noneed_pos=[noneed_pos;pos]; % 所有缺苗位置
        end
    end
end


start_base =[380 30];% 右边偏移
start_width = 4; % 补苗区X轴格子个数
start_long = 8; % 补苗区Y轴格子个数
start_num = start_width*start_long;
start_emptynum = 12;% 补苗区空格个数
start_v = ones(1,start_num); start_v(randi(start_num,1,start_emptynum))=0; start_Mat = reshape(start_v,start_long,start_width);
start_pos=[];

for i = 1:start_long
    for j = 1:start_width
        if start_Mat(i,j)==1
            pos = start_base+[(j-1+0.5)*size2 (start_long-i+0.5)*sizes ];
            start_pos=[start_pos;pos]; % 所有可使用苗的数目
        end
    end
end

% 线框位置计算
line1=[];line2=[];
for i = 1:end_long+1
    line1=[line1;end_base+[0 (i-1)*sizes]];
    line2=[line2;end_base+[size2*end_width (i-1)*sizes]];
end
 for j = 1:end_width+1
     line1=[line1;end_base+[(j-1)*size2 0]];
     line2=[line2;end_base+[(j-1)*size2 sizes*end_long]];
 end
data.start_pos=start_pos;
data.end_pos=end_pos;
data.end_num = size(end_pos,1);
data.select_num = size(start_pos,1);
% 线框位置计算
for i = 1:start_long+1
    line1=[line1;start_base+[0 (i-1)*sizes]];
    line2=[line2;start_base+[size2*start_width (i-1)*sizes]];
end
 for j = 1:start_width+1
     line1=[line1;start_base+[(j-1)*size2 0]];
     line2=[line2;start_base+[(j-1)*size2 sizes*start_long]];
 end
tic
start_pos_temp=start_pos;
POS_ORDER=[0 0]; % 起点
for i = 1:data.end_num
    POS_ORDER=[POS_ORDER;start_pos_temp(i,:);end_pos(i,:);]; % 添加
end
POS_ORDER=[POS_ORDER;0 0]; % 回到起点
toc
figure(3)
% 绘制结果
hold on
plot(end_pos(:,1),end_pos(:,2),'blues','MarkerSize',15)
plot(start_pos(:,1),start_pos(:,2),'blueo','MarkerSize',15)
plot(POS_ORDER(:,1),POS_ORDER(:,2),'-.*','LineWidth',1.6);
plot(noneed_pos(:,1),noneed_pos(:,2),'blacko','MarkerSize',15)
for i = 1:size(line1,1)
plot([line1(i,1) line2(i,1)],[line1(i,2) line2(i,2)],'black-');
end
for i = 1:size(POS_ORDER,1)-1
    text(POS_ORDER(i,1),POS_ORDER(i,2),num2str(i));
end

dist_ORDER=sqrt(sum((POS_ORDER(2:end,:)-POS_ORDER(1:end-1,:)).^2,2));
disp('固定次序法距离和')
title(['固定次序算法 总位移:' num2str(sum(dist_ORDER))])
disp(sum(dist_ORDER))
disp('固定次序距离')
disp(dist_ORDER)
disp('固定次序位置')
disp(POS_ORDER)
figure(4)
hold on
plot(start_pos(:,1),start_pos(:,2),'redo','MarkerSize',15)
plot(noneed_pos(:,1),noneed_pos(:,2),'blacko','MarkerSize',15)
for i = 1:size(line1,1)
plot([line1(i,1) line2(i,1)],[line1(i,2) line2(i,2)],'black-');
end

主函数代码：

rng default
rng (1)
%%
sizes = 500/8; % 格子Y轴长度
size2 = 250/4; % X轴长度
end_base =[30 30]; % 左边偏移
end_width = 4; % X轴格子个数
end_long = 8; % Y轴格子个数
end_num = end_width*end_long;
end_emptynum = 8; % 缺苗个数
end_v = ones(1,end_num); end_v(randi(end_num,1,end_emptynum))=0; end_Mat = reshape(end_v,end_long,end_width); % 矩阵
%end_Mat=[1 1 1 1;1 0 1 1;1 0 1 1;1 1 1 1] 指定的话直接覆盖矩阵就行。
end_pos = [];
noneed_pos=[];
for i = 1:end_long
    for j = 1:end_width
        if end_Mat(i,j)==0
            pos = end_base+[(j-1+0.5)*size2 (end_long-i+0.5)*sizes ];
            end_pos=[end_pos;pos]; % 所有缺苗位置
        end
        if end_Mat(i,j)==1
            pos = end_base+[(j-1+0.5)*size2 (end_long-i+0.5)*sizes ];
            noneed_pos=[noneed_pos;pos]; % 所有缺苗位置
        end
    end
end


start_base =[380 30];% 右边偏移
start_width = 4; % 补苗区X轴格子个数
start_long = 8; % 补苗区Y轴格子个数
start_num = start_width*start_long;
start_emptynum = 12;% 补苗区空格个数
start_v = ones(1,start_num); start_v(randi(start_num,1,start_emptynum))=0; start_Mat = reshape(start_v,start_long,start_width);
start_pos=[];

for i = 1:start_long
    for j = 1:start_width
        if start_Mat(i,j)==1
            pos = start_base+[(j-1+0.5)*size2 (start_long-i+0.5)*sizes ];
            start_pos=[start_pos;pos]; % 所有可使用苗的数目
        end
    end
end

% 线框位置计算
line1=[];line2=[];
for i = 1:end_long+1
    line1=[line1;end_base+[0 (i-1)*sizes]];
    line2=[line2;end_base+[size2*end_width (i-1)*sizes]];
end
 for j = 1:end_width+1
     line1=[line1;end_base+[(j-1)*size2 0]];
     line2=[line2;end_base+[(j-1)*size2 sizes*end_long]];
 end
data.start_pos=start_pos;
data.end_pos=end_pos;
data.end_num = size(end_pos,1);
data.select_num = size(start_pos,1);
% 线框位置计算
for i = 1:start_long+1
    line1=[line1;start_base+[0 (i-1)*sizes]];
    line2=[line2;start_base+[size2*start_width (i-1)*sizes]];
end
 for j = 1:start_width+1
     line1=[line1;start_base+[(j-1)*size2 0]];
     line2=[line2;start_base+[(j-1)*size2 sizes*start_long]];
 end
% 绘制背景
figure(1)
hold on
plot(end_pos(:,1),end_pos(:,2),'blacks','MarkerSize',15)
plot(start_pos(:,1),start_pos(:,2),'blueo','MarkerSize',15)
for i = 1:size(line1,1)
plot([line1(i,1) line2(i,1)],[line1(i,2) line2(i,2)],'black-');
end
% 遗传算法参数
dim = data.end_num+data.select_num;
lb = zeros(1,dim); % 下界
ub = ones(1,dim); % 上界
NP = 200; % 种群数目
maxiter = 200; % 迭代次数
Pc = 0.8; % 交叉概率
Pm = 0.1; % 变异概率
Pe = 0.1; % 精英选择比例
tic
res = GA(NP,dim,maxiter,@callfit,lb,ub,data,Pc,Pm,Pe); % 调用遗传优化
toc
figure(2) % 绘制迭代图
plot(res.trace,'LineWidth',1.6);
xlabel('Iteration')
ylabel('Fitness')
[dist,RES]=callfit(res.X,data)

figure(1) % 绘制遗传结果
plot(RES.trace(:,1),RES.trace(:,2),'-.*','LineWidth',1.6);
plot(noneed_pos(:,1),noneed_pos(:,2),'blacko','MarkerSize',15)
disp('遗传算法距离和')
disp(dist)
disp('遗传算法距离')
disp(RES.distlist)
disp('遗传算法位置')
disp(RES.trace)
title(['遗传算法 总位移:' num2str(dist)])
for i = 1:size(RES.trace,1)-1
    text(RES.trace(i,1),RES.trace(i,2),num2str(i));
end