禁忌搜索算法求解TSP旅行商问题Matlab实现

一. 禁忌搜索算法

禁忌搜索算法是一种全局性邻域搜索算法，模拟人类具有记忆功能的寻优特征。它通过局部邻域搜索机制和相应的禁忌准则来避免迂回搜索，并通过破禁水平来释放一些被禁忌的优良状态，进而保证多样化的有效搜索，以实现全局优化。

二. 禁忌搜索计算流程以及Matlab实现

1. 算法流程图

2. 代码实现

主程序

%% 禁忌搜索算法求解TSP问题程序


clear
clc

data = initData();                                          % 存放城市之间距离的矩阵
cityQty = size(data,1);                                     % 城市总数
otherCities = 2:cityQty;       
tabuTimes = 3;                                              % 禁忌次数

% 随机产生初始解
randCities = otherCities(randperm(cityQty - 1));
xnew = [1 randCities];                                      % 产生xnew
fitxnew = routeDistance(data,xnew);                         % 计算xnew解的路线长度
xbest = xnew;                                               % 让xbest = xnew     
fitxbest = fitxnew;                                         % 让fitxbest = fitxnew

% 定义禁忌表，1，2为置换的两点符号，3为禁忌剩余代数
tabuTable = zeros(0,3);

% 循环计算
isStop = 0;
while isStop < 4                                            % 设置迭代次数
    % 使用2-opt方式产生xnew的全部邻域解
    [neighbors,changePos] = neighborBy2opt(xnew);                       % 产生xnew的邻域解
    
    % 计算neighbors中每个解的值，并获取最小值
    neighborRows = size(neighbors,1);
    fitnesses = zeros(neighborRows, 2);
    for i = 1 : neighborRows
        fitnesses(i,1) = routeDistance(data, neighbors(i,:));
        minC = min(changePos(i,:));
        maxC = max(changePos(i,:));
        for j = 1 : size(tabuTable,1)
            if tabuTable(j,1) == minC && tabuTable(j,2) == maxC
                fitnesses(i,2) = 1;
                % 特赦准则
                if fitnesses(i, 1) < fitxbest 
                    fitnesses(i,2) = 0;
                end
            end
        end
    end
    
    [~, idx] = sortrows(fitnesses, [2,1]);
    xnow = neighbors(idx(1),:);
    fitnow = fitnesses(idx(1),1);               
    tabuObj = changePos(idx(1), :);                                         % 禁忌对象
    tabuTable = updataTabuTable(tabuTable, tabuObj, tabuTimes);             % 更新禁忌表
    
    % 进行解的更新和终止循环判断
    
    if fitnow < fitxbest
        xbest  = xnow;
        fitxbest = fitnow;
        isStop = 0;
    else
        isStop = isStop + 1;
    end
    xnew = xnow;
end

% 输出结果
disp('----------------------最短路径------------------------');
disp(xbest);
disp('---------------------最短路径长度---------------------');
disp(fitxbest);
    
    
    

initData()函数
初始化数据

%% 初始化数据: 各个城市之间的距离矩阵
function distData = initData()
    distData = [0 15 19 30 6 11 18 14 24
                15 0 34 44 14 23 29 4 38
                19 34 0 16 22 15 19 32 9
                30 44 16 0 35 30 18 44 7
                6 14 22 35 0 9 24 11 28
                11 23 15 30 9 0 24 20 23
                18 29 19 18 24 24 0 30 15
                14 4 32 44 11 20 30 0 37
                24 38 9 7 28 23 15 37 0];
end

routeDistance()函数
计算解的路线长度

function routeDist = routeDistance(cityDist, route)
% 根据城市间的距离计算总路径
    qty = max(size(route));
    sumDist = 0;
    for i = 1 : qty-1
        sumDist = sumDist + cityDist(route(i),route(i+1));
    end
    sumDist = sumDist + cityDist(route(qty),route(1));
    routeDist = sumDist;
end
    

neighborBy2opt（）函数
两点邻域法产生xnow的邻域解集

function [routes, changePos] = neighborBy2opt( route )
    % 根据一条路径route，采用2opt方式产生其全部邻域解集
    cityQty = max(size(route));
    if cityQty <= 3
        disp('cityQty is too small in neighborBy2opt.....');
    end
    pos = 2 : cityQty;
    changePos = nchoosek(pos, 2);
    rows = size(changePos, 1);
    routes = zeros(rows, cityQty);
    % 依次对两个点进行位置互换，形成新的解
    for i  = 1 : rows
        city1 = route(changePos(i,1));
        city2 = route(changePos(i,2));
        midRoute = route;
        midRoute(changePos(i,1)) = city2;
        midRoute(changePos(i,2)) = city1;
        routes(i,:) = midRoute;
    end
end

updataTabuTable（）函数
更新禁忌表里的对象和禁忌次数

function newTabuTable = updataTabuTable( tabuTable, tabuObj, tabuTimes )
% 更新禁忌表

oldRows = size(tabuTable,1);

% 先排序
if tabuObj(1) > tabuObj(2)
    midV = tabuObj(1);
    tabuObj(1) = tabuObj(2);
    tabuObj(2) = midV;
end

if oldRows == 0
    tabuTable = [tabuObj tabuTimes];
else
    % 将以前禁忌表中的禁忌次数-1
    for i = 1 : oldRows
        tabuTable(i, 3) = tabuTable(i,3) - 1;
    end
    
    % 查看当前禁忌对象是否存在于原有禁忌表中
    isExist = 0;
    for i = 1 : oldRows
        if tabuTable(i, 1) == tabuObj(1) && tabuTable(i, 2) == tabuObj(2)
            isExist = i;
            break;
        end
    end
    
    if isExist == 0
        midTabu = [tabuObj tabuTimes];
        tabuTable = [tabuTable; midTabu];
    else
        tabuTable(isExist, 3) = tabuTimes;
    end
    a = tabuTable(:, 3) == 0;
    tabuTable(a, :) = [];               % 删除禁忌次数为0的替换对
end
newTabuTable = tabuTable;
end

3. 实现结果