【VRP问题】基于模拟退火算法改进狼群算法求解带时间窗的车辆路径VRPTW问题附matlab代码...

最新推荐文章于 2023-03-16 09:29:20 发布

天天Matlab代码科研顾问

最新推荐文章于 2023-03-16 09:29:20 发布

阅读量8

点赞数

文章标签：算法模拟退火算法 matlab 机器学习人工智能

本文链接：https://blog.csdn.net/m0_57702748/article/details/140075290

版权

1 简介

针对城市物流配送和交通运输中广泛存在的带时间窗车辆路径问题,为寻求最佳路径规划,应用惩罚函数,构建了以总运输成本最小为目标的数学模型.在车辆路径优化求解方面,根据问题具体特征设计了1种二维编码方式,并采用近邻初始化方式构建初始解从而提升寻优速率;随后,结合狼群算法觅食行为中的游走、召唤及围攻3种行为,重新定义其智能行为,设计了一种求解带时间窗车辆路径问题的狼群算法.由于原始狼群算法的召唤行为引入距离判定因子来增大种群搜索空间,但也增加了算法复杂性且易陷入局部最优,故本研究舍弃了距离判定因子,采用猛狼1次奔袭便进入围攻状态来降低算法复杂度,并在算法中进一步增强了种群间信息交互.最后,应用该狼群算法求解多个测试算例.结果表明:基于模拟退火赛算法优化狼群算法在求解带时间窗的车辆路径问题时是可行的、有效的。

带时间窗的车辆路径问题 ( vehicle routingproblem with time windows，VRPTW) 是在基本车辆路径问题 ( vehicle routing problem，VRP) 的基础上增加了客户接受配送服务的时间窗要求，较 VRP 更贴近实际配送情况。与 VRP 类似，VRPTW 亦属于NP 难问题。由于问题规模较大时，传统精确算法难以求出 VRPTW 的最优解，因此，国内外很多学者利用智能启发式算法来寻找该问题的满意解。常见的求解 VRPTW 的智能优化算法包括: 蚁群算法、遗传算法、粒子群算法、禁忌搜索算法等。但是，由于遗传算法与禁忌搜索算法的早熟易收敛问题，蚁群算法和粒子群算法的易陷入局部最优缺陷，故目前仍未找到求解性能具有明显优势的 VRPTW 求解算法。因此，对 VRPTW 求解方法的研究仍属于运筹学、物流配送和交通运输工程等领域的研究重点和热点。狼群算法 ( Wolf Pack Algorithm，WPA) 作为一种新型智能启发式算法，是对自然界中狼群分工协作捕食行为的智能模拟，是一种基于种群的随机寻优算法。目前，该算法已被成功应用于 TSP 问题、0 － 1 高维背包问题和多维连续优化等多种问题的求解，并取得了良好的效果。但是，截至目前为止，尚未有文献将狼群算法应用于 VRPTW 的求解。基于求解 TSP 的离散狼群算法，结合带时间窗的车辆路径问题的具体特征，提出了求解 VRPTW的狼群算法，并与其他智能算法如蚁群算法、遗传算法等比较，不仅验证了本算法在求解VRPTW 问题上的有效性，而且拓宽了狼群算法的应用领域。

【VRP问题】基于模拟退火算法改进狼群算法求解带时间窗的车辆路径VRPTW问题附matlab代码_粒子群算法

【VRP问题】基于模拟退火算法改进狼群算法求解带时间窗的车辆路径VRPTW问题附matlab代码_优化算法_02

【VRP问题】基于模拟退火算法改进狼群算法求解带时间窗的车辆路径VRPTW问题附matlab代码_粒子群算法_03

【VRP问题】基于模拟退火算法改进狼群算法求解带时间窗的车辆路径VRPTW问题附matlab代码_优化算法_04

【VRP问题】基于模拟退火算法改进狼群算法求解带时间窗的车辆路径VRPTW问题附matlab代码_蚁群算法_05

2 部分代码

clc;clear

Location=[];

Location(:,1)=xlsread('Data.xlsx','sheet13','b2:b23');

Location(:,2)=xlsread('Data.xlsx','sheet13','c2:c23');

TW=xlsread('Data.xlsx','sheet13','e2:f23');

q=xlsread('Data.xlsx','sheet13','d2:d23');

workT=xlsread('Data.xlsx','sheet13','g2:g23');

n=1:size(Location,1)-1;

Q=10;

PAIXU=12;

v_min=35;v_mid=40;

v_max=50;

Ship=300;Fuel_Cost=3;Labor=20;

D=zeros(length(n)+1);

for i=1:length(n)+1

for j=i:length(n)+1

D(i,j)=6371.004*acos(sind(90-Location(i,1))*sind(90-Location(j,1))*cosd(Location(i,2)-Location(j,2))+cosd(90-Location(i,1))*cosd(90-Location(j,1)));

end

D=D+D';T=D/v_max;

yuanbao=cell(70,13);

TIME=zeros(1,70);

% for cunchu=1:70

popsize=100;

numWolf=1;

all_chromosome_cell=cell(100,12);

while numWolf<=popsize

n_n=n;

vehicle=1;

chromosome_cell=cell(length(n),4);

chromosome_cell{vehicle,1}=0;

chromosome_cell{vehicle,2}=0;

chromosome_cell{vehicle,3}=0;

chromosome_cell{vehicle,4}=0;

ArrivalWait_time=zeros(2,1);chromosome_cell{vehicle,5}=ArrivalWait_time;

while ~isempty(n_n)

R=randi(length(n_n));

customer=n_n(R);

i=1;

while i<=vehicle

if chromosome_cell{i,2}+q(customer+1)>Q

i=i+1;

if i>vehicle

vehicle=vehicle+1;

chromosome_cell{vehicle,1}=0;chromosome_cell{vehicle,2}=0;chromosome_cell{vehicle,3}=0;chromosome_cell{vehicle,4}=0;chromosome_cell{vehicle,5}=ArrivalWait_time;

end

elseif sum(chromosome_cell{i,5}(:,end))+workT(chromosome_cell{i,1}(end)+1)+T(chromosome_cell{i,1}(end)+1,customer+1)>TW(customer+1,2)

i=i+1;

if i>vehicle

vehicle=vehicle+1;

chromosome_cell{vehicle,1}=0;chromosome_cell{vehicle,2}=0;chromosome_cell{vehicle,3}=0;chromosome_cell{vehicle,4}=0;chromosome_cell{vehicle,5}=ArrivalWait_time;

end

else

if sum(chromosome_cell{i,5}(:,end))+workT(chromosome_cell{i,1}(end)+1)+T(chromosome_cell{i,1}(end)+1,customer+1)<TW(customer+1,1)

chromosome_cell{i,5}(1,end+1)=sum(chromosome_cell{i,5}(:,end))+workT(chromosome_cell{i,1}(end)+1)+T(chromosome_cell{i,1}(end)+1,customer+1);

chromosome_cell{i,5}(2,end)=TW(customer+1,1)-chromosome_cell{i,5}(1,end);

else

chromosome_cell{i,5}(1,end+1)=sum(chromosome_cell{i,5}(:,end))+workT(chromosome_cell{i,1}(end)+1)+T(chromosome_cell{i,1}(end)+1,customer+1);

chromosome_cell{i,5}(2,end)=0;

end

chromosome_cell{i,1}(end+1)=customer;

chromosome_cell{i,2}=chromosome_cell{i,2}+q(customer+1);

chromosome_cell{i,4}=chromosome_cell{i,4}+D(chromosome_cell{i,1}(end-1)+1,chromosome_cell{i,1}(end)+1);

n_n(R)=[];

break;

end

Total_fuel=zeros(1,vehicle);

for k=1:vehicle

Single_fuel=0;

chromosome_cell{k,4}=chromosome_cell{k,4}+D(chromosome_cell{k,1}(end)+1,1);

chromosome_cell{k,3}=sum(chromosome_cell{k,5}(:,end))+workT(chromosome_cell{k,1}(end)+1)+T(chromosome_cell{k,1}(end)+1,1);

Singleton_gene=chromosome_cell{k,1};

Singleton_weight=0;

for OIL=1:numel(Singleton_gene)-1

Singleton_weight=Singleton_weight+q(Singleton_gene(OIL)+1);

Q_Z=(0.8+0.2*(Singleton_weight/Q))*0.174*450*((22/27)^3)*(D(Singleton_gene(OIL)+1,Singleton_gene(OIL+1)+1)/22);

Q_F=0.174*450*((22/27)^3)*0.4*workT(Singleton_gene(OIL)+1);

Single_fuel=Single_fuel+Q_Z+Q_F;

end

Single_fuel=Single_fuel+(0.8+0.2*((Singleton_weight+q(Singleton_gene(end)+1))/Q))*0.174*450*((22/27)^3)*(D(Singleton_gene(end)+1,1)/22)+0.174*450*((22/27)^3)*0.4*workT(Singleton_gene(end)+1);

Total_fuel(k)=Single_fuel;

end

all_chromosome_cell{numWolf,1}=[chromosome_cell{:,1}];

all_chromosome_cell{numWolf,1}(end+1)=0;

all_chromosome_cell{numWolf,2}=[chromosome_cell{:,2}];

all_chromosome_cell{numWolf,3}=[chromosome_cell{:,4}];

all_chromosome_cell{numWolf,4}=[chromosome_cell{:,5}];

all_chromosome_cell{numWolf,5}=[chromosome_cell{:,3}];

all_chromosome_cell{numWolf,6}=sum(cell2mat(chromosome_cell(:,4)));

all_chromosome_cell{numWolf,7}=vehicle;

all_chromosome_cell{numWolf,10}=Total_fuel;

all_chromosome_cell{numWolf,11}=sum(Total_fuel);

all_chromosome_cell{numWolf,12}=all_chromosome_cell{numWolf,7}*Ship+sum(all_chromosome_cell{numWolf,5})*Labor+all_chromosome_cell{numWolf,11}*Fuel_Cost;

numWolf=numWolf+1;

end

for i=1:popsize

SuDu=zeros(1,length(all_chromosome_cell{i,1})-1);

SuDu(:)=v_max;

all_chromosome_cell{i,8}=SuDu;

end

all_chromosome_cell=sortrows(all_chromosome_cell,12);

BEST=all_chromosome_cell(1,:);

TanWolf=all_chromosome_cell(1:popsize/2,:);

D_GL=repmat(1/8,1,8);D_LP=cumsum(D_GL);

R_GL=repmat(0.2,1,5);R_LP=cumsum(R_GL);

DestroyWeight_matrix=[zeros(2,8);D_GL];

InsertWeight_matrix=[zeros(2,5);R_GL];

temperature=500;

beta=0.98;

nnum=1;

tab=1;

while temperature>=0.01

TanWolf_Copy=TanWolf;

first_i=1;

TanWolf_Son={};

SIMP={};

SDEG={};

while first_i<=length(TanWolf)

a=randperm(size(TanWolf_Copy,1),1);

TanWolf_Rand=TanWolf_Copy(a,:);

TanWolf_Copy(a,:)=[];

second_i=1;

h=3;

while second_i<=h

[c,b,Insert,TWolf_One,DestroyWeight_matrix,InsertWeight_matrix,POHUA,CHARU]=ALNS_VRPTW_main(DestroyWeight_matrix,InsertWeight_matrix,TanWolf_Rand,n,D_LP,R_LP,Q,q,T,TW,workT,D,v_max);

TWolf_One{12}=TWolf_One{7}*Ship+sum(TWolf_One{5})*Labor+TWolf_One{11}*Fuel_Cost;

%TanWolf_Son(h*(first_i-1)+second_i,:)=TWolf_One;

second_i=second_i+1;

if TWolf_One{12}<TanWolf_Rand{12}

SIMP(size(SIMP,1)+1,:) = TWolf_One;

if TWolf_One{12} < BEST{12}

tab=0;

Bubian=1;

BEST = TWolf_One;

DestroyWeight_matrix(2,POHUA)=DestroyWeight_matrix(2,POHUA)+6;%加分6

InsertWeight_matrix(2,CHARU)=InsertWeight_matrix(2,CHARU)+6;

else

DestroyWeight_matrix(2,POHUA)=DestroyWeight_matrix(2,POHUA)+1;%加分3

InsertWeight_matrix(2,CHARU)=InsertWeight_matrix(2,CHARU)+1;

end

else %模拟退火判断

Num_rand=rand();

if Num_rand < exp(-(TWolf_One{12}-TanWolf_Rand{12})/temperature)

TanWolf_Son(size(TanWolf_Son,1)+1,:)=TWolf_One;

DestroyWeight_matrix(2,POHUA)=DestroyWeight_matrix(2,POHUA)+2;%加分3

InsertWeight_matrix(2,CHARU)=InsertWeight_matrix(2,CHARU)+2;

end

%TanWolf_Copy(first_i,:) = TWolf_One;%%%%%%

if size(SIMP,1)~=0

SIMP=sortrows(SIMP,12);

TanWolf(first_i,:)=SIMP(1,:);

SIMP={};

end

if size(TanWolf_Son,1)<=3 && size(TanWolf_Son,1)>0

SDEG=[SDEG;TanWolf_Son];

end

first_i=first_i+1;

TanWolf_Son={};

nnum=nnum+1;

if rem(nnum,500)==0

[DestroyWeight_matrix,D_LP]=DestroyWeight_update(DestroyWeight_matrix);

[InsertWeight_matrix,R_LP] = InsertWeight_update( InsertWeight_matrix);

end

tab=tab+1;

if tab>500

break

end

TanWolf_Son_H=[TanWolf;SDEG];

TanWolf_Son_H=sortrows(TanWolf_Son_H,12);

if size(TanWolf_Son_H,1)>120

TanWolf=TanWolf_Son_H(1:100,:);

end

temperature=temperature*0.98;

end

R=BEST{1,1};%最优路径

figure(3)

sz=100;

% scatter(Location(:,1),Location(:,2),50);%客户位置

hold on

[n,nn] = size(D);%节点个数

for i=2:n

plot(Location(i,1),Location(i,2),'ro','MarkerSize',5,'MarkerEdgeColor','k','MarkerFaceColor','b');

text(Location(i,1)-0.001,Location(i,2)+0.005,[num2str(i-1)],'Fontsize',10);hold on

end

hold on

plot(Location(1,1),Location(1,2),'p','MarkerSize',30,'MarkerEdgeColor','k','MarkerFaceColor','y');hold on%发货中心

ss=0;

cost=zeros(k,1);

for j=1:length(R)-1

if R(j)==0

ss=ss+1;

end

switch ss

case 1

line([Location(R(j)+1,1),Location(R(j+1)+1,1)],[Location(R(j)+1,2),Location(R(j+1)+1,2)]);hold on%%画出车辆1的路程

cost(ss,1) = cost(ss,1) + D(R(j)+1,R(j+1)+1);%计算车辆1的路程

case 2

line([Location(R(j)+1,1),Location(R(j+1)+1,1)],[Location(R(j)+1,2),Location(R(j+1)+1,2)],'Color','r');hold on%%画出车辆2的路程

cost(ss,1) = cost(ss,1) + D(R(j)+1,R(j+1)+1);%计算车辆2的路程

case 3

line([Location(R(j)+1,1),Location(R(j+1)+1,1)],[Location(R(j)+1,2),Location(R(j+1)+1,2)],'Color','g');hold on %%画出车辆3的路程