LINGO小练习——物流配送问题——混合整数规划模型MILP

贴吧里看到的题，稍微做了一下，虽然建的是MILP模型，但奇怪的是电脑运了一宿还是Feasible状态得不出Gobal optimal的结果，开了两个版本的LINGO 11和17 算了23小时，结果一直是318不变，感觉最优解就是318，可能是某项约束太宽了或者维数太大了，整数线性模型LINGO都算的很慢，这也不算是严格的旅行商模型吧，我在模型里并没有加TSP约束，因为从题目中看不出来要求小车是每天都回到基地的，仅要求2个供应商满足6个市场的需求即可。有知道的可以评论告诉我下。
题目的意思我整理了下，供应商1和2每天制造固定的商品(Production)，满足S1-S6每天对应的需求(Needs)，剩余的商品可以遗留到下一天使用，最后对每天的路线进行规划配送（线路上的数字代表成本）最优配送成本(Cost)。
6天、8个地点（2个基地，6个市场）、2台小车依次进行物流配送；
设：
$$\begin{gathered} x_{n,j,k}^{i,1}=\{\begin{array}{c}1\\ 0\end{array}\begin{array}{c}\text{小车}1\text{第}i\text{天第}n\text{次从}j\text{地去往}k\text{地}\\ \text{else}\end{array} \\ {x}_{n,j,k}^{i,2}=\{\begin{array}{c}1\\ 0\end{array}\begin{array}{c}\text{小车}2\text{第}i\text{天第}n\text{次从}j\text{地去往}k\text{地}\\ \text{else}\end{array} \end{gathered}$$
我们设P1为7地，P2为8地；每天小车1都从P1出发，小车2都从P2出发：
$$\sum _{k=1}^8{x}_{1,7,k}^{i,1}=\sum _{k=1}^8{x}_{1,8,k}^{i,2}=1$$
每次决策只有一个，只能走相邻，并且连续：
$$\begin{gathered} \sum _{k=1}^8{x}_{n,j,k}^{i,1}=\sum _{k=1}^8{x}_{n+1,k,j}^{i,1}\begin{array}{cc}& \end{array}\sum _{k=1}^8{x}_{n,j,k}^{i,2}=\sum _{k=1}^8{x}_{n+1,k,j}^{i,2} \\ \begin{array}{cc}& \end{array} \\ {x}_{n,j,k}^{i,1}+d_{i,j}⩽2\begin{array}{cc}& \end{array}{x}_{n,j,k}^{i,2}+d_{i,j}⩽2\begin{array}{cc}& \end{array}{d}_{i,j}=\{\begin{array}{c}1\\ 2\end{array}\begin{array}{c}\text{相邻}\\ \text{不相邻}\end{array} \end{gathered}$$
设置两个0-1变量$\begin{array}{cc}{z}_{i,j}^1& z_{i,j}^2\end{array}$表示小车$1、2$是否到达第$i$天$j$地，和正整数变量$\begin{array}{cc}{q}_{i,j}^1& q_{i,j}^2\end{array}$表示小车1/2第$i$天对$j$地的供给，满足需求约束：
$$\begin{gathered} \begin{array}{cc}{q}_{i,j}^1⩽100z_{i,j}^1& q_{i,j}^2⩽100z_{i,j}^2\end{array} \\ \begin{array}{cc}& \end{array} \\ {q}_{i,j}^1+q_{i,j}^2=Need{s}_{i,j} \end{gathered}$$
最后我们对剩余量(Residual)加到下一天的剩余量当中：
$$\begin{gathered} \begin{array}{cc}P{r}_i^1=P_i^1+r_i^1& P{r}_i^2=P_i^2+r_i^2\end{array} \\ \begin{array}{cc}& \end{array} \\ \begin{array}{cc}{r}_{i+1}^1=P{r}_i^1-{\sum }_{j=1}^8{q}_{i,j}^1& r_{i+1}^2=P{r}_i^2-{\sum }_{j=1}^8{q}_{i,j}^2\end{array} \end{gathered}$$
最后总的最小化目标函数就是一周配送成本最小化：
$$\min =\sum _{i=1}^6\sum _{n=1}^8\sum _{j=1}^8\sum _{k=1}^8\sum _{q=1}^{2}cos{t}_{i,j}{x}_{n,j,k}^{i,q}$$
将已知数据整理好就可以建立LINGO模型进行求解了。

代码附在这里：

model:
sets:
a1/1..8/;
a2/1..6/:r1,r2,P1,P2,Pr1,Pr2,Pcost1,Pcost2;
a3(a2,a1,a1,a1):x,y;
a4(a2,a1,a1);
a5(a2,a1):q1,q2,z1,z2,Needs;
a6(a1,a1):d,cost;
a7(a1,a1,a1);
endsets
@for(a3(i,n,j,k):@bin(x(i,n,j,k)));
@for(a3(i,n,j,k):@bin(y(i,n,j,k)));!0-1constraint;
@for(a5(i,j):@bin(z1(i,j)));
@for(a5(i,j):@bin(z2(i,j)));
@for(a3(i,n,j,k):x(i,n,j,k)+d(j,k)<=2);
@for(a3(i,n,j,k):y(i,n,j,k)+d(j,k)<=2);!dist constraint;
@for(a4(i,n,k)|n#lt#8:@sum(a1(j):x(i,n+1,k,j)-x(i,n,j,k))=0);
@for(a4(i,n,k)|n#lt#8:@sum(a1(j):y(i,n+1,k,j)-y(i,n,j,k))=0);!continous constraint;
@for(a5(i,n):@sum(a6(j,k):x(i,n,j,k))=1);!decision only;
@for(a5(i,n):@sum(a6(j,k):y(i,n,j,k))=1);
@for(a5(i,j):z1(i,j)<=@sum(a6(n,k):x(i,n,k,j)));!reach if;
@for(a5(i,j):z2(i,j)<=@sum(a6(n,k):y(i,n,k,j)));
@for(a5(i,j):q1(i,j)<=64*z1(i,j));!supply quantity;
@for(a5(i,j):q2(i,j)<=64*z2(i,j));!supply quantity;
@for(a5(i,j):@gin(q1(i,j)));
@for(a5(i,j):@gin(q2(i,j)));
@for(a2(i):P1(i)=Pr1(i)+r1(i));
@for(a2(i):P2(i)=Pr2(i)+r2(i));
@for(a2(i)|i#lt#6:r1(i+1)=P1(i)-@sum(a2(j):q1(i,j)));!residual calc;
@for(a2(i)|i#lt#6:r2(i+1)=P2(i)-@sum(a2(j):q2(i,j)));!residual calc;
r1(1)=0;
r2(1)=0;
@for(a5(i,j):q1(i,j)+q2(i,j)=Needs(i,j));
@for(a2(i):@sum(a1(j):x(i,1,7,j))=1);
@for(a2(i):@sum(a1(j):y(i,1,8,j))=1);
@for(a2(i):Pcost1(i)=@sum(a7(n,j,k):x(i,n,j,k)*cost(j,k)));
@for(a2(i):Pcost2(i)=@sum(a7(n,j,k):y(i,n,j,k)*cost(j,k)));
min=@sum(a2(i):Pcost1(i)+Pcost2(i));
data:
@text()=@writefor(a3(i,n,j,k)|x(i,n,j,k)#eq#1 :'供应商1第',i,'天第',n,'次从',j,'地去往',k,'地',@newline(1));
@text()=@writefor(a3(i,n,j,k)|y(i,n,j,k)#eq#1 :'供应商2第',i,'天第',n,'次从',j,'地去往',k,'地',@newline(1));
@text()=@writefor(a5(i,j)|z1(i,j)#eq#1:'供应商1第',i,'天在',j,'地提供',q1(i,j),'物资',@newline(1));
@text()=@writefor(a5(i,j)|z2(i,j)#eq#1:'供应商2第',i,'天在',j,'地提供',q2(i,j),'物资',@newline(1));
d=
1	1	1	2	2	2	1	2
1	1	2	1	2	2	1	1
1	2	1	1	2	1	2	2
2	1	1	1	2	1	2	1
2	2	2	2	1	1	2	1
2	2	1	1	1	1	2	2
1	1	2	2	2	2	1	2
2	1	2	1	1	2	2	1;
Needs=
50	40	38	23	34	22	0	0
60	44	38	23	34	23	0	0
59	43	45	19	32	34	0	0
64	44	38	23	34	23	0	0
34	43	45	29	32	34	0	0
50	23	28	23	34	22	0	0;
Pr1=200	232	234	235	190	180;
Pr2=59	56	55	67	78	45;
cost=
0	18	8	2	2	2	14	2
18	0	2	9	2	2	10	15
8	2	0	8	2	13	2	2
2	9	8	0	2	10	2	11
2	2	2	2	0	9	2	8
2	2	13	10	9	0	2	2
14	10	2	2	2	2	0	2
2	15	2	11	8	2	2	0;
enddata

有反馈说代码运行不了改了一下方便复制