C++调用cplex求解VRPTW模型

VRPTW数学模型

VRPTW——带时间窗约束的车辆路径规划问题。

（1）目标函数为最小化成本或者时间。
（2）约束1：每个顾客只能被访问一次；约束2：车辆的载重限制。
（3）约束3-5：车辆从depot(0)出发回到depot(n+1)。
（4）约束6：车辆数量限制。
（5）约束7-8：时间窗约束。

C++调用cplex求解

多维变量的创建

IloNumVarArray类，只能定义一维的数组，而这里的模型中需要用到三维和二维的变量。Cplex C++ 接口提供了一个lloArray类模版，我们可以利用此类模版定义多维数组。

比如二维变量数组，在头文件中做如下定义：

typedef IloArray<IloNumVarArray> NumVarMatrix; 

接下来，便可以用NumVarMatrix定义二维变量。

NumVarMatrix s(env, v_num);
for (IloInt k = 0; k < v_num; k++)
{
	s[k] = IloNumVarArray(env, c_num + 2);
	for (IloInt i = 0; i < c_num + 2; i++)
	{
		s[k][i] = IloNumVar(env, ILOINT);
	}
}

三维数组也是同样的方法：

typedef IloArray<IloNumVarArray> NumVarMatrix; 
typedef IloArray<NumVarMatrix> NumVar3Matrix;

接下来，便可以用NumVar3Matrix定义三维变量。

C++调用cplex求解VRPTW完整代码

#include <ilcplex/ilocplex.h>
#include <stdio.h>
#include <cstdlib>
#include <iostream>
typedef IloArray<IloIntVarArray> IntVarMatrix;  
typedef IloArray<IloNumVarArray> NumVarMatrix;  
typedef IloArray<NumVarMatrix> NumVar3Matrix; 
typedef IloArray<IloIntArray> IntMatrix;  
typedef IloArray<IloNumArray> NumMatrix;   
typedef IloArray<NumMatrix> Num3Matrix;    

ILOSTLBEGIN //#define ILOSTLBEGIN using namespace std;

int main(int argc, char **argv)   
{
	IloEnv env;

	IloModel model(env);

	const int M = 10000;
	const int v_num = 3, capacity = 500, c_num = 25;
	float xcord[c_num + 2], ycord[c_num + 2], t_service[c_num + 2];
	int demand[c_num + 2], lbtw[c_num + 2], ubtw[c_num + 2];

	//从txt文件中读取客户信息_____x,ycord，demand，tw，time_service
	ifstream inn;
	inn.open("vrptw.txt");
	if (inn)
	{
		for (int j = 0; j < c_num + 2; j++) {
			inn >> xcord[j], inn >> ycord[j], inn >> demand[j], inn >> lbtw[j], inn >> ubtw[j], inn >> t_service[j];
		}
	}
	inn.close();

	float distance[c_num + 2][c_num + 2], time[c_num + 2][c_num + 2];

	//数据初始化，计算客户之间的距离以及访问时长
	for (int i = 0; i < c_num + 2; i++) {
		for (int j = 0; j < c_num + 2; j++) {
			if (i == j) {
				distance[i][j] = 0;
				time[i][j] = 0;
			}
			else{
				distance[i][j] = floor(sqrt((xcord[i] - xcord[j]) * (xcord[i] - xcord[j]) + (ycord[i] - ycord[j]) * (ycord[i] - ycord[j]))*10)/10;
				time[i][j] = distance[i][j] + t_service[i];
			}	
		}
	}

	//三维决策变量x[k][i][j]，1 if a vehicle drives directly from customer i to customer j
	NumVar3Matrix x(env, v_num);
	for (IloInt k = 0; k < v_num; k++)
	{
		x[k] = NumVarMatrix(env, c_num + 2);
		for (IloInt i = 0; i < c_num + 2; i++)
		{
			x[k][i] = IloNumVarArray(env, c_num + 2);
			for (IloInt j = 0; j < c_num + 2; j++) {
				x[k][i][j] = IloNumVar(env, 0, 1, ILOBOOL);
			}
		}
	}

	//二维决策变量s[k][i],the time a vehicle starts to service a customer
	NumVarMatrix s(env, v_num);
	for (IloInt k = 0; k < v_num; k++)
	{
		s[k] = IloNumVarArray(env, c_num + 2);
		for (IloInt i = 0; i < c_num + 2; i++)
		{
			s[k][i] = IloNumVar(env, ILOINT);
		}
	}

	//约束1 对角元素为0
	for (int k = 0; k < v_num; k++) {
		for (int i = 0; i < c_num + 2; i++) {
			model.add(x[k][i][i] == 0);
		}
	}

	//约束2 Each customer is visited exactly once
	IloExpr sum_xkij(env);
	for (int i = 1; i < c_num + 1; i++) {
		sum_xkij.clear();
		for (int k = 0; k < v_num; k++) {
			for (int j = 0; j < c_num+2; j++) {
				sum_xkij += (x[k][i][j]);
			}
		}
		model.add(sum_xkij == 1);
	}

	//约束3 A vehicle can only be loaded up to it's capacity
	IloExpr sum_demand(env);
	for (int k = 0; k < v_num; k++) {
		sum_demand.clear();
		for (int i = 1; i < c_num+1; i++) {
			for (int j = 0; j < c_num + 2; j++) {
				sum_demand += (demand[i] * x[k][i][j]);
			}
		}
		model.add(sum_demand <= capacity);
	}

	//约束4 Each vehicle must leave the depot 0
	IloExpr sum_xk0j(env);
	for (int k = 0; k < v_num; k++) {
		sum_xk0j.clear();
		for (int j = 0; j < c_num + 2; j++) {
			sum_xk0j += (x[k][0][j]);
		}
		model.add(sum_xk0j == 1);
	}

	//约束5 After a vehicle arrives at a customer it has to leave for another destination
	IloExpr sumxi(env), sumxj(env);
	for (int h = 1; h < c_num + 1; h++) {
		for (int k = 0; k < v_num; k++) {
			sumxi.clear(); sumxj.clear();
			for (int i = 0; i < c_num + 2; i++) {
				sumxi += (x[k][i][h]);
				sumxj += (x[k][h][i]);
			}
			model.add(sumxi - sumxj == 0);
		}
	}

	//约束6 All vehicles must arrive at the depot n + 1
	IloExpr sumxki(env);
	for (int k = 0; k < v_num; k++) {
		sumxki.clear();
		for (int i = 0; i < c_num + 2; i++) {
			sumxki += (x[k][i][c_num + 1]);
		}
		model.add(sumxki == 1);
	}

	//约束7 The time windows are observed
	for (int i = 0; i < c_num + 2; i++) {
		for (int k = 0; k < v_num; k++) {
			model.add(s[k][i] <= ubtw[i]);
			model.add(s[k][i] >= lbtw[i]);
		}
	}

	//约束8 From depot departs a number of vehicles equal to or smaller than v
	IloExpr s1(env);
	for (int k = 0; k < v_num; k++) {
		for (int j = 0; j < c_num + 2; j++) {
			s1 += (x[k][0][j]);
		}
	}
	model.add(s1 <= v_num);

	//约束9 Vehicle departure time from a customer and its immediate successor
	for (int i = 0; i < c_num + 2; i++) {
		for (int j = 0; j < c_num + 2; j++) {
			for (int k = 0; k < v_num; k++) {
				model.add(s[k][i] + time[i][j] - M * (1 - x[k][i][j]) <= s[k][j]);
			}
		}
	}

	//目标函数
	IloExpr obj(env);
	for (int k = 0; k < v_num; k++) {
		for (int i = 0; i < c_num + 2; i++) {
			for (int j = 0; j < c_num + 2; j++) {
				obj += distance[i][j] * x[k][i][j];
			}
		}
	}
	model.add(IloMinimize(env, obj));

	//创建求解对象
	IloCplex cplex(model);

	//处理异常
	cout << "begin to solve." << endl;

	try
	{
		if (!cplex.solve())
		{
			cout << "求解失败" << endl;
			if ((cplex.getStatus() == IloAlgorithm::Infeasible) ||
				(cplex.getStatus() == IloAlgorithm::InfeasibleOrUnbounded))
			{
				cout << endl << "No solution - starting Conflict refinement" << endl;
			}

			env.error() << "Failed to optimize LP." << endl;
			throw(-1);
		}
		cout << "异常处理完毕" << endl;
		
		env.out() << "Solution status = " << cplex.getStatus() << endl;
		env.out() << "Solution value = " << cplex.getObjValue() << endl;
	}

	catch (IloException& e)
	{
		cerr << "Concert exception caught: " << e << endl;
		//save results
	}
	catch (...)
	{
		cerr << "Unknown exception caught" << endl;
	}

	//输出VRP路线
	cout << "Solutions: " << endl;
	for (int k = 0; k < v_num; k++) 
	{
		cout << "vehicle " << k + 1 << " :" << endl;
		for (int i = 0; i <= c_num + 1; i++) 
		{
			for (int j = 0; j <= c_num + 1; j++) 
			{
				IloInt c = cplex.getValue(x[k][i][j]);
				if (c == 1) {
					if (i == 0) cout << i << " --> " << j;
					else if (i != 0) cout << " --> " << j;
					i = j - 1;
					break;
				}
				if (j == c_num + 1) { cout << endl; break; }
			}
		}
	}

	env.end();
	//return 0;
	system("PAUSE");
	return EXIT_SUCCESS;
}

vrptw.txt 输入格式

小算例：25个客户，3辆车，载重上限500。

40	50	0	0	1236	0
45	68	10	912	967	90
45	70	30	825	870	90
42	66	10	65	146	90
42	68	10	727	782	90
42	65	10	15	67	90
40	69	20	621	702	90
40	66	20	170	225	90
38	68	20	255	324	90
38	70	10	534	605	90
35	66	10	357	410	90
35	69	10	448	505	90
25	85	20	652	721	90
22	75	30	30	92	90
22	85	10	567	620	90
20	80	40	384	429	90
20	85	40	475	528	90
18	75	20	99	148	90
15	75	20	179	254	90
15	80	10	278	345	90
30	50	10	10	73	90
30	52	20	914	965	90
28	52	20	812	883	90
28	55	10	732	777	90
25	50	10	65	144	90
25	52	40	169	224	90
40	50	0	0	1236	0

该算例求解结果：

结语

多维数组创建参照的这篇文章：CPLEX with C++求解多维问题
模型是参照github上的一个cplex opl代码给出的，这里只是将其中的模型照搬到C++里。cplex-vrptw-implementation
学习总结：初步学会C++调用Cplex变量创建、添加约束与目标、从txt简单地读取数据。