提出了一种求解多车场车辆路径问题(Multi—depot Vehicle Routing Problem,MDVRP)的分散搜索算法(Scatter Search,SS)。该算法基于顾客信息进行编码,采用扫描算法(Sweep algorithm)与最优划分过程产生待选解集;通过启发式规则选择不同解包含的弧来生成新解;并由2-交换、2-交换’及最坏移出-预测插入启发式三种局部搜索策略构成迭代下降算法对解进行改进。通过采用文献中随机生成的小规模数据及MDVRP的Benchmark问题对算法进行测试,验证了算法的有效性。
A Scatter search (SS) is proposed to solve the Multi-depot Vehicle Routing Problem (MDVRP). The code of the solution is based on the customers. Sweep algorithm and optimal splitting procedure are used to generate the trial solutions. New solutions are generated by heuristic choosing the arcs of different solutions. Three local search strategies, 2-OPT, 2-OPT' and Worst removal-Regret heuristic, are integrated into an iterative descending algorithm to improve the solutions. We test the random data with small scale from the literature and benchmark problems of MDVRP to illustrate that the SS algorithm is effective.