The Study On Meta-heuristic Algorithms Based On Two Improved Strategies To Solve Route-related Problems

Route-related problems are hot topics in operations research since it has both theoretical and practical significance to study route-related problems.In addition,meta-heuristic algorithms can solve route-related problems effectively,and thus meta-heuristic algorithms are also research hotspots in recent years.Route-related problems can be solved more efficiently and accurately through improving traditional meta-heuristic algorithms.Firstly,this paper gave a brief introduction on the backgrounds of route-related problems and their academic significance,reviewed the research status of algorithms used to solve route-related problems at home and abroad,and summarized shortcomings of the research status.Then,based on the shortcomings,two improved strategies for meta-heuristic algorithms—strategy of selective-directed mutation and strategy of limiting search scope — were proposed through analyzing common characteristics of generalized meta-heuristic algorithms,and application scopes for the two strategies were also clarified.Then,this paper studied two representative problems—traveling salesman problem(TSP)and many-to-many Milk-run routing problem with pipeline inventory cost(MM-MRP-PIC).Two meta-heuristic algorithms —adaptive hybrid simulated annealing-tabu search algorithm with a dynamic neighborhood structure based on a circle-directed mutation(AHSATS-D-CM)and two-stage simulated annealing algorithm with limited search scope(TSSALSS)—based on the two improved strategies were proposed to verify the feasibility and effectiveness of the two strategies in solving the two problems respectively.For TSP,the benchmarks in TSPLIB were employed to test the performance of the proposed algorithm.The results obtained by the proposed algorithm were compared with those obtained by traditional meta-heuristic algorithms and other heuristic algorithms derived from literatures to verify the superiority of the proposed algorithm.For MM-MRP-PIC,this paper firstly proposed a mathematical model to describe this problem.TSSALSS was developed to solve this Model.Then numerical examples were randomly generated by a computer to test the performance of the proposed algorithm.Subsequently,the test results were compared with those obtained by widely-used algorithms in literatures to demonstrate the advantage of the proposed algorithm.Finally,two cases were studied to demonstrate the superiority of the many-to-many Milk-run transportation mode.Finally,this paper declared that the ideas of proposed strategies have the possibility of being employed to improve other meta-heuristic algorithms to solve more other route-related problems.