Parameter Tuning-based Metaheuristics In Dynamic Network Flow With Flexible Arc Outages

Maximum Total Flow with Flexible Arc Outages(MaxTFFAO)is a class of optimization problems arising from the daily equipment maintenance operations of the port.In this problem,the whole process from mining coal from mine,transporting through railway,to the frontier of the terminal is abstracted into a network.When the equipment in the network is maintained,the state of the arc in the network changes,and the flow in the network changes accordingly.When a certain number of equipment is being maintained,it is necessary to schedule different maintenance operations so as to maximize the output of coal in the entire network.At the same time,MaxTFFAO is also a combinatorial optimization problem that combines network flow problems and scheduling problems,and has strong NP-hard characteristics.A large number of exact algorithms and metaheuristics are used in resolving MaxTFFAO problem.Exact algorithms such as decomposition algorithms,mixed integer linear programming,etc.,can obtain the optimal solution to the problem.But when the problem size becomes large,the solution cannot be obtained in a limited time and will result in a higher cost.During the process of solving the problem by using the metaheuristic algorithms,although the solution of the problem can be obtained within a limited time,due to the irrationality of parameter setting in the algorithm,the final solution of the problem is not necessarily a better solution to the problem.Because the key factor of metaheuristics to play its algorithm performance depends largely on the parameter setting of the algorithm,it is necessary to study the parameter setting problem while solving the MaxTFFAO problem with metaheuristic algorithm.This paper uses the parameter tuning tool irace to tune the parameters in the metaheuristic algorithms,which can improve the performance of the metaheuristic being applied to solve the MaxTFFAO problem.The main research contents and innovations of this paper are as follows:(1)Firstly,this paper makes a deep introduction about the background and research significance of the research problems,introduces the MaxTFFAO problem and the metaheuristics commonly used to solve optimization problems,and make a survey about parameter tuning methods of metaheuristic algorithms with parameters at home and abroad.The algorithm theory used in the research process of this subject and the latest parameter tuning tools are also introduced.(2)Taking MaxTFFAO problem as the research goal,the parameters of random greedy adaptive search algorithm and hybrid tabu search algorithm which have solved this problem are tuned.The optimal parameter settings of the two algorithms are obtained and verified through experiments.On the basis of the optimized parameter setting,the performance of the hybrid tabu search algorithm for solving the problem can be improved after making changes from three aspects.(3)A hybrid metaheuristic algorithm combining tabu search and iterative local search algorithm is designed.The optimal parameter setting of the hybrid metaheuristic algorithm is obtained after being tuned by irace.The parameter optimized metaheuristic algorithm solves the problem with a better solution.All tuning and algorithm improvement effects were verified through comparative experiments.This paper innovatively applies irace to the MaxTFFAO problem solving process,which makes the hybrid metaheuristic algorithm after being tuned be more suitable with the background of the problem and the implementation of the algorithm,and thus improves the performance of the algorithm in practical application.