Genetic algorithms with heuristics rules to solve multi source single product flexible multistage logistics network problems

Abstract

To be successful in today’s active business competition, enterprises need to design and build effective flexible logistics networks. Since the flexible multistage logistic network (fMLN) problem is NP-hard, many researchers have attempted to use Meta-heuristics methods such as Genetic Algorithms (GAs) to solve the problem. Previous research works using GA for fMLN only considered the problem as a single source, at least in the last network layer between retailer and customer. In real world, however, the problem is one of multi-source logistics network. In this research, the genetic algorithms with penalty method, called P-GA, is used to solve the multi source single product fMLN problem. It is shown however that the P-GA requires unreasonable elapsed time to obtain an acceptable solution. To speed up the algorithm, the research proceeds with the developments of heuristics rules for initialization, crossover and mutation within P-GA and named as HR-GA. This research shows the proposed HR-GA has substantially reduced the elapsed time to obtain better acceptable solution.