A variable neighbourhood search algorithm with novel archive update strategies for redundancy allocation problems

Abstract

This article proposes a variable neighbourhood search algorithm with novel archive update strategies (VNS_AU) to solve redundancy allocation problems with the series–parallel system configuration. When solving single-objective optimization problems with different constraint levels, traditional approaches can only deal with one constraint level at a time. This usually leads to the waste of search information and extra computational effort. Different from traditional VNS algorithms, the proposed VNS_AU algorithm employs an archive originally derived from the idea of Pareto optimal set in multi-objective optimization to keep track of the best solutions on different constraint levels. The performance of the proposed VNS_AU algorithm is tested on a well-known benchmark suite with 33 instances. Computational results show that VNS_AU outperforms seven other competing methods in the literature. The number of evaluations and the central processing unit (CPU) time needed by VNS_AU also surpass other methods.

Keywords:

• redundancy allocation problem
• series–parallel system
• variable neighbourhood search
• metaheuristic