Multi-objective Multi-Factorial Evolutionary Algorithm for Fuzzy Reliability Redundancy Allocation

Abstract

The Reliability Redundancy Allocation Problem (RRAP) seeks to optimize system reliability by enhancing component reliability and reducing redundancy under constraints such as cost, weight, and volume. Given dynamic manufacturing uncertainties, fuzzy modeling with the meta-heuristic-based solution approach is often preferred. While evolutionary multi-task optimization (EMTO) has been used to solve two single-objective RRAPs simultaneously, the recent multi-objective RRAPs (MO-RRAP) formulation incorporate both reliability and cost or weight for greater real-world applicability. Simultaneously optimizing multiple MO-RRAPs with similar system structures enables effective knowledge transfer, such as sharing genetic material between tasks leading to faster convergence, improved solutions, and reduced memory consumption. In this paper, we propose a multi-objective multi-factorial evolutionary algorithm (MO-MFEA) to simultaneously solve two fuzzy multi-objective RRAPs, one for a series system and one for a complex bridge system, across two test sets (reliability with cost and reliability with weight). Our results indicate that proposed MO-MFEA based approach provides superior solution quality and reduced computation time compared to the non-dominated sorting genetic algorithm (NSGA-II) when solving each problem independently.