A bi-objective simulation-based optimization algorithm for redundancy allocation problem in series-parallel systems. (1st July 2021)

• Record Type: Journal Article
• Authors: Chambari, Amirhossein; Azimi, Parham; Najafi, Amir Abbas
• Is Part Of: Expert systems with applications. Volume 173(2021)
• Publication Date: 2021-07-01
• Subjects: Simulation-based optimization -- Redundancy allocation problem -- NSGA-II -- Bi-objective -- Redundancy strategies; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2021.114745
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: A redundancy allocation problem (RAP) with a series-parallel structure is considered. Proposing a bi-objective simulation-based optimization algorithm for solving RAP. A customized NSGA-II is coupled with the simulation method. Each subsystem can be active, cold-standby, mixed, and k- mixed redundancy strategy. Statistical analysis is employed to test the performance of optimization methods. Abstract: The present study proposes a bi-objective simulation-based optimization model applicable to the redundancy allocation problem (RAP) with heterogeneous components for the objective functions of system reliability maximization and system cost minimization. Proposed RAP identifies the optimal component types, the redundancy level, and the redundancy strategy, comprising active, cold-standby, mixed, or K-mixed configurations, with imperfect switching. Based on the stochastic nature and NP-hard complexity of the problem, except for the active redundancy strategy, there is no analytical closed-form method for accurately assessing system reliability. Hence, earlier studies carried out system reliability optimization by single-stage stochastic techniques that estimate the lower Lagrangian function bound. This limitation adds to the design cost and hinders greater system reliability. Generally, one cannot analytically evaluate system reliability. The present study employs simulation sampling in order to make efficient and unbiased reliability estimates. 4Dscript interpreting programming is utilized to design the computerized simulation model. Since RAP has a combinatorial nature, the present study exploits the controlled elitist non-dominated sorting genetic algorithm (NSGA-II) to obtain Pareto-optimal fronts with properly-distributed optimal points. Various benchmark solutions of the literature are investigated to validate the developed model and evaluate the proposed NSGA-II method efficiency. The findings revealed satisfactory performance in system reliability enhancement and total system cost reduction compared to the earlier methods.