A robust loss function approach for a multi-objective redundancy allocation problem

Abstract

A redundancy allocation problem consists of the selection of a number of redundancies to be allocated in each subsystem of a series–parallel configuration to maximize system reliability, while minimizing overall system cost. Most existing studies either ignore system cost or assume deterministic component reliabilities. However, the selection of a network system design requires the estimation of its reliability and cost. Therefore, the uncertainty associated with such estimates must be considered in the decision process. In this study, an efficient method based on the loss function approach is proposed to solve the redundancy allocation problem. This approach not only optimizes the overall system reliability and cost estimates simultaneously, but also considers their associated variance–covariance structure. Experimental results obtained using a standard example show the effectiveness of the proposed method and its superior performance when compared to existing approaches.