RELIABILITY OPTIMIZATION OF A SERIES-PARALLEL K-OUT-OF-N SYSTEM WITH FAILURE RATE DEPENDS ON WORKING COMPONENTS OF SYSTEM

This paper presents a mathematical model for a redundancy allocation problem (RAP) with k-out-of-n subsystems and failure rate depends on working components of system. It means that failure rate of components increases when a component fails. The subsystems may use either active or cold-standby redundancy strategies which considered as a decision variable. Thus, the proposed model and solution methods are to select the best redundancy strategy among active or cold-standby, component type, and levels of redundancy for each subsystem. The objective function is to maximize the system reliability under cost and weight constraints. Since RAP belongs to Np-hard problems, four meta-heuristic algorithms named genetic algorithm, Memetic algorithm, simulated annealing and particle swarm optimization are proposed. The results shown that the MA is better than other algorithms. Finally, in order to determine whether there is any significant difference between the results of four algorithms or not, a statistical test is applied