OPTIMAL PARTICLE SWARM FOR RELIABILITY ALLOCATION TO COMPLEX SYSTEM

Abstract

In this work, a reliability function is computed for a complex system, along with an optimal reliability distribution among the components of the system to be built. The majority of research to date has adopted exponentially increasing closed functions to relate cost and reliability, despite the fact that it is generally known that component cost is an increasing function of its reliability. In fact, these functions are sometimes ill-defined or difficult to generate, and so it often makes sense to discuss cost-reliability connections between separate datasets. We consider scenarios where each component is offered in a range of reliability levels at varying prices. The nonlinear integer program is the result of a design optimization challenge. We introduce particle swarm optimization (PSO) to calculate system reliability assignment for each system component as well as to calculate overall system reliability since each system configuration has an equivalent representation as either serial communication of parallel subsystems or parallel communication of serial subsystems. The cost of each component of the system was determined using the exponential behavior feasibility factor cost function.