Reliability Optimization of Series-Parallel Systems Using Asynchronous Heterogeneous Hierarchical Parallel Genetic Algorithm
ZENG Wen-hua1, Yiannis Papadopoulos 2, David Parker 2 (1.School of Software, Xiamen University, Xiamen 361005, China; 2.Department of Computer Science, University of Hull, UK)
Redundancy allocation of series-parallel systems is a multiobjective optimization problem. In this paper, a new parallel genetic algorithm, named asynchronous heterogeneous hierarchical parallel genetic algorithm(AHHPGA), is proposed to solve reliability optimization of series-parallel systems. The AHHPGA is a hierarchical model, which uses coarse-grained model as upper layer, and fine-grained model as lower layer. The new parallel genetic algorithm is a heterogeneous model, different exploration/exploitation degree and different topology are assigned to each subpopulation. The migration model of AHHPGA is asynchronous, which includes asynchronous reception and asynchronous emigration. A new model-based reliability evaluation technique, which is called HiP-HOPS, is used to overcome the limitations imposed by reliability block diagrams. The simulation result of well known Fyffe’s problem is better than traditional genetic algorithms.