Multi-deme parallel genetic algorithm in reliability analysis of composite power systems

Intelligent search based techniques such as genetic algorithm (GA) have been proposed to deal with reliability evaluation of complex power systems recently. In this type of methods, the guided search is carried out on a population scale trying to find all the dominant failure states, based on which different reliability indices can be calculated accordingly. However, the process may be time-consuming when power flow analysis is involved in deciding the status of a system state in complex power systems such as composite systems. To speed up the computing process, parallel implementation of GA is proposed in this study by using multi-deme based search, where multiple subpopulations are distributed in different processors. In this way, simultaneous search is achieved through parallel implementation. An IEEE reliability test system is used for simulation studies. It turns out that the proposed parallel method is effective in increasing the computing efficiency of GA when it is used for reliability evaluation of composite power systems.