基于分布式并行遗传算法的电力系统无功优化

针对传统遗传算法寻优质量差、计算时间长的问题,提出了基于计算机集群的一种新的分布式并行遗传算法解决电力系统无功优化问题.采用遗传模拟退火算法和分布式并行计算MPI(Message Passing Interface)技术,实现多进程的分布式集群计算.该算法通过个体迁移策略来协调优化各个子种群,使用计算效率来判断计算负载状态,采用动态种群来进行负载平衡.通过运用标准测试算例IEEE14节点和一个实际电力系统的无功优化计算,结果表明这种算法具有很高的稳定性,有较好的并行效率,适合求解大规模电力系统的无功优化问题.

Optimal research of distributed parallel genetic algorithm for reactive power optimization

A distributed parallel genetic algorithm based on personal computer(PC) cluster was proposed to solve reactive power optimization,aiming at the disadvantage of traditional genetic algorithm,such as the bad searching quality and long computation time.It adopts the improved genetic simulated annealing algorithm and distributed parallel technique message passing interface(MPI),to implement the distributed computing on PC cluster.The algorithm uses the individual migration strategy to collaboratively optimize every process.The dynamic populations are adopted to balance the computing load.An IEEE 14 test system and a practical power system are tested.The results reveal that the algorithm has a good stable searching capacity and good parallel efficiency.The proposed method can be used to solve the reactive power optimization of large-scale power system.