基于RQGA和非支配排序的多目标混沌量子遗传算法

为了提高多目标优化算法的收敛性、分布性和减少算法的计算代价,借鉴实数编码遗传算法和多目标优化理论,构建一种多目标混沌量子遗传算法.在分析量子位概率的混沌特性、量子态干涉特性和量子位实数编码的基础上,采用量子位概率交叉和混沌变异的方式进化种群,以提高寻优能力和收敛速度,利用非支配排序、精英保留和分层聚类等多目标优化策略保持种群多样性的同时,保证进化向Pareto全局最优解集方向进行.通过混合算法性能对比测试验证了多算法集成的有效性,并分析关键参数对算法性能的影响.电力系统多目标无功优化的仿真结果验证了该算法的有效性和可行性.

Multi-objective chaotic quantum genetic algorithm based on RQGA and non-dominated sorting

In order to improve the convergence and distribution together with less computation cost of multi-objective algorithm,Multi-objective Chaotic Quantum Genetic Algorithm(MCQGA) was proposed,referencing to Real-coded Quantum Genetic Algorithm(RQGA) and multi-objective optimization theory.Based on analysis of chaotic feature of quantum bits probability,interference feature of quantum states and real-coded quantum bits,quantum bits probability crossover and chaos mutation were adopted to improve search efficiency and convergence rate.The population diversity and evolution direction of Pareto global optimal solution were ensured by multi-objective optimization strategies such as non-dominated sorting,elitist preserve and hierarchical clustering.The effectiveness of multi-algorithm integration and the influence of critical parameters on hybrid algorithm are analyzed through performance test and contrast.The conclusion drawn from multi-objective reactive power optimization confirms its validity and feasibility.