梯级水电站群优化调度多目标量子粒子群算法  

Multi-objective quantum-behaved particle swarm optimization for operation of cascade hydropower stations

作　　者：牛文静[1] 冯仲恺[1] 程春田[1] 

机构地区：[1]大连理工大学水电与水信息研究所,辽宁大连116024

出　　处：《水力发电学报》2017年第5期47-57,共11页Journal of Hydroelectric Engineering

基　　金：国家自然科学基金(91547201;51210014);国家重点基础研究发展计划(973计划)资助项目(2013CB035906)

摘　　要：为科学求解梯级水电站群多目标优化调度模型,提出一种基于量子行为进化机制的多目标量子粒子群算法(MOQPSO)。该方法以标准量子粒子群算法(QPSO)为基础,引入外部档案集合存储非劣粒子,利用个体支配关系实现档案集合的动态更新维护;依据个体领导能力优劣选择粒子历史最优位置与种群全局最优位置,维持搜索过程中个体进化方向的多样性;采用混沌变异算子对个体进行局部扰动,提升算法的全局收敛性能。乌江流域模拟调度结果表明,所提方法具有良好的收敛速度与寻优能力,可快速获得兼顾梯级水电系统经济性与可靠性要求的Pareto解集,能够为工程人员提供科学的决策依据。This paper describes a method of multi-objective quantum-behaved particle swarm optimization （MOQPSO） based on quantum evolutionary mechanism for multi-objective operation of cascade hydropower stations. On the basis of the quantum-behaved particle swarm optimization （QPSO）, this new method adopts external archive collection to store non-dominated particles and implements maintenance and dynamic update of the collection using individual dominance relations. It uses individual leadership to choose the previous best position of the whole particle population and the previous best position of each particle so that diversity in individual evolution directions can be maintained during the search. A chaos mutation operator can be added to the method to further enhance its local search capability and global convergence performance. The method has been applied in the operation of hydropower stations in the Wu River basin. The results indicate that MOQPSO could generate a Pareto solution set that combines the considerations of reliability and benefits and thus it would lay a theoretical basis for decision making.

关 键 词：梯级水电站群 优化调度 多目标优化 量子粒子群算法 混沌变异 外部档案集合 

分 类 号：TV697.1[水利工程—水利水电工程]