基于混合量子粒子群算法的梯级水电站群调度  

Hybrid quantum-behaved particle swarm optimization for operation of cascade hydropower plants

作　　者：夏燕 冯仲恺[1] 牛文静 覃晖[1] 蒋志强[1] 周建中[1] XIA Yan;FENG Zhongkai;NIU Wenjing;Q;JIANG Zhiqiang;ZHOU Jianzhong(School of Hydropower and Information Engineering,Huazhong University of Science and Technology,Wuhan 430074;Bureau of Hydrology,ChangJiang Water Resources Commission,Wuhan 430010)

机构地区：[1]华中科技大学水电与数字化工程学院,武汉430074 [2]长江水利委员会水文局,武汉430010

出　　处：《水力发电学报》2018年第11期24-35,共12页Journal of Hydroelectric Engineering

基　　金：国家重点研发计划(2017YFC0405900);国家自然科学基金(51709119);湖北省自然科学基金面上项目(2018CFB573)

摘　　要：针对标准量子粒子群(QPSO)算法在求解复杂梯级水电站群联合调度问题时存在的早熟收敛、停滞寻优等不足,提出了一种耦合两重改进策略优势的混合量子粒子群(HQPSO)算法:首先对个体极值按照一定的概率进行变异搜索操作,以增加个体多样性、强化种群全局开采能力;而后建立外部档案集合来存储进化过程中的部分精英个体,利用基于动态概率辨识机制的单纯形算子指导外部档案集中的个体开展邻域寻优,以提高算法搜索能力、避免陷入局部最优。乌江流域实践结果表明:HQPSO算法的收敛速度与全局搜索能力得到增强,有效克服了QPSO的缺陷与不足,具有一定的工程实际应用价值。In solving the joint scheduling problem of cascade hydropower stations, the standard quantumbehaved particle swarm optimization（QPSO） algorithm suffers from premature convergence and local trapping, among other shortcomings. This paper presents a hybrid QPSO（HQPSO） that combines the advantages of the two-fold improvement strategy. This new method first does mutation search for individual extremes at a given probability to increase the diversity of individuals and enhance the global exploiting capability of the population. Then, it establishes an external archive set to conserve certain particles found in the evolutionary process. Finally, it uses the Nelder-Mead operator for dynamic probability identification to help particles searching in the neighborhood, improving its searching capability and avoiding falling into a local optimum. Application to the Wu River shows that the HQPSO is faster in convergence and global searching and practically applicable, avoiding the shortcomings of QPSO.

关 键 词：梯级水电站群 优化调度 量子粒子群算法 单纯形算子 变异搜索 

分 类 号：TV697.1[水利工程—水利水电工程] TV213.9