水库群发电优化调度遗传算法整体改进策略研究  

Research on overall improved genetic algorithm applied in optimal generation dispatching of multi-reservoir system

作　　者：郑姣[1] 杨侃[1] 倪福全[2] 刘国帅[1] 

机构地区：[1]河海大学水文水资源学院,江苏南京210098 [2]四川农业大学信息与工程技术学院,四川雅安625014

出　　处：《水利学报》2013年第2期205-211,共7页Journal of Hydraulic Engineering

基　　金：国家重点基础研究发展规划973项目(2012CB417006);国家科技支撑计划项目(2009BAC56B03);江苏高校优势学科建设工程资金资助

摘　　要：针对水库群发电优化调度的高维性和复杂非线性,提出了一种收敛性全面改善的改进自适应遗传算法。改进算法采用了整体改进策略:初始种群生成方面,运用加入混沌优化的初始种群解空间生成法,改善初始种群质量;选择运算方面,采用对适应度函数进行非线性转换的三角函数选择算子,保持种群多样性;交叉和变异运算方面,采用随个体优劣和种群分散程度自适应调整的交叉概率和变异概率,提高算法收敛性。以典型入库流量下三峡梯级水库发电优化调度为实例,对比了上述整体改进策略和传统改进策略,结果表明:整体改进策略和传统改进策略相比,在克服遗传算法早熟和提高算法收敛性能方面有一定的优势,改进自适应遗传算法适用于求解水库群发电优化调度问题。For the high dimension and complex nonlinear characteristic of multi-reservoir system generation optimization, a comprehensive convergence improved adaptive genetic algorithm （AGA） is proposed. In this algorithm, initial population solution space generation method with chaos optimization is used to improve the quality firstly. Secondly, trigonometric selective operator, which does nonlinear transformation on fit- ness, is adopted to maintain the diversity of population. Finally, adaptive parameters of crossover and muta- tion, depending on the fitness value of individual and the dispersion degree of population, are adopted to improve the convergence speed of GA. Furthermore, the performance of these proposed improvement strate- gies are checked against with the historical improvement strategies by optimal operation of the Three Gorges cascade reservoirs premised on historical hourly inflows. The results show that the proposed improvement strategies have advantages in overcoming GA premature and improving convergence performance, and this improved AGA is feasible and effective for optimal generation dispatching of multi-reservoir system.

关 键 词：水库群发电优化调度 混沌优化 三角函数选择算子 参数自适应 

分 类 号：TV697.1[水利工程—水利水电工程]