水库优化调度遗传算法参数的灵敏性分析  

Sensitive Analysis on the Parameters of Genetic Algorithm Applied in Optimal Operation of Reservoir

作　　者：钟平安[1] 唐林[1,2] 

机构地区：[1]河海大学水文水资源学院,江苏南京210098 [2]黄河勘测规划设计有限公司国际工程设计院,河南郑州450003

出　　处：《水力发电》2010年第11期13-16,共4页Water Power

基　　金：国家科技支撑计划(2008BAB29B09);国家科技支撑计划(2009BAC56B03)

摘　　要：选取选择算子、交叉算子、变异算子、初始种群数、交叉率、变异率和终止条件等7个主要因素,以搜索能力、时效性、收敛性、稳定性为算法性能评价指标,运用正交试验方法,进行遗传算法参数的多指标、多因素、多水平的模拟试验。采用极差分析方法,确定因素对遗传算法性能影响,得到结论,遗传算法参数对各性能评价指标影响比模式的影响大;初始种群数、优值不变终止代数和变异率是影响算法综合性能的主要敏感参数。实例表明,采用确定式采样选择、均匀交叉和均匀变异模式遗传算法的综合性能较优。The multi-index,multi-factor and multi-level simulation tests for the parameters of Genetic Algorithm(GA) were carried out by orthogonal test method after selecting selection operator,crossover operator,mutation operator,the initial population size,crossover rate,mutation rate and termination conditions as major factors and search capabilities,timeliness,convergence and stability as performance evaluation indexes.The range analysis method was used to determine the impact of factors on the performance of GA.The results show that the GA's performance is more sensitive in parameters than operator selection,initial population,termination condition and mutation ratio are the most sensitive factors,and the fixed sampling selection,uniform cross and uniform mutation operators mode has best overall performance.

关 键 词：水库优化调度 遗传算法 算法参数 灵敏性分析 Optimal Operation Genetic Algorithm 性能评价指标 综合性能 种群数 正交试验方法 变异率 主要因素 选择算子 性能影响 搜索能力 变异模式 模拟试验 敏感参数 均匀交叉 交叉算子 

分 类 号：TV73[水利工程—水利水电工程] TM612[电气工程—电力系统及自动化]