梯级水库优化调度的改进ODDDP算法研究  

Study on Improved ODDDP Algorithm for Cascade Reservoirs Optimal Operation

作　　者：方国华[1] 郑旺 吴承君 颜敏 FANG Guo-hua;ZHENG Wang;WU Cheng-jun;YAN Min(College of Water Conservancy and Hydropower Engineering,Hohai University,Nanjing 210098,China;Yellow River Institute of Hydraulic Research,Zhengzhou 450003,China)

机构地区：[1]河海大学水利水电学院,江苏南京210098 [2]黄河水利委员会黄河水利科学研究院,河南郑州450003

出　　处：《水电能源科学》2024年第9期179-184,共6页Water Resources and Power

基　　金：国家自然科学基金项目(52179012)。

摘　　要：“维数灾”问题和易陷入局部最优问题是梯级水库优化调度模型求解中的热点和难点。正交离散微分动态规划算法依据正交试验原理,解决了离散微分动态规划算法在求解水库群联合优化调度问题中的维数灾问题,但并未克服其依赖初始解质量和易陷入局部最优的缺陷。对此,借鉴入侵杂草算法空间扩散机制,引入高斯随机变量,并基于三角余弦函数的周期性质,设计提出了M-IWO-ODDDP算法,从算法全局搜索能力和搜索深度两方面,提高算法全局收敛能力。采用Schaffer、Shubert两种测试函数,分析测试了M-IWOODDDP算法的收敛性和鲁棒性。并以乌江流域11级梯级水库为例开展实例应用研究,结果表明,M-IWOODDDP寻优效果显著且对初始解质量无明显依赖性,算法收敛能力强、鲁棒性好。The problems of curse of dimensionality and easily falling into local optimum are the hot and difficult points in solving the model of cascade reservoirs optimal operation.The problem of curse of dimensionality for Discrete Differential Dynamic Programming algorithm(DDDP)in solving the problem of cascade reservoirs optimal operation was solved by Orthogonal Discrete Differential Dynamic Programming algorithm(ODDDP).However,the ODDDP does not overcome the defects of relying on the initial solution quality and easily falling into local optimum.By introducing Gaussian random variables,the M-IWO-ODDDP was proposed by referencing the spatial diffusion mechanism of the invasive weed algorithm and the periodicity of the trigonometric cosine function.The global convergence of the algorithm can be improved by improving the global search ability and search depth of the ODDDP.The convergence and robustness of the MIWO-ODDDP were tested by using Schaffer and Shubert functions.The 11 cascade reservoirs in Wujiang River Basin was used as a case study.The results show that the optimization effect of M-IWO-ODDDP is significant and it not only has no obvious dependence on the quality of the initial solution,but also has good robustness.

关 键 词：水库优化调度 离散微分动态规划 正交试验 高斯分布 收敛性分析 

分 类 号：TV697.11[水利工程—水利水电工程]