水库群调度高维优化问题约束处理方法研究  

Research on Constraint Processing Method of High-dimensional Optimization Operation Problem of Cascade Reservoirs

作　　者：何中政 李树良 黄伟[1,2] 闫峰 付吉斯 熊斌 HE Zhongzheng;LI Shuliang;HUANG Wei;YAN Feng;FU Jisi;XIONG bin(School of Infrastructure Engineering,Nanchang University,Nanchang 330031,China;Key Laboratory of Poyang Lake Environment and Resources Utilization,Ministry of Education,Nanchang University,Nanchang 330031,China)

机构地区：[1]南昌大学工程建设学院,江西南昌330031 [2]南昌大学鄱阳湖环境与资源利用教育部重点实验室,江西南昌330031

出　　处：《工程科学与技术》2024年第6期230-238,共9页Advanced Engineering Sciences

基　　金：国家自然科学基金项目(52209024);江西省水科院开放基金项目(2022SKSH01);江西省自然科学基金项目(20224BAB204075,20212BAB214065);江西省水利厅科技项目(202324YBKT24,202223YBKT43)。

摘　　要：随着水库群优化调度的调度规模的增加和调度时间步长的精细化,水库群调度高维优化问题的决策变量维度逐渐增加到数百数千维。在具有高维决策变量的梯级水库优化调度中,往往需要考虑多重复杂约束。现有传统优化方法在处理此类问题时难以找到有效可行解;而智能优化算法的多维度联动随机搜索,寻优空间大但寻优效率低。为此,本文提出了一种结合罚函数的嵌套DPSA–POA和智能算法的约束处理方法,将罚函数与DPSA–POA和智能算法嵌套,一方面可克服DPSA–POA收敛结果容易受初值影响和寻优空间狭窄的缺陷,另一方面可提升智能算法随机搜索策略的寻优效率。随后,本文以决策变量高达2 196维的赣江中游梯级水库群防洪优化调度问题为例开展分析,相关分析结果表明:1)结合罚函数嵌套DPSA–POA智能算法的3种约束处理方式,在不同来水情形下均能得到高维优化问题可行解;2)3种约束处理方式中,嵌套优化得到可行解后只进行DE优化的方式2收敛精度最高,计算时间约10 h;嵌套优化得到可行解后只进行DPSA–POA优化的方式3收敛精度次之,计算时间约1~3 h;3)现有可行解优先策略(SF)、随机排序策略(SR)、罚函数策略(PF)和ε–松弛约束策略(EC)配合现代智能算法,无法在不同来水情形下稳定收敛到可行解,且可行解的收敛精度相比本文提出的方法有明显差距。综上,本文提出的高维优化问题约束处理方法可有效解决水库群调度高维优化问题。With the expansion of operational scales and the refinement of time steps in optimizing cascade reservoirs,the dimensionality of decision variables in such problems can range from hundreds to thousands.In the operational optimization of cascaded reservoirs with high-dimensional decision variables,it is often essential to consider multiple complex constraints.Traditional optimization methods struggle to effectively identify feasible regions when addressing these challenges.The intelligent optimization algorithm is a multidimensional linkage random search,which boasts a vast optimization space but suffers from low optimization efficiency.Therefore,this study introduces a constraint processing approach that integrates a penalty function with nested DPSA–POA and intelligent algorithms and applies it to the optimal flood control operation problem of cascade reservoirs in the middle reaches of the Ganjiang River,with decision variables extending to 2196 dimensions.The results of the correlation analysis indicated that:1)the nested DPSA–POA intelligent algorithm combined with a penalty function can address the high-dimensional optimization problem under varying water inflow conditions using three constraint processing methods;2)Of the three constraint processing methods,method 2,which involves DE optimization after securing a feasible solution through nested optimization,achieves the highest convergence accuracy,though the computation time is approximately 10 h;method 3,which involves DPSA–POA optimization after securing a feasible solution through nested optimization,achieves the second highest convergence accuracy,with a computation time of about 1~3 h;3)Existing SF,SR,PF,and EC constraint treatment strategies fail to consistently converge to a feasible solution under different water inflow conditions,and the convergence accuracy of the results,upon obtaining a feasible solution,is significantly lower than that of the method introduced in this study.Accordingly,the nested constraint processing method presented i

关 键 词：高维优化问题 约束处理方法 DPSA–POA 智能算法 水库群 

分 类 号：TV122[水利工程—水文学及水资源]