Mε-OIDE求解约束优化问题算法及其在水库群防洪调度中的应用  

Mε-OIDE algorithm for solving constrained optimization problems and its application in flood control operation of reservoir group

作　　者：王文川[1] 田维璨 徐雷 刘昌军[3] 徐冬梅[1] WANG Wenchuan;TIAN Weican;XU Lei;LIU Changjun;XU Dongmei(College of Water Resources,North China University of Water Resources and Electric Power,Zhengzhou 450046,China;College of Hydrology and Water Resources,Hohai University,Nanjing 210024,China;Research Center on Flood and Drought Disaster Reduction,China Institute of Water Resources and Hydropower Research,Beijing 100038,China)

机构地区：[1]华北水利水电大学水资源学院,河南郑州450046 [2]河海大学水文水资源学院,江苏南京210024 [3]中国水利水电科学研究院防洪抗旱减灾工程技术研究中心,北京100038

出　　处：《水利学报》2023年第2期148-158,共11页Journal of Hydraulic Engineering

基　　金：河南省重点研发与推广专项(202102310259);河南省高校科技创新团队项目(18IRTSTHN009)。

摘　　要：约束处理技术和初始种群代表性对约束优化算法的性能具有重要影响。针对ε约束处理法求解约束优化问题时结果不稳定、经验参数难取值等问题,本文首先从当前两种不同的ε约束处理法出发,通过对其优缺点的分析,将Z-ε约束处理法对等式约束额外进行δ放松的操作补充到TS-ε处理法的整体框架中,并增设一个用户自定义参数来处理多样的约束条件,从而提出了一种改进的ε约束处理法。基于原始的差分进化算法,将其与前述改进的ε约束处理法和经典的反向学习初始化种群策略耦合,提出一种轻量化的Mε-OIDE(Modifiedε-Opposition-based-learning Initialization Differential Evolution)约束优化算法。在CEC2006基准函数集上的测试结果验证了耦合策略的有效性,表明提出的Mε-OIDE算法具有高精度和强鲁棒性。此外,在水库群防洪调度问题上的优化结果进一步证明了Mε-OIDE优化算法处理实际约束优化问题的可行性和高效性。Constraint handling methods and a representative initial population have a significant impact on the performance of constrained optimization algorithms.Aiming at the problems of theεconstraint handling method in solving constrained optimization problems such as unstable capability and difficulties in selecting empirical parameter values,this paper starts from the current two differentεconstraint processing methods,through the analysis of their advantages and disadvantages,the Z-εconstraint processing method additionally performsδrelaxed operations on equality constraints are supplemented to the overall framework of the TS-εprocessing method,and adds a user-defined parameter to deal with various constraint conditions,so as to propose a modifiedεconstraint handling method.Based on the primary differential evolution algorithm,a lightweight constrained optimization algorithm named Mε-OIDE was proposed by coupling it with the aforementioned modifiedεconstraint handling method and classical opposition-based learning population initialization strategy.The test results on the CEC2006 benchmark function set verify the effectiveness of the coupling strategy,indicating that the proposed Mε-OIDE algorithm has high accuracy and strong robustness.In addition,the optimization of reservoir group flood control operation further prove that the Mε-OIDE algorithm is feasible and efficient in dealing with practical constrained optimization problems.

关 键 词：约束优化 ε约束处理法 差分进化算法 反向学习 水库群 防洪优化调度 

分 类 号：TP697[自动化与计算机技术—控制理论与控制工程]