水库单目标优化调度技术比较研究

Comparative Study on Single-objective Optimization Algorithms for Reservoir Operation

摘要: 优化算法的选择是水库优化调度中的一个难点。本文选取了传统优化算法POA、智能优化算法GA和DE等，从变量规模、算子选择、参数选定、约束处理方法等方面对算法在水库优化调度中的性能进行了深入的比较研究。结果表明：智能优化算法在水库优化调度中具有一定的适用性，不同算法的搜索效率差异较大，需要根据问题和算法特点对算子和参数进行精心设计和选择。对于变量较少的小规模水库优化调度问题，通过选取合适的算子和参数，智能优化算法GA和DE从耗时等方面会优于传统优化技术POA；但对于变量较多的复杂水库优化调度问题，POA在耗时和解的质量等方面仍然具有显著的优势。研究成果可为水库优化调度优化算法的选择提供参考。

Abstract: Selection of optimal algorithms is one of the most complex problems for reservoir operation. Progressive optimality algorithm (POA), genetic algorithm (GA) and differential evolution algorithm (DE) were selected in this paper, and the performance of these algorithms were compared from the aspects of the number of decision variables, selection of arithmetic operators, determination of parameter values, constraint handling, etc. Results show that modern intelligent algorithms were applicable to reservoir operation optimization with big differences in performance for different intelligent algorithms. So it is necessary to select appropriate operators and parameters when using modern intelligent algorithms in reservoir operation. GA and DE with proper operators and parameters may have an advantage over POA in performance for reservoir operation problems with less decision variables, but POA is still superior to GA and DE for complex reservoir operation problems with large number of decision variables. This study helps to select proper optimization algorithms and parameter values for reservoir operation.