基于多目标差分进化算法的水库多目标防洪调度研究

为求解水库多目标防洪优化调度问题，提出一种基于自适应柯西变异的多目标差分进化算法，克服了早熟收敛问题，提高了收敛精度；同时，根据多目标优化的特点对差分算子进行修正，并引入外部档案技术，提高了算法的收敛速度。本文以坝前最高水位最低、最大下泄流量最小和汛末水位最接近汛限水位为目标，对三峡水库展开多目标防洪调度研究，结果表明，该算法可在较短时间内生成大量在各目标分布均匀、分布范围广的非劣调度方案供决策者评价优选，为水库多目标防洪调度决策提供了一种新的调度方案生成方法。

Multi-objective optimization of reservoir flood dispatch based on multi-objective differential evolution algorithm

A multi-objective differential evolution algorithm based on the adaptive Cauchy mutation(MODE ACM) is proposed for optimization calculation of reservoir flood control. The adaptive Cauchy mutation strategy of this algorithm can effectively prevent from premature convergence and improve the convergence precision. The operators of differential evolution are modified according to the characteristics of multi-objective optimization. By using the archiving technique, the convergence rate of the proposed method can be obviously accelerated. The algorithm is applied to optimize the dispatch of the reservoir in Three Gorges Project for typical floods occurred in history. The result indicates that a lot of non-dominated dispatch schemes for decision-maker can be generated in a short time, these schemes distributed widely and uniformly on every optimized objective. This proposed algorithm provides a new way to generate feasible multi-objective flood control operation.