基于并行遗传算法的新安江模型参数优化率定方法

Parameter calibration of Xinanjiang rainfall-runoff model by using parallel genetic algorithm

本文结合新安江模型参数的特点,以洪峰流量、峰现时间和洪水总量的合格率为评价目标,定义了评价洪水性能目标的模糊合格率,提出了新安江模型参数率定的并行遗传算法,并在微机集群环境下,利用JAVA语言进行了算法编程。串行和并行遗传算法计算结果的比较表明,本文提出的并行遗传算法可以大大缩短优化过程的时间,得到较为稳定的模型参数。

A parallel genetic algorithm(PGA)for calculation of Xinanjiang rainfall-runoff model is proposed.The algorithm evaluates the fitness function based on the definition of fuzzy qualified ratios of floods.The PGA is written in Java language and executed in a cluster of PCs.The comparison of calculation results shows that the proposed method is remarkably better than serial genetic algorithm.The time consumption for optimization is greatly reduced and the more stable parameters can be attained.