论文部分内容阅读
在差分进化算法的基础上,受马尔可夫链蒙特卡罗方法的启发,建立了differential evolution adaptive metropolis(DREAM)算法.DREAM算法融合了马尔可夫链蒙特卡罗方法和差分进化算法的优势,较好地解决了马尔可夫链蒙特卡罗方法中搜索步长的恰当取值以及搜索方向的准确定位问题,并能有效解决差分进化算法的群体多样性和收敛速度问题.在DREAM算法基础上,引入多目标优化思想,提出了一种基于改进适应度分配策略和外部存档方案的多目标DREAM算法,并应用于岷江流域CMD-3PAR降雨-径流模型参数优选研究.结果表明:多目标DREAM算法能够找到一组范围宽广、分布均匀且数量充足的Pareto最优解供决策者评价优选.
Based on the differential evolution algorithm, inspired by the Markov chain Monte Carlo method, a differential evolution adaptive metropolis (DREAM) algorithm is established.DREAM algorithm combines the advantages of Markov chain Monte Carlo method and differential evolution algorithm, Which better solves the problem of the proper value of search step and the exact orientation of search direction in the Markov chain Monte Carlo method and can effectively solve the population diversity and convergence speed of the differential evolution algorithm.Based on the DREAM algorithm , A multi-objective DREAM algorithm based on improved fitness allocation strategy and external archive scheme is proposed and applied to the optimization of CMD-3PAR rainfall-runoff model parameters in Minjiang River Basin.The results show that the multi-target DREAM algorithm It is possible to find a set of Pareto optimal solutions with a wide range, uniform distribution and sufficient quantity for decision-makers’ evaluation.