识别河流水质模型参数的单纯形-差分进化混合算法  

Simplex-differential evolution hybrid algorithm for parameter identification of river water quality model

作　　者：付翠[1] 刘元会[1] 郭建青[2] 袁华[1] 

机构地区：[1]长安大学理学院,西安710064 [2]长安大学环境科学与工程学院,西安710051

出　　处：《水力发电学报》2015年第1期125-130,共6页Journal of Hydroelectric Engineering

基　　金：国家自然科学基金(1171043)

摘　　要：应用文中构造的单纯形-差分进化混合算法,分析了一维河流纵向离散和二维河流横向扩散两种情况下的水团示踪试验数据,识别了河流水质参数。这种混合算法是将具有局部搜索能力强、收敛速度快,而对初始值的选取依赖性较强等特点的单纯形算法与具有全局搜索能力强、受控参数少、而局部搜索能力弱等特点的差分算法进行混合。使其同时具有随机性搜索与确定性运算所具有的共同优点,能够较好地平衡局部搜索能力和全局搜索能力。数值实验结果表明:该算法可有效地应用于河流纵向离散和横向扩散两种情况下的水团示踪试验数据分析与水质参数识别,并且运算速度快、收敛性好,具有重要的应用价值。This paper describes a simplex-differential evolution hybrid algorithm to analyze experimental water quality data of river stream, and estimates water quality parameters for a 1-D river under longitudinal dispersion and a 2-D river under transverse diffusion. This algorithm combines the features of simplex method and differential evolution method： the former has a strong local search ability and fast convergence but higher dependence on the selection of initial parameters; the latter has an strong global search ability and fewer controlled parameters but weak local search ability. The combined algorithm has advantages in both random searching and deterministic calculation, and well balances the local against global search ability. Numerical experiments show that this new algorithm is of fast convergence and cost saving in analysis of experiment data of water quality and estimation of water quality parameters of river stream, and thus it has a good application potential.

关 键 词：环境水力学 河流水质模型 单纯形-差分进化算法 纵向离散系数 横向扩散系数 

分 类 号：TV211.12[水利工程—水文学及水资源]