Copula分布估计算法在瞬时单位线参数估计中的应用  

Parameter Estimation of Instantaneous Unit Hydrograph Using Estimation of Distribution Algorithm Based on Copula

作　　者：朱从飞[1] 陆宝宏[1,2] 庄德利[3] 刘欢[1] 左建[1] 

机构地区：[1]河海大学水文水资源学院,江苏连云港222300 [2]河海大学水文水资源与水利工程科学国家重点实验室,江苏南京210098 [3]东海县水务局,江苏连云港222300

出　　处：《水电能源科学》2015年第12期28-31,共4页Water Resources and Power

基　　金：国家自然科学基金项目(NSFC-50979023);水利部公益性科研专项(201201026);中国博士后科学基金资助项目(2013M531270);江苏省博士后基金资助项目(1302029C);浙江省科技计划项目(2013C33033);江苏省水利科技项目(2013062)

摘　　要：针对利用传统矩法估计瞬时单位线参数时还原的洪水过程与实测洪水过程之间往往存在较大差异的问题,结合Copula函数与分布估计算法的优点,以实测地面径流与还原地面径流的加权残差绝对值之和最小为准则,采用Copula分布估计算法进行流域瞬时单位线的参数估计,这样既保证了算法的全局搜索能力,又充分考虑了参数间的相关关系。实例应用结果表明,利用Copula分布估计算法估计单位线,还原后的洪水过程与实测洪水过程的吻合度较传统的矩法高,特别是对于洪峰,优势更为明显。Aiming at the problem that the difference between the computed flood process based on the traditional moment method and observed flood process is often great, with the principle of minimizing the summation of the weight absolute errors between the computed flood process and observed flood process, the estimation of distribution algorithm （EDA） based on Copula （Copula EDA） is presented to estimate the instantaneous unit hydrograph in this paper. This method, considering the relationship of parameters and ensuring global search capacity, takes the advantage of both copula function and EDA. The case study indicates that Copula EDA has an obvious advantage over traditional moment method in parameter estimation, especially for the flood peak, which means that this method is practical.

关 键 词：瞬时单位线 参数估计 分布估计算法 COPULA理论 

分 类 号：P333[天文地球—水文科学]