河道一维水动力学的卡尔曼滤波模式优化  

Kalman filtering mode optimization for 1D hydrodynamic model in river

作　　者：孙金[1,2] 王船海[1,2] 吴晓玲[1,2] 吴朱昊[1,2] 

机构地区：[1]河海大学水文水资源学院,南京210098 [2]河海大学水文水资源与水利工程科学国家重点实验室.南京210098

出　　处：《水动力学研究与进展（A辑）》2012年第6期679-686,共8页Chinese Journal of Hydrodynamics

基　　金：水利部公益性行业科研专项(201101024);海洋公益性行业科研专项(200905013-8);国家重点基础研究发展计划(973计划,2009CB421105);国家自然科学基金(51009045)~~

摘　　要：该文以水位和流量为状态量的交替滤波校正方法为基础,重点针对滤波模型噪声均值的时空分布和模型噪声方差阵计算方法进行研究。文中采用空间上线性分布及时间上动态统计的思路解决模型噪声均值空间分布不连续的问题;采用纯水动力学模型噪声方差阵与滤波状态转移矩阵相结合的处理方式选取具有相关关系的滤波模型噪声方差阵。选择长江上游干流寸滩至万县河段进行实例验证,并与多套卡尔曼滤波模式进行了对比计算,结果表明:在该验证实例中,改进后的滤波模式对提高洪水预报精度最有效。Based on the Alternate Kalman filtering method with state variables of water stage and discharge, this paper focus on studying the spatial and temporal distribution of the filter model noise mean and the calculation of the error covariance matrix of model noise. This paper use the way which is linear distribution on the space and dynamic statistical on the time to solve the problem of the uneven spatial distribution of the model noise mean. It adopts the approach which combines the covariancematrix of original hydraulic model noise and the filter state transition matrix to select the covariance matrix of filter model noise which contains correlation of model noise. It takes the Cuntan to Wanxian river in the upper reaches of the Yangtze River as an example of verification, and designs multiple sets of Kalman filter mode for comparison, the results show that the improved filter mode is the most effective to improve the accuracy of forecasting flood in this verification instance.

关 键 词：水动力学模型 实时校正 交替卡尔曼滤波 模型噪声均值时空分布 模型噪声方差阵 

分 类 号：P338[天文地球—水文科学]