基于卡尔曼滤波技术的河道汇流实时校正

基于卡尔曼滤波的河道汇流实时校正算法,采用非线性马斯京根矩阵解法,将河道汇流表达为一个时变线性系统,满足了卡尔曼滤波状态空间表达的要求.在进行滤波时,对系统参数矩阵的实时修正减小了河道汇流的非线性对线性滤波的影响.通过对每个子流域的“远程”和“本地”卡尔曼滤波校正,使全流域得到了全面合理的校正.提出一种新的模型噪声方差阵和量测噪声方差阵实时统计方法,该统计方法可有效地防止滤波发散现象,提高滤波预报的精度.通过在三峡区间流域上的应用,证明了基于卡尔曼滤波的河道汇流实时校正算法的有效性.

Real-time correction of flow concentration of river channels based on Kalman filter technique

The real-time correction algorithm based on Kalman filter technique was proposed to improve the accuracy of real-time hydrological forecasting under the circumstances that the nonlinearity of flow concentration of river channels is evident.In Muskingum matrix algorithm,the flow concentration was expressed as a linear system varying with time,which satisfied the requirement of spatial expression of Kalman filter.The real-time correction of the parameter matrix of the system decreased the influences of the nonlinearity of flow concentration on the linear filter.The advantage of the algorithm is that each sub-basin can be corrected twice by obtaining of the remote and local correction values,and then the whole basin can be rationally updated.Moreover,a real-time statistical method was put forward to estimate the system noise covariance matrix and measurement noise covariance matrix,which effectively prevented the filter divergence and improved the accuracy of forecasting.The effectiveness of the algorithm was verified by its application to some river channels in Three-Gorges region.