论文部分内容阅读
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 filter correction method of water level and flow rate, this paper focuses on the spatial and temporal distribution of noise mean and the calculation method of model noise variance matrix. In this paper, the problem of discontinuity of spatial distribution of model noise mean is solved by the idea of spatially linear distribution and temporal dynamic statistics. By using the combination of noise variance matrix of pure water dynamic model and filter state transition matrix, the filter model Noise variance matrix. The case study from the Cuntan to Wanxian reach of the upper reaches of the Yangtze River is carried out and compared with several sets of Kalman filter models. The results show that the improved filter model is most effective in improving the accuracy of flood forecasting.