基于误差自回归的洪水实时预报校正算法的研究

根据三水源新安江模型洪水预报误差信息，探讨了三种基于误差自回归模型的洪水实时预报校正算法，即固定遗忘因子的递推最小二乘算法，可变遗忘因子的递推最小二乘算法和辅助变量法，并将其应用于鲇鱼山水库的实时洪水预报。通过对三种实时校正方法进行分析比较，认为具有可变遗忘因子递推最小二乘算法效果最好。

Study on Adjustment Methods of Real-time Flood Forecasting in View of Autoregressive Model

According to flood forecasting error information of Three-water Source Xinanjiang model and autoregressive model, the paper deliberated three methods of adjustment techniques of real-time flood forecasting, namely, recursive least-squares algorithm with constant forgetting factor,recursive least-squares algorithm with variable forgetting factor and auxiliary variation method. The forecasting accuracy is satisfactory in concrete application in Nianyushan reservoir. Finally the three adjustment methods are analyzed and compared in detail. It is considered that the second method is best.