一类洪水预报的非线性时序模型——指数自回归模型

A Nonlinear Time Sery Model for Flood Forecast——exponential autoregressive model

作为雨洪系统的输出——洪水时间序列，它包含了系统中各种变量的过去信息，同时蕴含着大量关于系统演变的规律和趋势，这样的时间序列往往是不可逆的，非性线相依的偏态序列，并且存在着广泛的频幅相依特性。在进行洪水预报时，传统法多采用线性化技术，但预报精度并不理想，因此要提高预报精度，有必要考虑洪水的非线性特性。基于此，本文用指数自回归模型进行洪水预报研究，实例分析表明该模型可提高洪水预报精度。本文的尝试工作为洪水预报提供了一种可行的模型。

As the output of the flood-rainfall system, the flood time series which include lots of past information of all variations and which hide many laws and trends of system evolution are often time irreversibility, nonlinear dependence, asymmetry and amplitude-frequency dependence. Traditional flood forecast usually uses linear techology, but the forecastingprecision is dissatisfactory. In order to increase flood forccasting precision the nonlincar character of flood must be considered. Bccause of those reasons above, the exponential autoregressive model is studied for flood forecast in this paper. Case study shows that the model can improve flood forecasting precision which dependenced upon consideration of nonlinear character. As a result, the model is suitable for flood forecast.