首先构造出人工神经网络非线性时序模型 ,然后用该模型进行单变量和多变量时间序列预报研究。为了与传统的随机水文模型对比 ,选择了自回归模型。以日流量序列为例 ,研究结果表明 ,人工神经网络非线性时序模型预报效果不错 ,可以在水文预报中加以应用 Firstly, a nonlinear time series model of artificial neural network is constructed, and then the model is used to study single-variable and multi-variable time series forecasting. To compare with the traditional random hydrological model, an autoregressive model was chosen. Taking the daily flow sequence as an example, the results show that artificial neural network nonlinear time series model is effective and can be applied in hydrological forecasting