水文预测是水文学为经济和社会服务的重要方面。其预报结果不仅能为水库优化调度提供决策支持，而且对水电系统的经济运行、航运以及防洪等方面具有重大意义。自回归模型（AR模型）、人工神经网络（ANN）和自适应神经模糊推理系统（ANFIS）在日径流时间序列中应用广泛。将这三种模型应用于桐子林的日径流时间序列预测中，不仅采用纳什系数(NS系数)、均方根误差（RMSE）和平均相对误差（MARE）为评价指标，对三种模型的综合性能进行了比较。而且，在对三种模型预测结果的平均相对误差的阈值统计基础上，分析了三种模型的预测误差分布。同时，通过研究模型性能指标随预见期的变化过程评价了三种模型不同预见期下的预测能力。结果表明ANFIS相对于ANN和AR模型不仅具有更好的模拟能力、泛化能力，而且在相同的预见期下具有更优的模型性能，可以作为日径流时间序列预测的推荐模型。

Hydrological prediction is an important aspect of hydrology’s service for economic and society. The prediction result not only provides decision support for reservoir generation operation, but also is of great significance to the economical operation of hydropower systems, navigation, flood control and so on. The autoregressive model (AR model), artificial neural network (ANN) and adaptive neural fuzzy inference system (ANFIS) have been widely applied in the daily runoff time series prediction. In this paper, these three models were applied in daily runoff prediction at Tongzilin station. Nash–Sutcliffe efficiency coefficient (NS coefficient), root mean square error (RMSE) and mean absolute relative error (MARE) were used to evaluate the performances of three models. Threshold statistics index was used to analyze prediction error distribution of three models. At the same time, the prediction ability of three models was studied by gradually increasing the prediction period. The results showed that ANFIS had not only better simulation ability and generalization ability, but also better model performance in the same prediction period compared to ANN and AR model. As a result, ANFIS can be a recommended prediction model for daily runoff time series.

十二五科技支撑计划项目（2013BAB05B00）