时间序列分析在水文预报中起重要作用 ,其关键是要建立一个合适的预报模型 .文章提出基于 BP算法的单输出和多输出水文预报时间序列神经网络模型 ,克服了以往多种基于随机分析预报模型的缺点 ,不仅能实现快速灵活的信息处理 ,而且具有很强的非线性映射和自学习、自适应能力 ,这为更精确描述复杂非线性水文过程提供了可能 .通过对历史数据的学习 ,模型可对水文径流量时间序列进行预报 ,两个实例分析表明模型的可行性和有效性 Time series analysis plays an important role in hydrological forecasting, and the key is to establish a suitable forecasting model.This paper proposes BP neural network model for single-output and multi-output hydrological forecasting time series based on BP algorithm, The shortcomings of the model can not only achieve fast and flexible information processing, but also have strong nonlinear mapping and self-learning and self-adaptability, which makes it possible to describe complex non-linear hydrological processes more accurately. Through the study of historical data, The model can predict the time series of hydrological runoff. Two examples show that the model is feasible and effective