Abstract: Aiming at the different characteristics of water-bloom growing in different seasons，a predicting model combined with wavelet analysis and artificial neural network（WANN） is proposed based on the study in mechanism of water-bloom growing.The model which has great capabilities of both self-study of artificial neural network and multi-resolution analysis power of wavelet，is applied in forecasting water-bloom in lakes and rivers of Beijing in summer.Through the wavelet multi-resolution analysis，the influence is decreased efficiently from unnecessary noise brought by primary data，finally the performance of network is enhanced.Compared with the calculation results，WANN possesses high forecasting accuracy and has better performance than BP model.

摘要： 针对不同季节水华生长的不同特点，在对水华生长规律研究的基础上，运用小波分析对表征水华的叶绿素信号进行降噪处理，建立一种结合小波变换与神经网络相结合的水华预测模型（WANN模型），该模型既有神经网络的自学习能力特性，又有小波的局部特性，并将其应用到北京夏季河湖水华预测中。通过小波多分辨率分析，对样本包含的信息进行充分挖掘，提取反映其变化规律的成分，有效避免了原始数据中噪声对网络的干扰，提高网络的性能，WANN模型预测结果与BP网络预测结果对比，具有较高的预测能力，从而获得相对理想的预测效果。