考虑到噪声对径流分析的影响以及传统消噪技术在径流消噪过程中存在的不足,引入小波变换技术对含噪信号进行分解和重构,通过对小波分解后的小波系数限定阈值来消除噪声.在此基础上,以相空间重构理论为依据,探讨了混沌径向基函数网络径流预测模型的建模思路、特点及关键参数的选取.最后以宜昌水文站为例,对其月径流时间序列进行了混沌识别和预测.实例表明:该模型能较好地处理复杂的径流序列,具有较高的泛化能力和很好的预测精度. Considering the influence of noise on runoff analysis and the shortcomings of traditional denoising technology in runoff denoising, the wavelet transform is introduced to decompose and reconstruct the noisy signal. The threshold of the wavelet coefficient after wavelet decomposition is defined to eliminate the noise Based on this, based on the theory of phase space reconstruction, the modeling ideas, characteristics and key parameters of the chaos radial basis function network runoff prediction model are discussed.Finally, taking Yichang hydrological station as an example, the monthly runoff Time series chaotic identification and prediction.Examples show that this model can deal with complex runoff sequences well and has high generalization ability and good prediction accuracy.