【摘要】 基于小波消噪理论对水文序列中的噪声进行了处理,然后利用混沌理论中的相空间重构技术计算出饱和嵌入维数作为多元回归模型的可控变量个数,将小波、混沌和多元回归方法结合起来对日径流进行了预测。与消噪前相比,消噪处理后建立的模型预测精度有了明显提高:预测合格率提高8%,平均绝对百分比误差为9.53%。因此在对水文时间序列进行混沌分析和预测之前,对其进行小波消噪是完全必要的。更多还原

【Abstract】 The saturated embedding dimension as the member of the controlled variable of the multivariate autoregressive model was computed by the daily runoff time series which was de-noised by wavelet technology. The daily runoff was predicted with the built above model. Compared with the model gained by the original daily runoff time series, the prediction precision is increased obviously. The eligibility rate increases 8% and the mean absolute percentage error is 9.53%. So it is essential that the hydrological time series are de-noised by wavelet method before chaos identification and prediction of the hydrologic system.更多还原