Based On The Wavelet Theory And Grey Model Of Slope Deformation Monitoring Analysis And Research

It is inevitable that it contains a lot of noise for slope deformation monitoring data, then the problems must be solved which how to remove the noise and real information is maximum limit got. Wavelet analysis is a high performance signal analysis method, it have excellent de-noising ability and time-frequency localization analysis function, Signal singularities distribution can be detected, it has good analytical capability for the non-stationary time series. This paper’s main concepts are as follows:1.Monitoring data is processed by wavelet packet analysis, in which election criteria of the optimum wavelet packet base is searched, Threshold entropy is best by the simulation experiment.2. Selecting the biggest scale, it is not the smaller de-noising result better for RMSE of de-noising, it is very close with he measured data, Mean square error limit is mentioned. The smaller its value, the more stable de-noising data in corresponding scale, this scale is the best decomposition level. Size is solved by specific experimental analysis.3. Wavelet packet is decomposed with binary tree, along with the increase of decomposition layers, transforming coefficient is reduced gradually. It is not located in the same threshold the corresponding threshold values should be reduced. Amending threshold and its reasons are mentioned. This method is better by de-noising results.4. New information GM(1,1)、Metabolism GM(1,1)、ARMA、BP neural network methods are used to de-noising data. Metabolism GM(1,1) that new information is continuously added, Old information is timely removed, it makes the computer memory is not increased. Modeling computational cost is reduced too, the results show that metabolism is ideal, grey model is feasible for slope monitoring data.