Wavelet Analysis And Gray Forecasting For Deformation Monitoring Data

Because the structure of slopes is complex, with the cooperating of foundation and super structure, the slopes will have uneven settlements. This situation will bring the slopes tilt or cracks, and affect the normal exploration. More importantly, it will endanger the safety of the structures, so the deformation forecast of the slopes has become an important aspect in disaster prevention and disaster mitigation of the building operations. In order to ensure the security of the slopes exploration and construction, and to avoid economic losses and casualties, we have to regularly carry on the settlement observation to the high-rise slopes and the important slopes, to obtain the subsidence distortion data, and carry on analysis and forecast of it. So we can master the law of the slopes subsidence distorts, correctly predict the size of the deformation, and so as to promptly adopt the suitable measures to prevent and remedy to ensure the safe use of the slopes. This thesis proposes the application of grey system theory on the slopes subsidence forecast. It also takes subsidence distortion as a living example and establishes the grey forecast model to carry on the analysis to its subsidence data. The main contents are as follows:(1) Noise in deformation monitoring data is inevitable. How to extract diagnostic information from noise data sequence and improve deformation monitoring precision is the critical technology among the deformation monitoring systems. This thesis applies the de-noising theory to study the monitoring data and studies the influence of the method(algorism) of de-noising, wavelet base function and threshold value. The data processed by hard threshold value is rougher than soft threshold value. The threshold value and wavelet base function have an important influence. It achieves a good effect by using force de-noising. It also has an important influence of the selection of reconstruction scale. Through studing the slopes subsidence data of Dagushan, the results show that the wavelet de-noising theory can obtain satisfactory effect. So we can gain the more truthful data. They can express the law of the slopes subsidence distorts. The law of the data is more obvious after the wavelet de-noising. Therefore, using the obtainable data to establish the grey model is more accurate.(2) Through analyzing the characteristics of wavelet theory and grey model, the grey forecast model based wavelet analysis is proposed. This model uses wavelet analysis de-noising method to process the original deformation monitoring data, and then, builds the GM model to forecast. At the same time, the results are compared with the forecasted only by building grey model. The research shows that the new forecast method can improve forecast precision.(3) This thesis compares the forecast model based Kalman filtering with the forecast model based wavelet analysis. The research shows that the forecast model based wavelet analysis can increase predictive validity.