Application Of Wavelet-based Grey Markov Combination Model In Foundation Pit Deformation Prediction

Subways in major cities of our country has entered a period of accelerated construction.However,in the construction process of subway foundation pit,due to the influence of many factors,it will lead to a certain degree of deformation of the engineering body.If the deformation exceeds the safety range,it will lead to serious consequences of safety accidents.Therefore,it is particularly important to strengthen the monitoring of Metro deformation,timely and effective prediction of Metro deformation,and to ensure the safety of metro construction and operation.At present,the commonly used single prediction models have their own limitations,and the prediction accuracy is often low,so it is difficult to meet the needs of deformation monitoring in Metro engineering.Therefore,taking the foundation pit construction monitoring of Jinkai Road Station of Nanning Metro Line 5 as an example,aiming at the dynamic and noisy characteristics of foundation pit deformation,a grey Markov combination model based on wavelet transform is constructed to predict the deformation of foundation pit construction in Metro station.The main research contents of this paper include:1.According to the requirements of technical specifications for monitoring of foundation pit engineering,the monitoring scheme and method of foundation pit deformation construction are designed scientifically and reasonably,which provides a real and reliable data basis for the establishment of the combined model of foundation pit deformation prediction.2.The basic principle of wavelet transform and the specific process method of wavelet threshold de-noising as well as the accuracy evaluation index are briefly introduced.The selection of wavelet function,the determination of decomposition level and the specific threshold acquisition method are discussed in combination with the actual engineering data.Through the comparison of the denoising effect,it is concluded that the best denoising effect is achieved by using DB6 wavelet to select heuristic threshold for three-level decomposition in the process of data processing in this paper.3.Combining the characteristics of grey model and wavelet denoising,two kinds of modeling methods of wavelet grey combination model are proposed,and the prediction accuracy of the model is further improved by combining two improved methods of original GM(1,1)model.Experiments show that the combination model with improved GM(1,1)model is more accurate than the combination model with wavelet reconstruction after choosing appropriate wavelet denoising method to denoise the high-frequency data to obtain a new stable high-frequency sequence,and then using the improved GM(1,1)model to predict each sequence.4.The modeling process of grey Markov model is elaborated in detail.The metabolic principle is used to improve the grey Markov model to a certain extent.The rationality of the improved method is verified by an engineering example.5.A grey Markov combination forecasting model based on wavelet is established.First,wavelet analysis is used as a pretreatment tool for data analysis,which can effectively eliminate the noise in the original data.Then,grey forecasting model is used to fit the data and find out its changing trend.Finally,Markov model is used to modify the model.The experimental results show that the combination model can enhance the sensitivity of prediction data to change trend by deep mining the implicit relationship in model data,eliminate the data divergence caused by long-term prediction,and more accurate predictions can be obtained.