Application Of BP Wavelet Neural Network In Subway Tunnel Deformation Analysis

With the development of the science and technology, the traffic routes have been more complex. To relief the press of the traffic, the construction of subways has developed rapidly. It has been one of the most important parts of the construction of the metropolis. In order to ensure the safe operation of the subway, we need to know the tunnel deformation of the subway tunnel at any time, then we can deal with the possible security risks in the subway tunnel, so the deformation monitoring of subway tunnel becomes crucial. At present, several main data processing methods in the subway tunnel deformation analysis is regression analysis, time series analysis model, the gray system analysis model, kalman filter model, artificial neural network model.This paper mainly adopts the method of BP neural network, which is more widely used as a method of dynamic data processing. And it is an algorithm to adjust the weights and thresholds in the network. In the training and learning process of the BP network, there may be some problems, such as slow convergence of BP network model and easy to fall into local minimum. We need to make some improvements in view of traditional BP neural,such as additional momentum and adaptive learning rate adjustment method which effectively improve the convergence speed of BP neural network and avoid the local minimum point. The wavelet function of BP neural network is used to improve the speed of network convergence and avoid local minimum points. BP wavelet neural network model is established to deal with subway tunnel data. Based on a large amount of literature,first of all, this paper summarizes the significance of deformation monitoring, describes the purpose and significance of domestic and subway deformation monitoring technology and the methods of the data processing. In the wavelet analysis theory, it describes several commonly wavelet functions are described, for example, the Harr wavelet function,Daubechies wavelet, Symlets wavelet, Morlet wavelet function and Hat wavelet Mexican.Secondly this paper expatiates the basic principles, models and arithmetic of BP neural network; the way of combination of wavelet analysis and BP neural network, principle,model, arithmetic. Finally, combining with the case analysis, this article use the MATLAB software to write data processing procedures to deal with the subway tunnel observation data and the results of the two prediction models were analyzed. Finally, this article summarized the presented insufficient and development needs of BP wavelet neural network. At the last of this paper, it summarizes the insufficiency and proposed the shortage and development needs of BP wavelet neural network.