Scheme And Research On Deformation Monitoring Of Urban Subway

With increasing levels of urbanization, the urban population increased rapidly, and it has brought a series of problems, such as traffic congestion and environmental pollution. Urban subway as a tool to ease pressure on urban traffic congestion, has been rapid development. However, the city subway lines generally pass through the main roads and many of the population centers, during the construction of the subway will cause deformation of the underground tunnel itself, pipeline and surrounding buildings. Deformation monitoring is an important means to protect the construction safety of urban subway and reduce the impact of subway construction on the surrounding environment, and deformation analysis and prediction of subway monitoring data to provide accurate guidance for the safe construction in the late, so deformation monitoring is particularly important in urban subway construction process. This paper mainly study as follows contents about deformation monitoring of city subway:(1) Combined with practical engineering of Xiao Xihu Station of Lanzhou Rail Transit Line 1 project, it discusses the subway construction deformation monitoring network layout and organizational implementation of monitoring programs, and introduces the basic theory of deformation monitoring network design and the quality standards of optimization design. By changing the observation rights of level monitoring and control network and optimization, to improve the quality standards monitoring network.(2) This paper discusses the principle of deformation prediction common methods, and comparative analysis the forecast results of these different kinds of methods with specific examples. Time series forecasting model suitable for use in short-term forecasting, regression analysis is a static prediction method, and the anti-interference ability of gray theoretical prediction is relatively weak. On the basis of these methods, it proposed the main major BP neural network and wavelet neural network model in this paper.(3) Synthetically expound the basic principles and methods of wavelet denoising, combined with the deformation monitoring data of the Xiao Xihu deep pit, using MATLAB language, and choosing different wavelet function, threshold and the number of layers of wavelet, original observation data de-noising, then compared with the original observation data. And analysis of obtained dbN and symN wavelet function, fixed thresholding effect is relatively good. Finally, the denoised data is applied to the prediction of deformation, make the predictions more accurate.(4) Introduced artificial neural network and wavelet theory method, research is mainly focused on the wavelet network prediction model, it based on wavelet theory and BP neural network combined. And increased the learning rate momentum k based on wavelet neural network constituting an improved wavelet neural network prediction model. According to the basic theory, choose observation data of representative five points about the Xiao Xihu deep pit established wavelet neural network prediction model.(5) Combined with MATLAB language and the actual deformation monitoring data, and writing the program of BP neural network, neural network and improved wavelet neural network, then simulated & trained three prediction models. Cumulative settlement denoised data were predicted and compared with the measured value. The results showed that: the prediction accuracy of artificial neural networks and improved wavelet neural network prediction model is better than BP neural network. The improvement of wavelet neural network model prediction closer to the observed values, and the network model has good value in the city subway deformation prediction.