Application Of Wavelet Neural Network In High-speed Railway Subgrade Settlement Monitoring

With the continuous development and progress,the traffic conditions of social economy has become an important factor restricting social and economic development,so our country is building high-speed railway vigorously.To ensure high-speed train capable of safe,smooth and comfortable operation,it is necessary to ensure that the line track with high smoothness,the settlement must be high-speed railway line project for effective control,especially subgrade settlement.Meanwhile,we must assess systematic deformation and settlement prediction on the subgrade before laying the track to be laid.Used for settlement prediction method can be divided into two categories: measured data analysis and calculation method based on theory.There are many analysis methods based on the measured data,such as regression analysis,gray system theory,artificial neural network,but most of the excavation deformation prediction,dam construction(structures),the application of high-speed railway settlement monitoring the model is not much.Taking a long-Hangzhou passenger dedicated section Jiangxi section as the background,the establishment of the zone of deformation monitoring network,methods of observation,monitoring accuracy and so do discussed establishing wavelet neural network model for forecasting subgrade settlement analysis of the data to do a thorough the study.(1)Researching the method of signal de-noising by wavelet analysis,using wavelet threshold algorithm subgrade settlement data are removed,the original data processing carried out in the presence of noise,expectations for the future predictions more accurate.(2)Researching BP neural network algorithm,We analyzed the limitations of BP neural network algorithm and improved the traditional network model.,they improved model can overcome the existing problems in the traditional network model including the easy to form local minima and not the global optimum,low of training learning efficiency slow of convergence.And we applied the improved model to deformation prediction.(3)Researching the combination of wavelet analysis and neural network model.Combine respective advantages of wavelet transform and BP neural network,to build wavelet neural network model,improve the model and verify the effect.(4)We applied BP neural network model and wavelet neural network model to predictive value by using MATLAB.Then,we compared measured values and analyzed their performance.The performance analysis and experimental results show that wavelet neural network is better than the BP neural network,the high railway subsidence forecast analysis based more efficient and more precise.