Methods Of Evaluation And Early-warning Of The Railway Line Status Based On Dynamic Inspection Data Mining

With the increase of the speed in normal railway and the development of high speed railway, track irregularity becomes more and more important, because it directly affects the safety and comfort of the passengers in the train. At present, the railway bureau inspects the operational status of the railway line by using track inspection car, so a reasonable analysis of the data from track inspection car to guide the maintenance and repair of railway is becoming the focus of railway technicians’research at home and abroad.This article uses transfinite score, track quality index (TQI) and track irregularity power spectra to evaluate the irregularity of Wuhan-Guangzhou passenger dedicated line. Through the analysis found that the line is good in general, and problems mainly in the lateral acceleration limit of the vehicle body.In order to further analyze the detection data of track inspection car, the author gets the numerical value of each detection project in the waveforms of inspection (STE format) with the C language program, then calibrates the mileage in the data of inspection by using the gray correlation model. However, there are burrs in the waveform of inspection, so the author tries to reduce the noise in the data of inspection with the method of wavelet transform.We can do some early-warning according to the data of inspection with enough pretreatment. The author puts forward an evaluation method of cumulative update based on the existing one for track irregularity, and trains a BP neural network using the data of inspection to do the early-warning, it turns out that the network performs very well. Therefore the author tries to predict lateral acceleration of the vehicle body using speed and static inspection data, and trains a BP neural network to do the early-warning. By comparing the the warning results and measured results, it proves that the method is feasible.