Research On Data Denoising And TQI Prediction Algorithm For Track Irregularity

The geometrical smooth state of the track is a reflection of the comprehensive performance of the track structure.With the rapid increase in railway operating mileage and train load,the deterioration of the geometric state of the track line is getting faster and faster,which also puts forward higher requirements for the maintenance and maintenance.How to obtain the accurate track detection data and use the historical detection data to research the development law of the geometrical irregularity in orbit has important practical significance.We can reasonably predict the development trend of orbital geometric irregularity,manage the railway and ensure the safe and smooth operation of the train.This has become an important means for the current railway sector to ensure smooth line,improve maintenance efficiency and reduce maintenance costs.Aiming at the problem of noise in the dynamic detection data of the orbit,a novel denosing scheme which combined with the adaptive characteristic of the Aggregation Empirical Mode Decomposition(EEMD)algorithm is proposed.The original data is processed by the angle cosine and fuzzy threshold.The boundary position of the noise-dominated and signal-dominated intrinsic mode function(IMF)is obtained by the angle cosine method,and the noise-dominated IMF is processed by the fuzzy threshold method.Finally,all the IMFs are reconstructed to get the denoising signal.The three kinds of simulation signals are denoised under different signal to noise ratio.The results show that the proposed algorithm is superior to the wavelet half-soft threshold,EMD-Soft algorithm and EMD-IT denoising algorithm in the whole.De-noising experiment is carried out on the two-track irregular motion detection data.The results show that the denoising algorithm can reduce the noise of the orbital dynamic detection data and provide more realistic rush data for the management of track irregularity.Due to the lack of historical data of TQI,TQI forecasting is Typical small sample prediction.Based on the trend and randomness of the development of track quality index(TQI),the gray model and support vector machine model are studied and analyzed respectively.Combined with advantages of the gray model for long-term prediction with high precision and the support vector machine to fit the small sample data,A new model based on gray and support vector machine(SVM)is established,which can predict the trend of TQI in the future.In order to improve the prediction accuracy of the model,the traditional gray model is improved and the particle swarm optimization(PSO)algorithm is used to select the parameters.Respectively,track irregularity forecast results about two lines of Shanghai-Kunming line show higher precision of the combined model,and it's easy to program the calculation.