Study On Prediction Models Of The Track Vertical Profile Irregularity And TQI

With the rapid development of railway transport, high-speed and overloaded trains run generally, which exacerbating the deterioration of track condition and increasing the frequency of railway maintenance, at the same time, the gradual increase of railway operating mileage improves the workload of the railway repairman in a large scale, all of this are serious challenges to the works departments. The forecast for track irregularity can monitor the state of the track quality, arrange the maintenance plan reasonably, reduce the maintenance costs, and ensure safety and comfort. Therefore, forecasting the.law of track irregularity development has becomed the basic issues of the modern track mechanics, and is always the research focus at home and abroad.The forecast for track irregularity can be divided into the local track irregularity forecast and the section whole track irregularity forecast. Based on the comprehensive analysis of the track irregularity prediction models researched by the domestic and foreign scholars, this thesis focus on the track irregularity change characteristics and building the track irregularity forecast model.In addition, the local irregularity prediction models for different sections are built based on the maximum of track vertical profile irregularity, and the overall irregularity prediction models are built based on the track quality index (TQI) and the standard deviation of track vertical profile irregularity.Firstly, there are some problems with the field test data, such as mileage drift and burrs. In order to solve these problems, the data is preprocessed, including correcting mileage and smoothing data. Then, the maximum variation characteristics of the irregularity section are analyzed, and random exponential prediction models for different sections are created. The results show that the models can describe the maximum trends and stochastic volatility well. Secondly, the linear regression method and two-parameter exponential smoothing method are used to forecast the standard deviation of track vertical profile irregularity in straight and curve section. However, in order to weaken the random fluctuations of data, an improved two-parameter exponential smoothing method is presented in this paper, and the predictive effect is verified according to the measured values. Finally, this thesis analyzes the variation characteristics of the track quality index TQI, and proposes the modeling thought based on gray theory. After that, the honisometric GM(1,1) model and gray interval prediction model are established as prediction models for different sections. The results show that the gray interval prediction model can get a better accuracy, and it’s fitting curve can describe the development trend of TQI well. Compared with other prediction models in the elated references, it is apparent that the gray interval prediction model get a better practicability and easy to realize programming.