Research On The Key Methods For Railway Track Irregularity Prediction

Safety and efficiency are the two eternal topics of rail transport, but with the development of high-speed railway passenger transportation and freight overloading, the state of the quality of the track deteriorating faster and faster, which is a serious threat to life and property safety. We can carry on a scientific and rational analysis by using the history and irregularity data of the synthesis of track measuring car, and make a auxiliary decision about the plan of the line’s conversation on the basis of digging the variation of the time-series data, It has become an important means to improve the maintenance efficiency and reduce the economic costs on the promise of the safety of train operation.The thesis aiming at the problem of the existence of the abnormal points and the mileage drift, we use the absolute mean value method and Pauta criterion to process the raw data to remove the abnormal points, and use the method calculating the sum of the difference between two data checks to judge the offset of the mileage drif, the experimental results show that pre-treatment is a good way to reduce the detection error caused by various reasons.Then, we improve the research and analysis of the grey prediction model and the neural network prediction theory, use the advantages of the neural network has a good effect on data fitting and the grey prediction model has high accuracy of long-term forecasting, and establish a prediction method based on data selection vector, the experimental results show that the new model has a high prediction accuracy. Finally, we apply the prediction model to the arrangement of the annual Optimization Track Synthetical Maintenance Plan. In the prediction model, we use the maintenance time and maintenance location as the decision variables, and use the minimum annual average of the track quality index as the target function, then, in consideration of a series of constraint functions, we establish a aided decision model to get the optimum solution through the genetic algorithms. The experimental results show that the model can arrange out annual maintenance plan of the line quickly.