Scoring Of Railway Quality And Predicting Of Rail Grinding Cycle

Rail profile is the important characteristic of railway,it reflects the status of the rail train which is under the influence of normal and natural factors.Besides,it affects much to the security and normal running of the rail.Therefore,it is meaningful and instructive to research with rail profile for the grinding of rail,the cost of construction and maintenance has reduced as well,while the polishing work is instructive.Innovation point of this paper is to build a scoring model and cycle model,thinking of the actual situation,to optimize and improve the algorithm used.Paper outlines basic characteristics,and gives the solution of the discrete data curve node,according to this thinking,four indicators of profiles,normal values,grinding area,vertical grinding and side abrasion value,how to calculate them are given,according to the four criteria weights to build scores of models.Using decision theory methods,firstly,use the binomial coefficient method to determine the coefficient of weight,then through the expert evaluating method and entropy techniques to improve the weighting coefficient,small error rate of scoring models in the end.Combined with the existing second generation profile instrument,developed scoring modelprogram at the client,use to statisticsthe scores of profile data.Statistical rating data of history profiles,through linear regression and polynomial fitting to predictive modeling,combining rail wear pattern,firstlythrough data pre-processing to eliminate outliers,then interpolation,and finally fitting model for prediction and forecast when amendments of the value date parameters to improve results.It’s the first time to apply the decision theory to the rail on the assessment,thinking of the actual effect,this is a kind of innovation in rail profile field.When using the decisiontheory to makethe score model,there is not using a single method,instead of using the binomial coefficient and entropy technology approach to modeling.When building the cycle forecast model,using polynomial fitting technique,according to the rail profile quality factors,data preprocessing and predictions made by amendment,which in the forecast model is a theoretical innovation.Based on the studies of the score and profile grinding cycle,obtained profiles of scoring models and cycle model,throughpracticalverified,the applicability of the model in the range is good.The researchof the model is a guide to the actual grinding operation,consider grinding a section needs a long time,and with various sections where required grinding time is not completely uniform,so there are some errors in predicting the next Polish dates,but these errors are within a reasonable range.Becausethere are many factors which are not taken into account,the results in models have its limitations,more data is needed to compensate for and correct model and experiment.