Research On Mileage Error Correction Algorithm And Evaluation Indexes Considering The Characteristics Of Track Geometry Data From The Rail Track Detection Vehicle

Track geometry data from the rail track detection vehicle inevitably have significant mileage error.The existence of mileage error not only affects the accurate positioning of the track disease location and delays the maintenance of the track disease,but also makes the track geometry data detected by the rail track detection vehicle in different periods have mileage offset and are not comparable,it directly affects the evaluation and prediction results of track irregularity data and the effectiveness of maintenance plan.At present,the mileage error correction technology of track geometry data is mainly based on least squares matching(LSM)and correlation coefficient function(CCF)to match the waveform of single-channel track geometry data,the mileage error is regarded as a constant to be spread into the mileage of track geometry data,so this way of the correction accuracy is low.In addition,dynamic time warping(DTW)is used to realize waveform matching between two sequences with different amount of data,Although the correction accuracy of DTW is high,DTW will destroy the original waveform characteristics of track geometry data and affect the analysis of track irregularity.Therefore,it is necessary to explore a high-precision mileage error correction algorithm considering the characteristics of track geometry data.To achieve the accurate correction of mileage error of track geometry data,Firstly,based on the rail track detection vehicle of the simulation model,the influencing factors of mileage error are comprehensively analyzed by using the control variable method.Then this thesis analyzed the characteristics of mileage error of track geometry data.Secondly,based on multi-channel track geometry data,the mileage information of the key point in the cant data is used as the correction point to correct the mileage error.Then,in order to reduce the mileage error of other sampling points between the two key points and improve the mileage correction accuracy of track geometry data,this thesis analyzed the shortcomings of least squares matching(LSM)and dynamic time warping(DTW).On the basis of the existing mileage correction methods,according to the changing trend of the waveform and its pointto-point matching relationship,a correlation interpolation optimal matching method(CIOM)is proposed,which takes into account the original track waveforms.CIOM contains two variables: the length of segment(S)and the movable amount of the segment boundary(k),that dynamically moves the segment boundary to realize point-to-point waveform matching,and to further correct the random error of sampling points while retaining the original waveform characteristics.an adaptive parameter adjustment algorithm is proposed for the CIOM,which greatly shortens the search time of input parameters(S,k)and realizes the automatic process of mileage error correction.Finally,in view of the shortcomings of the existing correction methods which only display the correction effect by drawing or calculating the correlation coefficient and difference of the correction result,it does not consider the waveform change of track geometry data.To improve the shortcomings of the evaluation system of the existing methods,the precision factor,shape factor and comprehensive factor are introduced as the quantitative evaluation indexes of the mileage correction result.After the simulation data and engineering measurement data experiments,the results show that:(1)When the track detection vehicle is in a variable speed state or the wheels are worn,the mileage error will be caused.(2)On the basis of one-time correction of the key point,the accuracy factor of CIOM can reach 0.92,and the correlation coefficient is increased to 0.88.(3)Compared with the existing mileage correction methods,CIOM not only accurately corrects the mileage error,but also retains the original waveform characteristics of track geometry data,and its precision factor and shape factor can reach about 0.9.In addition,CIOM makes up for the lack of LSM correction accuracy and the shortcomings of DTW that maybe changes the original waveform shortcomings.(4)The precision factor can reflect the alignment accuracy of the two-phase track geometry data,and the shape factor can make up for the shortage of existing evaluation indexes that cannot evaluate waveform changes.(5)The correction value of mileage error solved based on single-channel track geometry data is not reliable,and there is error.The correction value of mileage error should be solved by multi-channel track geometry data.CIOM provides a new idea for mileage error correction of track geometry data,it provides accurate mileage information for scientific analysis and research of track quality,and ensures the comparability of multi-period track geometry data.