| When the track geometry irregularity detection system performs track detection,due to the influence of complex detection environment,such as wheel idling and slipping,detection speed,whether to pass curve,track surface cleaning circumstance,and rail wear degree,the track detection data will inevitably generate mileage deviation,which will result in misalignment of waveforms and make it difficult to apply the data.Therefore,it becomes an important research issue to deeply carry out the research on the mileage deviation correction for the track detection data.In order to achieve mileage deviation correction,the identification and correction of mileage deviation become the key issue.Using waveform similarity as the breakthrough point,the use of the characteristics of different channel data using the same mileage system,using multi-channel data combination,following the principle of step-by-step control step-by-step correction,for the identification of large mileage deviation,using the curve of the main points of cant data change with a strong characteristics.It is proposed to use the normalized correlation coefficient as the waveform similarity coarse registration model,and use the template matching idea to extract large mileage deviation.On this basis,taking into account the nonlinear variation of the mileage deviation between the sample points,the mileage deviation correction model and amplitude resampling model are established respectively according to the basic principles of Lagrange and polynomial interpolation.In order to further accurately correct the mileage deviation,according to the correct waveform feature points with similar geometric positional relationship,the geometric constraint maximum vector angle and the maximum angle difference criterion are introduced to establish a precise registration model of the waveform features.According to the waveform variation law,the waveform feature extraction algorithm is proposed to process the waveform features.Then,the exact matching model establishes a one-to-one correspondence between the feature points,and obtains the mileage deviation at the feature points,and then realizes the accurate correction of mileage deviation.In order to verify the correctness and applicability of the algorithm,the track detection data under different line conditions measured by the same track geometric irregularity detection system are selected for algorithm verification.The results showed that the use of curve of the main points of cant data variation characteristics can effectively implement mileage deviation processing inside the long curve,the waveform feature can better control the short interval mileage deviation correction.The mileage deviation correction model can effectively correct the mileage deviation extracted by the waveform matching,and has a good correction effect,and the mileage correction accuracy is less than 0.635 m.The waveform feature extraction algorithm can accurately extract the key feature points of the waveform and highlight the key information of the waveform.The accurate regis tration model of waveform features can screen out the correct matching point pairs,establish the correspondence between feature points,and ensure that the waveform of the reference sequence is highly consistent with that of the observation sequence.Therefore,the algorithm has good correction effect and strong universality,and is applicable to the range deviation correction of detection data in the same track detection system. |
