Study On The Data Processing Method Of Laser Non-contact Dynamic Detection Of Wheel Tread Abrasion

With the rapid development of China’s railway industry,the density of railway transportation is increasing,and the running speed is also increasing.Ensuring the safe operation of trains is becoming more and more important for railway development.In the process of train operation,the tread of train wheel set is in close contact with the rail.Due to the influence of operation conditions and environment,the tread of train wheel set is vulnerable to damage.According to statistics,the proportion of train operation accidents caused by train wheel set is high.As one of the most common wheel set tread faults,tread abrasion is harmful to train operation safety.Therefore,the research on wheelset fault,especially the research method,technology and equipment of tread scratch,and timely detection of wheel tread fault are the important premise to ensure the safety of train operation,and the research has great application value.In this paper,the research status of wheel set tread fault detection methods and technologies at home and abroad is investigated and analyzed.At present,the wheel set tread fault detection is mainly divided into two categories: manual static detection and online dynamic detection.Static detection is mostly contact type,which needs to send the train to the detection point regularly for artificial detection.This detection method greatly wastes human and material resources,and can not find the wheel tread fault in time,so it can not meet the needs of the rapid development of China’s railway.The on-line dynamic detection of wheelset tread fault has become an inevitable trend.The main advantage of this method is that it can realize the non-contact dynamic detection of wheel tread scratch at different train speeds.Based on this method,the detection system is constructed in the laboratory,and the experimental data collection is completed.In the aspect of data analysis,this paper first analyzes the simulation normal data and scratch data,as the basis of distinguishing normal signal and abnormal signal.Then the peak value method is used to count the simulation data and experimental data,and the conclusion that the peak value increases with the increase of scratch depth is obtained.In order to overcome the influence of external noise and improve the stability of detection,wavelet transform modulus maximum method and wavelet packet energy spectrum method are selected to process the simulation data and experimental data respectively.Simulation results show that: wavelet transform modulus maximum method can effectively suppress noise in the process of signal processing,and the basis function with time resolution can effectively and accurately realize the analysis an processing of non-stationary signal.By comparing the variance and offset of multiple measurement results under the same scratch depth,it is concluded that the wavelet transform modulus maximum method is more stable than the peak identification tread fault analysis method in processing the experimental data.The analysis method of wavelet packet energy spectrum is a further improvement of the wavelet transform modulus maximum method,which can effectively decompose the low-frequency and high-frequency parts of the signal.This method has no redundancy or distortion in the signal processing results,which is the same as the verification method of wavelet transform modulus maximum method,By comparing the variance and offset of multiple measurement results under the same scratch depth between this method and the peak method,it is concluded that this method is more stable than the traditional peak method,which provides a feasible data processing method for non-contact dynamic detection of wheel tread scratch based on laser.