Research On Chirplet Transform Method For Synchronous Extraction Of Time-frequency Characteristics Of Train Wheel Set Bearing Fault Signals

Wheel set bearings are an important component of high-speed train bogies,and their health condition has a significant impact on the safe operation of trains.During the actual operation of trains,wheel bearings are usually in variable speed and strong noise conditions,and the collected fault vibration signals are often complex and variable in the time domain.When diagnosing faults in bearings,traditional signal processing methods cannot meet practical needs well due to their own limitations.Therefore,how to effectively extract relevant information of fault features from vibration signals is a key issue in fault diagnosis.On this basis,this thesis takes train wheelset bearings as the research object,and addresses the problems of frequency aliasing,low time-frequency resolution,and weak early fault signals in fault diagnosis.Starting from time-frequency features,an improved method for extracting fault signal features of train wheelset bearings based on Chirplet transform is studied.The main research content of this thesis is as follows:(1)Aiming at the problem of frequency aliasing when traditional synchronous extraction algorithm based on short time Fourier transform processes multi-component signals with similar frequencies,this paper introduces the synchronous extraction Chirplet transform method.This method combines Chirplet transform with synchronous extraction algorithm with high time-frequency resolution,which can effectively avoid frequency aliasing.The simulation signal experimental results have verified the effectiveness of this method,laying a theoretical foundation for the subsequent improved method based on synchronous extraction of Chirplet transform.(2)In response to the low accuracy of instantaneous frequency in synchronous extraction transformation and the difficulty in identifying time-frequency features caused by noise in actual train wheel bearing fault signals,a second-order synchronous extraction Chirplet transformation method with block neighborhood denoising is introduced.In order to improve the accuracy of instantaneous frequency,a complex frequency operator is constructed to obtain the second-order instantaneous frequency,and at the same time,a block neighborhood denoising algorithm is combined to improve the noise resistance performance.The simulation signal and engineering signal experiments show that the proposed method has better time-frequency characterization ability and can accurately identify the types of bearing faults.(3)In response to the problem that the parameters of time-frequency analysis methods are greatly influenced by subjectivity and the vibration signals of early bearing faults are weak,a second-order synchronous extraction Chirplet transform kurtosis map method using the optimal weighted sliding window is introduced.Combining the spectral kurtosis principle with synchronous extraction transformation algorithm to determine the position of impulse components in the signal,and then combining with the optimal weighted sliding window algorithm to achieve the extraction of weaker feature frequencies.The experimental results of simulation signals and engineering signals indicate that this method can accurately identify the characteristic frequencies of wheel bearing fault signals,and is feasible.