Research On Fault Diagnosis Of Rolling Bearing Based On Intelligent Method And Synchrosqueezing Transform

With the rapid development of modern industry,rotating equipment,one of the most important equipment structures,is increasingly becoming large-scale and intelligent.And the application field is also expanding.It covers industries such as industry,energy,transportation,shipping,and aviation,and is closely related to national production safety.Once these devices fail,they may cause accidents to a certain extent,cause huge economic losses and social impact,and even endanger personal safety.The “Made in China 2025” plan is one of the eight strategic action plans for enhancing product quality and improving the reliability of major equipment(quality stability).Therefore,fault diagnosis for rotating equipment becomes significant.Bearings is one of the most common structures in rotating equipment.This subject takes this as the research object,deeply studies the accumulation of existing technologies,summarizes the experience of previous people,and develops relevant principles about new methods and new technologies for fault diagnosis of rotating machinery.1.This paper researches the synchrosqueezing wavelet transform algorithm and introduces it into the rolling bearing fault diagnosis.Firstly,I build a rolling bearing fault experimental platform to collect the rolling bearing fault datas,analyze it using wavelet transform,and then use synchrosqueezing transform to compress the coefficients after wavelet transform.Compared with short-time Fourier transform and wavelet transform,the experimental results show that the synchrosqueezing wavelet transform can effectively extract the characteristic frequency of rolling bearing.2.The project use rolling bearing test platform to collects fault signals of rolling bearings affected by strong noise,and reduces the size of wounds in rolling bearing faults.It uses empirical mode decomposition,local characteristic-scale decomposition,variational mode decomposition to process signals to reduce noise,and simultaneously uses maximum Kurtosis to optimize the VMD algorithm,then uses the ynchrosqueezing wavelet transform algorithm to process the effective component analysis,and compares the results of the envelope and ynchrosqueezing wavelet transform algorithm.It is proved that this method can effectively suppress the noise and extract the characteristic frequency of the rolling bearing.3.Approximate entropy is used as the characteristic parameter of the rolling bearing fault signal,and the local characteristic-scale decomposition algorithm is introduced into the nuclear limit learning machine.Using local characteristic-scale decomposition algorithm to decompose the signal,calculate the approximate entropy values of each component,select a part of approximate entropy for training,and use the remaining data as test datas to test.The experimental results verify that the method can effectively judge the common fault phenomena of rolling bearings.