Feature Enhancement Extraction Of Rolling Element Bearing Faults Based On Instantaneous Angular Speed Signals

Rolling element bearings(REBs)are crucial components in rotating machinery and equipment as they provide support for the rotating shaft and reduce friction,resulting in improved rotational accuracy.Partial failure of REBs can cause economic losses and major accidents that endanger human life.Therefore,the extraction of REB fault characteristics is of great significance to ensure safe production and improve economic efficiency.Vibration analysis is a common method to extract REB fault characteristics,but it requires external installation of vibration sensors on the bearing or equipment housing,which poses difficulties for equipment that needs to perform a large range of compound motion,such as industrial robots.Additionally,strict periodic components like rotating shafts and gears make it challenging to extract REB fault characteristics from vibration signals.Compared with vibration sensors,rotary encoders are usually mounted coaxially,and the Instantaneous Angular Speed(IAS)signal based on the encoder is directly related to the dynamic characteristics of the rotating shaft,which has the advantages of less interference and short transmission path.Therefore,this paper carries out the enhanced extraction of weak fault characteristics of rolling bearings based on the IAS signal.The main work contents are as follows:(1)To address these challenges,this paper proposes an enhanced method to extract the weak fault characteristics of REBs based on the Instantaneous Angular Speed(IAS)signal from rotary encoders.The method uses the forward difference method to estimate the acquisition of the IAS signal,suppresses or eliminates interference from strict periodic components using the DPA algorithm,and uses the Multipoint Optimization Minimum Entropy Deconvolution Adjusted(MOMEDA)algorithm to enhance the REB fault impact in the random signal.Fault features are then extracted by envelope order spectroscopy.(2)The paper also proposes a REB fault feature extraction method based on parameter adaptive Maximum Correlation Kurtosis Deconvolution(MCKD)and Local Mean Decomposition(LMD)algorithms.The envelope entropy is used as the fitness function of the Gray Wolf Optimization(GWO)algorithm to adaptively determine the filter length and shift number optimization combination of the MCKD algorithm.LMD is used to adaptively decompose the signal,and the PF component containing rich fault information is selected using Correlated Kurtosis as the evaluation index.Fault characteristics are then revealed by envelope order spectroscopy.The proposed methods are verified through the analysis of simulation signals and measured REB outer ring fault data of the corresponding experimental bench.The results demonstrate that the proposed methods can effectively extract the fault characteristics of REBs,providing a reference for solving REB fault diagnosis in places where vibration detection is limited.