Morphological Signal Processing Methodsand Its Application In Rolling Bearing Fault Diagnosis

Rolling bearing is an important component in rotating machinery,widely used in rotating machinery system,but the bearing failure is often affect the mechanical operation,may lead to paralysis of mechanical system in serious condition,resulting in huge economic losses and casualties accident.According to incomplete statistics,the proportion of rotating machinery fault caused by bearing at about 30%.Consequently,early fault monitoring and identification of bearing faults are of great significance.Effective vibration signal analysis is the key to bearing fault diagnosis.Its essence is to use signal processing method to extract the weak fault characteristic frequency from vibration signals,so as to identify the fault position and fault type.As a nonlinear signal processing method,mathematical morphology has achieved good results in mechanical fault diagnosis,but there are still many problems.In order to solve the problems of unsatisfactory feature extraction and unreasonable scale selection of structural elements in traditional morphological signal processing methods,an adaptive generalized gradient product morphological filtering algorithm and an optimal scale generalized morphological undecimated wavelet algorithm are proposed in this paper.The main contents of this paper are as follows:1、An algorithm for rolling bearing fault feature extraction based on Generalized Gradient Product Morphological Filtering is proposed.By referring to relevant literature at home and abroad,the existing morphological operators are realized.Through comparative analysis,the operators with strong noise reduction and feature extraction ability are summarized.On this basis,a generalized gradient product operator is proposed.Based on the Feature Amplitude Ratio,the Local Feature Amplitude Ratio is proposed to select the optimal structural elements adaptively.The Adaptive Generalized Gradient Product Morphological Filter is proposed.Compared with the existing methods,the proposed method has better feature extraction ability.The effectiveness and superiority of the proposed method are verified by analyzing and processing the simulation signals and the measured vibration signals of the fault bearing.2、Aiming at the problems of the traditional Morphological Undecimated Wavelet algorithm,such as unclear filtering result and poor feature frequency extraction,Generalized Morphological Undecimated Wavelet algorithm was proposed to construct the Optimal Scale Generalized Morphological Undecimated Wavelet Filter combining with Local Feature Amplitude Ratio.The simulation signals and the measured vibration signals of the fault bearing verify the feature extraction effect of the method,and the meaning of the method is more clear and more explanable.3、The above two morphological filters were applied to the composite fault diagnosis of high-speed train wheel bearings,and the fault features were extracted step by step based on the different sensitivity of different structural element scales to different fault types,so as to realize multi-fault diagnosis of bearings.The effectiveness of the proposed Adaptive Generalized Gradient Product Morphological Filter and Generalized Morphological Undecimated Wavelet Filter is proved by analyzing the measured compound fault signals of bearings.