Research On Fault Diagnosis Method Of Rolling Bearing Based On Fractal Geometry

Rolling bearing is widely used and plays an important role in mechanical equipment.Its state is the key to the noramal operation of the equipment.Because of its open structure,bad working environment and stress condition,70%of the faults of rotating machinery are related to rolling bearings.Therefore,fault diagnosis of rolling bearings has always been a research hotspot.It is of great significance to find new and more effective fault diagnosis methods to reduce the failure rate of mechanical equipment.The most widely used fault diagnosis method of rolling bearing is the time-domain linear analysis method based on Fourier transform technology.This method can deal with linear and stable signals well,but it can not be well applied to the processing of non-linear and non-stable signals and fault feature extraction.The vibration signal of rolling bearings is a complex non-linear signal.Therefore,starting with the non-linear analysis method,this paper analyses the vibration signal of rolling bearings and extracts the fault signal characteristics and can provide reference for fault diagnosis methods which are non-linear and non-stable.Aiming at the key point that the vibration signals of rolling bearing have different fractal characteristics under different working conditions.,a series of fault diagnosis methods of rolling bearing based on fractal geometry are explored in this paper.The key technology of rolling bearing fault diagnosis is to extract fault features from vibration signals.So this paper uses fractal dimension to describe and extract fault features.On this basis,three methods of rolling bearing fault diagnosis based on single fractal,rolling bearing fault diagnosis based on fractal dimension and EMD and rolling bearing fault diagnosis based on multi-fractal are studied.A variety of machine learning models are validated by experiments..Firstly,the single fractal method is used to extract fault features of rolling bearing signals.The experimental results show that the diagnostic accuracy is about 75%.Through analysis and research,noise and other factors may affect the accuracy of diagnosis,and single fractal can not deal with such problems well.Therefore,a fault diagnosis method for rolling bearings based on box dimension method and EMD algorithm is proposed in this paper.The experimental verification and comparative analysis of the validation results show that SVM and ELM models are effective,and the diagnostic accuracy is 88.9%and 93.1%respectively.In view of the defect that single fractal can not describe the signal characteristics completely,this paper introduces the rolling bearing fault diagnosis method based on multi-firactal MF-DFA algorithm.The experimental verification shows that the diagnostic accuracy can reach 94.4%.The results show that the fractal geometry method can be effectively applied to the fault diagnosis of rolling bearings with non-linear and unstable signals.