Research On Rolling Bearing Fault Diagnosis Method Based On Feature Mode Decomposition

Rolling bearings as an important component of mechanical equipment plays a core rotational support role,is widely used in various types of industrial equipment machinery.Due to the complex working conditions in which the bearings are located,the harsh environment,high speed and heavy load,etc.may cause the bearings to fail.According to the incomplete statistics of rotating machinery and equipment damage records,about 30% of the failures occur in the bearings,so bearing failure is the main factor causing the failure of rotating machinery.If the faulty bearing cannot be found and replaced in time,it will not only affect the productivity of industrial equipment,but may even cause serious safety problems,so it is important to conduct fault diagnosis research on rolling bearings to ensure the safety of equipment production.However,usually the bearings are located in a complex working environment,easy to interact with the surrounding parts,resulting in the collected bearing signal often contains other parts of the vibration shock signal and noise interference,etc.,can not intuitively determine the type of bearing failure.This paper takes rolling bearing as the research object,and the main work completed is as follows:(1)According to the basic structure of the bearing and the main failure form,build rolling bearing inner and outer ring and composite fault vibration simulation signal.Build a bearing failure experiment platform to collect the vibration data of bearing inner ring,outer ring,rolling body and compound failure bearing at different speed to provide experimental data for the later algorithm.(2)A noise reduction algorithm combining improved feature mode decomposition(CKFMD)and singular value decomposition(SVD)is proposed.The CKFMD algorithm can fully consider the impulsivity and periodicity of the signal,will not be limited by the filter shape and bandwidth,and is robust to other disturbances and noise.Compared with the feature mode decomposition algorithm,CKFMD achieves adaptive selection of parameters,avoiding the non-objectivity caused by manual selection and improving the decomposition effect of the algorithm.Through the analysis and verification of the inner and outer ring simulation signals and the bearing data collected from the comprehensive fault diagnosis experiment bench,it is proved that the CKFMD-SVD algorithm can highlight the bearing fault impact characteristics and has a better noise reduction effect.(3)An adaptive resonance demodulation bearing fault diagnosis method based on jellyfish search algorithm is proposed.On the basis of the jellyfish search algorithm,in order to find the effective bearing fault characteristic band,the ratio of the fault characteristic frequency amplitude in the envelope spectrum to the mean value of the upper envelope on the interval is proposed as the fault characteristic judgment index.The larger the value of the objective function,the clearer the fault characteristics on the envelope spectrum and the more accurate the fault band found.Through the analysis and verification of the signals at the early stage of failure occurrence in the whole-life cycle experiments of bearings at Xi’an Jiaotong University and the bearing signals collected by the MSF-MG test bench,this method can accurately extract the demodulation bands of bearing failures and realize the diagnosis of multiple types of bearing failures.(4)A Multi-fault jellyfish search algorithm(MFJS)is proposed.This method sets classification labels for individuals by the fault feature frequencies searched on the envelope spectrum,and sets multiple search populations to search multiple demodulation bands of composite fault bearings respectively according to the label types to realize the separation and extraction of bearing composite fault features.Then the MFJS algorithm is combined with the CKFMD-based noise reduction algorithm,and it is verified by analyzing the bearing composite fault data from Xi’an Jiaotong University and the bearing composite fault data collected from the MSF-MG test bench that this method can realize the separation of the bearing composite fault resonance bands and extract the multiple fault impact features existing in the bearing,which is an effective bearing composite fault diagnosis method.