Research On Damage Identification Of Rolling Bearing Based On Improved LMD And Multi-scale Scatter Entropy

Rolling bearing is one of the most widely used core parts in rotating machinery equipment.Whether the mechanical equipment can operate normally and complete the scheduled functions depends to a large extent on the performance and reliability of its basic components.Rolling bearings are prone to failures due to high speed,heavy load or harsh environment,which leads to unnecessary economic losses.Therefore,fault diagnosis of rolling bearings have important engineering practical value and theoretical significance for the effective implementation of condition-based maintenance and health management of mec hanical equipment.In recent years,there are many research results on fault diagnosis methods of rolling bearing fault vibration signals.The local mean and envelope estimation functions in the traditional local mean decomposition method are smoothed seve ral times using the sliding average,which results in local errors in the function,which the accuracy of the rolling bearing fault feature frequency is affect.Meanwhile,from the angle of extracting the fault characteristic frequency of rolling bearings,the fault diagnosis of rolling bearings is achieved and good diagnostic results are obtained.However,in reality,it is not enough only to know whether there is fault and the type of fault to prevent and maintain the bearing.Only to understand the fault evolution process and grasp the severity of the damage,can we effectively guide the bearing maintenance and save the production cost.On this basis,this paper focuses on envelope construction method,fault diagnosis and damage assessment of rolling bear ing.The main research work is as follows:(1)The threshold of nonstationary coefficient is lack of adaptive selection when envelope is constructed by composite interpolation.It is impossible to describe the non-linear and non-stationary process with constant value.To this end,an improved composite interpolation envelope construction method(ICIE)based on fractal box dimension D is proposed.The envelopes performance of different non-stationary coefficient thresholds D_n in the CIE method and ICIE is compared by using strong non-stationary analog signals.The absolute error of the envelopes is further calculated to verify the effectiveness of the ICIE method.Afterwards,the envelope of rolling bearing fault vibration signal is constructed by CIE method with different non-stationary coefficient thresholds and ICIE method.The actual engineering value of ICIE envelope construction method in rolling bearing fault vibration signal and the correctness of introducing fractal box dimension as non-stationary coefficient threshold are verified.(2)The construction of the envelope in LMD is inaccurate,which results in local error of the local mean and envelope estimation function.At the same time,improper screening of PF components leads to failure pulse compone nts that can not highlight vibration signals well,thus reducing the accuracy and reliability of identification.In this connection,an improved LMD method based on ICIE method is proposed,and a rolling bearing fault detection method based on ICIELMD and kurtosis criterion is further proposed.Firstly,an improved LMD method based on ICIE method is proposed,and the decomposition performance of LMD method before and after improvement is analyzed by combining the mean orthogonal index IO_ave and energy conservation index IEC.Secondly,fault vibration signals such as outer ring,inner ring and rolling body of rolling bearing are decomposed by ICIELMD method.The effective PF component is selected by the steepness criterion and reconstructed.Finally,The reconstructed signal is demodulated by envelope demodulation method to get the spectrum,and the spectrum characteristic frequency is extracted to realize the fault detection of rolling bearing.(3)In order to master the fault evolution process and damage deg ree of rolling bearing.A method for evaluating rolling bearing fault damage degree based on ICIELMD,fine composite multi-scale scatter Entropy and GK clustering is proposed.Firstly,the implementation process and performance of fine composite multi-scale dispersion entropy are introduced,and the ICIELMD and GK clustering algorithm are used to identify different damage degree of rolling bearing.Finally,compared with CIELMD-RCMDE-GK,and LMD-RCMDE-GK,the effectiveness and feasibility of the proposed method are verified.