Bearing Fault Diagnosis Based On Esmd And Modified SVM

As the size and complexity of the machine increase exponentially,the stability of a part of the machine more and more affects the overall machine.Bearings are the key components of rotating machinery,however,due to its special structure,working environment and other reasons,the failure rate is high.Therefore,it is very important to study mechanical fault diagnosis technology to carry out early warning and diagnosis of rolling bearing faults,and it is of great significance to ensure the reliability and safety of industrial production.A fault diagnosis method of rolling bearing based on Extreme-Point Symmetric Mode Decomposition(ESMD)and Modified Gravitational Search Algorithm Support Vector Machine(MGSA-SVM)is proposed in this paper.First,based on the analysis of the causes of the typical failure of the bearing,aiming at the shortages of the current bearing fault diagnosis technology,the improvement proposal of the technology is put forward.Then,aiming at the nonlinear and non-stationarity of the bearing vibration signal,the ESMD decomposition vibration signal method is proposed.Based on the inherited EMD algorithm,the signal can be decomposed according to the frequency.The optimal adaptive global curve is introduced to determine the number of decompositions and the interpolation algorithm is fundamentally improved.Comparing to the EMD algorithm,it performs better in some cases.Then,from different perspectives,the characteristics of the vectors decomposed by the vibration signals are calculated,and three kinds of variables,sample entropy,energy and kurtosis,are introduced to make the features of each component more prominent.In the fault classification,the kernel function coefficients and penalty coefficients that are not easy to be determined manually for Support Vector Machine(SVM),as well as the gravitation search algorithm is with the common problems such as the metaheuristic optimization algorithms,an improved optimization method MGSA is proposed.Through the simulation test,comparing MGSA with the original GSA algorithm and PSO algorithm,the training time and the classification accuracy have been improved.Finally,based on the fault data published by Case Western Reserve University and data obtained from Shanghai Baosteel,the effectiveness of rolling bearing diagnostics based on ESMD and an modified GSA-optimized SVM was tested.Finally,good results were obtained.