Rolling Bearing Fault Diagnosis Based on Improved VMD And GA-ELM

When the measured vibration signals of rolling bearings are decomposed using Variational Modal Decomposition (VMD), the number of modes K and penalty factor α should be manually set, otherwise they will influence the accuracy of the signal decomposition result, which has certain contingency. Meanwhile, Extreme Learning Machine (ELM) has the problem of unstable network structure. In view of above problems, a successive approximation method to optimize the VMD algorithm was proposed in this article, combining ELM optimized by the genetic algorithm (GA-ELM) to diagnose and classify the four states signals of rolling bearings. First, a set of initial parameters K and α are determined according to the amplitude-frequency characteristics. Then, the optimal parameters are determined using the successive approximation method. Finally, the information entropy of each component is calculated and sent into GA-ELM as the feature for modal recognition. Experimental results show that the method used in this article can classify more than 95% of faults in rotating bearings.