Research on Fault Diagnosis of Rotational Automaton Based on VMD-ELM

Abstract

Due to complex operating environment of automat, superposition of various response signals, in order to accurately, efficiently extract fault characteristics of automat signal, a automat fault analysis method using VMD and ELM was proposed. First automat signal was analyzed using VMD and compared with the result of EMD; meanwhile energy percentage of every modal component and sample entropy of different samples under various operating condition were extracted as characteristic values; extracted characteristic values were input into ELM for fault diagnosis and compared with the diagnostic result of traditional double-spectrum analysis. Finally, VMD method achieved adaptive subdivision of every component in frequency domain of signal and concluded that accuracy rate of ELM fault diagnosis is 87.5%. result of the experiment showed that VMD can effectively avoid mode aliasing and test feasibility and effectiveness of proposed method.