Research Of Rotating Machinery Fault Diagnosis Based On Resonance-based Sparse Signal Decomposition

Rotors,bearings and gears are the major components of rotating machines,their working conditions are significant to the overall operation.The traditional signal processing methods have only concerned about the running of some specific component,neglecting the analysis of the total system and compound failures.Under this circumstance,this paper surveys the fault signal of the system,intending to improve the extracting and splitting competence of the signal components.Conclusions are shown as below.(1)The optimizing method for high resonance Q factor in Resonance-based Sparse Signal Decomposition(RSSD)is proposed based on PSO.The Tunable Q-factor Wavelet Transform(TQWT)pre-decomposition method is designed for the selection of RSSD high resonance Q factor.Mutual Correlation of the decomposition and original signal is set as the fitness function searching for appropriate Q factor.Simulation signal of rotor bearing system is applied for providing the distinction efficiency of redundant and impulse signal.(2)The optimizing method of RSSD weight parameter is proposed based on fault feature ratio,which is used here for the weight parameter selection of RSSD.Aiming at confirming the energy assignment of RSSD.Inner and outer race failure signal of bearing weak faults are chosen here.Comparison with the original RSSD is shown to verify the efficiency of the proposed method.(3)Minimum Entropy Deconvolution(MED)is used for improvement of low resonance component,solving the matching problem of wavelet base function with amplitude modulation and frequency modulation gear fault signal.The detection ability of impulse signal is proved by the comparison with original sub-band analysis.(4)The combination method of high resonance Q factor selection with low resonance MED processing is proposed for gear bearing system diagnosis.Test bed signal and field data are chosen here for proving the efficiency of the proposed method.