Incipient Fault Diagnosis of Rolling Bearing Based on ACMD and Parametric Optimized MNAD

Abstract

For the detection of incipient bearing faults, formulating decoupling noise reduction of vibration signal and fault enhancement strategy is the key to overcoming noise interference and recovering periodic pulses from the original signal. This paper's method of achieving the above diagnosis is based on adaptive chirp mode decomposition (ACMD) and parametric optimized minimum noise amplitude deconvolution (POMNAD). Firstly, ACMD is used to reduce the influence of noise components. Secondly, MNAD, which has an outstanding ability to recover periodic pulses from the original signal, is performed on the noise-reduced component. However, practical applications have consistently shown that the fault characteristic frequency (FCF) parameter of MNAD must be selected adaptively to ensure MNAD works at its best. To solve the parameter setting problem, POMNAD is proposed in this paper. Taking the Gini index of the squared envelope spectrum (GISES) as the fitness function, and iteratively searching through the gray wolf optimization (GWO) algorithm, POMNAD can adaptively obtain the optimal FCF parameter of the filter. After ACMD and POMNAD, the envelope analysis is performed to determine the fault condition. Simulation and experiment indicate that the proposed method correctly detects the incipient fault of rolling bearings.