163 / 2023-10-22 10:41:57

Parameter-optimized Adaptive Local Iterative Filtering for Bearing Fault Diagnosis

Parameter-optimized adaptive local iterative filtering (ALIF), Local minimum envelop entropy, Singular value difference spectrum (SVDS), Fault diagnosis

Special Sessions > SS13 - Signal Decomposition-Integrated Mechanical Fault Detection and Remaining Useful Life Prediction

A novel methodology is proposed to analyze roller-bearing vibration signals in the presence of strong background noise, especially when detecting weak faulty features obscured by the noise. The approach involves two main steps. Firstly, it employs parameter-optimized adaptive local iterative filtering (ALIF) to obtain Intrinsic Mode Functions (IMFs) of the vibration signal. Particle swarm optimization (PSO) is used to determine the best parameters for ALIF, optimizing the combination based on the local minimum envelope entropy (LMEE) as a fitness function. This helps identify IMF components with distinct fault features via correlation analysis. In the second step, the singular value difference spectrum (SVDS) is applied to the selected IMF component to reduce noise and determine the optimal rank truncation for identifying fault feature frequencies. Experimental and numerical models confirm the effectiveness of this approach in identifying early roller bearing faults in noisy environments. In summary, the method integrates ALIF and SVDS for accurate detection of weak faulty features in complex settings.