Defect Detection in Rotor-Bearing-Gear Systems for Experimental Frequency-Domain Fault Diagnosis using FFT and Machine Learning Techniques

Abstract

Rotor bearing gear system is mostly used in rotating machinery in industries and has considerable susceptibility to defects such as wear, pitting, and failure in gears, corrosion of components, and damage to bearings under different operating conditions. This paper presents an experimental study of fault diagnosis implemented in the frequency domain using the Fast Fourier Transform (FFT) and Machine Learning Techniques. In the Experimental study, the bevel gearbox test rig was exercised under various operating conditions (200 rpm to 1000 rpm, loading (20 N to 100 N)), and six types of faults were exercised for each case, including mutually imposed multiple fault conditions. For each fault condition, FFT analysis identified characteristic fault frequencies, harmonics, sidebands, and modulation effects were found by FFT analysis. We observed that as the fault level, speed, and load increased, vibration amplitude increased significantly. Frequency-domain vibration features were used in this study with machine learning classifiers to determine fault conditions, demonstrating the effectiveness of the proposed method for condition monitoring and predictive maintenance applications.