Vibration Analysis Techniques for Automotive Bearing Fault Diagnosis: A Literature Review of Conventional and Intelligent Methods

Abstract

Bearing fault diagnosis is critical to the reliability of wheel hubs, electric traction motors, alternators, and transmissions because incipient defects generate weak vibration signatures that can be masked by changes in speed, load, and structural noise. This review aimed to analyze vibration techniques published between 2022 and 2026 for detecting and characterizing bearing faults, emphasizing their transferability to automotive applications. A narrative literature review with a structured search was conducted, and an analytical corpus of 19 DOI-verified works was organized into conventional and envelope processing, time-frequency and decomposition techniques, machine/deep learning, sensor fusion, and predictive-maintenance reviews. The findings indicate that time-domain indicators are suitable for trending and alarms, whereas envelope analysis retains an interpretability advantage by linking impacts to bearing kinematic frequencies. Adaptive and time-frequency techniques are more capable under non-stationary signals but remain sensitive to parameter selection. Intelligent models improve classification capacity, although generalization decreases when training and testing do not represent different vehicles, sensors, and operating regimes. Recent evidence therefore supports hybrid strategies that combine speed reference, physically interpretable preprocessing, and intelligent classification, particularly when vibration is fused with current, sound, or temperature.