Dynamic Model-Aided Intelligent Fault Diagnosis Method for Brushless DC Motor-Reducers

Abstract

Brushless DC (BLDC) motor-reducers, valued for their high efficiency and controllability, are widely used in industrial automation, consumer electronics, and new energy vehicles. Reliable operation and accurate fault diagnosis are of great significance using artificial intelligence (AI) based fault diagnosis. However, the practical performance is often hampered by the scarcity of fault samples, leading to undesirable diagnostic models. To address this challenge, this article proposes a novel intelligent diagnostic method driven by dynamic model-based fault sample generation. First, a dynamic model of a healthy BLDC motor-reducer is established and then iteratively updated by improving the consistency between the simulation signals and the measured signals. Second, typical fault patterns of the motor-reducer are mathematically modeled and injected into the dynamic model of the healthy motor-reducer to generate abundant simulation signals as fault samples. Third, experiments are conducted on representative AI models in the scene of insufficient fault samples. Results demonstrate that the proposed method significantly enhances diagnostic performance when certain fault samples are scarce or even missing.