Fault prediction and damage determination method for brushless DC motor based on ES-GM

Brushless DC (BLDC) motor has been widely used in various fields like automotive, aerospace etc., due to its low noise, high efficiency and reliability. However, it is because of BLDC motor's high reliability and longevity that degradation problems arising from the deterioration of its structures/components in the long-term operation are often overlooked. This problem will directly affect its application systems' stability, reliability and safety, so it is necessary to perform health state evaluation and fault prediction for BLDC motor. To achieve this goal, vibration monitoring technique is adopted to analyze failure modes of BLDC motor based on fault tree analysis. Frequency-domain energy of vibration is taken as characteristic to represent degradation state and time series of motor's degraded state is obtained by feature extraction and other approaches. To rapidly and accurately predict BLDC motor's health condition, a comprehensive prediction model (Exponential Smoothing-Gray Model, ES-GM) is proposed in this paper on the basis of full use of their respective advantages. Through testing, the proposed model has higher prediction accuracy than two single models, which can be well used in early fault prediction for BLDC motor. Finally, a quantitative damage determination model is proposed to quantitatively analyze the damage degree of faults in BLDC motor based on predicted results of GM(1,1) and ES-GM. This research can greatly help to evaluate motor's health state, formulate maintenance strategies and reduce safety risks.