Remaining Useful Life Estimation of the Motor Shaft Based on Feature Importance and State-Space Model

摘要

Induction motor is widely used in industry as a prime mover of the machine or mechanical equipment. The monitoring of this component is needed to assure that it works in its optimal performance and to prevent its sudden failure. The potential failures of the induction motor can be an electrical failure or mechanical failure. One of the mechanical failures is shaft failure. The failure of the motor shaft will cause the motor will not function properly. The objective of this paper is to estimate the RUL of the motor shaft based on Feature Importance (FI) and the state-space model. The FI of the feature is determined from the monotonicity and trendability criteria. Features with the high FI score then be used as the health indicator for RUL prediction. The RUL estimation was performed to the health indicator using the state-space model. The result shows that the state-space model can be used for the motor shaft RUL estimation satisfactorily before actual failure happened.