The dominant components in the stator current of a typical induction motor are deterministic (i.e., predictable) and the substantial amount of the information they present is not related to bearing faults. In this sense, these deterministic components as “noise” to bearing fault detection problems may create the possibility of missed and false alarms. Which regression and machine learning methods can be used for modeling deterministic components of induction motor current signal?
Dear Weijia Lu,
Thanks for your answer,
The bearing fault signal is non-stationary, thus , in many studies tried to use time-frequency methods. The main idea in our project is to eliminate (cancel out) deterministic components by subtracting the estimated or predicted deterministic components from the measured (actual)signal. We need a suitable neural network, regression , or machine learning method to model the deterministic components.
- Badges
- Science topic