Torque estimation in three-phase induction motors under broken bars and voltage variations using airgap flux measurement and a machine learning model

Abstract

This paper proposes a new approach for load torque estimation in squirrel cage induction motors using airgap flux measurement, by means of a Hall effect sensor installed between two stator slots of the machine. The rotor speed was estimated from the Hall sensor signal using a previous method designed by the authors, and the root mean square and mean value were also computed from the Hall signal to serve as inputs to a multi-layer perceptron model.The positioning of the Hall sensor inside the machine was also investigated during different experiments for torque prediction and considered as an additional input for the regression task. The present methodology does not require motor parameters for torque prediction. This research has been validated using some simulated and experimental results for different motor operational conditions. The performance metrics disclosed a good torque prediction for the machine fed not only by a rated voltage, but also by overvoltage/undervoltage power supply and with broken rotor bars, even for low loads. Both experimental and simulation results showed a mean absolute percentage error lower than 4% for a 7.5 kW induction motor.