Sensorless Direct Field-Oriented Control of Induction Motor Drive Using Artificial Neural Network-Based Reactive...

Area of Science:

• Electrical Engineering
• Control Systems
• Artificial Intelligence

Background:

• Sensorless control of induction machines (IM) is crucial for efficient electric drives.
• Conventional model reference adaptive system (MRAS) estimators face challenges with parameter variations and regenerative modes.
• Artificial neural networks (ANNs) offer potential for enhancing adaptive system performance.

Purpose of the Study:

• To present an advanced sensorless control method for IM using a Q-MRAS estimator integrated with a feedforward artificial neural network.
• To demonstrate the improved robustness and parameter independence of the proposed control strategy.
• To validate stable operation of the proposed method, especially in the regenerative mode.

Main Methods:

• Development of a Q-MRAS estimator incorporating a feedforward artificial neural network for IM speed estimation.
• Simulation studies using MATLAB/Simulink to validate the proposed control algorithm.
• Real-time implementation on a TMS320F28335 Digital Signal Controller with a LabVIEW interface.

Main Results:

• The proposed Q-MRAS with ANN demonstrated superior robustness and reduced dependence on IM parameters compared to conventional Q-MRAS.
• Stable operation was achieved in the regenerative mode, a common challenge for sensorless IM drives.
• Experimental validation on a 2.2 kW three-phase IM drive confirmed the simulation findings.

Conclusions:

• The integration of ANNs into the Q-MRAS estimator significantly enhances the performance of sensorless IM control.
• The proposed method provides a robust and reliable solution for IM drives, particularly in applications involving regenerative braking.
• This advanced control strategy holds promise for improving the efficiency and stability of electric motor systems.