Robust backstepping control of induction motors using neural networks

Area of Science:

• Electrical Engineering
• Control Systems
• Artificial Intelligence

Background:

• Induction motors require robust control techniques to maintain performance despite parameter variations.
• Existing methods may necessitate complex dynamical analysis or offline learning phases.

Purpose of the Study:

• To develop a novel, systematic, and robust control technique for induction motors utilizing neural networks (NNs).
• To address challenges posed by parameter variations, unknown load torque, and rotor resistance.

Main Methods:

• A backstepping-inspired approach using two-layer neural networks (NNs) for fictitious controller design and signal realization.
• A new tuning scheme is proposed to ensure boundedness of tracking error and weight updates.
• The method avoids the need for regression matrices and preliminary dynamical analysis.

Main Results:

• The proposed NN control technique demonstrates robustness to parameter variations in induction motors.
• The method guarantees boundedness of tracking error and weight updates.
• Implementation requires full state feedback, but load torque and rotor resistance can be unknown but bounded.

Conclusions:

• The presented NN-based robust control offers a systematic and efficient solution for induction motor control.
• The technique simplifies implementation by eliminating the need for regression matrices and offline learning.
• This approach enhances the reliability and adaptability of induction motor systems.