Novel technique for precise derating torque of induction motors using ANFIS

Area of Science:

• Electrical Engineering
• Artificial Intelligence

Background:

• Induction motors (IMs) face operational abnormalities like voltage unbalance and harmonic distortions.
• These issues necessitate torque de-rating for safe and efficient operation, but traditional methods are complex and time-consuming.

Purpose of the Study:

• To introduce the Adaptive Neuro-Fuzzy Inference System (ANFIS) as a predictive tool for de-rated torque estimation.
• To quantify de-rating factors (DFs) and analyze their impact on IM performance.
• To offer a simplified, real-time solution for torque de-rating assessment.

Main Methods:

• Utilized MATLAB/Simulink for performance analysis and simulation of de-rating factors.
• Developed and implemented an ANFIS controller integrating neural networks and fuzzy logic.
• Trained the ANFIS controller offline and validated its predictions against Simulink outcomes.

Main Results:

• ANFIS demonstrated reliable and accurate de-rated torque predictions with minimal deviations from simulation results.
• Identified voltage unbalance and harmonic distortions as primary causes of reduced torque.
• Showcased temperature rise as a factor exacerbating power losses and thermal stress.

Conclusions:

• ANFIS provides a simplified, rapid, and precise method for assessing IM torque de-rating under abnormal conditions.
• The ANFIS controller optimizes motor efficiency and extends operational longevity.
• This research offers a practical framework for de-rating torque evaluation and management in industrial settings.