Application of machine learning for inter turn fault detection in pumping system.

Area of Science:

• Engineering
• Electrical Engineering
• Mechanical Engineering

Background:

• Pump fault diagnosis is critical for operational safety and cost reduction in industrial applications.
• Inter-turn faults in three-phase induction motors significantly impact pump performance and energy consumption.
• Timely fault detection minimizes maintenance costs and prevents energy waste.

Purpose of the Study:

• To analyze the effects of inter-turn faults on three-phase induction motor parameters in centrifugal pumps.
• To develop and compare machine learning models for accurate fault detection in pump systems.
• To evaluate the suitability of Artificial Neural Network (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) for pump fault diagnosis.

Main Methods:

• Utilized a Simulink model based on mathematical equations to simulate inter-turn fault conditions.
• Employed machine learning algorithms, specifically ANN and ANFIS, for fault detection and analysis.
• Validated simulation results using a hardware-in-the-loop (HIL) simulator.

Main Results:

• Inter-turn faults were shown to cause substantial current increases, affecting motor and pump parameters.
• ANN and ANFIS models demonstrated satisfactory accuracy in training and testing phases.
• Performance metrics including RMSE, R-squared, and prediction accuracy were used for model comparison.

Conclusions:

• ANN and ANFIS models are effective tools for detecting inter-turn faults in centrifugal pump motors.
• The developed models offer accurate fault diagnosis, contributing to improved pump reliability and maintenance.
• Further comparison with various supervised algorithms confirmed the suitability of ANN and ANFIS for this application.