MONITORING AND DIAGNOSTICS OF FAULTS IN THREE-PHASE INDUCTION MOTORS USING NARX NEURAL NETWORK
Faults, Artificial intelligence, Induction motor
Three-phase induction motors play a crucial role in industrial operations. However, their failures can lead to serious operational issues. This study focuses on the early detection of failures through accurate diagnostics and the classification of faults in three-phase induction motors using Artificial Intelligence (AI) techniques. The analysis involved current, temperature, and vibration signals. Experiments were conducted on a test bench simulating real-world operating conditions, including stator phase imbalance, bearing damage, and shaft imbalance. For fault classification, a Nonlinear Autoregressive Network with Exogenous Inputs (NARX), a type of predictive network, was developed and adapted through parameter adjustments to perform classification tasks. The optimal network configuration was determined through a selection process using a grid search method with multiple training iterations, followed by the introduction of new data for validation of its efficiency. The test results for new data demonstrated that the network performed excellently and generalized across all evaluated scenarios, achieving the following accuracy rates: 'No Defect Motor' = 94.2%, 'Imbalance Defect' = 95%, 'Stator Defect' = 98%, and 'Stator Failure' = 95%