The defect in the stator winding turns causes a fluctuation in stator resistance, which leads to an incorrect assessment of the stator flux location, which can lead to the failure of the entire drive system. This study describes a novel artificial neural network (ANN) approach for identifying stator short-circuit failures in three-phase induction motors utilizing feature extraction and categorization. Delayed stator current signals are used in the first stage to estimate the mutual information, which is then used as input to decision trees and multilayer perceptron neural networks in the second step. This paper also employs a direct Torque Control (DTC) based fault-tolerant operation (FTO) for the induction motor drive. Voltage imbalance, load torque variations, and short-circuit levels ranging from 1% to 10% are reported in the offline and online experimental tests.