Saturation-Enhanced Multiple-Coupled Circuit Model for Simulation-Based Fault Diagnosis of Induction Motors Across Load Conditions

Multiple-Coupled Circuit Models (MCCMs) increasingly generate training data for current-based fault diagnosis of induction motors (IMs). Standard MCCMs use one constant saturation factor, identified at a single operating point, which limits the accuracy of simulated current amplitudes across loads. This paper extends a classical MCCM with a load-torque-dependent saturation factor calibrated at two operating points (no load and nominal load) and validates it on a 1.1 kW IM at 0–120% of the rated load. The extension keeps the simulated fundamental current within about 0.1 A of the measurement at all loads, whereas the standard model deviates by up to 0.86 A. A speed regression from the stator current, trained on simulated data, achieves errors below 5 min−1 on measured data at all loads, compared with several hundred min−1 at low load for the standard model. Relative fault-sideband amplitudes (dBc) differ by less than 0.5 dB between both models. Classifiers using sideband features alone perform equally for both models, whereas adding an RMS-current feature degrades classifiers trained on standard-model data by 1.6 to 19.0 percentage points but changes those trained on extended-model data by, at most, 2.4 points. The extension matters for amplitude-based features and estimators, not for sideband-based fault detection.