Abstract:

Modified Winding Function Theory (MWFT) offers a practical compromise between the simplicity of equivalentcircuit model and the high fidelity of finite-element analysis for induction motor fault analysis. However, the model requires a large number of geometric, electrical, magnetic, and mechanical parameters, many of which are unavailable for inservice machines. This paper presents a two-stage parameteridentification workflow for MWFT models when only nameplate data and a small set of benchmark measurements in healthy motor state are available. First, physically plausible initial values are derived from standard induction motor design rules. Second, the parameter set is refined by Differential Evolution to minimize the mismatch between simulated and expected current and speed. The method is evaluated on two squirrel-cage induction motors and then used to simulate rotor eccentricity as well as brokenbar faults. The results show that nameplate-based initialization already reproduces the characteristic fault frequencies, while measurement-assisted refinement improves agreement in both frequency location and amplitude. The proposed workflow makes MWFT-based fault modeling more practical for motors whose detailed design data are not available.