Common-mode inductors are critical components of conducted emission filters used to ensure electromagnetic compatibility in electrical systems. In compact filter designs, the leakage inductance of the common-mode inductor can be exploited to assist the filtering of differentialmode noise. However, accurately estimating this inductance is challenging due to the nonlinear behavior of the associated leakage magnetic fields. This work investigates the use of machine learning models to estimate the leakage inductance of three-phase common-mode inductors. The development followed the Cross-Industry Standard Process for Data Mining (CRISP-DM) methodology and used a dataset generated through finite element simulations. Two machine learning models were developed using geometric parameters as input features: a linear regression model and an artificial neural network model. The models were evaluated using out-ofsample test data and compared with a commonly used analytical formulation. Both machine learning approaches demonstrated improved predictive performance for the configurations analyzed, with the artificial neural network achieving the highest accuracy while the linear regression model also provided satisfactory results. A prototype web-based software tool embedding the trained neural network model was developed to demonstrate a practical way of applying the proposed approach.     «

Common-mode inductors are critical components of conducted emission filters used to ensure electromagnetic compatibility in electrical systems. In compact filter designs, the leakage inductance of the common-mode inductor can be exploited to assist the filtering of differentialmode noise. However, accurately estimating this inductance is challenging due to the nonlinear behavior of the associated leakage magnetic fields. This work investigates the use of machine learning models to estimate...     »