Magnetic Field Emission Reduction Through Machine Learning Optimization of an Active Compensation Coil

Dynamic wireless power transfer systems for electric vehicles rely on magnetic field coupling between a roadway-embedded transmitter (Tx) and a vehicle-mounted receiver (Rx). While enabling efficient energy transfer, the radiated magnetic field raise concerns related to electromagnetic field (EMF) exposure and electromagnetic interference, particularly with implantable medical devices. Existing mitigation approaches such as conductive shields, magnetic materials, and passive reactive loops are limited by tradeoffs involving efficiency loss, added mass, cost, or insufficient control. This article presents a comprehensive methodology for active magnetic field compensation, based on a machine learning-assisted optimization of the electrogeometrical configuration of an additional compensation (or shielding) coil, connected in series with the Tx coil. The surrogate model guides the design to minimize leaked magnetic flux density in critical exposure zones while preserving power transfer efficiency above 90% under nominal load. Simulation results reveal that the proposed solution reduces peak flux density by up to 70% with a negligible efficiency penalty. The approach is validated through three-dimensional finite-element analysis and experimental prototypes.