AbstractThe induction heating model described herein couples the standard heat conduction equation with electro‐magnetic proximity‐skin equations. An Inverse Finite Element procedure, which is based on prior deterministic and probabilistic concepts, has been designed to solve the inherent inverse equation model with respect to the unknown coil current parameter. Simulated experiments using different noises in the input data have been performed in order to determine their influence on the estimated parameter. The IFEM has shown its capability to predict the optimal location for the temperature sensors, together with their numbers, consistently with a pre‐specified estimate accuracy. Specifically, only one temperature sensor, located in the middle of the two turns of the coil, results to be sufficient to estimate the unknown parameter to a satisfactory accuracy degree. This, may significantly help to design optimal experiments.