With ITER's re-baseline, the Electron Cyclotron Heating system has become of even higher importance with the multiple purposes of plasma heating, current drive, and instability suppression. ITER's Electron Cyclotron Heating system is based on several gyrotrons, such as radio-frequency sources supposed to provide on-demand power in the order of megawatts in the millimeter-wave domain. Due to the non-linear dynamics involved in the gyrotron operation and the structure of the machine itself, controlling a gyrotron and preparing it in the shortest time possible to provide radiofrequency power represents a challenging task for the tokamak operations. Reliable models of the power source are essential to design effective control strategies. To this aim, we further developed a semi-physical thermal lumped model of the ITER European Gyrotron's Magnetron Injection Gun available in literature. The model has been re-implemented in C++ for better computational performances and therefore making it particularly suitable to be exploited by data-driven model identification algorithms and real-time control codes. In this paper the model parameters are identified by using a Particle Swarm Optimization algorithm on experimental data and the model validation is finally carried out on a dedicated dataset, aiming at providing a tool to design and to be integrated into the control system of the machine.