The calculation of core loss of power electronics converters requires a large number of core loss data, including the excitation frequency, flux density ripple, and DC magnetic field bias characteristic under rectangular voltage excitation. However, in the case of the conventional method (that employs a B-H analyzer with a chopper circuit), a very long time is required to collect this core loss base data, because circuit parameter tuning is needed for each measurement point. Therefore, we propose a high-speed core loss base data collection method for core loss calculation under power electronics converter excitation. The core loss base data collection time was substantially reduced upon usage of a pulse width modulated (PWM) inverter and a neural network because various core loss characteristics were expressed upon excitation of the PWM inverter, and these core losses were learned by the neural network. The core loss base data, as measured via the proposed and conventional methods, were compared to validate the measurement accuracy of the proposed method, and the difference was found to be lesser than 12%. Thus, the proposed method achieves high-speed core loss base data collection.