Prediction of dynamic hysteresis under highly distorted exciting fields by neural networks and actual frequency transplantation

Neural Network (NN) and actual frequency transplantation (AFT) are combined for prediction of dynamic hysteresis when the exciting field, H(t), is highly polluted by harmonics. The NN forecasts the Fourier Series for flux density for well-known H(t) waveforms (i.e., triangular, square wave fields etc.). The task of AFT is to approach the arbitrary distortion of H(t) by exploiting loop predictions by NN under pure sinusoidal excitations and then by transplanting loop branches related to frequencies detected in short time-windows of the H(t) period. These actual frequencies will be evaluated by an appropriate time-frequency analysis of H(t). Model validations will be presented in comparison with experimental data.