Hysteresis Characterization Dataset

To simulate the hysteresis behavior of oriented silicon steel sheets under mixed complex excitation conditions, we measured the corresponding hysteresis data and constructed the dataset using an experimental measurement platform. The hysteresis dataset is obtained by superimposing heterodyne voltage signals with varying amplitudes onto the industrial frequency voltage excitation. The amplitude of the heterodyne signal voltage is determined by the magnetic flux density amplitude (Bm), which ranges from 0.2 to 1.8 T with increments of 0.1 T. The frequency of the heterodyne signal voltage is measured within the range of 20-100 Hz with intervals of 10 Hz, and the modulation depth (m) of the intermediate frequency (IF) signal amplitude relative to the heterodyne signal varies at 5%, 10%, 20%, and 50%, respectively.

The dataset consists of the magnetic induction and field strengths B(k-1) and H(k-1) at the previous time step, the magnetic induction strength B(k) at the current time step, flux amplitude Bm, magnetization frequency f , and the percentage m of the industrial frequency signal. The output is the magnetic field strength H(k) at the current time step. Each sampling window comprises 1,000 data pairs, resulting in a total of 514,000 samples in the dataset. The dataset will be partitioned into training and test sets for curve prediction and evaluation, respectively. Specifically, the test set comprises flux amplitudes of 0.4, 0.6, and 1.0 T, while the training set encompasses the remaining flux amplitudes. Additionally, the training employed six hysteresis loop families with frequencies ftrain of 20, 40, 60, 80, 90, and 100 Hz, while model testing utilized frequencies ftest of 30 and 70 Hz.