Dataset for Early Detection of ITSC Faults in PMSMs

The published dataset contains the following: faulted_PMSM_data.mat - identified flux maps of a faulted Prius PMSM extracted from FEA-simulation (4-D with the size of 11x11x11x61) simulation_data.zip - transient simulation data (6667 records of dq-currents) training_data.zip - DWT features extracted from the simulation data and experimental data ( Wavelet levels for 6667 sample) Transformer_Classifier.ipynb - Transformer model with training loop In the transient simulation the naming convention is the following: healthy: simulation_results_hlth_Mload_XX.X_n_YYY_step_time_ZZZ_speed_step.mat, where X is the load, Y is the speed and Z is the time where the transient happened faulted: simulation_results_Rf_XX_Mload_YYY_n_ZZZ_phW_speed_step.mat, where X is the fault resistance, Y is the load, Z is the speed and W is the faulted phase index ( 1 - U, 2 -V, 3 - W) In the training data, each csv is a given wavelet level for every sample . The "exp" notation means that it is extracted from experimental data. Detailed description can be found in the paper: " Early Detection of ITSC Faults in PMSMs Using Transformer Model and Transient Time-Frequency Features" (https://dx.doi.org/XXX).