Mitigating Erroneous PMU measurements via an ANN to Enhance ML-based Transient Stability Prediction Accuracy

In our study, datasets of two simulators, namely phasor-based simulator and hybrid-type simulator are used. In the hybrid environment, first, the outputs of the phasor-based simulator are converted to instantaneous waveforms, then based on instruction, distortions and noises are added (superimposed) to these waveforms, and finally, the distorted waveforms are fed to the detailed model of PMUs simulated in EMT domain. Outputs of both simulators can be found in the submitted file. The outputs of the phasor-based simulator including voltage magnitude, voltage angle, and voltage frequency were named “Bus_Mag”, “Bus_Ang”, and “Bus_Freq”, respectively. The format of these files is cell array and each cell comprises a 39*17 single-precision array, where 39 is the number of buses that will be equipped with PMUs and 17 is the number of phasors quantities including pre-fault, during-fault, and post-fault data. On the other hand, the outputs of the hybrid simulator including voltage magnitude, voltage angle, and voltage frequency were named Bus_Mag_Hybrid, Bus_Ang_Hybrid, and Bus_Freq_Hybrid respectively. This file only comprises the first 5000 samples. The format of the aforementioned datasets is cell array and each cell comprises a 39*17 single-precision array, where 39 is the number of PMUs and 17 is the number of PMU measurements including pre-fault, during-fault, and post-fault measurements. The double-precision array “TSAT_Index” contain power angle-based stability index for each scenario, which can be obtained by observing the maximum angle separation between any two generators at the same time in post-fault response, to determine if any generator in the system is out of synchronism.