Context-aware voltage sag waveform decomposition method using hybrid models for complex power quality detection

Abstract

To analyze the voltage sag characteristics of distribution networks, automotive companies need an accurate and efficient non-stationary signal analysis tool to extract voltage sag feature information from power quality monitors and fault recorders, which is crucial for automotive companies to analyze grid characteristics and improve localized charging strategies. We innovatively propose a hybrid decomposition method to decompose voltage sag waveforms from composite disturbance data. Subsequently, we design a context-aware mechanism to adapt the proposed decomposition method to the complex power quality data in different regions. In addition, tests were conducted on an integrated power quality simulation system to verify that this method can accurately decompose the voltage sag waveform from the composite disturbance signals. Finally, the experimental simulation results also show the better performance of the method in decomposing voltage sag waveforms with different causes in different composite disturbance signals compared with other mainstream methods.