A Data-Driven Framework for Incipient Fault Detection in Transmission Systems: Implementation and Insights

This paper proposes a data-driven incipient fault detection framework based on voltage waveform data collected from GPS-enabled power quality monitors deployed across Taiwan’s southern science parks. The framework integrates a multi-stage process, including waveform-based incipient fault detection, fault identification, feature extraction, and incipient fault classification. By leveraging time-synchronized waveform events and historical fault records, the system identifies incipient fault events and establishes the correlation between incipient and permanent fault events. Based on the reliable classification performance, case studies are conducted to investigate the waveform characteristics of incipient faults across different equipment types. These findings contribute to a deeper understanding of incipient fault behaviors and offer insights for the implementation of incipient fault detection and classification framework.