Robust classification model for PMU-based on-line power system DSA with missing data

Wide-area measurement system (WAMS) has been widely deployed in modern power systems to achieve data-driven dynamic security assessment (DSA). Considering the unavailability of measurement data during unintentional events, a robust classification model for phasor measurement unit (PMU)-based DSA is proposed. The proposed approach identifies the observability of different PMU combinations in the system and an ensemble of random vector functional link networks is trained with observability-constrained feature subsets to ensure its viability under any PMU missing condition. Meanwhile, the available operating features are comprehensively utilised by the proposed classification model to support accurate DSA under both normal and abnormal data collection conditions. The proposed classification model is tested on New England 39-bus system with high accuracy and strong robustness. © The Institution of Engineering and Technology.