A novel predictor for areal blackout in power system under emergency state using measured data

Power systems are always under the risk of various disturbances threating their stability and integrity with the potential for blackouts. Due to unforeseen contingencies, uncertain renewable resources and insufficient corrective measures, the pre-fault dynamic security assessment (DSA) methods cannot guarantee full security under alert operating state of power systems, and pulling to emergency state becomes inevitable. Furthermore, these methods provide a global view of the security of whole power systems. To address the limitations of DSA-based methods, this paper proposes an agile approach to prevent the propagation of blackouts in the power network, requiring a local identification strategy based on blackout areas. An online power system blackout predictor (PSBP) is designed to predict the amount of future potential blackouts (AFPB) under emergency conditions as a supplement to traditional DSA methods. This approach introduces a new paradigm for system vulnerability assessment. For each blackout area, the PSBP employs decision tree-based ensemble algorithms. By continuously tracking incremental changes in the dominant transient operating variables that reflect disturbances and system vulnerability, the PSBP can predict the potential for a blackout area to progress toward a full blackout. The vulnerability index, calculated based on incremental changes observed in the moving window of snapshots (MWS), is updated incrementally. This index serves as input to the PSBP to predict the likelihood of a blackout in each area. The method is demonstrated using the IEEE 39-bus test system, and the results highlight its potential as a valuable tool for assessing the future vulnerability of power systems based on historical behavior data.