Blackout prediction in interconnected electric energy systems considering generation re-dispatch and energy curtailment

Blackouts or cascading outages are costly events that threaten the integrity of electric energy systems around the world. Controlled splitting is executed as the last countermeasure to reduce the undesired economic and social consequences of a blackout. In this paper, a new two-stage scheme is proposed to predict the risk of a blackout in electric energy systems. In the first stage, the boundaries of electric islands are determined using a Mixed Integer Non-Linear Programming model that minimizes the cost of generation re-dispatch and load curtailment. In the second step, a data-mining technique is perfected to predict the risk of electrical separation of an electric island from the rest of the network. Each predictor is trained based on the phasor-measurement data taken at the synchronous generator terminals. Using a wide-area measurement system, the required phasor measurements are gathered and processed in the Energy Management System. Various scenarios, including the island and non-island conditions, are generated and then utilized by the decision-tree classification technique to predict the risk of a blackout. The proposed algorithm is simulated over the IEEE 39-bus test system to demonstrate its performance in online applications.