DNN-based Contingency Screening Module for Voltage Stability analysis

Abstract

Fast and accurate contingency screening (CS) has become a key enabler for secure operation of the power system. This is due to market activities, complex controls, and power supply intermittency that is caused by the integration of renewable energy sources. This paper proposes an online CS scheme for power systems voltage stability analysis (VSA) using deep neural networks (DNNs). The DNN model receives a snapshot of the power system status from state estimator. This snapshot contains information about the current topology of the system, the voltages at different buses and the loading of lines and generators. The model is trained to classify the state of the system as secure (stable) or insecure (unstable) under different system loading and contingency conditions. Three power system security constraints were considered: (1) the MVA loading of lines and generators is less than 110% of its rated value; (2) the voltage magnitude at the buses is within limits; and (3) the power flow solution is converged. Violating any of these conditions, the power system is considered insecure. Contingencies that lead to insecure operation are sorted in a list based on the number of violated conditions for further analysis. The proposed scheme is tested on the ISO New-England IEEE 39 bus system. The test results show that the proposed scheme is suitable for online applications.