CNN‐LSTM based power grid voltage stability emergency control coordination strategy

Abstract

Abstract The occurrence of cascading failure and small probability contingency in hybrid AC/DC power grid aggravates the mismatch risk of traditional emergency control with offline‐predetermination–online‐practice (ODOP) mode. To ensure the voltage stability of power grid under large disturbances, this paper proposes a voltage stability emergency control coordination strategy based on convolutional neural network (CNN) and long short‐term memory (LSTM) network. First, the mismatch mechanism of traditional emergency control is revealed under the variation of operating conditions. Second, the start criterion of complementary emergency control with ODOP mode is proposed, and the CNN‐LSTM network is established to quantitatively evaluate the voltage stability margin. Finally, the emergency control sensitivity index is proposed to predict the stability margin enhancement under alternative emergency control measures, and the optimal ODOP‐based emergency control strategy is determined to coordinate multiple voltage stability emergency control measures. Case studies are performed in the actual Northwest China local region hybrid AC/DC power grid with voltage instability problems, and the simulation results verify the effectiveness of the proposed method.

Keywords

• artificial intelligence
• power system control
• power system stability