Ultrafast Active Response Strategy against Malfunction Attack on Fault Current Limiter

Fault current limiter (FCL) is a crucial device to guarantee the corresponding breaker can clear the fault with over-limit short-circuit current (SCC). If a FCL is disabled by malicious cyber-attackers, the breaker will fail to clear the fault. Under this circumstance, the breaker failure protection system will be activated to trip all the connected transmission lines, resulting in a severe N-k contingency. In order to deal with the breaking failure caused by FCL malfunction attack, we propose an ultrafast active response strategy (UFARS) based on deep neural network framework. Firstly, an Long Short-Term Memory (LSTM) network is designed to extract features from the time sequence of SCC. Then, a deep reinforcement learning (RL) based approach is proposed to determine the optimal strategy to clear the fault. More specifically, with the features from the LSTM network, the deep RL will select and trip a combination of transmission lines connected to the faulty line, such that the SCC will become smaller than the breaking capacity. Thus, the fault can be cleared with the corresponding breaker. After the fault is cleared, the tripped connected transmission lines will be reclosed to recover the integrity of power system. An environment is established to simulate the power system dynamics and obtain the essential data to train the proposed deep neural network. Numerical results show that the proposed strategy can clear the fault with the minimum impact on power system stability.