Research on cascading fault-guided search of high-voltage lines based on GCN-LSTM

Searching for possible cascade failures in advance can be used for preventive control to prevent power outages. However, the randomness and unpredictability of faults make it difficult to search and predict cascaded faults. On the other hand, guided search can efficiently find out potential high-risk routes and improve fault prediction and search efficiency. Based on the grid graph model and the temporal and spatial characteristics of cascade fault propagation, a GCN-LSTM based cascade fault guided search method for high-voltage lines is proposed in this paper, which can be used to locate highly vulnerable nodes and vulnerable branches. The details are as follows: Firstly, the improved OPA model describing the cascade fault is given, which can be used to construct the cascade fault data; Secondly, a cascade fault search framework and a high-vulnerability line search guidance algorithm are proposed. Thirdly, the GCN-LSTM structure of high-voltage line cascade fault guided search is designed. In conclusion, the guided search performance of GCN, LSTM, CNN, CNN-LSTM, and GCN-LSTM models was analyzed and compared using the IEEE-RTS39 test system. The findings demonstrate that the GCN-LSTM structure exhibits superior efficiency and accuracy compared to the other models. Consequently, the GCN-LSTM model holds greater practical application value in guided search scenarios.