电网智能化调度与控制教育部重点实验室(山东大学), 山东省济南市 250061
Key Laboratory of Power System Intelligent Dispatch and Control of Ministry of Education(Shandong University), Jinan 250061, China
Rapid development of large-capacity DC transmission systems and emergence of multi-infeed DC systems increase the risk of cascading failures. A fast judgment model for DC blocking based on deep learning is established and a fast search method for high-risk cascading failures is proposed. The steady-state features related to the network structure and the fault location are selected as inputs. The stacked denoising autoencoder(SDAE)is utilized to extract high-order features of the inputs. The influence of an AC failure on other lines is measured according to the load rate of normal lines after this failure and the line outage risk is defined based on the influence degree. The depth first search(DFS)strategy is adopted in the search process, which takes the high outage risk as the search direction. AC/DC cascading failures with high failure probabilities can be screened preferentially. Simulation results show that the proposed method can quickly provide the propagation paths and failure probabilities of high-risk AC/DC cascading failures, which can be used for the online security early warning and preventive control decisions.
ZHU Yuanzhen, LIU Yutian. Fast Search for High-risk Cascading Failures Based on Deep Learning DC Blocking Judgment[J]. Automation of Electric Power Systems, 2019, 43(22):59-66. DOI:10.7500/AEPS20190429001.