In the analysis of cascading outages and blackouts in power systems, risky cascading fault chains should be accurately identified in order to do further block or alleviate blackouts. However, the huge computational burden makes online analysis difficult. In this paper, an online search method for representative risky fault chains based on reinforcement learning and knowledge transfer is proposed. This method aims at promoting efficiency by exploiting similarities of adjacent power flow snapshots in operations. After the “representative risky fault chain” is defined, a framework of tree search based on Markov Decision Process and Q-learning is constructed. The knowledge in past runs is accumulated offline and then applied online, with a mechanism of knowledge transition and extension. The proposed learning based approach is verified on an illustrative 39-bus system with different loading levels, and simulations are carried out on a real-world 1000-bus power grid in China to show the effectiveness and efficiency of the proposed approach.

中文翻译：

---

基于强化学习和知识转移的代表性风险故障链在线搜索方法

在分析电力系统级联停电和停电事故时，应准确识别有风险的级联故障链，以进一步阻断或缓解停电事故。然而，巨大的计算负担使得在线分析变得困难。本文提出了一种基于强化学习和知识转移的代表性风险故障链在线搜索方法。该方法旨在通过利用操作中相邻潮流快照的相似性来提高效率。在定义了“代表性风险故障链”之后，构建了基于马尔可夫决策过程和Q-learning的树搜索框架。过去运行中的知识离线积累，然后在线应用，具有知识转移和扩展的机制。