Fault Diagnosis for Smart Grids over Knowledge Graph and Abductive Reasoning

Smart grids require structured, precise fault diagnosis knowledge, as well as fast and intelligent fault reasoning methods. This paper proposes a novel fault diagnosis framework for power grids based on knowledge graphs and abductive reasoning. The knowledge graphs for fault diagnosis are first extracted from CIM/XML and IEC 61850/SCD files, containing information on system topology, protection logic, and protection configurations. Temporal constraints are incorporated into the knowledge graph, with graph queries utilized to manage the knowledge effectively. Additionally, an abductive reasoning method is introduced to infer faulted components across the knowledge graphs, involving three key steps: backward reasoning, forward reasoning, and evidence fusion. Finally, a test system featuring multiple faults and protection device malfunctions is examined to illustrate the method's efficacy.