Abstract:In view of the fact that the massive data in the current power grid operation has not fully mined effective information, and the dispatching information is heterogeneous and unstructured from multiple sources, power dispatching cannot obtain effective fault information in a timely manner, resulting in a reduction in the efficiency of dispatching and processing of power dispatching during emergency repair work, and fault information The delivery is not timely or accurate, etc. This paper proposes a power grid fault handling knowledge graph construction technology based on the attention mechanism BiLSTM-CRF model. Relying on the constructed power grid fault handling knowledge graph, it can efficiently screen effective information about power grid faults, form an intelligent fault auxiliary decision-making application, and effectively improve the efficiency of power grid fault repair.