Abstract: Scheduling decision knowledge is usually stored in text files or databases on scheduling procedures, or in the experts' experience. Dispatchers need to rely on the large amount of professional knowledge support, the history data and the real-time grid situation awareness to make optimal decisions in a short time in the light of situation changes. In view of the complexity of the scheduling knowledge and the accuracy of the scheduling decision, an auxiliary decision method for the distribution network faults based on knowledge graphs is proposed. The fault scheduling knowledge map including the scheduling knowledge, the fault processing knowledge and the business process knowledge is constructed by using the power grid scheduling rules, the fault planning and the manual experience knowledge, and the knowledge representation formed by the power grid topology structure is constructed, and the fault planning and the fault processing cases are correlated in the form of event clusters. Then the Artificial Intelligence Markup Language (AIML) and the graph algorithm are combined to realize auxiliary knowledge question answering, case matching or business recommendation for distribution network scheduling faults, and generate the multi-objective distribution network reconstruction strategies through the fault feedback information and the real-time decision-making scenarios. Finally, a friendly and interactive application system for auxiliary decision making for fault dispatching is developed and put into online operation in the distribution network in Changsha city, Hunan Province. The recommendation algorithm and interactive strategy have been effectively verified, which shows that the system can provide fast, intelligent and accurate auxiliary decision support for dispatchers.