With the development of the wind power industry in China, data from the field has reached a new stage in terms of sources and magnitude. These data are scattered over each business system, and if they can be managed intensively and applied, they will be helpful to industry. A wind power data management method based on a knowledge graph is proposed. First, knowledge graph elements such as entity, relation and attribute are extracted from multi-source wind power data, and the extraction methods for structured and unstructured data are given. Then those elements are used to construct a panoramic knowledge graph of wind power data in the graph database Neo4j to achieve the connection between different types and business data. Finally, typical cases such as fault reasoning, business query, etc. are compared with the traditional relational database-based data management mode. The results show that the proposed method is good in application.