Abstract:The rapid development and wide application of smart grid provide power enterprises with massive data with diverse structures and complex sources, making smart grid become one of the most important application fields of big data. How to apply big data technology to achieve data collection, storage and mining, and then improve the level of safe production of power enterprises has become an important research topic. This article first elaborated the basic concept of the smart grid, large data sources, characteristics and big data technology. Then draw lessons from the thought of the traditional "heinrich rule", from the perspective of hidden trouble, use big data technology to process the massive data generated in the safe production of power in a province of China, and then through these data to establish safety accident proportion model of electric power enterprises in the province. Then, through regression analysis, predict the number of potential hidden dangers in the future. Finally, according to the proportion model of safety accidents, the number of possible accidents and accidents in the future is predicted, and determine the control target of the number of hidden dangers, and in the end a set of production safety early warning model is formed in order to eliminate hidden dangers and reduce accidents.