The invention discloses a typical daily load characteristic-based power load forecasting method, including: collecting power load data of users in a set industry in a set area, and performing preprocessing on the data; obtaining typical daily load curves of the users and an industry overall load curve; obtaining a correlation coefficient of the typical daily load curves of the users; classifying the users according to a power utilization laws of the users; and selecting a user whose power load change trend has the highest consistency with an industry overall load change, and forecasting power load data of an overall industry through the power utilization behavior laws of the user. The typical daily load characteristic-based power load forecasting method provided by the invention has the beneficial effects of being capable of deriving a forecast of an abnormality of a specific industry and even a whole society power load abnormality from an abnormality of a certain user power load, improving accuracy of a power load forecasting result, and providing a power guarantee for reasonable economic dispatching of an electric power system, reduction of production cost, and prevention of a large-area power failure of a power grid or a power grid breakdown.