In today’s society, with the continuous increase in various production activities and residential electricity consumption, the load pressure on the power system is increasing. Therefore, an accurate forecast of the power load can improve the economic cost, safety, and maintainability of the power generation industry. The focus of the research on the load forecasting of the power system is the collection of forecast data and the establishment of accurate forecasting models. At present, various forecasting methods proposed at home and abroad have different characteristics. This paper focuses on the fluctuation characteristics of electricity consumption in different industries and extracts key influencing factors. Then, it uses these extracted influencing factors and historical electricity consumption data to train and predict the Support Vector Regression models, and compares the prediction effects. The validation data set in this paper uses the electricity consumption of various industries in Shenyang. This research contributes to environmental and low-carbon initiatives by accurately predicting power loads, optimizing energy usage, promoting renewable energy development, and facilitating the establishment of intelligent energy systems.