Electric energy is a very important energy source, and the economic development of a country cannot be separated from the production of electric energy. This also indirectly reflects the stability of the power system during operation. By predicting the load of the power system, the stability problem of the system can be effectively solved. With the continuous improvement of load forecasting accuracy in the power system, the operating status of the power system has also had a significant impact. Therefore, accurate load forecasting must be carried out considering various factors. This article provided an overview of new power systems under intelligent digital technology. It introduced a new prediction principle and method in the existing prediction model, analyzed the input data of the forecasting model, proposed a load forecasting algorithm, and validated the load forecasting algorithm using a certain region as an example. The accuracy of using the SVR (support vector regression) model load forecasting method was as low as 90.3% and as high as 98%. The experimental results showed that compared with traditional load forecasting methods, load forecasting based on SVR model improves accuracy while reducing the number of training samples. By analyzing and predicting load data, it is possible to predict load characteristics and their changing patterns, which can effectively solve the problems of inaccurate prediction results in traditional prediction methods and lay the foundation for future intelligent digital technology in new power system load forecasting.