The short-term power load forecasting of the power system is an important part of the power grid management system. The prediction accuracy directly affects the stable and safe operation of the power grid, and the weather becomes more and more important in the load forecasting. This paper takes a city load as an example to study the short-term load forecast, and establishes the SVR prediction model which consider the relevant factors of daily characteristics. The results show that the SVR forecasting model considering real-time meteorological factors has higher forecasting accuracy for short-term power load. The forecasting error considering meteorological and date type is smaller than that without considering meteorological and date. The embedded dimension and time delay have greater impact on the accuracy of the forecasting model. The research results can provide reference for short-term load forecasting of power grid.