Examining the problems of strong load randomness, poor short-term load forecasting accuracy and long calculation time in intelligent power environment, a forecasting method combining the improved Drosophila optimization algorithm IFOA and generalized regression neural network GRNN is proposed. The input factors of the model are load data and meteorological information. By improving the search distance of the fruit fly optimization algorithm, the model can be used to optimize the smooth factor of the generalized regression neural network GRNN, thereby improving the performance and prediction accuracy of the network. The accuracy and validity of the proposed prediction method are verified by simulation. The results show that the improved method can reduce prediction error and increase the stability of the algorithm. This study provides a reference for the development of a short-term power load forecasting system in China. This work is supported by National “Thirteen-Five Year” Research and Development Program of China (No. 2017YFB0602500).