Abstract: With the continuous increase of the proportion of renewable energy sources in the power energy sector, precise meteorological information has become an indispensable technological support for the health and safety of the new power system. However, traditional meteorological products have significant gaps in terms of temporal and spatial resolution, medium- and long-term forecasting, abrupt and extreme weather forecasting, etc., compared with the requirements of the power energy department. The accuracy, timeliness and certainty of meteorological information and forecasts are still far from meeting the requirements of power users. In recent years, the rapid development of artificial intelligence (Artificial Intelligence, AI) in the field of meteorology has promoted the widespread application of data-driven methods in meteorological forecasting, providing new technical paths to solve the problems of numerical weather prediction (Numerical Weather Prediction, NWP) such as the accumulation of errors over time, limited characterization of uncertainty, high resolution and high computational cost of large sample forecasting. This paper, in response to the demand for meteorological simulation and forecasting by renewable energy sources and the existing bottleneck issues, takes the GAP generative assimilation and prediction model (Generative Assimilation and Prediction, GAP) developed by the Institute of Atmospheric Physics of the Chinese Academy of Sciences as an example, and introduces the advantages of AI models in precise forecasting, extreme event prediction, data construction in sparse observation areas, and medium- and long-term forecasting in energy meteorology.