With the development of smart grids, distributed generation will be an increasingly important part of the grid. Distributed generations which lack of management will lead to heavy losses. Combining with geospatial information grid, the paper carries out research of information acquisition for distributed generations. To begin with, the use of existing power infrastructure to reduce development costs as the precondition, the information acquisition model of distributed generation based on GSIG is given, and functions corresponding to different layer are analyzed. Besides, problems of multi-source heterogeneous and interoperability during information acquisition are tackled. And compared with the existing schemes, the superiority of the model is analyzed. At last, in the experimental platform, the function of real time statistics and power generation capability prediction is carried out, and the real time analysis and simulation of the model are tested.