The output of wind farm and photovoltaic power station is intermittent and fluctuating, and reasonable wind and solar capacity ratio can fully realize their complementarity. Inaccurate theoretical power calculation of power station will affect the extraction of true features of wind and solar, and then lead to larger capacity matching error. In this paper, based on the model-generator method commonly used in theoretical power calculation of new energy power station, the dynamic information window is selected according to the actual operation of non-model-generators. By using the measured power of model-generators and non-model-generators in the dynamic information window, the non-model-generators are grouped and the scale coefficient of each group of non-model-generators is dynamically identified, thus calculating the theoretical power of new energy power station. Based on the theoretical power of many years of history, this paper analyzes the characteristics of new energy power station, simulates random output scenarios, screens the scenarios, and establishes a wind-solar ratio optimization method based on the risk of source-load mismatch. The accuracy of the improved model-generator method is verified by an example. By using this method, the improved theoretical power data of wind farm and photovoltaic power station in a certain area of northwest China are obtained for many years, and the optimal ratio of wind and solar in this area is optimized.