The calculation of actual emission reduction of grid-connected photovoltaic（PV）stations plays a very important role in the planning of PV stations and the development of PV industry.With the steady expansion of PV stations，the adverse effects of power output fluctuation in the traditional power system will，in turn，become more conspicuous.To mitigate such adverse effects，it is necessary to have some spinning reserve on the system side and the effect should be counted in the carbon emission reduction of PV.This paper analyzes the measured power data of Longyuan photovoltaic station in Geermu area of Qinghai province，gives the classification algorithm for PV power curves based on the characteristics of power fluctuation.The curves are divided into four categories based on the fluctuation，corresponding to the sunny，overcast，hardly cloudy and cloudy weather conditions.The statistics of power generation of PV station under each weather condition is also given.Based on this，the carbon emission reduction calculation model and method considering the influence of reserve are also proposed，which can accurately calculate the carbon emission reduction effects according to the capacity and meteorological information.This algorithm is also able to calculate the carbon reduction benefits for PV station planning.