Provided is a weather type identification method based on an irradiance characteristic parameter. The weather type identification method based on the irradiance characteristic parameter is used for identifying lacked weather types in photovoltaic power station day-to-day historical data. The technical scheme is that the method is based on an internal relevance relationship between an irradiance day-to-day change rule and different weather types. The irradiance characteristic parameter serves as input, and the weather types serve as output. A weather type identification model is established through a support vector machine method. The weather type identification model is trained by utilizing weather type information integrated historical data. A nonlinear mapping relation between the input and the output is matched, and then the lacked weather types in a day-to-day data record are identified and obtained through the weather type identification model. According to the weather type identification method based on the irradiance characteristic parameter, effective identification to the lacked weather types is achieved through a support vector machine model. Adverse impacts of the weather type information loss in the photovoltaic power station day-to-day data record to classification prediction algorithm implementation are eliminated, and favorable conditions are created for photovoltaic power station irradiance improvement and the accuracy of generated power classification forecasting.