A Classified Forecasting Approach of Power Generation for Photovoltaic Plants Based on Weather Condition Pattern Recognition

Accurate photovoltaic(PV) power forecasting is very important for the dispatching of power grid and optimal operation of PV plants. It is very difficult for the unified model to forecast the output power of PV plant precisely under multiple weather conditions. Taken the total classification number, representativeness and equilibrium of distribution into consideration based on the analysis of variation law of solar irradiance, four generalized weather types were obtained by summarizing the meteorological professional weather types. The classification forecast approach of PV power was proposed subsequently. A weather status pattern recognition model based on support vector machine(SVM) was constructed with the input feature parameters extracted from solar irradiance data to identify the missing weather type label of part historical data. The accuracy of weather status pattern recognition and the validity of classification power forecast approach for PV plant are verified by the simulation using actual operating data of PV plant.