Research On Short-term Power Forecasting Correction Technology For Large-scale Photovoltaic Power Plants

Photovoltaic(PV)is given a lot of attention because of its advantages of clean,no pollution and sustainable,which has a profound impact on social development.With the continuous advancement of science and technology,PV is developing rapidly and its percentage in generate electricity is increasing.However,there is obvious intermittent and fluctuation in PV due to the influence of irradiance and temperature,which affects power output.Once this kind of unstable power is connected to the grid,it may cause changes in the steady-state and transient characteristics of the power system.Therefore,Therefore,it is necessary to forecast the power of PV for a period of time in the future and the PV power forecasting is also very important for the planning,coordination,dispatch and operation of the power grid.However,the short-term forecasting is often hard to satisfy the demands due to the fluctuation and intermittence of PV output.Therefore,this paper proposes short-term forecasting correction methods of PV power based on the analysis of PV output characteristics,influencing factors and error characteristicsFirstly,the various influencing factors of PV output are analyzed and the influence mechanism of irradiance and temperature on PV output is clarified.The irradiance mainly influences the short-circuit current and the temperature mainly influences the open circuit voltage of the PV cell,so the PV output will be significantly affected by these two factors.In addition,PV output has different characteristics in different weather conditions and different seasons.Secondly,the principle and structure of commonly used PV short-term forecasting models are introduced,and the error of irradiance and temperature in numerical weather prediction(NWP)in different regions and different months are illustrated.When analyzing the power forecasting error,considering different time scales,including year,season,month and so on,the distribution characteristics of the error are analyzed from many aspects,and considering the relationship between PV output and irradiance,the error between the two is compared and analyzed.Then,a multi-model forecasting method considering seasonal characteristics is proposed to avoid the disadvantages of single model and enhance the precision of the forecasting model.After the preliminary forecasting results are obtained,the forecasted power is corrected by non-iterative method using the measured value which close to the forecasted power.In the process of correction,the influence of the number of inputs measured values and correction time on the forecasting accuracy is considered,which further improves the final forecasting accuracy.Eventually,the validity of the method is proved by comparing with other methods.Finally,starting from different aspects of power forecasting,the forecasting accuracy is improved from each stage.In the stage of data preprocessing,the measured irradiance is used to improve the accuracy of NWP irradiance.In the stage of forecasting model,particle swarm optimization(PSO)algorithm is used to optimize model parameters and reduce model errors.In the stage of data post-processing,the error forecasting model is established to forecast the error,which is combined with the power forecasting value to further improve the forecasting accuracy.The validity of the method is proved by comparing the error of each stage.