Prediction Of Photovoltaic Power Generation Combined With Climate Characteristics In Liaoning

Liaoning is rich in photovoltaic resources and has huge potential.In recent years,photovoltaic power generation in Liaoning has made rapid progress,and multiple photovoltaic power stations have been built throughout the province.However,photovoltaic power generation is highly unstable.In order to maintain the stable operation of the power grid,the dispatching department needs to accurately predict the power of photovoltaic power stations.There are many factors that affect photovoltaic power generation,among which the most direct and most influential is the local realtime weather conditions.Due to the late start of the research and development of the photovoltaic energy technology in Liaoning,the power control of these photovoltaic power plants is still in the stage of direct output power and lack of photovoltaic power prediction system.Therefore,this paper designs a comprehensive similar day-based prediction model of photovoltaic power founded on weather characteristics in Liaoning.First,this paper introduces the current status of photovoltaic forecasting research at home and abroad,analyzes the huge photovoltaic energy potential of Liaoning area,and confirms the necessity of developing photovoltaic power forecasting research in Liaoning area.The power generation of photovoltaic cells Principles and the composition of grid-connected photovoltaic power generation system is introduced,and the analysis of photovoltaic power prediction and the main factors affecting the predicted power of photovoltaic power generation is discussed.Second,there may be local vacancies,out of range and abnormal fluctuations in the historical data of the output power of photovoltaic power plants.According to the situation,an abnormal data detection method is adopted to improve the accuracy of photovoltaic power station power prediction.Third,a comprehensive weather similar day-based program is designed.Based on the climate characteristics of Liaoning,36 professional weather types are classified into four generalized weather types,and the training samples are classified into four generalized weather small sample data sets using gray correlation algorithm.Considering the advantages of the small sample data training set,the RBF neural network prediction is selected as the main body of the prediction system model after analyzing the advantages and disadvantages of BP neural network and RBF neural network.Fourth,the prediction model is established.The real-time photovoltaic power generation data of the sample photovoltaic power station in 2020 is classified and devided into four small sample training data sets with generalized weather characteristics.Using the Matlab software,The RBF neural network is programmed,and the four sample sets is trained and saved as prediction models separately.Fifth,according to the historical data of photovoltaic power generation of a photovoltaic power station in Gongchangling District,Liaoyang City,Liaoning Province in 2020,the BP neural network model and the small sample RBF neural network model designed in this paper with Liaoning climate characteristics are used to perform power prediction.The four generalized weather forecast days selected from January to April 2021 are used for power forecasting.The comparison images of actual and forecasted values are drawn,the pros and cons of the two algorithms are compared,and the absolute and relative errors are calculated.Finally,the simulation prediction results are analyzed based on the accuracy index parameters of the photovoltaic power generation power prediction given by the State Grid document.Calculate the mean square error,qualified rate,root mean square error and standard deviation of the predicted data,which proves that the fitting degree of the RBF neural network model to the nonlinear function is much higher than that of the BP neural network.Comparing the index requirements of the State Grid enterprise documents(root mean square error less than 0.2,pass rate greater than80%),it proves that the prediction results of the scheme meet the required standards,and then verifies the small sample Liaoning photovoltaic power prediction based on the RBF neural network designed in this paper.Availability of strategy.