Open AccessPrediction of Short-term PV output power Based on PCA-Stacking under different weather conditions
Author(s) -
Xiaofei Zhu
Publication year - 2020
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/853/1/012023
Subject(s) - photovoltaic system , computer science , principal component analysis , support vector machine , artificial neural network , data mining , term (time) , stacking , grid , power (physics) , artificial intelligence , machine learning , engineering , mathematics , physics , quantum mechanics , electrical engineering , geometry , nuclear magnetic resonance
With the continuous expansion of photovoltaic scale, the accurate prediction of photovoltaic power generation is increasingly important for grid dispatching and grid optimization operations. In this paper, the photovoltaic power generation mainly uses meteorological factors and historical data as the input and output of the neural network. The input quantity is large, the data is redundant, and the network is difficult to converge, which always has a great adverse effect on the accuracy of photovoltaic output prediction. Firstly, different weather types are classified according to the trend graphs of different weather types. Principal components analysis (PCA) is used to analyze less comprehensive features from multiple meteorological factors and reduce the input of predictive models. At the same time, aiming at the problem that the prediction accuracy of a single prediction model such as the existing neural network and wavelet analysis method is limited, the idea and method of integrated learning are introduced, and a short-term prediction method based on Stacking method combined with SVM and Xgboost is proposed. Compared with the single model of SVM and Xgboost, the results show that the proposed method has a significant improvement compared with the accuracy of a single prediction model.