Photovoltaic power generation probabilistic prediction based on a new dynamic weighting method and quantile regression neural network

• Published in: Chinese Control Conference pp. 6445 - 6451
• Main Authors: Cheng, Ze, Zhang, Wen, Liu, Chong
• Format: Conference Proceeding
• Language: English
• Published: Technical Committee on Control Theory, Chinese Association of Automation 01.07.2019
• Subjects: dynamic weighting method; Forecasting; Heuristic algorithms; K-Nearest Neighbor (KNN) algorithm; Neural networks; photovoltaic power generation; Photovoltaic systems; Prediction algorithms; probabilistic prediction; quantile regression neural network (QRNN)
• ISSN: 1934-1768
• DOI: 10.23919/ChiCC.2019.8866208

In order to allow efficient planning of electric power system, the reliable prediction of photovoltaic power generation is very important. This paper proposes a new solar power probabilistic forecasting method based on dynamic weighting method, K-Nearest Neighbor (KNN) algorithm and quantile regression neural network (QRNN). Firstly, a new dynamic weighting method is used to tune the optimal weights of meteorological factors dynamically. Then based on the optimal weighted Euclidean distance metric method, KNN algorithm is used to find the similar examples more accurately. Finally, QRNN model is established to obtain different quantiles and approximately estimate the probability distribution of solar power output. The data from IEEE Working Group on Energy Forecasting is used to valid ate the effectiveness of proposed method and the experimental results show that the proposed model has reliable and accurate prediction ability.