To comprehensively explore the correlation information between different factors affecting photovoltaic output,the paper proposes a short term photovoltaic output interval prediction method based on integrated machine learning model to further improve the accuracy of the short-term photovoltaic output interval prediction with the machine learning model.Firstly,the fast correlation based filter(FCBF)is used to extract the optimal features from multidimensional historical photovoltaic data and meteorological data.Then on the basis of integrating multiple machine learning models,the prediction errors during the training process are collected,and the probability distribution of the prediction errors is obtained through maximum likelihood estimation,thereby obtaining the upper and lower limits of the prediction interval.Finally,the photovoltaic output curve is got by combining the integrated learning model prediction,and the final day-ahead photovoltaic output prediction interval is obtained.The reliability and superiority of the proposed model are verified through the examples.
关键词
短期光伏功率预测/特征选择/机器学习/区间预测
Key words
short-term photovoltaic power prediction/feature selection/machine learning/interval prediction
维普
万方数据