基于样本双重选择的马尔科夫链光伏功率预测  

Markov chain photovoltaic power forecasting based on sample dual selection

作　　者：郭婕 厉虹[1] 王丽婕[1] GUO Jie;LI Hong;WANG Lijie(School of Automation,Beijing Information Science&Technology University,Beijing 100192,China)

机构地区：[1]北京信息科技大学自动化学院

出　　处：《北京信息科技大学学报（自然科学版）》2019年第6期14-18,共5页Journal of Beijing Information Science and Technology University

基　　金：国家自然科学基金资助项目(51607009)

摘　　要：为了提高光伏发电功率预测的精度,选择相似日作为预测模型的训练样本,提出了一种基于加权灰色关联度和格拉布斯准则的马尔科夫链预测方法。通过计算预测日与历史日气象因素的加权灰色关联度,初步选择相似日,再采用格拉布斯准则对相似日的功率数据进行异常判断,完成二次筛选,得到最终的相似日样本并建立马尔科夫链预测模型;利用MATLAB对实际光伏电站的功率进行预测分析。结果表明,双重筛选能够有效提取相似日样本,提高预测的精度。In order to improve the accuracy of photovoltaic power prediction, by selecting similar days as the training samples of the prediction model, a Markov chain prediction method based on weighted grey relational grade and Grubbs criterion is proposed. By calculating the weighted grey relational grade of meteorological factors between the forecasting day and the historical day, the similar days are preliminarily selected. Then the power data of the similar days are judged by the Grubbs criterion, and the secondary screening is completed. The final similar day samples are used to establish Markov chain prediction model, and the power of the actual photovoltaic power station is predicted and analyzed by using MATLAB. The results show that double screening can effectively extract similar days and improve the accuracy of prediction model.

关 键 词：光伏发电 马尔科夫链 灰色关联度 格拉布斯准则 

分 类 号：TM615[电气工程—电力系统及自动化]