摘要
风电功率数据中的离群点检测有助于提高风电场的运行维护效率,保障发电稳定性。针对风电功率数据进行分析,提出了一种结合堆叠随机森林(SCRF)和异常检测变换-流式超阈值峰值(AT-SPOT)算法的组合算法。将数据分为4类,分别采用SCRF算法和AT-SPOT算法,检测3类不符合时间序列规律的数据和1类符合时间序列规律的数据。实验结果表明,在多个风电功率数据集中,该组合算法性能良好,显著优于其他异常检测算法。
Wind farm anomaly detection can improve operational efficiency,ensure power generation stability,and reduce risks.By analyzing wind power data,a combined algorithm incorporating stacked random forest(SCRF)and anomaly transformer-streaming peaks-over-threshold(AT-SPOT)algorithm is proposed.Firstly,the data is divided into four categories,with three categories not following time series patterns detected using the SCRF algorithm,and one category that follows time series patterns detected using the AT-SPOT algorithm.Finally,the method’s effectiveness is validated using real-world datasets.Experimental results show that the combined algorithm performs exceptionally well,significantly outperforming various anomaly detection methods.
作者
于子洋
刘丹丹
YU Ziyang;LIU Dandan(School of Electronic and Information Engineering,Shanghai University of Electric Power,Shanghai 200090,China)
出处
《上海电力大学学报》
2025年第4期389-395,共7页
Journal of Shanghai University of Electric Power
作者简介
通信作者:刘丹丹(1980-),女,博士,副教授。主要研究方向为人工智能在电力系统中的应用、嵌入式人工智能。E-mail:liudandan@shiep.edu.com。
