摘要
为了对风电机组实测数据中的异常数据进行有效识别和剔除,通过分析风电机组的风速-功率异常数据,提出了基于分功率区间的自适应密度聚类(DBSCAN)异常数据识别算法,采用无标签聚类轮廓系数对DBSCAN算法中关键参数半径(ε)和邻域密度阈值(Z)进行自适应最佳选择,并利用该算法识别实验机组的异常数据。结果表明:该算法避免了人为主观设定导致的误差,能够对风电机组异常数据进行有效识别。
In order to effectively identify and eliminate the actual abnormal data of wind turbines, an adaptive density clustering(density-based spatial clustering of applications with noise, DBSCAN) abnormal data recognition algorithm based on sub-power interval was proposed by analyzing the abnormal wind speed-power data of wind turbines. The key parameters such as radius(ε) and neighborhood radius threshold(Z) in DBSCAN clustering were adaptively and optimally selected by using unlabeled clustering profile coefficients, and the abnormal data of experimental units were identified by the algorithm. Results show that the algorithm avoids the errors caused by human subjective setting and can effectively identify abnormal data of wind turbines.
作者
雷萌
郭鹏
刘博嵩
LEI Meng;GUO Peng;LIU Bosong(School of Control and Computer Engineering,North China Electric Power University,Beijing 102206,China)
出处
《动力工程学报》
CAS
CSCD
北大核心
2021年第10期859-865,共7页
Journal of Chinese Society of Power Engineering
基金
国家自然科学基金资助项目(51677067)。