基于RANSAC-DBSCAN的风速功率曲线异常数据清洗方法  

ABNORMAL DATA CLEANING METHOD OF WIND SPEED-POWER CURVE BASED ON RANSAC-DBSCAN

作　　者：罗朗川 李汝辉 曾东 邹明衡 Luo Langchuan;Li Ruhui;Zeng Dong;Zou Mingheng(Guangdong Yudean Zhanjiang Wind Power Generation Co.,Ltd.,Zhanjiang 524043,China;school.of Energy&Environment,Southeast University,Nanjing 211189,China)

机构地区：[1]广东粤电湛江风力发电有限公司,湛江524043 [2]东南大学能源与环境学院,南京211189

出　　处：《太阳能学报》2025年第4期445-453,共9页Acta Energiae Solaris Sinica

摘　　要：针对海上风电机组在运行中不可避免地产生大量噪声、故障、弃风限电等异常数据,导致运行数据可用性差的问题。梳理和分析风功率曲线中异常数据的分布特征,提出基于随机采样一致(RANSAC)回归与含噪声的基于密度的空间聚类(DBSCAN)融合算法的风功率曲线异常数据清洗,并从算法的清洗效果、清洗效率以及数据删除合理性进行对比验证。结果表明,所提方法能够快速、简便、合理地识别异常数据范围,具有工程应用价值。Addressing the challenge of offshore wind turbines inevitably producing a significant volume of abnormal data—such as noise,faults,wind abandonment,and power limitation,which compromises the availability of operational data—this paper analyzes the distribution characteristics of anomalous data within wind power curves.We propose a novel method for cleaning this data,utilizing a fusion algorithm that combines Random Sample Consensus(RANSAC)and Density-Based Spatial Clustering of Applications with Noise(DBSCAN).This approach is rigorously evaluated for its cleaning effectiveness,efficiency,and the justification of data exclusion.Our findings reveal that the proposed methodology can swiftly,effortlessly,and logically delineate the boundaries of anomalous data,demonstrating its significant potential for engineering applications.

关 键 词：海上风电 数据分析 异常诊断 随机采样一致 基于密度的空间聚类 

分 类 号：TK89[动力工程及工程热物理—流体机械及工程]