摘要
根据风功率异常数据分布特点及其产生原因提出一种改进贝叶斯变点-Thompson tau异常数据清洗方法。首先,对小于切入风速的非零功率值进行剔除;接着基于改进贝叶斯变点-Thompson tau法剔除功率曲线底部堆积和周围离散型异常值,得到清洗后的正常数据;最后,应用该文所提方法,对某风场6台机组的实际功率风速数据进行清洗,用清洗时间、数据删除率、均方根误差及决定系数等评价指标对所提方法进行验证。分析结果显示:与四分位-k-均值聚类算法、最优组内方差算法、贝叶斯变点-四分位法及Thompson tau-四分位法比较,该文方法能有效识别和剔除各类功率风速异常值,所建功率曲线精度高,且清洗时间短,清洗效果好,通用性较强。
According to the distribution characteristics and causes of abnormal wind power data,an improved Bayesian change point-Thompson tau abnormal data cleaning method is proposed.First,the non-zero power value less than the cut-in wind speed is eliminated.Then,abnormal values accumulated at the bottom of the power curve and the surrounding discrete outliers are removed based on the improved Bayesian change point-Thompson tau method,and the cleaned normal data are obtained.Finally,the proposed method is applied to clean the actual power wind speed data of six units in a wind field,and the evaluation indexes such as cleaning time,data deletion rate,root mean square error and determination coefficient are used to verify the proposed method.The results show that compared with quartile k-means clustering algorithm,optimal intra-group variance algorithm,Bayesian change point-quartile method and Thompson tau-quartile method,the proposed method can effectively identify and eliminate all kinds of power and wind speed outliers,and the constructed power curve has high accuracy.The cleaning time is short,cleaning effect is good and versatility is strong.
作者
王智明
陈小国
王领军
鲁文彬
Wang Zhiming;Chen Xiaoguo;Wang Lingjun;Lu Wenbin(School of Mechanical and Electrical Engineering,Lanzhou University of Technology,Lanzhou 730050,China)
出处
《太阳能学报》
北大核心
2025年第3期687-696,共10页
Acta Energiae Solaris Sinica
基金
国家自然科学基金(52365063)。
