摘要
在发电厂电力多源异构数据异常值辨识的过程中,过于复杂的数据特征为数据特征分析的过程造成了较大的困难,进而影响数据异常值辨识的准确度。为了缓解这一问题,提出发电厂电力多源异构数据异常值安全辨识方法。针对电力数据的多源异构特性,进行归一化处理,构建平衡的电力数据集。在数据特征空间中,计算重要性度量值,提取出有效的数据特征。在集成学习下,分析目标域数据特征,通过计算平均相对密度,辨识出目标域数据集的异常值。实验结果表明,该方法能够有效提高发电厂电力数据异常值辨识的准确度,表现出了较高的Top-k异常点覆盖率,在发电厂电力数据异常分析领域中有着良好的实践应用前景。
In the process of outlier identification of multi-source heterogeneous data in power plants,too complex data characteristics cause great difficulties for the process of data feature a-nalysis,which further affects the accuracy of outlier identification.In order to alleviate this problem,a safety identification method for outliers of multi-source heterogencous data in power plants is proposed.According to the multi-source heterogeneous characteristics of power data,normalization is carried out to construct a balanced power data set.In the data feature space,the importance measure is calculated and effective data features are extracted.Under ensemble learn-ing,the characteristics of target domain data are analyzed,and the abnormal values of target do-main data set are identified by calculating the average relative density.The experimental results show that this method can effectively improve the accuracy of power plant power data outlier i-dentification,showing a high coverage rate of Top-k outliers,and has a good practical application prospect in the field of power plant power data anomaly analysis.
作者
林岑伟
LIN Cenwei(Chn energy i&c technology co,Itd,Xicheng District,Beijing,100032)
出处
《长江信息通信》
2025年第8期137-139,共3页
Changjiang Information & Communications
关键词
数据异常辨识
电力数据异常
发电厂电力数据
多源异构数据
安全辨识
异常数据
data anomaly identification
power data anomaly
power plant power data
multi-source heterogeneous data
security identification
abnormal data
