基于模式识别的WAMS有功功率错误数据处理  

Wrong Active Power Data Identification and Correction for WAMS Based on Pattern Recognition

作　　者：万楚林 陈皓勇[1] 郭曼兰 

机构地区：[1]华南理工大学电力学院,广东省广州市510640

出　　处：《电网技术》2017年第3期922-930,共9页Power System Technology

摘　　要：目前,国内广域量测系统(WAMS)在实际应用中出现较多错误数据,由此衍生出大量错误的电网设备告警记录。为解决传统方法在辨识和还原WAMS错误数据时时效性差、漏报率高的问题,文章重点分析了在有功功率量测过程中所出现的异常情况,归纳和总结出典型错误数据的特征,提出一种基于模式识别的WAMS错误数据快速辨识和恢复方法。该方法根据数据特点提前设定错误数据和正常数据的标准特征向量,并将实测数据以同样方法转换成判断向量,从而进行模式匹配,实现错误数据的辨识。最后文中设计了3个算例分别展示了暂态下正确数据、稳态下错误数据以及大量数据下的辨识和恢复结果。结果表明,因为无需输入大量网架信息和没有迭代计算,模式识别的方法无论是准确度还是计算时间上都具有较强优势,能够为电网状态监测和分析等提供较可靠的数据支持。At present, error warning problems are produced in wide area measurement system（WAMS） because of wrong data injection. To overcome problem of time delay and high non-response rates when using common methods to process wrong data, a rapid identification and recovery method based on pattern recognition concept is put forward in this paper. According to WAMS data characteristics, standard feature vectors are set in advance. Then all WAMS measured data are changed to unique feature vectors in the same way and matched with the standard feature vectors, so that wrong data can be easily and quickly identified. Finally, three cases are designed to show identification and recovery results of three kinds of data including correct data in transient state, wrong data in steady state and a large amount of real data. Result demonstrates that, without repeated iterations and injection of power grid parameters, pattern recognition has strong advantages on both accuracy and computational time to provide data support for power grid state monitoring and analysis.

关 键 词：广域测量系统 模式识别 错误数据辨识 错误数据恢复 数据相关性 

分 类 号：TM734[电气工程—电力系统及自动化]