一种新的电压暂降事故源识别方法研究
|
摘要
电压暂降发生频率高、影响范围广、造成危害大。针对电力监测系统中带有事故源信息的电压暂降监测数据非常有限且不易获得的问题,提出了一种基于半监督支持向量机的电压暂降源识别方法。分析了各种电压暂降事故源,利用短时傅里叶变换(STFT)对电压暂降信号进行时频分析,提取出各类暂降特性参数,运用半监督支持向量机对其进行训练与识别,实现在少量带事故源标签电压暂降监测数据下电压暂降源的可靠识别。算例结果显示,在少量标签数据下半监督支持向量机比传统支持向量机具有更高的暂降源识别精度。
Voltage sag has the characteristics of high frequency, wide influence and great harm, etc. This paper proposed a voltage sag source identification method based on the semi-supervised support vector machine(SVM) in view of the situation that the labeled data with accident source information was very limited and not easy to obtain in the power monitoring system. All kinds of voltage sag sources were analyzed. The short time Fourier transform(STFT) was used for time-frequency analysis. All kinds of voltage sag characteristic parameters were extracted and the semi-supervised SVM was adopted for training and identification to realize the reliable identification of voltage sag sources under the conditions that there was a small number of labeled voltage sag monitoring data. Example results show that the semisupervised SVM has higher identification accuracy than the traditional SVM in the case of a small number of labeled data.
Voltage sag has the characteristics of high frequency, wide influence and great harm, etc. This paper proposed a voltage sag source identification method based on the semi-supervised support vector machine(SVM) in view of the situation that the labeled data with accident source information was very limited and not easy to obtain in the power monitoring system. All kinds of voltage sag sources were analyzed. The short time Fourier transform(STFT) was used for time-frequency analysis. All kinds of voltage sag characteristic parameters were extracted and the semi-supervised SVM was adopted for training and identification to realize the reliable identification of voltage sag sources under the conditions that there was a small number of labeled voltage sag monitoring data. Example results show that the semisupervised SVM has higher identification accuracy than the traditional SVM in the case of a small number of labeled data.
引文
[1]ZHANG Lidong,BOLLEN M H J.Characteristic of voltage dips in power system[J].IEEE Transactions on Power Delivery,2000,15:827-832.
[2]张波.电压暂降特征提取与扰动原因分析[D].北京:中国电力科学研究院,2005.
[3]刘奇,周雒维,卢伟国.基于广义S变换的暂态电能质量扰动定位与识别[J].电力系统保护与控制,2012,40(7):60-65.
[4]吕干云,方奇品,蔡秀珊.基于多分类支持向量机的电压暂降源识别[J].电力系统保护与控制,2010,38(22):151-155.
[5]李国栋,丁宁,徐永海.基于Mamdani型模糊推理的电压暂降源识别[J].华北电力大学学报,2010,37(2):43-48.
[6]刘建伟,刘媛,罗雄麟.半监督学习方法[J].计算机学报,2015,38(8):1592-1617.
[7]王世旭,吕干云.基于标签传播半监督学习的电压暂降源识别[J].电力系统及其自动化学报,2013,25(4):34-38.
[8]徐志超,杨玲君,李晓明.基于聚类改进S变换与直接支持向量机的电能质量扰动识别[J].电力自动化设备,2015,35(7):50-58.
[9]王学伟,张宏财.基于S变换和最小二乘支持向量机的电能质量扰动识别[J].电测与仪表,2009,46(8):1-4.
[10]MELHORN C J,DAVIS T D,BEAM G E.Voltage sags:their impact on the utility and industrial customers[J].IEEE Transactions on Industry Applications,1998,34(2):549-558.
[11]李昆仑,张伟,代运娜.基于Tri-training的半监督SVM[J].计算机工程与应用,2009,45(22):103-106.
[12]CHEN K,WANG S.Semi-supervised learning via regularized boosting working on multiple semisupervised assumptions[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33(1):129-143.
[13]WANG Fei,ZHANG Changshui.Label propagation through linear neighborhoods[J].IEEETransactions on Knowledge and Date Engineering,2008,20(1):55-67.
[14]徐健,张语勍,李彦斌.短时傅里叶变换和S变换用于检测电压暂降的对比研究[J].电力系统保护与控制,2014,42(24):44-48.
[2]张波.电压暂降特征提取与扰动原因分析[D].北京:中国电力科学研究院,2005.
[3]刘奇,周雒维,卢伟国.基于广义S变换的暂态电能质量扰动定位与识别[J].电力系统保护与控制,2012,40(7):60-65.
[4]吕干云,方奇品,蔡秀珊.基于多分类支持向量机的电压暂降源识别[J].电力系统保护与控制,2010,38(22):151-155.
[5]李国栋,丁宁,徐永海.基于Mamdani型模糊推理的电压暂降源识别[J].华北电力大学学报,2010,37(2):43-48.
[6]刘建伟,刘媛,罗雄麟.半监督学习方法[J].计算机学报,2015,38(8):1592-1617.
[7]王世旭,吕干云.基于标签传播半监督学习的电压暂降源识别[J].电力系统及其自动化学报,2013,25(4):34-38.
[8]徐志超,杨玲君,李晓明.基于聚类改进S变换与直接支持向量机的电能质量扰动识别[J].电力自动化设备,2015,35(7):50-58.
[9]王学伟,张宏财.基于S变换和最小二乘支持向量机的电能质量扰动识别[J].电测与仪表,2009,46(8):1-4.
[10]MELHORN C J,DAVIS T D,BEAM G E.Voltage sags:their impact on the utility and industrial customers[J].IEEE Transactions on Industry Applications,1998,34(2):549-558.
[11]李昆仑,张伟,代运娜.基于Tri-training的半监督SVM[J].计算机工程与应用,2009,45(22):103-106.
[12]CHEN K,WANG S.Semi-supervised learning via regularized boosting working on multiple semisupervised assumptions[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33(1):129-143.
[13]WANG Fei,ZHANG Changshui.Label propagation through linear neighborhoods[J].IEEETransactions on Knowledge and Date Engineering,2008,20(1):55-67.
[14]徐健,张语勍,李彦斌.短时傅里叶变换和S变换用于检测电压暂降的对比研究[J].电力系统保护与控制,2014,42(24):44-48.