摘要
现有的变电站局部放电定位方法在定位时存在较大的定位误差,为此提出基于融合聚类的220 kV变电站局部放电定位方法。采用奇异值分解(Singular Value Decomposition,SVD)的方式对放电信号进行非线性滤波降噪处理,结合信号成分的含量越高,对应的奇异值越大的特点,根据奇异值差分谱中任意2个相邻奇异值参量的变化程度确定保留或是剔除。以时域波形幅值为基准对降噪后的信号进行划分,根据各个划分结果对应的信号特征与变电站的额定输出信号的相似度对信号聚类,信号数据最多的聚类对应的位置即为最终计算得出的放电位置。测试结果表明,设计方法的定位结果与实际放电位置的误差稳定在0.30~0.50 m,具有较高的准确性。
The existing partial discharge location methods in substations have large location errors.Therefore,this paper proposes a 220 kV substation partial discharge location method based on fusion clustering.The Singular Value Decomposition(SVD)method is used for nonlinear filtering and noise reduction of the discharge signal.Combined with the characteristics that the higher the content of the signal component,the larger the corresponding singular value is,the retention or elimination is determined according to the change degree of any two adjacent singular value parameters in the singular value difference spectrum.The signal after noise reduction is divided based on the amplitude of time domain waveform,and the signal is clustered according to the similarity between the signal characteristics corresponding to each division result and the rated output signal of the substation.The position corresponding to the cluster with the most signal data is the finally calculated discharge position.The test results show that the error between the positioning result of the design method and the actual discharge position is stable between 0.30~0.50 m,which has high accuracy.
作者
单强
SHAN Qiang(State Grid Xinjiang Electric Power Co.,Ltd.,Urumqi 830000,China)
出处
《通信电源技术》
2022年第12期92-94,共3页
Telecom Power Technology
关键词
融合聚类
变电站局部放电
奇异值分解(SVD)
非线性滤波降噪
奇异值差分谱
时域波形幅值
fusion clustering
partial discharge of substation
Singular Value Decomposition(SVD)
nonlinear filtering and noise reduction
singular value difference spectrum
time domain waveform amplitude
