变电站多局部放电源分离的选择双谱算法
发布时间：2018-09-14 11:55
【摘要】：抑制现场噪声干扰、有效提取信号特征是局部放电信号检测和分析的关键。给出了利用Fisher可分离度选择具有最强类可分离度的双谱作为信号的特征参数,并利用特征参数训练径向基神经网络来判断信号的类型的算法。通过混有高斯白噪声的电磁波仿真软件得到的模拟不同局部放电源辐射的电磁波信号,利用该算法进行信号分离,验证了该算法的有效性。最后在变电站现场未知局部放电源的情况下,对采集到的局部放电辐射电磁波信号利用该算法进行处理得到信号类型数,并训练用于信号分离的径向基神经网络。基于现场实测信号分离结果,并结合基于时延序列的局部放电源定位结果验证了该算法在变电站现场干扰情况下分离多局部放电源的有效性。
[Abstract]:The key of PD signal detection and analysis is to suppress the field noise interference and extract the signal features effectively. This paper presents an algorithm to select bispectrum with the strongest separability by using Fisher separability as the characteristic parameter of the signal and to train the radial basis function neural network to judge the type of the signal by using the characteristic parameter. By mixing the electromagnetic wave simulation software with Gao Si white noise, the electromagnetic wave signals of different local discharge power sources are simulated. The algorithm is used to separate the signals, and the validity of the algorithm is verified. Finally, in the case of unknown partial discharge power supply in substation, the acquired partial discharge electromagnetic wave signal is processed by this algorithm to obtain the number of signal types, and the radial basis function neural network for signal separation is trained. Based on the field measured signal separation results and the local discharge location results based on time-delay sequence, the effectiveness of the proposed algorithm for separating multi-local discharge power sources in substation field interference is verified.
【作者单位】： 上海交通大学电气工程系;国网山东省电力公司聊城供电公司;
【基金】：国家863高技术基金项目(SS2012AA050803)~~
【分类号】：TM855
[Abstract]:The key of PD signal detection and analysis is to suppress the field noise interference and extract the signal features effectively. This paper presents an algorithm to select bispectrum with the strongest separability by using Fisher separability as the characteristic parameter of the signal and to train the radial basis function neural network to judge the type of the signal by using the characteristic parameter. By mixing the electromagnetic wave simulation software with Gao Si white noise, the electromagnetic wave signals of different local discharge power sources are simulated. The algorithm is used to separate the signals, and the validity of the algorithm is verified. Finally, in the case of unknown partial discharge power supply in substation, the acquired partial discharge electromagnetic wave signal is processed by this algorithm to obtain the number of signal types, and the radial basis function neural network for signal separation is trained. Based on the field measured signal separation results and the local discharge location results based on time-delay sequence, the effectiveness of the proposed algorithm for separating multi-local discharge power sources in substation field interference is verified.
【作者单位】： 上海交通大学电气工程系;国网山东省电力公司聊城供电公司;
【基金】：国家863高技术基金项目(SS2012AA050803)~~
【分类号】：TM855
【参考文献】
相关期刊论文 前7条
1 李剑;王小维;金卓睿;孙才新;程昌奎;;变压器局部放电超高频信号多尺度网格维数的提取与识别[J];电网技术;2010年02期
2 弓艳朋;刘有为;吴立远;;采用分形和支持向量机的气体绝缘组合电器局部放电类型识别[J];电网技术;2011年03期
3 王彩雄;唐志国;常文治;郑书生;李成榕;;局部放电超高频检测抗干扰与多源放电信号分离方法[J];电网技术;2012年03期
4 张晓星;王震;唐炬;刘蕾;魏燕;;气体绝缘变压器局部放电超高频在线监测系统[J];高电压技术;2010年07期
5 侯慧娟;盛戈v，
本文编号：2242642
本文链接：https://www.wllwen.com/kejilunwen/dianlilw/2242642.html
