小波去噪对提高GIS超声局部放电信号识别率的研究

doi:10.11985/2017.11.007

摘要/Abstract

摘要：

在气体绝缘组合电器（GIS）实体模型内部分别放置针-板、悬浮金属颗粒和绝缘子表面固定金属颗粒三种缺陷模型,用超声波传感器在相同电压下采集到良好的局部放电波形,将从现场运行设备上测得的背景噪声叠加到原放电波形上。对叠加噪声后的放电波形采用小波去噪,针对波形特点选取了7个特征参数,分别用去噪前后波形的特征参数对BP_Adaboost分类器进行训练和测试,结果表明用去噪后波形提取的特征量作为分类器输入的识别率更高。

关键词: 气体绝缘组合电器, 超声波传感器, 小波去噪, BP_Adaboost分类器, 放电类型识别

Abstract:

Needle-plate, suspended metal particles and metal particles fixed on insulator surface were placed separately in GIS entity model. The discharge waveforms were detected by using ultrasonic sensor under the same voltage. In order to be consistent with the field noise, we added the noise which was detected from the field equipment to the discharge waveforms. The waveforms were processed by wavelet de-noising. Then Aiming at the waveforms’ chara- cteristics, seven characteristic parameters were chosen. The characteristic parameters before and after de-noising were used separately to train and test BP_Adaboost classifier. The results showed that by using characteristic vectors grasped from waveforms after de-noising, the recognition result is higher.

Key words: GIS, ultrasonic sensor, wavelet de-noising, BP_Adaboost classifier, recognition of discharge types

中图分类号:

TM855

张波,刘成国,徐忠,林涛. 小波去噪对提高GIS超声局部放电信号识别率的研究[J]. 电气工程学报, 2017, 12(11): 41-45.

Bo Zhang,Chengguo Liu,Zhong Xu,Tao Lin. Research on Improving the Recognition Rate of GIS Ultrasonic Discharge Signal by Wavelet Denoising[J]. Journal of Electrical Engineering, 2017, 12(11): 41-45.

图/表 13

图1

图2

图3

图4

图5

表1

表2

表3

表4

表5

表6

表7

表8

参考文献 11

[1]	高凯, 倪浩, 杨凌辉 . GIS局部放电检测的技术发展和分析[J]. 华东电力, 2012,40(8):1384-1387.
	Gao Kai, Ni Hao, Yang Linghui . GIS partial discharge testing technology development and analysis[J]. East China Electric Power, 2012,40(8):1384-1387.
[2]	邱毓昌 . GIS装置及其绝缘技术[M]. 北京: 水利电力出版社, 1994: 34-89.
[3]	谭向宇, 杨卓, 赵现平 , 等. GIS内金属性缺陷引起的局部放电超声信号分布特性研究[J]. 高压电器, 2012,48(12):65-69.
	Tan Xiangyu, Yang Zhuo, Zhao Xianping , et al. Ultrasonic signal distribution characteristics of partial discharge due to metallic defects in GIS[J]. High Voltage Apparatus, 2012,48(12):65-69.
[4]	刘君华, 吴晓春, 徐敏骅 . 超高频与超声波法在GIS局部放电现场检测中的应用研究[J]. 华东电力, 2011,39(12):2096-2099.
	Liu Junhua, Wu Xiaochun, Xu Minhua . Application research on on-line detection of GIS partial discharge based on ultrahigh frequency method and ultrasonic method[J]. East China Electric Power, 2011,39(12):2096-2099.
[5]	张晓星, 唐炬, 孙才新 , 等. 一种基于线性鉴别分析的GIS局部放电模式识别[J]. 重庆大学学报, 2006,29(10):1-4.
	Zhang Xiaoxing, Tang Ju, Sun Caixin , et al. PD pattern recognition based on linear discriminant analysis in GIS[J]. Journal of Chongqing University, 2006,29(10):1-4.
[6]	胡云鹏, 陈焕新, 周诚 , 等. 基于小波去噪的冷水机组传感器故障检测[J]. 华中科技大学学报(自然科学版), 2013,41(3):16-19. doi: 10.1007/s11596-006-0614-4 pmid: 17357488
	Hu Yunpeng, Chen Huanxin, Zhou Cheng , et al. Chiller sensor fault detection using wavelet de-noising[J]. J. Huanzhong Univ. of Sci. & Tech. (Natural Science Edition), 2013,41(3):16-19. doi: 10.1007/s11596-006-0614-4 pmid: 17357488
[7]	王立地, 朱楠, 韩春成 . 小波去噪在动态负荷建模实测信号中的应用[J]. 电测与仪表, 2010,47(9):18-21. doi: 10.1038/s41598-019-55333-1 pmid: 31836756
	Wang Lidi, Zhu Nan, Han Chuncheng . Application of wavelet denoising on measurement signals of dynamic load model[J]. Electrical Measurement & Instrumentation, 2010,47(9):18-21. doi: 10.1038/s41598-019-55333-1 pmid: 31836756
[8]	邱选兵, 魏计林, 崔小朝 , 等. 异构双核构架下的脉冲涡流信号实时小波去噪[J]. 计算机应用, 2013,33(3):866-870. doi: 10.3724/SP.J.1087.2013.00866
	Qiu Xuanbing, Wei Jilin, Cui Xiaozhao , et al. Real-time wavelet denoising for pulsed eddy current signal based on heterogeneous dual-core[J]. Journal of Computer Applications, 2013,33(3):866-870. doi: 10.3724/SP.J.1087.2013.00866
[9]	张燕, 倪远平 . 基于AdaBoost.M2-NN的变压器故障诊断[J]. 甘肃科学学报, 2012,24(1):97-101.
	Zhang Yan, Ni Yuanping . Fault diagnosis of transformer based on adaBoost. M2-NN[J]. Journal of Gansu Sciences, 2012,24(1):97-101.
[10]	周英, 尹邦德, 任铃 , 等. 基于BP神经网络的电网短期负荷预测模型研究[J]. 电测与仪表, 2011,48(2):68-71. doi: 10.1038/s41598-019-55333-1 pmid: 31836756
	Zhou Ying, Yin Bangde, Ren Ling , et al. Study of electricity short-term load forecast based on BP neural network[J]. Electrical Measurement & Instrumentation, 2011,48(2):68-71. doi: 10.1038/s41598-019-55333-1 pmid: 31836756
[11]	彭湘陵 . 基于BP_Adaboost和HMM的湖南方言识别方法研究[D]. 长沙:湖南师范大学, 2012.

M	SD	MAD	S	K	1N	2N
-0.007 113	0.027 83	0.019 56	0.009 2	5.809 2	410.5	4.064
-0.007 141	0.022 47	0.016 54	0.012 8	5.430 4	351.2	3.332
-0.007 174	0.024 70	0.017 06	0.010 6	7.378 1	363.1	3.637
-0.007 276	0.028 23	0.020 85	0.006 5	4.480 6	433.4	4.122
-0.007 382	0.029 07	0.020 88	0.036 2	7.626 7	437	4.237

M	SD	MAD	S	K	1N	2N
-0.006 623	0.020 12	0.015 11	0.010 2	4.706 5	322.3	2.992
-0.006 675	0.019 87	0.015 58	0.004 5	3.502 8	327.7	2.962
-0.006 759	0.020 10	0.015 75	0.004 7	3.466 4	332.9	3.013
-0.006 646	0.023 14	0.018 2	0.005 7	3.678 2	376.5	3.404
-0.006 659	0.020 76	0.015 62	0.018 2	3.538 4	330.6	2.983

M	SD	MAD	S	K	1N	2N
-0.006 686	0.019 5	0.013 61	0.022	6.414 0	297	2.982
-0.006 738	0.019 07	0.014 80	0.013 2	5.205 0	304	2.928
-0.006 812	0.020 25	0.014	0.014 0	5.704 3	321	3.012
-0.006 801	0.020 09	0.014 57	0.026 4	6.685 2	311	2.997
-0.006 735	0.016 21	0.011 22	0.074 2	12.353 6	259	2.49

M	SD	MAD	S	K	1N	2N
-0.007 112	0.017 31	0.009 25	0.017 7	10.548 3	253.6	2.640
-0.007 142	0.012 02	0.006 518	0.028 2	13.766 8	204.5	1.982
-0.007 171	0.014 5	0.007 247	0.015 1	16.224 4	223.1	2.329
-0.007 272	0.017 07	0.010 09	0.016 1	7.705 2	262.5	2.627
-0.007 388	0.018 31	0.009 959	0.073 4	19.495 5	261.2	2.791

M	SD	MAD	S	K	1N	2N
-0.006 632	0.009 65	0.005 391	0.011 3	13.477 4	176	1.659
-0.006 675	0.008 692	0.005 415	0.000 9	7.856 5	174.2	1.56
-0.006 752	0.008 922	0.005 648	0.014 1	7.733 4	178.2	1.582
-0.006 646	0.011 42	0.007 255	0.011 2	8.442 4	199.2	1.879
-0.006 657	0.008 771	0.005 42	0.080 8	8.810 1	174.2	1.555