Page 125 - 《应用声学》2021年第4期
P. 125
第 40 卷 第 4 期 吴国鑫等: 声纹的变压器放电与机械故障诊断研究 609
FastICA算法成功将模拟的故障声与变压器本体噪 [9] Shatnawi Y, Al-khassaweneh M. Fault diagnosis in inter-
声分离,分别采用改进小波包 -BP 神经网络算法和 nal combustion engines using extension neural network[J].
IEEE Transactions on Industrial Electronics, 2014, 61(3):
梅尔对数频谱 -卷积神经网络算法识别不同声音。
1434–1443.
实验模拟了放电声和机械故障声,采集了不同变电 [10] Rzeszucinski P, Orman M, Pinto C T, et al. Bearing
站的变压器声和 6 种干扰声。实验表明,本文提出 health diagnosed with a mobile phone: acoustic signal
方法可在有干扰声的条件下辨别放电故障、机械故 measurements can be used to test for structural faults in
motors[J]. IEEE Industry Applications Magazine, 2018,
障和变压器本体噪声。改进小波包-BP神经网络算
24(4): 17–23.
法的识别率较高,可达 99.6%;梅尔对数频谱 -卷积 [11] Liu Y, Qian Q, Fu Y, et al. Wayside acoustic fault diagno-
神经网络算法识别率为97.57%,泛化性较好。 sis of railway wheel-bearing paved with Doppler effect re-
duction and EEMD-based diagnosis information enhance-
参 考 文 献 ment[C]// Nanjing: International Conference on Sensing
Technology, 2016: 1–5.
[1] 周东旭, 王丰华, 党晓婧, 等. 基于压缩观测与判别字典学习
[12] Tian G, Shao X, Ma Q, et al. Study on fault diagnosis of
的干式变压器声纹识别 [J]. 中国电机工程学报, 2020, 40(19):
gear wearing based on non-stationary acoustic signal[C]//
6380–6390.
Beijing: International Conference on Electronic Measure-
Zhou Dongxu, Wang Fenghua, Dang Xiaojing, et al. Dry
ment & Instruments, 2009: 4-783–4-786.
type transformer voiceprint recognition based on com-
[13] 孙 汉 文, 李 喆, 林 睿, 等. 基 于 新 奇 检 测 的 两 级 电 气 故
pressed observation and discrimination dictionary learn-
障声纹识别算法 [J/OL]. [2021-02-26]. 电网技术: 1–8,
ing[J]. Proceedings of The Chinese Society for Electrical
https://doi.org/10.13335/j.1000-3673.pst.2020.1429.
Engineering, 2020, 40(19): 6380–6390.
[14] 田昊洋, 马文嘉, 王丰华, 等. 基于稀疏自动编码器的特高压
[2] 蔡鋆, 袁文泽, 张轩瑞, 等. 基于特高频自感知的变压器局
变压器可听噪声分析 [J]. 高压电器, 2020, 56(9): 67–73, 79.
部放电检测方法 [J/OL]. [2021-02-26]. 高电压技术: 1–10,
Tian Haoyang, Ma Wenjia, Wang Fenghua, et al. Re-
https://doi.org/10.13336/j.1003-6520.hve.20201164.
search on audible noise of UHV power transformer based
[3] 朱辉, 王煜, 孙涛. 变压器运行期间声音异常处理方法探析 [J].
on sparse autoencoder[J]. High Voltage Apparatus, 2020,
技术与市场, 2019, 26(11): 148–149.
56(9): 67–73, 79.
[4] 吴晓文, 周年光, 彭继文, 等. 电力变压器噪声特性与相关因
[15] 周东旭, 王丰华, 党晓婧, 等. 基于稀疏表示理论的特高压
素分析 [J]. 电力科学与技术学报, 2018, 33(3): 81–85, 146.
交流变压器声信号盲分离研究 [J]. 电网技术, 2020, 44(8):
Wu Xiaowen, Zhou Nianguang, Peng Jiwen, et al. Noise
3139–3148.
characteristic and relevant factors analysis of power trans-
Zhou Dongxu, Wang Fenghua, Dang Xiaojing, et al. Blind
formers[J]. Journal of Electric Power Science and Technol-
separation of UHV power transformer acoustic signal
ogy, 2018, 33(3): 81–85, 146.
preprocessing based on sparse representation theory[J].
[5] 余长厅, 黎大健, 汲胜昌, 等. 变电站声纹成像测试系统的应
Power System Technology, 2020, 44(8): 3139–3148.
用分析 [J]. 广西电力, 2020, 43(5): 63–66.
[16] 杜一明. 基于声信号的变压器故障诊断系统研究 [D]. 武汉:
Yu Changting, Li Dajian, Ji Shengchang, et al. Applica-
华中科技大学, 2013.
tion analysis of the voiceprint imaging test system in sub-
[17] 吴松. 基于声学特征的变压器故障诊断研究 [D]. 武汉: 华中
stations[J]. Guangxi Electric Power, 2020, 43(5): 63–66.
[6] 余长厅, 黎大健, 陈梁远, 等. 基于声纹及振动的变压器故障 科技大学, 2012.
诊断技术研究 [J]. 高压电器, 2019, 55(11): 248–254. [18] 张瑞琪. 基于声学信号的变压器放电故障诊断方法研究 [D].
Yu Changting, Li Dajian, Chen Liangyuan, et al. Trans- 武汉: 华中科技大学, 2018.
former fault diagnosis technique based on voiceprint [19] 耿琪深, 王丰华, 金霄. 基于 Gammatone 滤波器倒谱系数与
and vibration[J]. High Voltage Apparatus, 2019, 55(11): 鲸鱼算法优化随机森林的干式变压器机械故障声音诊断 [J].
248–254. 电力自动化设备, 2020, 40(8): 191–196, 224.
[7] 赵书涛, 李沐峰, 王亚潇, 等. 断路器操动状态声音辨识的优 Geng Qishen, Wang Fenghua, Jin Xiao. Mechanical
化算法的研究 [J]. 电测与仪表, 2017, 54(10): 26–31. fault sound diagnosis based on GFCC and random for-
Zhao Shutao, Li Mufeng, Wang Yaxiao, et al. Research est optimized by whale algorithm for dry type trans-
of optimization algorithm for sound signal recognition of former[J]. Electric Power Automation Equipment, 2020,
circuit breaker operating state[J]. Electrical Measurement 40(8): 191–196, 224.
& Instrumentation, 2017, 54(10): 26–31. [20] 王丰华, 王邵菁, 陈颂, 等. 基于改进 MFCC 和 VQ 的变
[8] 赵莉华, 徐舒蓉, 谢荣斌, 等. 变压器可听声诊断技术中传声 压器声纹识别模型 [J]. 中国电机工程学报, 2017, 37(5):
器的选择 [J]. 电测与仪表, 2018, 55(1): 102–108. 1535–1542.
Zhao Lihua, Xu Shurong, Xie Rongbin, et al. The selec- Wang Fenghua, Wang Shaojing, Chen Song, et al.
tion of microphone in power transformer diagnosis tech- Voiceprint recognition model of power transformers based
nology based on the audible sound[J]. Electrical Measure- on improved MFCC and VQ[J]. Proceedings of The
ment & Instrumentation, 2018, 55(1): 102–108. Chinese Society for Electrical Engineering, 2017, 37(5):