|
|
|
| Vibration Voiceprint Characteristics and Looseness Detection Method of UHVDC Converter Valve Saturable Reactor |
| Liu Yunpeng1, Lai Tingyu1, Liu Jiashuo1, Wei Xiaoguang2, Pei Shaotong1 |
1. Hebei Provincial Key Laboratory of Power Transmission Equipment Security Defense North China Electric Power University Baoding 071003 China;
2. State Key Laboratory of Advanced Power Transmission Technology Global Energy Interconnection Research Institute Co. Ltd Beijing 102209 China |
|
|
|
Abstract As the key equipment in converter valves, the saturable reactor is used to protect the normal opening and closing of the thyristor. However, in the process of microsecond conduction of the thyristor, the high harmonics will produce high frequency mechanical vibration and thermal shock vibration on the saturable reactor, increasing the probability of the fault of the saturable reactor. Therefore, it is of great significance to evaluate its status. Recently, some methods were proposed to monitor the state of the saturable reactor, but most of them can only monitor the abnormal temperature rise of the core, and cannot identify the mechanical failure of the core. This paper presents a voiceprint identification model of saturable reactor core looseness based on optimized S-transform (OST) and improved deep residual shrinkage network (IDRSN). The saturable reactor vibration sound is used to identify the core looseness accurately.
Firstly, the vibration test of saturable reactor at high frequency pulse excitation was carried out, and the voiceprint signals in different core looseness were measured. Secondly, in the principal value range of acoustic signal spectrum, the Gaussian window parameters were optimized by using the energy concentration formula to improve the time-frequency resolution of voiceprint spectrum. Thirdly, the voiceprint characteristics after loosening were analyzed, it was found that two characteristic indexes of high-low frequency ratio and dominant frequency ratio in low-frequency component can only give early warning to the state with high degree of loosening. Finally, the core loosening data of five different azimuth measuring points were brought into improved deep residual shrinkage network based on adaptively parametric rectifier linear unit for training, to eliminate the redundant information in the voiceprint spectrum and map the features in different degrees of looseness independently, so as to enhance the learning ability of common features.
The results of saturable reactor vibration voiceprint test show that the vibration period of saturable reactor under high-frequency pulse excitation is the same as the electrical signal, which is 0.02 s. After the core is loosened, the frequency spectrum complexity and odd even subharmonic ratio do not change significantly, and the basic frequency changes greatly when the looseness is high, but it cannot give early warning in case of slight looseness. The proportion of dominant frequency of low frequency component and the ratio of high frequency to low frequency can sensitively reflect different loose states of iron core, but the threshold is difficult to set. The basic frequency proportion of low frequency component and the ratio of high frequency to low frequency can reflect different core loosening degrees sensitively, but the threshold is difficult to set. The recognition results of test data show that the convergence time of the proposed model is 262.13 s, and the average recognition accuracy of core looseness is 95.93 %, realizing the unified recognition of five different measuring points. Compared with short time Fourier transform (STFT), the accuracy of OST is improved by 5.56 %. The comparison between IDRSN and other neural networks such as residual network50 and deep residual shrink network shows that the calculation accuracy of IDRSN is improved by 2.82 % and 4.67 % respectively, and there is no false or missing judgment.
The results show that: (1) Saturable reactor vibration voiceprint spectrum contains a large number of high order harmonics, and the ratio of 50 Hz odd and even frequency multiplication is equal. After the core clamp is loosened, the proportion of 50~1 200 Hz component decreases, and the proportion of 1 250~18 000 Hz component increases. (2) The optimized S-transform method is proposed, which solves the problems of frequency band aliasing and spectrum leakage by adjusting the frequency domain window parameters with energy concentration. Compared with STFT, it has higher time-frequency resolution. (3) An improved deep residual shrinkage network based on adaptive parameter modified linear unit is built. The uniform recognition of voiceprint signals collected in different directions is realized by independent feature mapping, which has higher recognition accuracy compared with other traditional models.
|
|
Received: 21 December 2021
|
|
|
|
|
|
[1] 辛保安, 郭铭群, 王绍武, 等. 适应大规模新能源友好送出的直流输电技术与工程实践[J]. 电力系统自动化, 2021, 45(22): 1-8.
Xin Baoan, Guo Mingqun, Wang Shaowu, et al.Friendly HVDC transmission technologies for large-scale renewable energy and their engineering practice[J]. Automation of Electric Power Systems, 2021, 45(22): 1-8.
[2] 许汉平, 杨炜晨, 张东寅, 等. 考虑换相失败相互影响的多馈入高压直流系统换相失败判断方法[J]. 电工技术学报, 2020, 35(8): 1776-1786.
Xu Hanping, Yang Weichen, Zhang Dongyin, et al.Commutation failure judgment method for multi-infeed HVDC systems considering the interaction of commutation failures[J]. Transactions of China Electrotechnical Society, 2020, 35(8): 1776-1786.
[3] 孟沛彧, 王志冰, 迟永宁, 等. 适应多能源基地远距离输送电能的混合四端直流输电系统控制策略研究[J]. 电工技术学报, 2020, 35(增刊2): 523-534.
Meng Peiyu, Wang Zhibing, Chi Yongning, et al.Control strategy of hybrid four-terminal HVDC transmission system dedicated for long-distance power delivery from multiple energy bases[J]. Transactions of China Electrotechnical Society, 2020, 35(S2): 523-534.
[4] 赵畹君. 高压直流输电工程技术[M]. 北京: 中国电力出版社, 2004.
[5] 纪锋, 曹均正, 陈鹏, 等. 高压直流输电系统逆变侧阀饱和电抗器电气应力研究[J]. 高电压技术, 2014, 40(8): 2579-2585.
Ji Feng, Cao Junzheng, Chen Peng, et al.Research of electrical stress on saturable reactor in inverter valve of high voltage direct current transmission system[J]. High Voltage Engineering, 2014, 40(8): 2579-2585.
[6] 张鹏宁, 李琳, 纪锋, 等. HVDC阳极饱和电抗器阻尼弹性体降振降噪试验研究[J]. 电网技术, 2017, 41(12): 3839-3845.
Zhang Pengning, Li Lin, Ji Feng, et al.Test study on reduction of vibration and noise to damping elastomer in HVDC anode saturable reactor[J]. Power System Technology, 2017, 41(12): 3839-3845.
[7] 陶敏, 姚舒, 董妍波, 等. 特高压换流阀用饱和电抗器的振动研究与优化方案[J]. 高压电器, 2019, 55(12): 200-204.
Tao Min, Yao Shu, Dong Yanbo, et al.Vibration research and optimization on UHVDC converter valve saturable reactor[J]. High Voltage Apparatus, 2019, 55(12): 200-204.
[8] 王丰华, 王邵菁, 陈颂, 等. 基于改进MFCC和VQ的变压器声纹识别模型[J]. 中国电机工程学报, 2017, 37(5): 1535-1543.
Wang Fenghua, Wang Shaojing, Chen Song, et al.Voiceprint recognition model of power transformers based on improved MFCC and VQ[J]. Proceedings of the CSEE, 2017, 37(5): 1535-1543.
[9] Yi Lu, Mak M W.Adversarial data augmentation network for speech emotion recognition[C]//2019 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC), Lanzhou, China, 2019: 529-534.
[10] Qian Yanmin, He Tianxing, Deng Wei, et al.Automatic model redundancy reduction for fast back-propagation for deep neural networks in speech recognition[C]//2015 International Joint Conference on Neural Networks (IJCNN), Killarney, Ireland, 2015: 1-6.
[11] 王卓, 王玉静, 王庆岩, 等. 基于协同深度学习的二阶段绝缘子故障检测方法[J]. 电工技术学报, 2021, 36(17): 3594-3604.
Wang Zhuo, Wang Yujing, Wang Qingyan, et al.Two stage insulator fault detection method based on collaborative deep learning[J]. Transactions of China Electrotechnical Society, 2021, 36(17): 3594-3604.
[12] 谢庆, 杨天驰, 裴少通, 等. 基于多尺度协作模型的电气设备红外图像超分辨率故障辨识方法[J]. 电工技术学报, 2021, 36(21): 4608-4616.
Xie Qing, Yang Tianchi, Pei Shaotong, et al.Super-resolution identification method of electrical equipment fault based on multi-scale cooperation model[J]. Transactions of China Electrotechnical Society, 2021, 36(21): 4608-4616.
[13] 张重远, 罗世豪, 岳浩天, 等. 基于Mel时频谱-卷积神经网络的变压器铁芯声纹模式识别方法[J]. 高电压技术, 2020, 46(2): 413-423.
Zhang Zhongyuan, Luo Shihao, Yue Haotian, et al.Pattern recognition of acoustic signals of transformer core based on Mel-spectrum and CNN[J]. High Voltage Engineering, 2020, 46(2): 413-423.
[14] 刘云鹏, 王博闻, 岳浩天, 等. 基于50Hz倍频倒谱系数与门控循环单元的变压器偏磁声纹识别[J]. 中国电机工程学报, 2020, 40(14): 4681-4694, 4746.
Liu Yunpeng, Wang Bowen, Yue Haotian, et al.Identification of transformer bias voiceprint based on 50Hz frequency multiplication cepstrum coefficients and gated recurrent unit[J]. Proceedings of the CSEE, 2020, 40(14): 4681-4694, 4746.
[15] 黄南天, 徐殿国, 刘晓胜. 基于S变换与SVM的电能质量复合扰动识别[J]. 电工技术学报, 2011, 26(10): 23-30.
Huang Nantian, Xu Dianguo, Liu Xiaosheng.Identification of power quality complex disturbances based on S-transform and SVM[J]. Transactions of China Electrotechnical Society, 2011, 26(10): 23-30.
[16] Tang Qiu, Qiu Wei, Zhou Yicong.Classification of complex power quality disturbances using optimized S-transform and kernel SVM[J]. IEEE Transactions on Industrial Electronics, 2020, 67(11): 9715-9723.
[17] 朱叶叶, 汲胜昌, 张凡, 等. 电力变压器振动产生机理及影响因素研究[J]. 西安交通大学学报, 2015, 49(6): 115-125.
Zhu Yeye, Ji Shengchang, Zhang Fan, et al.Vibration mechanism and influence factors in power transformers[J]. Journal of Xi'an Jiaotong University, 2015, 49(6): 115-125.
[18] Stockwell R G, Mansinha L, Lowe R P.Localization of the complex spectrum: the S transform[J]. IEEE Transactions on Signal Processing, 1996, 44(4): 998-1001.
[19] 刘宇舜, 周文俊, 李鹏飞, 等. 基于广义S变换模时频矩阵的局部放电特高频信号去噪方法[J]. 电工技术学报, 2017, 32(9): 211-220.
Liu Yushun, Zhou Wenjun, Li Pengfei, et al.Partial discharge ultrahigh frequency signal denoising method based on generalized S-transform modular time-frequency matrix[J]. Transactions of China Electrotechnical Society, 2017, 32(9): 211-220.
[20] 李盼, 娄钊瑜, 马康, 等. 一种自适应S变换在电能质量特征提取中的应用[J]. 中国电机工程学报, 2021, 41(22): 7660-7668.
Li Pan, Lou Zhaoyu, Ma Kang, et al.Application of adaptive S-transform in power quality feature extraction[J]. Proceedings of the CSEE, 2021, 41(22): 7660-7668.
[21] 刘云鹏, 王博闻, 周旭东, 等. 基于162台超、特高压变压器的声纹特征预警阈值划定研究[J]. 华北电力大学学报(自然科学版), 2021, 48(5): 45-53.
Liu Yunpeng, Wang Bowen, Zhou Xudong, et al.Threshold delineation research for early warning of voiceprint eigenvalues based on 162 sets of EHV and UHV transformers[J]. Journal of North China Electric Power University (Natural Science Edition), 2021, 48(5): 45-53.
[22] Zhao Minghang, Zhong Shisheng, Fu Xuyun, et al.Deep residual networks with adaptively parametric rectifier linear units for fault diagnosis[J]. IEEE Transactions on Industrial Electronics, 2021, 68(3): 2587-2597.
[23] Zhao Minghang, Zhong Shisheng, Fu Xuyun, et al.Deep residual shrinkage networks for fault diagnosis[J]. IEEE Transactions on Industrial Informatics, 2020, 16(7): 4681-4690.
[24] He Kaiming, Zhang Xiangyu, Ren Shaoqing, et al.Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 2016: 770-778. |
|
|
|
2023, Vol. 38 
