基于改进TCN模型的变压器故障诊断方法研究

doi:10.3969/j.issn.2097-0706.2024.11.005

摘要/Abstract

摘要：

针对传统机器学习的变压器诊断方法受限于精度较低、信息数据来源单一和故障样本稀缺等问题，提出一种多源异构数据融合技术和泥环优化算法（MRA）优化时间卷积网络（TCN）的变压器故障诊断模型。通过选择油色谱分析数据、红外高压套管检测图谱、超声波放电检测图谱以及局部放电特高频检测图谱作为变压器故障诊断模型的输入信息，采用Informer网络和残差网络（ResNet）对不同类型的数据分别进行特征提取与学习，同时对多类型数据的特征进行融合，结合MRA算法对TCN网络参数进行优化，综合结果进行故障诊断分类。试验结果表明，本文方法纳什效率系数为0.82且准确率达到了94.83％，收敛速度更快，由此证明本文所提方法可以有效提高变压器故障诊断性能。

关键词: 变压器, 故障诊断, 泥环优化算法, 多源异构数据融合, 时间卷积网络

Abstract:

Transformer diagnosis methods based on traditional machine learning have been limited by low accuracy， single-source data， and a scarcity of fault samples. This paper proposes a transformer fault diagnosis model utilizing multi-source heterogeneous data fusion and the Mud Ring Algorithm （MRA） to optimize the Temporal Convolutional Network （TCN）. Oil chromatography data， infrared high-voltage bushing detection images， ultrasonic discharge detection images， and ultra-high frequency partial discharge detection images were selected as input information for the transformer fault diagnosis model. The Informer network and ResNet （Residual Network） were applied to extract and learn features from different data types， followed by feature fusion of multi-type data. The MRA algorithm was used to optimize the parameters of the TCN network， and the integrated results were used for fault classification. Experimental results showed that the proposed method achieved a Nash efficiency coefficient of 0.82 and an accuracy of 94.83%， with faster convergence， demonstrating its effectiveness in enhancing transformer fault diagnosis performance.

Key words: transformer, fault diagnosis, mud ring algorithm, multi-source heterogeneous data fusion, temporal convolutional network

中图分类号:

TM41

徐波, 魏艺君, 邓芳明. 基于改进TCN模型的变压器故障诊断方法研究[J]. 综合智慧能源, 2024, 46(11): 38-45.

XU Bo, WEI Yijun, DENG Fangming. Research on transformer fault diagnosis method based on improved TCN model[J]. Integrated Intelligent Energy, 2024, 46(11): 38-45.

图/表 11

图1

图2

表1

油色谱数据预处理

日期	$φ$ （H₂）/ (μL $⋅$ L^-1)	$φ$ (CO)/ (μL $⋅$ L^-1)	$φ$ (CO₂)/ (μL $⋅$ L^-11)	$φ$ (CH₄)/ (μL $⋅$ L^-1)	$φ$ (C₂H₆)/ (μL $⋅$ L^-1)	$φ$ (C₂H₄)/ (μL $⋅$ L^-1)	$φ$ (C₂H₂)/ (μL $⋅$ L^-1)	$φ$ (THC)/ (μL $⋅$ L^-1)	$w$ (H₂O)/ (mg $⋅$ kg^-1)
2023-07-10	150.75	157.56	2 305.79	10.40	1.70	1.58	0.87	15.55	8.50
2023-07-24	194.24	185.34	2 470.92	11.12	2.07	1.58	0.92	15.69	12.90
2023-07-31	209.96	176.00	3 235.13	11.73	2.58	1.90	3.36	19.57	13.40
2023-08-02	231.70	207.86	3 606.76	12.73	3.27	2.12	3.92	22.04	10.70
2023-08-07	187.64	201.50	2 463.52	11.21	3.24	1.98	1.22	17.65	12.80
2023-08-09	178.61	200.00	1 873.92	10.78	1.80	1.32	0.36	14.28	19.00
2023-08-14	157.10	178.46	1 968.24	10.42	3.06	1.67	0.41	15.56	6.80
2023-08-16	188.72	176.23	1 915.65	11.06	1.62	1.43	0.41	14.52	13.60
2023-08-21	178.92	197.50	2 386.50	11.06	2.18	1.61	0.00	14.85	8.20
2023-08-24	163.61	175.78	1 922.36	10.09	2.37	2.18	0.42	15.06	18.10
2023-08-30	170.59	188.61	1 902.63	10.83	1.80	1.30	0.34	14.26	10.00
2023-09-06	202.30	199.62	2 798.32	11.46	2.36	1.80	1.43	17.05	11.30
2023-09-13	174.64	167.46	2 604.40	10.53	2.84	1.56	1.32	16.25	7.70
2023-09-21	202.76	186.22	2 761.99	11.94	2.46	1.99	1.54	17.93	10.10
2023-09-27	196.88	177.07	2 890.30	11.81	1.90	1.76	1.76	17.23	7.70

表1

图3

图4

图5

表2

表3

图6

表4

训练数据对比

方案	Huber	$P$ /%	$R$ /%	$A$ /%	$N$
1	0.34	82.32	83.51	79.92	0.46
2	0.18	89.65	88.92	89.72	0.64
3	0.24	87.16	86.61	86.34	0.57
4	0.15	91.68	92.52	94.83	0.82
5	0.36	80.16	79.18	75.50	0.41

表4

图7

参考文献 24

[1]	叶利振. 基于DGA的变压器绝缘油分析与故障诊断在澳门电网中的应用研究[D]. 广州: 华南理工大学, 2016.
	YE Lizhen. Research on the application of transformer insulating oil analysis and fault diagnosis based on DGA in the Macau Power Grid[D]. Guangzhou: South China University of Technology, 2016.
[2]	蒲丽娟, 刘念, 刘航宇, 等. 基于油浸变压器故障数据的IEC三比值统计分析[J]. 陕西电力, 2015, 43(5): 41-44, 49.
	PU Lijuan, LIU Nian, LIU Hangyu, et al. Statistical analysis of IEC three-ratio method based on fault data of oil-immersed transformers[J]. Shanxi Electric Power, 2015, 43(5): 41-44, 49.
[3]	WANG Y, ZHU Y L, ZHANG L G. Credibility-based fault fusion diagnosis method for power transformer[J]. IETE Journal of Research, 2022: 1-8.
[4]	胡晨, 尹恩韬, 乐健. 基于多源数据融合的变压器典型故障诊断模型研究[J]. 电工电气, 2023,(7): 41-45, 65.
	HU Chen, YIN Entao, LE Jian. Research on typical fault diagnosis model of transformers based on multi-source data fusion[J]. Electrotechnics Electric, 2023,(7): 41-45, 65.
[5]	孙川. 基于多源异构数据融合的变压器故障诊断研究[D]. 北京: 北京邮电大学, 2023.
	SUN Chuan. Research on transformer fault diagnosis based on multi-source heterogeneous data fusion[D]. Beijing: Beijing University of Posts and Telecommunications, 2023.
[6]	赵伟杰. 基于多源信息融合的变压器状态评估方法研究[D]. 沈阳: 沈阳工业大学, 2023.
	ZHAO Weijie. Research on transformer condition evaluation method based on multi-source information fusion[D]. Shenyang: Shenyang University of Technology, 2023.
[7]	吴琼, 崔昊杨, 周坤, 等. 多源检测数据融合的变压器故障诊断模型[J]. 现代电子技术, 2022, 45(1): 181-186.
	WU Qiong, CUI Haoyang, ZHOU Kun, et al. Transformer fault diagnosis model based on multi-source detection data fusion[J]. Modern Electronics Technique, 2022, 45(1): 181-186.
[8]	肖宏磊, 留毅, 夏红军, 等. 基于PCA与SSA-LightGBM的油浸式变压器故障诊断方法[J]. 综合智慧能源, 2023, 45 (3): 9-16.. doi: 10.3969/j.issn.2097-0706.2023.03.002
	XIAO Honglei, LIU Yi, XIA Hongjun, et al. A fault diagnosis method for oil-immersed transformers based on PCA and SSA-LightGBM[J]. Integrated Intelligent Energy, 2023, 45(3): 9-16. doi: 10.3969/j.issn.2097-0706.2023.03.002
[9]	陈月, 许秀英, 梁春英, 等. 多源信息融合在变压器故障诊断领域中的应用[J]. 农机使用与维修, 2023, (6): 108-112.
	CHEN Yue, XU Xiuying, LIANG Chunying, et al. Application of multi-source information fusion in transformer fault diagnosis[J]. Agricultural Machinery Using and Maintenance, 2023, (6): 108-112.
[10]	WANG H, ZHANG Z. TATCN: Time series prediction model based on time attention mechanism and TCN[C]// 2022 IEEE 2nd international conference on Computer Communication and Artificial Intelligence (CCAI). IEEE, 2022: 26-31.
[11]	竺筱晶, 薛睿萌. 基于小波变换的Bi-LSTM-TCN短期电价预测[J]. 电工电能新技术, 2023, 42(12): 60-68. doi: 10.12067/ATEEE2208070
	ZHU Xiaojing, XUE Ruimeng. Wavelet transform-based Bi-LSTM-TCN method for short-term electricity price forcasting[J]. Advanced Technology of Electrical Engineering and Energy, 2023, 42(12): 60-68. doi: 10.12067/ATEEE2208070
[12]	付豪, 邹花蕾, 张腾飞. 基于长短期记忆循环神经网络的变电站监控系统智能故障推理方法[J]. 综合智慧能源, 2022, 44 (12): 11-17. doi: 10.3969/j.issn.2097-0706.2022.12.002
	FU Hao, ZOU Hualei, ZHANG Tengfei. Intelligent fault reasoning method for the substation monitoring system based on LSTM[J]. Integrated Intelligent Energy, 2022, 44(12): 11-17. doi: 10.3969/j.issn.2097-0706.2022.12.002
[13]	杨海晶. 基于TCN与DBN组合模型的变压器状态预测和故障诊断的研究[D]. 镇江: 江苏大学, 2022.
	YANG Haijing. Research on transformer state prediction and fault diagnosis based on TCN and DBN combined model[D]. Zhenjiang: Jiangsu University, 2022.
[14]	DESUKY A S, CIFCI M A, KAUSAR S, et al. Mud Ring Algorithm: A new meta-heuristic optimization algorithm for solving mathematical and engineering challenges[J]. IEEE Access, 2022, 10: 50448-50466.
[15]	黄星华, 吴天舒, 杨龙玉, 等. 一种面向旋转机械的基于Transformer特征提取的域自适应故障诊断[J]. 仪器仪表学报, 2022, 43(11): 210-218.
	HUANG Xinghua, WU Tianshu, YANG Longyu, et al. Domain adaptive fault diagnosis based on Transformer feature extraction for rotating machinery[J]. Chinese Journal of Scientific Instrument, 2022, 43(11): 210-218.
[16]	何志铧, 熊祖强. 基于Informer神经网络的工作面矿压预测研究[J]. 矿业研究与开发, 2024, 44(7): 142-148.
	HE Zhihua, XIONG Zuqiang. Research on Ming pressure prediction of working face based on informer neural network[J]. Mining Research and Development, 2024, 44(7): 142-148.
[17]	颜劲夫, 何其骏, 瞿业峰, 等. 基于CNN-BiLSTM及ResNet网络的板中损伤TFM定位与检测研究[J]. 南京大学学报(自然科学), 2024, 60(4): 566-576.
	YAN Jinfu, HE Qijun, QU Yefeng, et al. Research on TFM localization and detection of damage in plates based on CNN-BiLSTM and ResNet networks[J]. Journal of Nanjing University (Natural Science), 2024, 60(4): 566-576.
[18]	宋诚, 夏翔, 王鑫一, 等. 基于MFCC和CNN的变压器声学特征提取及故障识别[J]. 电工电气, 2023(6): 49-54.
	SONG Cheng, XIA Xiang, WANG Xinyi, et al. Transformer acoustic feature extraction and fault identification based on MFCC and CNN[J]. Electrotechnics Electric, 2023(6): 49-54.
[19]	范焱, 刘乔, 袁笛, 等. 空域和频域特征解耦的红外与可见光图像融合[J]. 红外与激光工程, 2024, 53(8): 222-237.
	FAN Yan, LIU Qiao, YUAN Di, et al. Spatial and frequency domain feature decoupling for infrared and visible image fusion[J]. Infrared and Laser Engineering, 2024, 53(8): 222-237.
[20]	毕福昆, 李子静, 王彦平, 等. 基于Transformer和CNN特征融合的非接触式心率检测算法[J]. 信号处理, 2023, 39(11): 2062-2070.
	BI Fukun, LI Zijing, WANG Yanping, et al. Non-contact heart rate detection algorithm based on transformer and CNN feature fusion[J]. Journal of Signal Processing, 2023, 39(11): 2062-2070.
[21]	滕志军, 付雨珊, 谷梁岑, 等. 融合动态权重系数与Levy飞行的黏菌优化算法[J]. 陕西科技大学学报, 2024, 42(4): 191-198.
	TENG Zhijun, FU Yushan, GU Liangcen, et al. An optimized slime mould algorithm combining dynamic weight coefficient and Levy flight[J]. Journal of Shanxi University of Science and Technology, 2024, 42(4): 191-198.
[22]	薛睿超, 牛富强. 一种适用于被动声学监测系统的海豚回声定位信号快速检测方法[J]. 声学技术, 2023, 42(6): 733-740.
	XUE Ruichao, NIU Fuqiang. A fast detection method of dolphin echolocation signal for passive acoustic monitoring system[J]. Technical Acoustics, 2023, 42(6): 733-740.
[23]	李杰. 基于仿生海豚的水下目标识别技术研究[D]. 哈尔滨: 哈尔滨工程大学, 2023.
	LI Jie. Research on underwater target recognition technology based on bionic dolphin[D]. Harbin:Harbin Engineering University, 2023.
[24]	付乐乐, 陈宏, 叶镇豪, 等. 基于模型预测控制的仿生海豚路径跟踪研究[J]. 传感器与微系统, 2024, 43(3): 31-34.
	FU Lele, CHEN Hong, YE Zhenhao, et al. Path tracking research of bionic dolphin based on MPC[J]. Transducer and Microsystem Technologies, 2024, 43(3): 31-34.

方案	输入数据集	预测模型
1	仅DGA数据	TCN+MRA
2	DGA+IHVBDI+UDDI+UHFPDDI数据	TCN
3	DGA+IHVBDI数据	TCN+MRA
4	DGA+IHVBDI+UDDI+UHFPDDI数据	TCN+MRA
5	仅DGA数据	TCN