基于SMOTE与Bayes优化的LSTM网络变压器故障诊断

doi:10.11930/j.issn.1004-9649.202210046

摘要/Abstract

摘要：

随着电力信息化的提高，智能算法结合历史数据进行变压器故障诊断的方法越来越受到关注。在溶解气体分析法基础上借助少数类样本过采样（SMOTE）算法合成新样本，实现样本多维度扩充，并以贝叶斯优化算法寻找长短期记忆（LSTM）网络模型参数的最优设置值，以降低训练集错误率，进而建立了变压器故障诊断模型。结果表明：样本扩充后的变压器故障诊断模型过拟合度降低约20%，测试集准确率提升约10%。

关键词: 变压器, 故障诊断, 采样, 长短时记忆网络

Abstract:

With the improvement of power informatization, the method of transformer fault diagnosis based on intelligent algorithm and historical data has been paid more and more attention. On the basis of dissolved gas analysis, synthetic minority oversampling technique (SMOTE) algorithm was used to synthesize new samples, realize multi-dimensional expansion of samples, and use Bayes optimization algorithm to find the best setting value of long short term memory (LSTM) network model parameters to reduce the error rate of training set, and then establish transformer fault diagnosis model. The results show that the overfitting degree of the transformer fault diagnosis model after sample expansion is reduced by about 20%, and the accuracy of the test set is increased by about 10%.

Key words: transformer, fault diagnosis, sampling, long short-term memory network

张宏杰, 陈贵凤, 闫宏伟, 杨晓龙, 侯天仁, 张伟. 基于SMOTE与Bayes优化的LSTM网络变压器故障诊断[J]. 中国电力, 2023, 56(10): 164-170.

Hongjie ZHANG, Guifeng CHEN, Hongwei YAN, Xiaolong YANG, Tianren HOU, Wei ZHANG. Fault Diagnosis of LSTM Network Tansformer Based on SMOTE and Bayes Optimization[J]. Electric Power, 2023, 56(10): 164-170.

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://www.electricpower.com.cn/CN/10.11930/j.issn.1004-9649.202210046

https://www.electricpower.com.cn/CN/Y2023/V56/I10/164

图/表 8

参考文献 25

1	廖伟涵, 郭创新, 金宇, 等. 基于四阶段预处理与GBDT的油浸式变压器故障诊断方法[J]. 电网技术, 2019, 43 (6): 2195- 2203.
	LIAO Weihan, GUO Chuangxin, JIN Yu, et al. Oil-immersed transformer fault diagnosis method based on four-stage preprocessing and GBDT[J]. Power System Technology, 2019, 43 (6): 2195- 2203.
2	夏飞, 罗志疆, 张浩, 等. 混合神经网络在变压器故障诊断中的应用[J]. 电子测量与仪器学报, 2017, 31 (1): 118- 124.
	XIA Fei, LUO Zhijiang, ZHANG Hao, et al. Application of mixed neural network in transformer fault diagnosis[J]. Journal of Electronic Measurement and Instrumentation, 2017, 31 (1): 118- 124.
3	张奎, 王建南, 王肖峰. 基于神经网络的变压器故障诊断[J]. 电子测量技术, 2017, 40 (12): 98- 101.
	ZHANG Kui, WANG Jiannan, WANG Xiaofeng. Fault diagnosis of transformer based on neural netwo[J]. Electronic Measurement Technology, 2017, 40 (12): 98- 101.
4	赵峰, 刘君. 基于DGA与神经网络融合的变压器故障诊断方法[J]. 机电工程, 2008, 25 (11): 27- 30, 34.
	ZHAO Feng, LIU Jun. Transformer fault diagnosis approach based on DGA and neural network[J]. Mechanical & Electrical Engineering Magazine, 2008, 25 (11): 27- 30, 34.
5	缪芸, 缪翼军, 陈红坤, 等. 基于模糊层次分析法与支持向量机的变压器风险评估[J]. 现代电力, 2014, 31 (6): 64- 69.
	MIAO Yun, MIAO Yijun, CHEN Hongkun, et al. Risk assessment of transformer based on fuzzy analytic hierarchy process method and support vector machine[J]. Modern Electric Power, 2014, 31 (6): 64- 69.
6	熊铠. 基于故障树的电力变压器故障分析[J]. 机电信息, 2019, (30): 25, 28.
	XIONG Kai. Fault analysis of power transformer based on fault tree[J]. Mechanical and Electrical Information, 2019, (30): 25, 28.
7	张邵杰. 基于烟花算法优化SVM的变压器故障诊断[J]. 机电信息, 2021, (22): 30- 31.
	ZHANG Shaojie. Transformer fault diagnosis based on SVM optimized by fireworks algorithm[J]. Mechanical and Electrical Information, 2021, (22): 30- 31.
8	夏洪刚, 郭红兵, 肖金超. 基于CNN的电力变压器故障诊断方法[J]. 电子设计工程, 2020, 28 (13): 189- 193.
	XIA Honggang, GUO Hongbing, XIAO Jinchao. Power transformer fault diagnosis method based on CNN[J]. Electronic Design Engineering, 2020, 28 (13): 189- 193.
9	陈尔奎, 韩清春, 周栾. 基于布谷鸟算法和BP神经网络的矿用变压器故障诊断[J]. 煤炭技术, 2018, 37 (6): 223- 224.
	CHEN Erkui, HAN Qingchun, ZHOU Luan. Fault diagnosis of mine-used transformer based on cuckoo search and BP neural network[J]. Coal Technology, 2018, 37 (6): 223- 224.
10	陈欢, 彭辉, 舒乃秋, 等. 基于蝙蝠算法优化最小二乘双支持向量机的变压器故障诊断[J]. 高电压技术, 2018, 44 (11): 3664- 3671.
	CHEN Huan, PENG Hui, SHU Naiqiu, et al. Fault diagnosis of transformer based on LS-TSVM optimized by bat algorithm[J]. High Voltage Engineering, 2018, 44 (11): 3664- 3671.
11	崔宇, 侯慧娟, 胥明凯, 等. 基于双重注意力机制的变压器油中溶解气体预测模型[J]. 中国电机工程学报, 2020, 40 (1): 338- 347, 400.
	CUI Yu, HOU Huijuan, XU Mingkai, et al. A prediction method for dissolved gas in power transformer oil based on dual-stage attention mechanism[J]. Proceedings of the CSEE, 2020, 40 (1): 338- 347, 400.
12	孙先亮, 谭小波. 基于LSTM的动态网络异常行为检测方法[J]. 沈阳理工大学学报, 2021, 40 (6): 22- 26.
	SUN Xianliang, TAN Xiaobo. Detection method of abnormal behavior of dynamic network based on LSTM[J]. Journal of Shenyang Ligong University, 2021, 40 (6): 22- 26.
13	陈飞, 程合彬, 王伟光. 基于CNN-LSTMs混合模型的人体行为识别方法[J]. 信息技术与信息化, 2019, (4): 32- 34.
	CHEN Fei, CHENG Hebin, WANG Weiguang. Human behavior recognition method based on CNN-LSTMs mixed model[J]. Information Technology and Informatization, 2019, (4): 32- 34.
14	徐宇颂, 邹山花, 卢先领. 基于特征选择和组合模型的短期电力负荷预测[J]. 中国电力, 2022, 55 (7): 121- 127.
	XU Yusong, ZOU Shanhua, LU Xianling. Short-term load forecasting based on feature selection and combination model[J]. Electric Power, 2022, 55 (7): 121- 127.
15	张文斌, 王博翔, 赵长颖. 基于贝叶斯优化的选择性热光伏辐射器设计[J]. 工程热物理学报, 2021, 42 (11): 2965- 2971.
	ZHANG Wenbin, WANG Boxiang, ZHAO Changying. Designing selective thermophotovoltaic emitter through Bayesian optimization[J]. Journal of Engineering Thermophysics, 2021, 42 (11): 2965- 2971.
16	石怀涛, 尚亚俊, 白晓天, 等. 基于贝叶斯优化的SWDAE-LSTM滚动轴承早期故障预测方法研究[J]. 振动与冲击, 2021, 40 (18): 286- 297.
	SHI Huaitao, SHANG Yajun, BAI Xiaotian, et al. Early fault prediction method combining SWDAE and LSTM for rolling bearings based on Bayesian optimization[J]. Journal of Vibration and Shock, 2021, 40 (18): 286- 297.
17	杨如民, 陈敏, 余成波. 基于贝叶斯概率优化的Wi-Fi室内定位算法[J]. 计算机应用与软件, 2021, 38 (2): 97- 102, 144.
	YANG Rumin, CHEN Min, YU Chengbo. A Wi-Fi indoor positioning algorithm based on Bayesian probability optimization[J]. Computer Applications and Software, 2021, 38 (2): 97- 102, 144.
18	石洪波, 陈雨文, 陈鑫. SMOTE过采样及其改进算法研究综述[J]. 智能系统学报, 2019, 14 (6): 1073- 1083.
	SHI Hongbo, CHEN Yuwen, CHEN Xin. Summary of research on SMOTE oversampling and its improved algorithms[J]. CAAI Transactions on Intelligent Systems, 2019, 14 (6): 1073- 1083.
19	李亮, 范瑾, 闫林, 等. 基于混合采样和支持向量机的变压器故障诊断[J]. 中国电力, 2021, 54 (12): 150- 155.
	LI Liang, FAN Jin, YAN Lin, et al. Transformer fault diagnosis based on hybrid sampling and support vector machines[J]. Electric Power, 2021, 54 (12): 150- 155.
20	张明龙, 张振宇, 罗翔, 等. 基于多核支持向量机的混合扰动波形辨识算法研究[J]. 电力系统保护与控制, 2022, 50 (15): 43- 49.
	ZHANG Minglong, ZHANG Zhenyu, LUO Xiang, et al. Complex disturbance waveform recognition based on a multi-kernel support vector machine[J]. Power System Protection and Control, 2022, 50 (15): 43- 49.
21	县国成, 王永攀, 高俊, 等. 基于ELM-SVM模型与电能计量大数据的窃电识别技术研究[J]. 智慧电力, 2022, 50 (9): 82- 89.
	XIAN Guocheng, WANG Yongpan, GAO Jun, et al. Power theft identification technology based on ELM-SVM model and big data of electric energy measurement[J]. Smart Power, 2022, 50 (9): 82- 89.
22	刘海涛, 叶筱怡, 吕干云, 等. 基于BAS-SVM的配电网电压暂降源识别[J]. 中国电力, 2022, 55 (5): 128- 133.
	LIU Haitao, YE Xiaoyi, LÜ Ganyun, et al. Identification of voltage sag source in distribution network based on BAS-SVM[J]. Electric Power, 2022, 55 (5): 128- 133.
23	王志强, 董忠涛, 王骁龙, 等. 考虑建筑物遮蔽的电力应急通信基站选址双层优化方法[J]. 电力系统保护与控制, 2022, 50 (17): 107- 116.
	WANG Zhiqiang, DONG Zhongtao, WANG Xiaolong, et al. Two-layer optimization method for site selection of a power emergency communication base station considering building shadowing effect[J]. Power System Protection and Control, 2022, 50 (17): 107- 116.
24	吉兴全, 陈金硕, 张玉敏, 等. 基于CNN-SVM的配电网故障分类研究[J]. 智慧电力, 2022, 50 (1): 94- 100.
	JI Xingquan, CHEN Jinshuo, ZHANG Yumin, et al. Fault classification in distribution network based on CNN-SVM[J]. Smart Power, 2022, 50 (1): 94- 100.
25	郑业爽, 李世春, 鲁玲. 基于多策略ISOA优化SVM的变压器故障诊断研究[J]. 智慧电力, 2023, 51 (2): 38- 44.
	ZHENG Yeshuang, LI Shichun, LU Ling. Transformer fault diagnosis based on multi-strategy ISOA optimized SVM[J]. Smart Power, 2023, 51 (2): 38- 44.

序号	H₂	CH₄	C₂H₆	C₂H₄	C₂H₂
1	78.0	28.00	13.00	29.00	110.00
2	0.0	5.20	5.12	9.58	14.60
3	66.0	8.27	8.21	9.21	8.21
4	49.1	12.20	0.30	3.90	4.80
5	59.0	10.40	4.00	10.00	12.70
6	30.0	7.40	8.50	1.80	19.00
7	176.0	206.00	47.70	75.70	68.70
8	345.0	112.30	27.50	51.50	58.80
9	61.5	24.60	1.33	5.60	20.50
10	58.0	44.90	11.00	20.60	23.50
11	86.0	112.00	67.00	46.00	85.80
12	20.0	13.00	35.00	27.00	36.30
13	2781.0	1293.00	248.00	23.00	84.00

序号	H₂	CH₄	C₂H₆	C₂H₄	C₂H₂
1	78.0	28.00	13.00	29.00	110.00
2	0.0	5.20	5.12	9.58	14.60
3	66.0	8.27	8.21	9.21	8.21
4	49.1	12.20	0.30	3.90	4.80
5	59.0	10.40	4.00	10.00	12.70
6	30.0	7.40	8.50	1.80	19.00
7	176.0	206.00	47.70	75.70	68.70
8	345.0	112.30	27.50	51.50	58.80
9	61.5	24.60	1.33	5.60	20.50
10	58.0	44.90	11.00	20.60	23.50
11	86.0	112.00	67.00	46.00	85.80
12	20.0	13.00	35.00	27.00	36.30
13	2781.0	1293.00	248.00	23.00	84.00

故障类型	H₂	CH₄	C₂H₆	C₂H₄	C₂H₂
高温过热	3.76~274	22.7~766	7.22~727.56	56~2434.25	0~153.36
局部放电	180.85~ 2587.2	0.574~125	0.234~137	0.188~52	0~51.31
中温过热	19~679	27.8~4992	20.2~1823	30~3671	0~95
低温过热	16~565	11.8~170	3~597	0.6~50.7	0~0.7
低能放电	0~2781	5.2~1293	0.3~248	1.8~75.7	4.8~110
高能放电	31~1678	5.5~652.9	0.1~80.7	4.7~1005.9	10~419.1

故障类型	H₂	CH₄	C₂H₆	C₂H₄	C₂H₂
高温过热	3.76~274	22.7~766	7.22~727.56	56~2434.25	0~153.36
局部放电	180.85~ 2587.2	0.574~125	0.234~137	0.188~52	0~51.31
中温过热	19~679	27.8~4992	20.2~1823	30~3671	0~95
低温过热	16~565	11.8~170	3~597	0.6~50.7	0~0.7
低能放电	0~2781	5.2~1293	0.3~248	1.8~75.7	4.8~110
高能放电	31~1678	5.5~652.9	0.1~80.7	4.7~1005.9	10~419.1

采样过程	根样本					新样本
采样过程	H₂	CH₄	C₂H₆	C₂H₄	C₂H₂	H₂	CH₄	C₂H₆	C₂H₄	C₂H₂
1	18.6	110.2	162.3	50.7	0.7	42.2	110.1	160.6	50.5	0.5
2	101.0	169.0	595.0	34.0	0.0	100.3	169.3	595.0	34.0	0.0
3	565.0	53.0	34.0	47.0	0.0	435.1	56.6	36.1	45.5	0.0
4	95.0	110.0	160.0	50.0	0.0	29.7	50.9	50.3	25.8	0.0
5	16.0	38.4	70.0	28.0	0.0	16.0	38.4	62.3	24.5	0.0
6	16.0	38.4	61.5	15.8	0.0	16.0	38.4	62.0	19.2	0.0
7	23.0	11.8	3.0	0.6	0.0	19.0	20.3	3.2	14.9	0.0