Research on power transformer fault diagnosis algorithm based on fuzzy reinforcement learning

doi:10.3969/j.issn.2097-0706.2024.10.007

Abstract

Abstract:

Currently， there are common problems of low accuracy and limited recognition of fault types in power transformer fault diagnosis. An adaptive transformer fault diagnosis model based on fuzzy reinforcement learning and decision tree algorithm is proposed. Firstly， by conducting Dissolved Gas Analysis（DGA） on real transformers， a series of variables that can reflect the status of transformers are extracted. Then， the decision tree J48 algorithm is used to screen these variables and select the 8 most representative variables， aiming to achieve high classification accuracy with the least number of input variables. Finally， the selected variables are input into the fuzzy reinforcement learning classifier for fault diagnosis. The experimental results show that the constructed fault diagnosis model is more accurate， with an accuracy of up to 99.7%.Compared to traditional DGA fault recognition algorithms， the diagnostic algorithm based on fuzzy reinforcement learning proposed has a higher accuracy in diagnosing power transformer faults.

Key words: transformer, DGA, fuzzy learning, decision tree J48, fault diagnosis

CLC Number:

TM273

ZHANG Kao, HE Kailin, YANG Peihao. Research on power transformer fault diagnosis algorithm based on fuzzy reinforcement learning[J]. Integrated Intelligent Energy, 2024, 46(10): 48-55.

Figures/Tables 11

Table 1

Comparison of traditional DGAs

诊断方法	输入向量	可识别故障数量
IEC	$[C H 4 / H 2, C 2 H 2 / C 2 H 4, C 2 H 4 / C 2 H 6] T$	9
Rogers	$[C H 4 / H 2, C 2 H 2 / C 2 H 4, C 2 H 4 / C 2 H 6, C 2 H 6 / C H 4] T$	12
Doernenburg	$[C H 4 / H 2, C 2 H 2 / C 2 H 4, C 2 H 2 / C H 4, C 2 H 6 / C 2 H 2] T$	3
Modified Rogers	$[C H 4 / H 2, C 2 H 2 / C 2 H 4, C 2 H 4 / C 2 H 6] T$	6
Duval Triangle	$[C H 4 C H 4 + C 2 H 2 + C 2 H 4, C 2 H 2 C H 4 + C 2 H 2 + C 2 H 4, C 2 H 4 C H 4 + C 2 H 2 + C 2 H 4] T$	6
Key-Gases	$[H 2, C H 4, C 2 H 2, C 2 H 4, C 2 H 6, C O] T$	4

Table 1

Fig.1

Table 2

Table 3

Table 4

Table 5

Action intensity vectors

故障类型	$a i *$ ，维度为1×6 561
正常工况	2	4	1	5	…	2	2	6	1
局部放电	5	4	3	5	…	2	6	1	5
低能放电	1	4	6	5	…	3	4	1	2
高能放电	6	4	2	5	…	2	5	3	1
低温热故障	5	4	3	5	…	2	6	1	5
高温热故障	1	5	2	3	…	4	6	6	1

Table 5

Table 6

Rule intensity vector

故障类型	$β (a *)$ $∀ a ∈ A$ ，维度为1×6
正常工况	0.644 5	0.173 0	0.000 0	0.000 0	0.000 0	0.182 4
局部放电	0.119 3	0.744 4	0.000 1	0.000 0	0.000 1	0.136 1
低能放电	0.093 1	0.000 0	0.744 5	0.000 0	0.000 2	0.162 3
高能放电	0.036 2	0.000 0	0.000 0	0.719 2	0.000 3	0.244 3
低温热故障	0.103 5	0.000 0	0.000 0	0.000 1	0.744 5	0.151 9
高温热故障	0.118 4	0.000 2	0.000 0	0.144 4	0.000 0	0.737 0

Table 6

Table 7

Table 8

Classifier output

故障类别	$β (a *)$
$a i d e n (y k)$ =1	$β (a 1 *) = 0.644 5$
$a i d e n (y k)$ =2	$β (a 2 *) = 0.744 4$
$a i d e n (y k)$ =3	$β (a 3 *) = 0.744 5$
$a i d e n (y k)$ =4	$β (a 4 *) = 0.719 2$
$a i d e n (y k)$ =5	$β (a 5 *) = 0.744 5$
$a i d e n (y k)$ =6	$β (a 6 *) = 0.737 0$

Table 8

Table 9

Fig.2

References 16

[1]	黄艺璇, 杨世海, 曹晓冬, 等. 基于分段电流特征和随机森林算法的低压配电网拓扑识别方法[J]. 电力需求侧管理, 2023, 25(2):63-69.
	HUANG Yixuan, YANG Shihai, CAO Xiaodong, et al. Topology identification method of low voltage distribution network based on segmented current characteristics and random forest algorithm[J]. Power Demand Side Management, 2023, 25(2): 63-69.
[2]	刘钧迪, 韦德福, 王飞鸣, 等. 油浸式配电变压器入网检测可拓综合评价[J]. 东北电力技术, 2023, 44(7):34-39.
	LIU Jundi, WEI Defu, WANG Feiming, et al. Comprehensive evaluation of oil immersed distribution transformer network access detection extenics[J]. Northeast Electric Power Technology, 2023, 44(7):34-39.
[3]	谭珊, 赵仲勇, 杨建, 等. 基于动态时间规整的变压器绕组变形故障诊断方法研究[J]. 高压电器, 2024, 60(2):108-118.
	TAN Shan, ZHAO Zhongyong, YANG Jian, et al. Research on diagnosis method of transformer winding deformation fault based on dynamic time warping[J]. High Voltage Electrical Appliances, 2024, 60(2):108-118.
[4]	朱亮亮, 滕姗姗, 徐辰冠, 等. 基于自编码多层感知器的电力负荷管理终端运行故障诊断方法[J]. 电力需求侧管理, 2024, 26(2):113-118.
	ZHU Liangliang, TENG Shanshan, XU Chenguan, et al. Operating fault diagnosis method for power load management terminals based on auto-encoder multilayer perception[J]. Power Demand Side Management, 2024, 26 (2): 113-118.
[5]	潘志城, 周海滨, 邓军, 等. 一起500 kV油浸纸绝缘套管油色谱数据异常原因分析[J]. 变压器, 2020(6):67-72.
	PAN Zhicheng, ZHOU Haibin, DENG Jun, et al. Analysis of abnormal oil chromatography of 500 kV oil immersed paper insulation bushing[J]. Transformer, 2020(6): 67-72.
[6]	沈琪, 倪钱杭, 章岸, 等. 变压器油色谱在线监测在缺陷发现中的案例分析[J]. 变压器, 2024, 61(3):72-75.
	SHEN Qi, NI Qianhang, ZHANG An, et al. Case analysis of transformer oil chromatography online monitoring in defect detection[J]. Transformer, 2024, 61 (3): 72-75.
[7]	龙玉江, 姜超颖, 钟掖, 等. 极端气象条件下基于深度学习网络特征的变压器故障预测[J]. 现代电子技术, 2024, 47(4):91-96.
	LONG Yujiang, JIANG Chaoying, ZHONG Ye, et al. Transformer fault prediction based on deep learning network features under extreme weather conditions[J]. Modern Electronic Technology, 2024, 47 (4): 91-96.
[8]	王科, 苟家萁, 彭晶, 等. 基于LSTM网络的变压器油中溶解气体浓度预测[J]. 电子测量技术, 2020, 43(4):81-87.
	WANG Ke, GOU Jiaqi, PENG Jing, et al. Prediction of dissolved gas concentration in transformer oil based on LSTM network[J]. Electronic Measurement Technology, 20020, 43(4):81-87.
[9]	XIAO Y, PAN W G, GUO X M, et al. Fault diagnosis of traction transformer based on Bayesian network[J]. Energies, 2020, 13(18):4966.
[10]	ZHAI S L, CHEN X W, WEI L, et al. Research on identification methods of gas content in transformer insulation oil based on deep transfer network[J]. Journal of Materials Science: Materials in Electronics, 2020, 31(18):15764-15772.
[11]	刘云鹏, 许自强, 董王英, 等. 基于经验模态分解和长短期记忆神经网络的变压器油中溶解气体浓度预测方法[J]. 中国电机工程学报, 2019, 39(13):3998-4008.
	LIU Yunpeng, XU Ziqiang, DONG Wangying, et al. Prediction method of dissolved gas concentration in transformer oil based on empirical mode decomposition and long short-term memory neural network[J]. Proceedings of the CSEE 2019, 39 (13): 3998-4008.
[12]	陈浈斐, 章黄勇, 马宏忠, 等. 基于深度学习的电力设备故障诊断方法研究综述[J]. 电气自动化, 2022, 44(1):1-2,6.
	CHEN Zhenfei, ZHANG Huangyong, MA Hongzhong, et al. A review of deep learning based fault diagnosis methods for power equipment[J]. Electrical Automation, 2022, 44 (1): 1-2,6.
[13]	王鑫明, 付永波, 李俊涛. 基于决策树算法的电力设备故障诊断辅助决策方法[J]. 自动化应用, 2023, 64(12):128-131.
	WANG Xinming, FU Yongbo, LI Juntao. Auxiliary decision method for power equipment fault diagnosis based on decision tree algorithm[J]. Automation Application, 2023, 64 (12): 128-131.
[14]	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. Oil-immersed transformer fault diagnosis method 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
[15]	IEEE. Guide for the interpretation of gases generated in oil-immersed transformers:IEEE Std C57.104—1991[S].
[16]	熊真真. 基于决策树的上海商业建筑虚拟电厂执行力分析[J]. 综合智慧能源, 2023, 45(6): 66-72. doi: 10.3969/j.issn.2097-0706.2023.06.009
	XIONG Zhenzhen. Analysis on execution of VPPs for commercial buildings in Shanghai based on decision tree[J]. Integrated Intelligent Energy, 2023, 45(6): 66-72. doi: 10.3969/j.issn.2097-0706.2023.06.009

故障类型	C₂H₄	CH₄	C₂H₆	H₂	C₂H₂
正常工况	6	2	12	<5	<0.5
局部放电	18	1	25	29	<0.5
低能放电	2	2	2	<5	<0.5
高能放电	4	<1	<1	6	<0.5
低温热故障	17	3	6	<5	<0.5
高温热故障	327	171	106	7	<0.5

项目	不同算法识别出的样本数
项目	ANN	FL	SVM	FRL
2个样本	1	0	2	2
4个样本	4	2	4	4
6个样本	3	2	5	6
8个样本	7	5	8	8
10个样本	8	6	9	9
38个样本	34	33	37	38
实验室模拟油样 58个样本	53	51	56	57