飞机主电源系统关键器件健康状态评估研究

doi:10.11985/2023.04.021

摘要/Abstract

摘要：

采用数据驱动的方法实现飞机主电源系统关键器件的健康状态评估。选取飞机主电源系统中故障率较高的旋转整流器整流二极管和调压器功率(Metal oxide semiconductor field effect transistor, MOSFET)管作为系统关键器件，并分析确定了关键器件的老化特性；利用仿真数据筛选出励磁机的励磁电压平均值和励磁电流平均值作为表征系统关键器件健康状态的电源系统特征变量；采用主成分分析法对系统特征变量数据进行解耦，得到可以分别表征两个关键器件健康状态的特征参数；利用高斯混合模型建立了关键器件的健康基准模型，并计算系统老化模型与健康基准模型的马氏距离；将马氏距离结果经过K-means聚类分析得到包含关键器件健康状态等级信息的训练样本集数据，将样本集数据经过神经网络训练得到系统关键器件的健康状态评估分类函数。采用蒙特卡罗仿真获取大量数据对分类函数进行验证，关键器件不同健康状态等级数据的原始分类和分类函数评估结果的匹配度达到90%以上，说明了该健康评估方法的有效性。

关键词: 飞机主电源系统, 健康状态评估, 主成分分析, K-means聚类, BP(Back propagation)神经网络

Abstract:

Data-driven method is used to evaluate the health status of the key components in the aircraft main power system. The rectifier diodes in the rotating rectifier and the metal oxide semiconductor field effect transistor(MOSFET) in the voltage regulator are selected as the key components of the aircraft main power system with high failure rate, and their aging characteristics are analyzed and determined. Using the simulation data, the average excitation voltage and the average excitation current of the exciter are selected as the characteristic variables of the power system to represent the health status of the key components of the system. Principal component analysis(PCA) is used to decouple the data of system characteristic variables, and the characteristic parameters which can represent the aging status of two key components are obtained. The Gaussian mixture model is used to establish the health reference model of key components, and the Mahalanobis distance between the system aging model and the health reference model is calculated. The Mahalanobis distance data is processed by K-means clustering analysis to obtain the data training sample set containing the health status level information of key components, and neural network is used to train the data of the sample set to obtain the health status evaluation classification functions of the key components of the system. Monte Carlo simulation is used to obtain a large amount of data to verify the classification function. The matching degree between the original data and the evaluation results of the classification function of different health status level data of key components is more than 90%, which shows the effectiveness of this health assessment method.

Key words: Aircraft main power system, health assessment, principal component analysis, K-means clustering, back propagation neural network

中图分类号:

TM76

李小宁, 高朝晖, 王爽, 汤孝, 李一卓. 飞机主电源系统关键器件健康状态评估研究[J]. 电气工程学报, 2023, 18(4): 188-198.

LI Xiaoning, GAO Zhaohui, WANG Shuang, TANG Xiao, LI Yizhuo. Research on Health Assessment Method of Key Components in Aircraft Main Power System[J]. Journal of Electrical Engineering, 2023, 18(4): 188-198.

图/表 15

图1

表1

图2

图3

图4

表2

表3

图5

图6

图7

图8

图9

图10

表4

表5

参考文献 28

[1]	秦海鸿, 严仰光. 多电飞机的电气系统[M]. 北京: 北京航空航天大学出版社, 2016.
	QIN Haihong, YAN Yangguang. Electrical system of multi-electric aircraft[M]. Beijing: Beihang University Press, 2016.
[2]	PAN Haiyang, DONG Ensheng, JIANG Yilin, et al. Prognostic and health management for aircraft electrical power supply system[C]// Proceedings of the IEEE 2012 Prognostics and System Health Management Conference (PHM-2012 Beijing),Beijing,China. IEEE, 2012:1-5.
[3]	BATZEL T D, SWANSON D C. Prognostic health management of aircraft power generators[J]. IEEE Transactions on Aerospace and Electronic Systems, 2009, 45(2):473-482. doi: 10.1109/TAES.2009.5089535
[4]	张敬, 李颖晖, 朱喜华, 等. 三级式发电机旋转整流器故障特征提取[J]. 微电机, 2011(7):99-103.
	ZHANG Jing, LI Yinghui, ZHU Xihua, et al. Fault feature extraction of three-stage generator rotating rectifier[J]. Micromotor, 2011(7):99-103.
[5]	沈颂华, 朱新宇. 基于故障字典的旋转整流器故障检测方法[C]// 中国航空学会航空电气工程第三届学术年会论文集, 1999:22-26.
	SHEN Songhua, ZHU Xinyu. Fault detection method of rotating rectifier based on fault dictionary[C]// Proceedings of the Third Annual Academic Conference of Aeronautical Electrical Engineering of Chinese Aeronautical Society, 1999:22-26.
[6]	吴克雄, 王振华. 基于专家系统和数据驱动的健康评估分析方法[J]. 黑龙江科学, 2019, 10(16):64-65.
	WU Kexiong, WANG Zhenhua. Health assessment analysis method based on expert system and data driven[J]. Heilongjiang Science, 2019, 10(16):64-65.
[7]	龙凤, 王忠, 程绪建, 等. 一种基于信息融合的军用电子产品PHM方案设计[J]. 微电子学与计算机, 2010, 27(9):86-90.
	LONG Feng, WANG Zhong, CHENG Xujian, et al. A PHM scheme design of military electronic products based on information fusion[J]. Microelectronics and Computer, 2010, 27(9):86-90.
[8]	崔江, 唐军祥, 龚春英, 等. 一种基于改进堆栈自动编码器的航空发电机旋转整流器故障特征提取方法[J]. 中国电机工程学报, 2017, 37(19):5696-5706.
	CUI Jiang, TANG Junxiang, GONG Chunying, et al. A fault feature extraction method of aerogenerator rotating rectifier based on improved stack automatic encoder[J]. Proceedings of the CSEE, 2017, 37(19):5696-5706.
[9]	崔江, 郭瑞东, 张卓然, 等. 基于改进DBN的发电机旋转整流器故障特征提取技术[J]. 中国电机工程学报, 2020, 40(7):326-333,372.
	CUI Jiang, GUO Ruidong, ZHANG Zhuoran, et al. Fault feature extraction technology of generator rotating rectifier based on improved DBN[J]. Proceedings of the CSEE, 2020, 40(7):326-333,372.
[10]	崔江, 冯赛, 张卓然, 等. 基于BLS的无刷发电机旋转整流器特征提取技术研究[J]. 中国电机工程学报, 2020, 40(12):312-321.
	CUI Jiang, FENG Sai, ZHANG Zhuoran, et al. Research on feature extraction technology of brushless generator rotating rectifier based on BLS[J]. Proceedings of the CSEE, 2020, 40(12):312-321.
[11]	张静, 李柠, 李少远, 等. 基于数据的风电机组发电机健康状况评估[J]. 信息与控制, 2018, 47(6):58-65,76-77.
	ZHANG Jing, LI Ning, LI Shaoyuan, et al. Health assessment of wind turbine generator based on data[J]. Information and Control, 2018, 47(6):58-65,76-77.
[12]	王娜, 李号彩, 张德利. 混合聚类分析算法在发电设备故障模式识别中的应用[J]. 电力信息与通信技术, 2017, 15(12):32-35.
	WANG Na, LI Haocai, ZHANG Deli. Application of hybrid cluster analysis algorithm in fault pattern recognition of power generation equipment[J]. Power Information and Communication Technology, 2017, 15(12):32-35.
[13]	XIA Chao, CHENG Xinhong, WANG Zhongjian, et al. On-resistance degradation induced by hot-carrier injection in SOI SJLDMOS[J]. IEEE Transactions on Electron Devices, 2013, 60(3):1279-1281. doi: 10.1109/TED.2013.2242077
[14]	CELAYA J R, PATIL N, SAHA S, et al. Towards accelerated aging methodologies and health management of power MOSFETs (technical brief)[C]// Annual Conference of the Prognostics and Health Management Society,2009,San Diego,CA, 2009:1-6.
[15]	CELAYA J R, SAXENA A, KULKARNI C S, et al. Prognostics of power MOSFETs under thermal stress accelerated aging using data-driven and model-based methodologies[C]// Annual Conference of the Prognostics and Health Management Society,2011,Montreal, 2011:2.
[16]	杨超. 600 V碳化硅结势垒肖特基二极管电学应力下的退化机理及模型研究[D]. 南京: 东南大学, 2014.
	YANG Chao. Study on degradation mechanism and model of 600 V silicon carbide junction barrier Schottky diode under electrical stress[D]. Nanjing: Southeast University, 2014.
[17]	YOSHIDA T, KAWAHARA H, OGAWA S. A new mechanism for degradation of Al-Si-Cu/TiN/Ti contacted p-n junction[C]// 30th Annual Proceedings Reliability Physics,1992,San Diego,CA,USA. IEEE, 1992:344-348.
[18]	陈勇, 陈章勇, 陈燕武. Saber仿真软件的设计与应用[M]. 北京: 科学出版社, 2017.
	CHEN Yong, CHEN Zhangyong, CHEN Yanwu. Design and application of Saber simulation software[M]. Beijing: Science Press, 2017.
[19]	汤孝, 高朝晖, 郗展, 等. 航空发电机健康特征参数与老化模式分析[J]. 航空科学技术, 2021, 32(6):27-35.
	TANG Xiao, GAO Zhaohui, XI Zhan, et al. Analysis of aero-generator health characteristic parameters and aging mode[J]. Aviation Science and Technology, 2021, 32(6):27-35.
[20]	陈佩. 主成分分析法研究及其在特征提取中的应用[D]. 西安: 陕西师范大学, 2014.
	CHEN Pei. Research on principal component analysis and its application in feature extraction[D]. Xi’an: Shaanxi Normal University, 2014.
[21]	韩同瑞. 基于主元分析法的工业锅炉故障诊断的研究[D]. 大连: 大连海事大学, 2012.
	HAN Tongrui. Research on industrial boiler fault diagnosis based on principal component analysis[D]. Dalian: Dalian Maritime University, 2012.
[22]	解素雯. 基于主成分分析与因子分析数学模型的应用研究[D]. 淄博: 山东理工大学, 2016.
	XIE Suwen. Application research of mathematical model based on principal component analysis and factor analysis[D]. Zibo: Shandong University of Technology, 2016.
[23]	华志颖. 基于四分之一超球SVM和主成分分析的WSN异常值检测算法研究[D]. 南京: 南京邮电大学, 2020.
	HUA Zhiying. Research on WSN outlier detection algorithm based on quarter hypersphere SVM and principal component analysis[D]. Nanjing: Nanjing University of Posts and Telecommunications, 2020.
[24]	CARREIRA-PERPINAN M A. Mode-finding for mixtures of Gaussian distributions[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2000, 22(11):1318-1323.
[25]	陈英. 高斯混合模型聚类及其优化算法研究[D]. 南昌: 华东交通大学, 2015.
	CHEN Ying. Research on clustering and optimization algorithm of Gaussian mixture model[D]. Nanchang: East China Jiaotong University, 2015.
[26]	张宏东. EM算法及其应用[D]. 济南: 山东大学, 2014.
	ZHANG Hongdong. EM algorithm and its application[D]. Jinan: Shandong University, 2014.
[27]	BURNHAM K P. Multimodel inference understanding AIC and BIC in model selection[J]. Sociological Methods & Research, 2004, 33(2):261-304.
[28]	林彬, 宋东, 和麟. 基于马氏距离与组距估计的复杂系统健康评估[J]. 仪器仪表学报, 2016, 37(9):2022-2028.
	LIN Bin, SONG Dong, HE Lin. Complex system health assessment based on Mahalanobis distance and group distance estimation[J]. Chinese Journal of Scientific Instrument, 2016, 37(9):2022-2028.

参数	第1阶段分布范围	第2阶段分布范围	第3阶段分布范围	第4阶段分布范围
VF/V	(1.8,0.1)	(2.4,0.1)	(2.9,0.08)	(3.4,0.06)
IR/A	(25μ,0.99)	(75μ,0.33)	(125μ,0.2)	(175μ,0.14)

特征参数	第一阶段	第二阶段	第三阶段
I_ef/A	0.379 4	0.388 6	0.402 3
df_Ief	10.343 3	10.180 0	9.673 3
V_ef/V	4.642 9	4.808 7	4.642 9
df_Vef	54.783 3	53.803 3	48.010 0

仿真组	二极管正向电压/V	二极管反向漏电流/A	MOS管导通电阻/Ω	循环仿真次数
1	(1.8,0.1)	(25μ,0.99)	(0.6,0.5)	30
2			(1.2,0.25)	30
3			(1.8,0.17)	30
4			(2.4,0.12)	30
5	(2.4,0.1)	(75μ,0.33)	(0.6,0.5)	30
6			(1.2,0.25)	30
7			(1.8,0.17)	30
8			(2.4,0.12)	30
9	(2.9,0.08)	(125μ,0.2)	(0.6,0.5)	30
10			(1.2,0.25)	30
11			(1.8,0.17)	30
12			(2.4,0.12)	30
13	(3.4,0.06)	(175μ,0.14)	(0.6,0.5)	30
14			(1.2,0.25)	30
15			(1.8,0.17)	30
16			(2.4,0.12)	30

仿真组	二极管老化阶段	MOS管老化阶段	整流二极管健康等级分类结果
1	优	优	30次“优”
2		良
3		中
4		差
5	良	优	30次“良”
6		良
7		中
8		差
9	中	优	30次“中”
10		良
11		中
12		差
13	差	优	30次“差”
14		良
15		中
16		差

仿真组	二极管老化阶段	MOS管老化阶段	功率MOS管健康等级分类结果
1	优	优	30次“优”
2		良	6次“优”, 22次“良”, 2次“中”
3		中	25次“中”, 5次“差”
4		差	30次为“差”
5	良	优	30次为“优”
6		良	4次“优”, 26次“良”
7		中	1次“良”, 23次“中”, 6次“差”
8		差	30次为“差”
9	中	优	30次“优”
10		良	6次“优”, 23次“良”, 1次“中”
11		中	26次“中”, 4次“差”
12		差	30次为“差”
13	差	优	30次为“优”
14		良	1次“优”, 26次“良”, 3次“中”
15		中	25次“中”, 5次“差”
16		差	30次为“差”