基于元学习的航空电子设备特征选择算法推荐方法

doi:10.12305/j.issn.1001-506X.2021.07.34

摘要/Abstract

摘要：

为了对航空电子设备的测试数据进行有效约简, 去除冗余信息和不相关特征, 基于机器学习领域现有的特征选择算法, 提出了一种元学习框架下的航空电子设备特征选择算法推荐方法。所提方法旨在根据不同航空电子设备测试数据所蕴含的信息, 推荐合适的特征选择算法。首先, 分析了数据集特征的描述方法。然后, 介绍了采用综合度量指数的算法性能评价方法。最后, 给出了特征选择算法推荐方法的框架。使用42个航空电子设备的测试数据和13个过滤型特征选择算法建立了元数据库, 采用留一法进行交叉验证, 推荐命中率达到了90%以上, 推荐性能比例达到97%以上。

关键词: 故障诊断, 元学习, 特征选择, 算法推荐, 航空电子设备

Abstract:

In order to reduce the test data of avionics effectively and remove redundant information and irrelevant features, based on the existing feature selection algorithms in the field of machine learning, a recommendation method for avionics feature selection algorithm under the meta-learning framework is proposed. The proposed method aims to recommend appropriate feature selection algorithms according to the information contained in the test data of different avionics. Firstly, the description method of data set feature is analyzed. Then, the algorithm performance evaluation method based on the multi-metric index is introduced. Finally, the framework of recommendation method for feature selection algorithm is given. A metadata database is established on 42 avionics data sets and 13 filtering feature selection algorithms, the leave-one-out method is used for cross validation. The recommended hit radio reaches more than 90% and the recommended performance radio reaches more than 97%.

Key words: fault diagnosis, meta-learning, feature selection, algorithm recommendation, avionics

中图分类号:

V243
TP206.1

李睿峰, 许爱强, 孙伟超, 王树友. 基于元学习的航空电子设备特征选择算法推荐方法[J]. 系统工程与电子技术, 2021, 43(7): 2011-2020.

Ruifeng LI, Aiqiang XU, Weichao SUN, Shuyou WANG. Recommendation method for avionics feature selection algorithm basedon meta-learning[J]. Systems Engineering and Electronics, 2021, 43(7): 2011-2020.

图/表 10

图1

表1

表2

图2

图3

表3

表4

图4

图5

表5

参考文献 30

1	张伟, 许爱强, 平殿发. 基于一种自适应核学习的KECA子空间故障特征提取[J]. 北京理工大学学报, 2017, 37 (8): 864- 868.
	ZHANG W , XU A Q , PING D F . A method for feature extraction in KECA feature subspace based on adaptive kernel learning[J]. Transactions of Beijing Institute of Technology, 2017, 37 (8): 864- 868.
2	SU Z Q , TANG B P , DENG L , et al. Fault diagnosis method using supervised extended local tangent space alignment for dimension reduction[J]. Measurement, 2015, 62, 1- 14. doi: 10.1016/j.measurement.2014.11.003
3	DEMETGUL M , YILDIZ K , TASKIN S , et al. Fault diagnosis on material handling system using feature selection and data mining techniques[J]. Measurement, 2014, 55, 15- 24. doi: 10.1016/j.measurement.2014.04.037
4	WANG G T , SONG Q B , XU B W , et al. Selecting feature subset for high dimensional data via the propositional FOIL rules[J]. Pattern Recognition, 2013, 46 (1): 199- 214. doi: 10.1016/j.patcog.2012.07.028
5	SHILBAYEH S, VADERA S. Feature selection in meta learning framework[C]//Proc.of the Science and Information Conference, 2014: 269-275.
6	姚旭, 王晓丹, 张玉玺, 等. 特征选择方法综述[J]. 控制与决策, 2012, 27 (2): 161- 166, 192.
	YAO X , WANG X D , ZHANG Y X , et al. Summary of feature selection algorithms[J]. Control and Decision, 2012, 27 (2): 161- 166, 192.
7	ZHAN H. Application of rough set and support vector machine in fault diagnosis of power electronic circuit[C]//Proc.of the IEEE 2nd International Conference on Information Management & Engineering, 2010: 289-292.
8	GAO W F , HU L , ZHANG P , et al. Feature selection consi-dering the composition of feature relevancy[J]. Pattern Recognition Letters, 2018, 112, 70- 74. doi: 10.1016/j.patrec.2018.06.005
9	ZHOU F Y, XUE B, CHEN K. Particle swarm optimization for feature selection with adaptive mechanism and new updating strategy[C]//Proc.of the 31st Australasian Joint Conference Artificial Intelligence, 2018: 419-431.
10	WANG Z X, ZHAO M, CHEN S Y. Feature selection based on extreme learning machine[C]//Proc. of the 3rd International Conference on Deep Learning Technologies, 2019: 57-63.
11	TANG X, DONG M, BI S. Feature selection algorithm based on K-means clustering[C]//Proc.of the IEEE 7th Annual International Conference on Cyber Technology in Automation, Control, and Intelligent Systems, 2018: 1522-1527.
12	曾子林, 张宏军, 张睿, 等. 基于元学习思想的算法选择问题综述[J]. 控制与决策, 2014, 29 (6): 961- 968.
	ZENG Z L , ZHANG H J , ZHANG R , et al. Summary of algorithm selection problem based on meta-learning[J]. Control and Decision, 2014, 29 (6): 961- 968.
13	FILCHENKOV A, PENDRYAK A. Datasets meta-feature description for recommending feature selection algorithm[C]//Proc.of the Artificial Intelligence and Natural Language and Information Extraction, Social Media and Web Search Fruct Conference, 2016: 11-18.
14	WANG G T , SONG Q B , SUN H L , et al. A feature subset selection algorithm automatic recommendation method[J]. Journal of Artificial Intelligence Research, 2014, 47 (1): 1- 34.
15	PARMEZAN A R S , LEE H D , WU F C . Metalearning for choosing feature selection algorithms in data mining: proposal of a new framework[J]. Expert Systems with Applications, 2017, 75, 1- 24. doi: 10.1016/j.eswa.2017.01.013
16	LORENA A C , MACIEL A I , PÉRICLES B C , et al. Data complexity meta-features for regression problems[J]. Machine Learning, 2018, 107, 209- 246. doi: 10.1007/s10994-017-5681-1
17	SONG Q B , WANG G T , WANG C . Automatic recommendation of classification algorithms based on data set characteristics[J]. Pattern Recognition, 2012, 45 (7): 67- 81.
18	高明哲. 基于相关向量机的载电子设备PHM关键技术研究[D]. 烟台: 海军航空大学, 2017.
	GAO M Z. Research on key technologies of PHM for airborne electronic equipment based on relevance vector machine[D]. Yantai: Naval Aviation University, 2017.
19	SUN Q , PFAHRINGER B . Pairwise meta-rules for better meta-learning based algorithm ranking[J]. Machine Learning, 2013, 93 (1): 141- 161. doi: 10.1007/s10994-013-5387-y
20	BRAZDIL P B , SOARES C , DA COSTA C J . Ranking learning algorithms: using IBL and meta-learning on accuracy and time results[J]. Machine Learning, 2003, 50 (3): 251- 277. doi: 10.1023/A:1021713901879
21	NGUYEN H B, XUE B, ANDREAE P, et al. Particle swarm optimisation with genetic operators for feature selection[C]//Proc.of the IEEE Congress on Evolutionary Computation, 2017: 286-293.
22	SMETANNIKOV I, DEYNEKA A, FILCHENKOV A. Meta learning application in rank aggregation feature selection[C]//Proc.of the 3rd International Conference on Soft Computing and Machine Intelligence, 2016: 120-123.
23	WANG J , HEDAR A R , WANG S , et al. Rough set and scatter search metaheuristic based feature selection for credit scoring[J]. Expert Systems with Applications, 2012, 39 (6): 6123- 6128. doi: 10.1016/j.eswa.2011.11.011
24	LUCIEN M, LAETITIA J, MARIE-ELÉONORE K M, et al. Feature selection using Tabu search with learning memory: learning Tabu search[C]//Proc.of the International Confe-rence on Learning and Intelligent Optimization, 2016.
25	SINGH S , SINGH A K . Correlation-based feature subset selection technique for web spam classification[J]. International Journal of Web Engineering and Technology, 2019, 13 (4): 363.
26	KOWSHALYA A M , MADHUMATHI R , GOPIKA N . Correlation based feature selection algorithms for varying datasets of different dimensionality[J]. Wireless Personal Communications, 2019, 108 (3): 1977- 1993. doi: 10.1007/s11277-019-06504-w
27	SONG Q B , NI J J , WANG G T . A fast clustering-based feature subset selection algorithm for high-dimensional data[J]. IEEE Trans.on Knowledge and Data Engineering, 2013, 25 (1): 1- 14. doi: 10.1109/TKDE.2011.181
28	DE S C , FONTANELLA F , MARROCCO C , et al. A GA-based feature selection approach with an application to handwritten character recognition[J]. Pattern Recognition Letters, 2014, 35, 130- 141. doi: 10.1016/j.patrec.2013.01.026
29	PALMA-MENDOZA R J , RODRIGUEZ D , DE-MARCOS L . Distributed reliefF-based feature selection in spark[J]. Knowledge & Information Systems, 2018, doi: 10.1007/s10115-017-1145-y
30	TSAMARDINOS I , BORBOUDAKIS G , KATSOGRIDAKIS P , et al. A greedy feature selection algorithm for big data of high dimensionality[J]. Machine Learning, 2019, 108, 149- 202. doi: 10.1007/s10994-018-5748-7

序号	设备名称	实例数	特征数	故障类	序号	设备名称	实例数	特征数	故障类
1	中央控制设备(单座)	377	25	2	9	后舱备份控制系统	516	30	3
2	中央控制设备(双座)	559	31	2	10	超短波语音电台	687	77	6
3	电源启动转换盒(单座)	455	23	4	11	无线电高度表	547	26	5
4	电源启动转换盒(双座)	490	31	4	12	短波电台收发信机	472	65	6
5	微波着陆机载设备	526	63	5	13	短波电台数字天调	496	90	6
6	空管应答机	176	26	4	14	数传1/导航兼备设备	724	66	7
7	备份控制设备(单座)	375	11	4	15	信标接收机	568	10	5
8	前舱备份控制系统	578	26	3	16	无线电罗盘	714	37	5
17	组合接收设备	193	93	10	30	数字视频记录仪	505	19	3
18	数据传输器	468	10	2	31	CD接收机	178	29	4
19	数据传输卡	395	42	4	32	管理控制计算机	308	40	5
20	维护监控板	132	9	2	33	低功率射频单元	456	36	6
21	数字地图计算机	314	44	3	34	ECM处理机	421	81	4
22	航空电子启动板	470	58	4	35	前端接收机A	858	27	4
23	平视显示器	351	47	4	36	前端接收机B	510	27	4
24	彩色多功能显示器	675	65	5	37	中央接收机	528	36	5
25	单色多功能显示器	536	86	8	38	前向中功率发射机	569	23	5
26	数据传输数字地图	336	29	3	39	前向高功率发射机	683	23	5
27	正前方控制面板	583	58	5	40	后向中功率发射机	599	23	5
28	前舱正前方控制板	540	59	6	41	箔条/红外弹控制器	484	229	10
29	后舱正前方控制板	495	58	6	42	视频监视器	404	8	3

序号	搜索策略	评价方法
1	前向最佳优先搜索	相关性指标
2	后向最佳优先搜索	相关性指标
3	双向最佳优先搜索	相关性指标
4	遗传搜索	相关性指标
5	秩搜索	相关性指标
6	散射搜索	相关性指标
7	禁忌搜索	相关性指标
8	前向最佳优先搜索	一致性指标
9	后向最佳优先搜索	相关性指标
10	双向最佳优先搜索	相关性指标
11	遗传搜索	相关性指标
12	秩搜索	相关性指标
13	散射搜索	相关性指标

方法	分类器
方法	C4.5	NaiveBayes	PART	IB1
考虑PRC面积	95.24	90.48	92.86	92.86
不考虑PRC面积	90.48	80.95	88.10	83.33

推荐结果排序	分类器
推荐结果排序	C4.5	NaiveBayes	PART	IB1
1	95.24	90.48	92.86	92.86
2	97.62	97.62	95.24	95.24
3	100	97.62	95.24	97.62

方法	分类器
方法	C4.5	NaveBayes	PART	IB1
考虑PRC面积	98.31	97.48	98.17	97.25
不考虑PRC面积	97.94	97.33	97.92	97.18