| Automation Technology, Control Technology |
|
|
|
|
| Fault classification based on modified active learning and weighted SVM |
| ZHU Dong-yang, SHEN Jing-yi, HUANG Wei-ping, LIANG Jun |
| College of Control Science and Engineering,Zhejiang University,Hangzhou 310027,China |
|
|
|
Abstract
A new method based on modified Best versus Second-Best (BvSB) active learning and weighted support vector machine for fault classification in real-world industrial process was presented in order to solve the problems that large-scale labeled fault samples are not easy to acquire, labeling cost is expensive, datasets are usually imbalanced and contaminated with outliers. An improved BvSB selection method was proposed to iteratively select the most valuable data and query their labels by comprehensively measuring the informativeness and representativeness of unlabeled instances and reducing the impact of outliers. WeightedSVM was introduced to tackle the impact of imbalanced class distribution on active learning and classification accuracy, using different weight factors for classes and individual samples. A new efficient method of determining the penalty coefficient was presented. Case study on TE process verifies that the proposed approach can achieve superior classification accuracy while reducing the labeling cost.
|
|
Published: 25 April 2017
|
|
|
|
Cite this article:
ZHU Dong-yang, SHEN Jing-yi, HUANG Wei-ping, LIANG Jun. Fault classification based on modified active learning and weighted SVM. JOURNAL OF ZHEJIANG UNIVERSITY (ENGINEERING SCIENCE), 2017, 51(4): 697-705.
|
基于主动学习和加权支持向量机的工业故障识别
针对流程工业过程中有标签故障样本少,样本标注代价昂贵,样本集存在类不平衡以及样本孤点问题,研究基于最优次优标号（BvSB）和加权支持向量机（WSVM）的工业故障分类方法.通过综合考虑样本的信息度和代表性以及样本中可能存在的孤立点,提出改进的主动学习算法,用于挖掘那些对当前分类器模型最有价值的样本进行标注.在支持向量机训练学习中,对不同样本采用不同的权重系数,不同类别赋予不同的惩罚因子,减少了样本分布不平衡时对主动学习和分类精度的影响,充分考虑样本点在特征空间的分布情况,提出新的惩罚系数选取方法.以TE过程为例,实验结果证明,提出的方法能够在获得较高故障分类准确率的情况下减少标注负担.
|
|
| ［1］ JAIN P, KAPOOR A. Active learning for large multi-class problems\[C\]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Miami: IEEE, 2009.
［2］ DASGUPTA S. Coarse sample complexity bounds for active learning ［J］. Neural Information Processing Systems, 2005, 18: 235-242.
［3］ 唐明珠,阳春华,桂卫华.基于改进的QBC和CS-SVM的故障检测［J］.控制与决策,2012, 27(10): 1489-1493.
TANG Ming-zhu, YANG Chun-hua, GUI Wei-hua. Fault detection based on modified QBC and CS-SVM \[J\]. Control and Decision, 2012, 27(10): 1489-1493.
［4］ YAN Z, CHEN J. Enhancing quality of statistic monitoring models by training set eesign with active learning approach ［J］. Chemometrics and Intelligent Laboratory Systems, 2015, 151(2016): 201-218.
［5］ TONG S, KOLLER D. Support vector machine active learning with applications to text classification ［J］. Journal of Machine Learning Research, 2012, 2(1):45-66.
［6］ WANG R, KWONG S, CHEN D. Inconsistency-based active learning for support vector machines ［J］. Pattern Recognition, 2012, 45(10): 3751-3767.
［7］ 徐海龙,别晓峰,冯卉,等.一种基于QBC的SVM主动学习算法［J］.系统工程与电子技术,2015, 37(12): 2865-2871.
XU Hai-long, BIE Xiao-feng, FENG Hui, et al. Active learning algorithm for SVM based on QBC \[J\]. System Engineering and Electronics, 2015, 37(12): 2865-2871.
［8］ COHN D A, GHAHRAMANI Z, JORDAN M I. Active learning with statistical models ［J］. Journal of Artificial Intelligence Research, 1996, 4(1): 705-712.
［9］ 陈荣,曹永锋,孙洪.基于主动学习和半监督学习的多类图像分类 ［J］. 自动化学报, 2011, 37(8): 954-962.
CHEN Rong, CAC Yong-feng, SUN Hong. Multiclass image classification with active learning and semi-supervised learning \[J\]. Acta Automatica Sinica, 2011,37(8): 954-962.
［10］ JOSHI A J, PORIKLI F, PAPANIKOLOPOULOS N. Multi-class active learning for image classification\[C\]∥IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Miami: IEEE, 2009.
［11］ PERRY T, BADER-EL-DEN M, COOPER S. Imbalanced classification using genetically optimized cost sensitive classifiers\[C\]∥2015 IEEE Congress on Evolutionary Computation. Sendai: IEEE, 2015.
［12］ BURNAEV E, EROFEEV P, PAPANOV A. Influence of resampling on accuracy of imbalanced classification \[C\]∥Proceedings of the 8th International Conference on Machine Vision. Barcelona: \[s.n.\], 2015.
［13］ YANG Y, MA Z, NIE F, et al. Multi-class active learning by uncertainty sampling with diversity maximization ［J］. International Journal of Computer Vision, 2015, 113(2): 113-127.
［14］ 施化吉,周书勇,李星毅,等.基于平均密度的孤立点检测研究［J］.电子科技大学学报,2007, 36(6): 1286-1288.
SHI Hua-ji, ZHOU Shu-yong, LI Xing-yi, et al. Average density-based outliers detection \[J\]. Journal of University of Electronic Science and Technology of China, 2007, 36(6): 1286-1288.
［15］ 金鑫,李玉鑑.面向非平衡数据处理的样例惩罚支持向量机 ［J］. 武汉大学学报:理学版, 2012, 58(2): 139-143.
JIN Xin, LI Yu-jian. Support vector machine with example dependent costs for dealing with imbalanced data \[J\]. Journal of Wuhan University: Natural Science Edition, 2012, 58(2): 139-143.
［16］ GE Z, SONG Z, GAO F. Review of recent research on data-based process monitoring ［J］. Industrial and Engineering Chemistry Research, 2013, 52(10):3543-3562.
［17］ LIU Y, LIAN J, BARTOLACCI M R, et al. Density-based penalty parameter optimization on C-SVM ［J］. The Scientific World Journal, 2014(2014):851814.
［18］ 金鑫,李玉鑑.不平衡支持向量机的惩罚因子选择方法［J］.计算机工程与应用,2011,47(33): 129-133.
JIN Xin, LI Yu-jian. Error-cost selection for biased support vector machines \[J\]. Computer Engineering and Applications, 2011, 47(33): 129-133.
［19］ TAX D M, DUIN R P. Support vector data description ［J］. Machine Learning, 2004, 54(1): 45-66.
［20］ DOWNS J J, VOGEL E F. A plant-wide industrial process control problem ［J］. Computers and Chemical Engineering, 1993, 17(3): 245-255.
［21］ 石向荣.面向过程监控的非线性特征提取方法研究［D］.杭州：浙江大学,2014.
SHI Xiang-rong. Study on nonlinear feature extraction for process monitoring \[D\]. Hangzhou: Zhejiang University, 2014. |
|
Viewed |
|
|
|
Full text
|
|
|
|
|
Abstract
|
|
|
|
|
Cited |
|
|
|
|
| |
Shared |
|
|
|
|
| |
Discussed |
|
|
|
|
