Mechanical fault Severity Identification Methods under Unbalanced Datasets
DUAN Li-xiang1, GUO Han1, 2, WANG Jin-jiang1
1. School of Mechanical and Transportation Engineering, China University of Petroleum, Beijing 102249, China;
2. China Special Equipment Inspection and Research Institute, Beijing 100013, China
In mechanical fault diagnosis, the samples under fault condition are often difficult to obtain, also called unbalanced dataset issue, which will lead to very low classification accuracy with the conventional algorithm, such as support vector machine (SVM). Weighted C-support vector machine shows the improved performance, however, due to the small sample space caused by close fault severities, the classification accuracy of weighted C-support vector machine is still not up to satisfactory. To improve the classification accuracy for close fault severity cases under unbalanced dataset, this paper presents an approach integrating weighted c-support vector machine algorithm with binary tree structure, named as BT-CSVM. The binary structure is then optimized taking account of sample space of class-to-class, sample space of inter-class, and unbalance degree. Experimental results show that the proposed method can effectively deal with unbalanced dataset problem by greatly improving the classification accuracy for close fault severity cases.
[1] 张龙, 黄文艺, 熊国良. 基于多尺度熵的滚动轴承故障程度评估[J].振动与冲击, 2014,33(9): 185-189.
Zhang Long, Huang Wen-yi, Xiong Guo-liang. Assessment of rolling element bearing fault severity using multi-scale entropy[J]. Journal of Vibration and Shock,2014,33(9): 185-189.
[2] 江瑞龙, 陈进, 刘韬, 等. 设备性能退化评估在巡检系统中的应用[J]. 振动与冲击,2012, 31(16) : 134-137．
JIANG Rui-long, CHEN Jin, LIU Tao, et al. Performance degradation assessment with the application of patrol inspection system[J]. Journal of Vibration and Shock, 2012,31(16) : 134 -137．
[3] 陶新民, 刘福荣, 童智靖, 等. 不均衡数据下基于SVM的故障检测新算法[J]. 振动与冲击, 2010,29(12):8-12.
Tao Xin-min, Liu Fu-rong, Tong Zhi-jing, et al. A new fault detection method of unbalanced data based on SVM[J]. Journal of Vibration and Shock, 2010, 29(12):8-12.
[4] 姚培, 王仲生, 姜洪开, 等. 不均衡数据下基于CS-Boosting
的故障诊断算法[J]. 振动、测试与诊断, 2013, 33(1):111-115.
Yao Pei, Wang Zhong-sheng, Jiang Hong-kai, et al. Fault Diagnosis Method Based on CS-Boosting for Unbalanced Training Data[J]. Journal of Vibration, Measurement & Diagnosis, 2013, 33(1):111-115.
[5] Pahlo C, Eva A, Jose B, et al. Multiclass Support Vector Machines With Example-Dependent Costs Applied to Plankton Biomass Estimation[J]. IEEE Transaction on Neural Networks and Learning Systems, 2013,24(11):1901-1905.
[6] 石志标, 苗莹. 基于FOA-SVM的汽轮机振动故障诊断[J].振动与冲击, 2014,33(22): 111-114.
Shi Zhi-biao, Miao Ying. Vibration fault diagnosis for steam turbine by using support vector machine based on fruit fly optimization algorithm[J]. Journal of Vibration and Shock, 2014,33(22): 111-114．
[7] Wang Xi-ping, Tan Wen-xue, Wu Hua-rui. An innovative SVM for wheat seed quality estimation[J]. Journal of Information & Computational Science, 2015,12(1): 223-233.
[8] 刘进军. 基于惩罚的SVM和集成学习的非平衡数据分类算法研究[J].计算机应用与软件, 2014,31(1): 186-190.
Liu Jin-jun. Research on classificating unbalanced data based on penalty-based SVM and ensemble learning[J]. Computer Applications and Software, 2014,31(1): 186-190.
[9] 赵海洋, 徐敏强, 王金东. 改进二叉树支持向量机及其故障诊断方法研究[J].振动工程学报, 2013,26(5):764-769.
Zhao Hai-yang, Xu Min-qiang, Wang Jin-dong. An improved binary tree SVM and application for fault diagnosis[J]. Journal of Vibration Engineering, 2013,26(5):764-769.
[10] Liu Xiao-feng, Lin Bo. Identification of resonance states of rotor-bearing system using RQA and optimal binary tree SVM[J]. Neurocomputing, 2015(152):36-44.
[11] 许永建. 变压器故障诊断技术研究[D].南京:南京理工大学,
2010.
Xu Yong-jian. Research on fault diagnosis technology of transformer[D]. Nanjing: Nanjing University of Science and Technology,2010.
