改进决策的带异常样本1-SVM算法及应用

PDF(917 KB)

振动与冲击 ›› 2015, Vol. 34 ›› Issue (10) : 84-87.

论文

王涛，李艾华，王旭平，蔡艳平，张敏龙

作者信息 +

An improved decision-making 1-SVM algorithm with abnormal samples and its application

WANG Tao，LI Ai-hua，WANG Xu-ping，CAI Yan-ping，ZHANG Min-long

Author information +

文章历史 +

摘要

针对正常类样本多、异常类样本缺乏问题，基于异常样本加入能提高分类能力及分类精度考虑，提出改进决策的带异常样本1-SVM算法，并用于机械设备异常状态检测。用两类样本同时训练1-SVM模型可改善1-SVM算法对异常样本的描述能力；通过调整决策边界提高1-SVM算法的分类精度。柴油机气阀机构故障检测实验结果表明，该算法对正常类及故障类样本的识别率均高于标准1-SVM算法及带异常样本的1-SVM算法。

Abstract

Aiming at normal samples abundance and fault samples deficiency, based on the consideration of abnormal samples to join can improve classification ability and classification accuracy, an improved decision-making 1-SVM algorithm with abnormal samples is put forward and applied to abnormal condition detection of mechanical equipments. On the one hand, 1-SVM model is trained with two kinds of samples to improve description ability of 1-SVM algorithm for abnormal samples. On the other hand, the decision boundary is adjusted to improve classification accuracy of 1-SVM algorithm. The improved 1-SVM algorithm is applied to fault detection of diesel valve train. The experimental results show that recognition rate of the improved algorithm to normal class and fault class samples is higher than of standard 1-SVM algorithm and 1-SVM algorithm with abnormal samples.

导出引用

王涛，李艾华，王旭平，蔡艳平，张敏龙. 改进决策的带异常样本1-SVM算法及应用[J]. 振动与冲击, 2015, 34(10): 84-87

WANG Tao，LI Ai-hua，WANG Xu-ping，CAI Yan-ping，ZHANG Min-long. An improved decision-making 1-SVM algorithm with abnormal samples and its application[J]. Journal of Vibration and Shock, 2015, 34(10): 84-87

参考文献

[1]李卫鹏,李凌均,孔维峰,等. 正交小波变换支持向量数据描述在故障诊断中的应用[J]. 机械科学与技术,2011, 30(3):466-470.
LI Wei-peng, LI Ling-jun, KONG Wei-feng, et al. Support vector data description in orthogonal wavelet transform for fault diagnosis[J]. Mechanical Science and Technology for Aerospace Engineering, 2011,30(3):466-470.

[2] 陈斌,阎兆立,程晓斌. 基于SVDD和相对距离的设备故障程度预测[J]. 仪器仪表学报,2011,32(7):1558-1563.
CHEN Bin, YAN Zhao-li, CHENG Xiao-bin. Machinery fault trend prediction based on SVDD and relative distance[J]. Chinese Journal of Scientific Instrument, 2011,32(7):1558-1563.
[3] McBain J,Timusk M. Feature extraction for novelty detection as applied to fault detection in machinery[J]. Pattern Recognition Letters,2011(32):1054-1061.
[4] Tax D M J. One-class classification[D]. Delft: Delft University of Technica1, 2001.
[5] 蒲晓丰，雷武虎，汤俊杰，等. 基于带野值的SVDD 的高光谱图像异常检测[J]. 光电工程,2010,37(12):83-87.
PU Xiao-feng, LEI Wu-hu, TANG Jun-jie,et al. Anomaly detection for hyperspectral image based on SVDD with negative examples[J]. Opto-Electronic Engineering, 2010,37(12):83-87.
[6] Scholkopf B, Platt J C, Shawe-Taylor J, Smola,et al. Estimating the support of a high-dimensional distribution [J]. Neural Comput,2001,13(7):1443-1471.
[7] Tax D M J, Duin R P W. Support vector data description [J]. Pattern Recognition Letters, 1999, 20(11/13): 1191- 1199.
[8] Chandola V, Banerjee A, Kumar V. Anomaly detection:a survey[J]. ACM Computing Surveys,2009, 41(3):1-58.
[9] Schölkopf B,Williamson R, Smola A, et al.Support vector method for novelty detection[C]. Advances in Neural Information Processing Systems 12[A]. Solla S A, Leen T K, Müller K R.MIT Press, 2000:582-588.
[10] 王涛,李艾华,姚良,等. 采用多层核学习机的柴油机气阀机构故障诊断[J]. 振动、测试与诊断,2010, 30(4): 462-464.
WANG Tao, LI Ai-hua, YAO Liang, et al. Fault diagnosis of diesel valve train based on multi-layer kernel learning machine[J]. Journal of Vibration, Measurement & Diagnosis, 2010, 30(4):462-464.
[11] 缪志敏,胡谷雨,丁力,等. SVDD在类别不平衡学习中的应用[J]. 应用科学学报,2008, 26(1):79-84.
MIAO Zhi-min, HU Gu-yu, DING Li,et al. Support vector date description implemented in class-imbalance learning[J]. Journal of Applied Sciences, 2008, 26