| 基于OCSVM和MCSVM的新故障检测与诊断技术研究 |
| |
| 引用本文： | 梁四洋,段修生.基于OCSVM和MCSVM的新故障检测与诊断技术研究[J].军械工程学院学报,2013(3):40-43. |
| |
| 作者姓名： | 梁四洋 段修生 |
| |
| 作者单位： | [1]北京理工大学信息与电子学院,北京100081 [2]军械工程学院电子与光学工程系,河北石家庄050003 |
| |
| 摘 要： | 针对SVM故障诊断方法无法对新故障模式进行有效检测的问题，在阐述OCSVM原理的基础上，详细分析了各种参数对其性能的影响，提出了基于OCSVM和多类分类MCSVM相结合的新故障检测与诊断方法．实验结果表明，该算法在故障诊断及新故障检测方面有较高的精度，具有良好的推广性．
|
| 关 键 词： | OCSVM 故障检测 MCSVM 故障诊断 参数分析 |
Method to Detect and Diagnose New Fault Based on OCSVM and MCSVM |
| |
| LIANG Si-yang,DUAN Xiu sheng.Method to Detect and Diagnose New Fault Based on OCSVM and MCSVM[J].Journal of Ordnance Engineering College,2013(3):40-43. |
| |
| Authors: | LIANG Si-yang DUAN Xiu sheng |
| |
| Institution: | 1. School of Information and Electronics, Beijing Institute of Technology, Beijing 100081, China; 2. Electronic and Optical Engineering Department,Ordnance Engineering College, Shijiazhuang 050003, China) |
| |
| Abstract: | OCSVM is an unsupervised machine learning method,widely used in the detection of the abnormal state. To solve the problem that SVM cannot detect new fault at work,based on elaborating OCSVM principle and detailed analysis of its various parameters＂ influences on the performance,we put forward a new method to detect and diagnose faults, based on OCSVM combined with the multi-class classification MCSVM. The experimental results show that the algorithm can ensure high accuracy in diagnosing and detecting fault and has a good prospect for wider use. |
| |
| Keywords: | OCSVM fault detection MCSVM fault diagnosis parametric analysis |
| 本文献已被 维普 等数据库收录！ |
