小波和多核SVM方法在UVA传感器故障诊断的应用  

Application of wavelet and multi-kernel SVM in UAV sensors fault diagnosis

作　　者：叶慧[1] 罗秋凤[2] 李勇[1] 

机构地区：[1]南京航空航天大学自动化学院,南京210016 [2]南京航空航天大学无人机研究院,南京210016

出　　处：《电子测量技术》2014年第1期112-116,共5页Electronic Measurement Technology

摘　　要：为了提高无人机传感器故障诊断的准确性,提出一种基于小波与多核支持向量机的诊断方法。采用小波处理信号,不依赖于系统的数学模型,直接利用信号模型,分析可测信号,提取频率等特征值,保存了原始信号的特征,提高故障的可分性。多核映射能够解决单核映射核函数及其参数选择的难题,增加故障的可区分性,提高分类器的精度。提出多核学习方法改进核函数的性能,将该方法对某无人机的传感器故障诊断,分别利用单核和多核支持向量机进行仿真,仿真结果表明了多核学习方法的有效性,提高了诊断精度。In order to improve the accuracy of sensor fault diagnosis in unmanned aerial vehicle（UAV）, a diagnose method based on wavelet packet and multi kernel support vector machine（SVM） is proposed. Using wavelet to process the signal dose not depend on the mathematical model of system and use the signal model direetly to analyze the signal and extract frequency characteristic. Because of keeping characteristic of the original, this diagnose method improve the separability of fault. Multi-kernel mapping can solve the selection problem of kernel functions and parameter of single kernel. It is easy to distinguish the fault and increase precision of classification. In this paper a multi-kernel learning method is proposed to improve the performance of kernel function. This method has been used in sensor fault diagnose of a certain UAV. Use the single kernel and multi-kernel support vector machine simulation respectively and simulation results show that it is efficiency and improves the diagnose precision.

关 键 词：传感器 小波包 多核支持向量机 故障诊断 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]