基于小波OCSVM的无人机气压高度传感器故障诊断研究  

Research on Fault Detection of Pressure-altitude Sensor of UAV Based on Wavelet One-class Support Vector Machine

作　　者：史岩[1] 李小民[1] 连光耀[2] 

机构地区：[1]军械工程学院无人机工程系,石家庄050003 [2]军械工程学院军械技术研究所,石家庄050003

出　　处：《计算机测量与控制》2013年第10期2658-2660,共3页Computer Measurement &Control

摘　　要：无人机气压高度传感器故障样本很少,故障数据采集一直是限制其故障诊断效果的一个瓶颈问题,而一类支持向量机(One Class Support Vector Machine,OCSVM)仅需要足够多正常样本,就能够达到很好的检测效果,故文章采用一类支持向量机对其进行故障检测研究,同时采用小波核函数提高OCSVM的抗噪能力和诊断精度;实验表明,在有足够数量正常训练样的情况下,小波OCSVM能够达到令人满意的诊断效果。There are few fault samples of pressure--altitude sensor of UAV, the acquisition of fault sample is a bottleneck of its fault di agnosis effect. Because One Class Support Vector Machine （OCSVM） can obtain the better diagnosis result with enough samples, OCSVM is utilized to fault detection research of pressure altitude sensor of UAV. And the wavelet kernel function is utilized to OCSVM in order to improve its diagnosis accuracy and noise immunity. Experiment result testifies that wavelet OCSVM can obtain satisfied diagnosis effects with enough normal samples.

关 键 词：无人机 气压高度传感器 一类支持向量机 故障检测 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]