摘要
针对目前航空发动机滑油滤的检查仍处于目视定性检查水平,检查结果依赖人员的经验,既主观又无定量依据的现状,研究了基于图像识别的航空发动机滑油滤磨屑检测技术。首先构造了油滤图像检测硬件系统;然后提出了利用核主成分分析(kernel principle component analysis,KP-CA)对滑油滤图像进行特征提取的方法;最后,利用实际采集的滑油滤图像进行了实例分析,并与普通的主成分分析(principle component analysis,PCA)方法进行比较。结果表明,KPCA方法可以更为有效地提取出滑油滤图像的磨损状态特征,能够有效地提高发动机磨损故障预报的准确率。
Inspection of the aero-engine oil filters still relies on the experience of the engineers.We present a quantitatively method for aero-engine oil filter inspection.First,we construct the oil filter image detection hardware system.Then,Kernel Principle Component Analysis ( KPCA) is used to extract the oil filter image features.Finally,we analyze the actual acquisition oil filter images with KPCA and compare the results with those by the method of traditional Principle Component Analysis ( PCA) .Comparison result indicates that the method of using KPCA can extract the wear condition features in oil filter images more efficiently.It will improve the prediction accuracy of aero-engine wear faults efficiently.
出处
《机械科学与技术》
CSCD
北大核心
2010年第6期731-736,共6页
Mechanical Science and Technology for Aerospace Engineering