摘要
孔探检测是航空发动机视情维修的重要技术,对于正确评估发动机的内部损伤,适时进行发动机修理具有重要意义,但是由于孔探图像的损伤评估往往依赖丰富的专家知识,因此在航空公司飞机多、分布广、专家少的情况下,难于实现发动机损伤的及时评估。本文研究了一种基于孔探图像纹理特征的航空发动机内部损伤评估方法,利用结构自适应神经网络模型,实现了航空发动机孔探图像损伤的自动识别,并进行了必要的验证,结果表明了本文方法的有效性。
Borescope detection is an important means widely applied in the on-condition maintenance of aero-engine. It is very important to evaluate the engine interior damages and repair the faulty engine on time. But the evaluation depends greatly on the experts' experiences so that it is very difficult to carry out evaluation in time because the experts are very few and the aircrafts spread widely. This paper researches an auto-evaluation method for aero-engine interior damages based on borescope image texture features, in which the neural network model with self-adaptive structure is applied to carry out automatic recognition of the damages. Finally, some typical borescope images with damages are used to verify the new method, and the results show its correctness and effectiveness.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2008年第8期1709-1713,共5页
Chinese Journal of Scientific Instrument
关键词
航空发动机
孔探检测
图像纹理
神经网络
aero-engine
borescope detection
image texture
neural network