神经网络技术在检测某型直升机旋翼胶结质量中的应用  

Application of Neural Network Technology in Testing for the Rotor Wing's Bonding Quality of Certain Helicopter

作　　者：程文[1] 孙金立[1] 

机构地区：[1]海军航空工程学院青岛分院,山东青岛266041

出　　处：《青岛大学学报（工程技术版）》2010年第2期20-23,共4页Journal of Qingdao University(Engineering & Technology Edition)

摘　　要：一般采用声阻法检测飞机蒙皮的胶结质量,但该方法难以对缺陷进行定量描述。为了解决胶结缺陷的定量问题,采用神经网络技术对直升机旋翼大梁与后段件蒙皮胶结缺陷进行定量分析,给出了缺陷大小与信号幅度的关系特征曲线,由此设置了神经网络的输出值,经网络反复训练,得到对应样品缺陷大小的估计值。实验结果证明,神经网络技术对大梁与后段件蒙皮胶结缺陷定量评价满足实际检测的要求,为胶结缺陷和类似缺陷的定量描述提供了很好的途径。Generally the acoustic resistance testing is used to the plane＇s skin bonding quality.But the method is difficult to quantify the bonding defect.In order to solve the sizing problem of bonding defect,this paper uses the neural network technology to the rotor wing＇s girder and the posterior segment skin bonding quality of certain helicopter.The curve of detect area and echo height is given,and the output of neural network is set.The estimative value of corresponding detect＇s size is gained by reiterative examination.Experiments results indicate that the neural network technology is able to achieve the inspection request in sizing evaluation for the girder and the posterior segment skin bond zone of certain helicopter.This method can provide a good path to quantify the bonding defect and the similar defect.

关 键 词：神经网络 声阻检测 大梁与后段件蒙皮胶结 缺陷定量评价 

分 类 号：V267.2[航空宇航科学与技术—航空宇航制造工程]