摘要
文中介绍了一种对构件粘接强度可以进行自动在线无损检测的超声检测系统,该系统可以对构件整体粘接强度合格与否进行预报。该系统采用BP神经网络对训练样本进行训练,并对测试样本进行测试,试验结果表明,该系统检测速度快,识别精度达到了91.3%,且无漏判现象出现,达到了该类产品在线检测的要求。
An automatic and nondestructive adhesion strength inspection ultrasonic system for halved joint cartridges is introduced in the paper, the system can predict whether the whole adhesion strength is qualifiable. The training model is trained by using BP neural network in the system, and test the data that not used to be trained, the result show that： the system works quickly, the recognition correction is arrival to 91.3 % and no leak appears, which satisfies the testing requesting of this kind of product.
出处
《弹箭与制导学报》
CSCD
北大核心
2007年第1期369-371,共3页
Journal of Projectiles,Rockets,Missiles and Guidance
关键词
粘接强度
无损检测
神经网络
adhesion strength
nondestructive test
neural network
