摘要
分析了变厚度曲面复合工件的超声检测和波形自动跟踪原理,利用改进型的BP神经网络来获取曲面采样点位置信息与灵敏度之间的映射关系,生成各采样点的灵敏度模板,实现了灵敏度的补偿与实时控制和超声波底波信号的自动跟踪。仿真结果和实验证明,该方法有效解决了复合材料A波信号的“多峰”现象给检测带来的困难,实现对变厚度曲面复合工件的波形自动跟踪,对提高类似大衰减率的材料的瞳面复杂工件的超声检测效率与精度具有普遍意义。
The principles of ultrasonic test and automatic wave tracking were analyzed and the improved BP neural network was proposed to acquire the mapping function between the plus of sensitivity and sampling points position information, which was made into template, in order to realize compensation and real-time control of the plus of sensitivity and automatic tracking of the bottom wave during the ultrasonic test for the curved surface composite parts with variable thickness. The simulated results and experiments have shown that the proposed method overcome the difficulty of multi- peeks successfully in automatic uhrasonic test for curved surface composite parts with variable thickness. It is in common significance to improve the efficiency and accuracy of ultrasonic test for similar complex curved surface parts with high attenuation rate.
出处
《计量学报》
CSCD
北大核心
2012年第5期400-404,共5页
Acta Metrologica Sinica
基金
国家自然科学基金项目(51075358)
浙江省自然科学基金项目(LQ12E05018)
关键词
计量学
超声检测
波形跟踪
变厚度曲面
改进型BP神经网络
Metrology
Ultrasonic test
Wave tracking
Variable thickness curved surface
Improved BP neural network