摘要
研究了以自回归模型谱为特征,进行焊接缺陷的超声回波模式识别的方法,根据线性判别原理建立了判别函数,利用所提取的特征值对焊接缺陷进行了模式分类,结果表明自回归模型谱在超声缺陷的模式识别中是一良好的特征量,各类目标具有比较清晰的类边界。
The pattern recognition method of ultrasontic echo of weld flaw has been analyzed with tbe features of auto-regression model spectrum. Based on the linear discriminatory principle, the discriminatory function is established. The pattem classification of weld flaw is carried out with the features extracted . The results show that, the auto-regression model spectra are good fea- tures in pattem recognition of weld flaws, and the flaw of every type has a clear class boundary.
出处
《哈尔滨工业大学学报》
EI
CAS
CSCD
北大核心
1997年第2期15-17,共3页
Journal of Harbin Institute of Technology
基金
航天基金
关键词
模式识别
特征向量
焊接缺陷
检测
超声波检测
Pattem recognition
feature extraction
weld flaw: ultrasonic detection