摘要
针对在线超声智能随机性识别研究中希望通过对裂纹试样尺寸的定量识别进而得到反映裂纹尺寸分布规律的目的 ,充分考虑在一定置信度下进行缺陷参数估计时对试样数的要求 ,以及智能识别的神经网络处理过程中对学习样本和测试样本的数量要求 。
A kind of random small sampling method is presented aimed at the purpose of on-line Ultrasonic intelligent recognition, which results the distribution law of the height and size of cracks can be obtained in this research. Considered the requirement of distribution parameter estimation in certain degree of confidence and the requirement of the artificial neural network of intelligent recognition, the small sampling method of on-line Ultrasonic intelligent recognition is proposed.
出处
《压力容器》
2003年第10期14-16,33,共4页
Pressure Vessel Technology
基金
国家自然科学基金资助项目 (5 0 0 0 5 0 0 6)
关键词
超声检测
在线智能识别
试样
小子样
ultrasonic testing
online intelligent recognition
samples
small samples
