摘要
既有钢轨探伤车超声检测系统均为国外引进。中国铁道科学研究院自主化钢轨探伤系统目前已经通过试验验证,所采用的伤损模式分类技术是其中的一项关键技术。根据自主化钢轨探伤系统的探轮设置,规定伤损的类别设置,并将伤损类别划分为四大类:核伤类、螺孔裂纹类、轨底裂纹类、水平裂纹类。归属于不同大类的伤损采用不同神经网络进行识别。以螺孔裂纹类为例进行说明,采用标定线数据进行测试,验证算法的有效性。
Currently the ultrasonic inspection systems equipped on domestic rail flaw detection car are all imported from foreign countries. The rail flaw detection system independently developed by China Academy of Railway Sciences has passed the test and verification, of which the flaw classification system is a key technology. The probe of the rail flaw detection system is set to define different defect types, including transverse defect, bolt hole crack, rail base crack and horizontal crack. Different neural networks are used to identify different defects. The bolt hole defect is also taken as an example to verify the effectiveness of the algorithm using calibration line data.
出处
《中国铁路》
2018年第3期82-87,共6页
China Railway
基金
国家自然科学基金重大科研仪器研制专项(61527803)
国家重点研发计划重大科学仪器设备开发专项(2016YFF0103700)
中国铁路总公司科技研究开发计划项目(2017G003-D)
关键词
钢轨探伤车
B显数据
BP神经网络
伤损模式分类
超声波探伤
rail flaw detection car
B-scan data
BP neural network
defect classification
ultrasonic flaw detection
