基于涡流超声数据融合的复合无损检测技术.pdf

中国科技论文在线 基于涡流/超声数据融合的复合无损检测 技术# 李国厚，张武波，黄平捷，周泽魁** 5 （浙江大学控制科学与工程学系，杭州 310027 ） 摘要：涡流和超声都是无损检测中应用较多的检测方法，两者都有各自的缺点和不足，为了 更好地进行可靠的、准确的无损检测，多种检测方法的融合不失为一途径。本文主要研究了 结合涡流和超声的无损检测方法，基于D-S 证据理论、模糊C-均值聚类算法和BP 神经网 络，完成异源数据的融合，从而提高缺陷分类的准确性，降低了对专家知识和大量实验数据 10 的依赖程度。 关键词：涡流；超声；无损检测；D-S 证据理论；模糊C-均值聚类；BP 神经网络 中图分类号：TM154 A research on quantitative evaluation of defects Application 15 of eddy current testing in the quantitative evaluation of the rail cracks LI Guohou, ZhangWubo, HuangPingjie, ZhouZekui (Department of control Science & Engineering,Zhejiang University,Hangzhou 310027, China) Abstract: In order to overcome the disadvantages of the commonly-used ultrasonic testing method, 20 the application of the eddy current testing (ECT) in the quantitative nondestructive testing of the rail cracks was investigated in this paper, the structure and the principle of the ECT experiment system and the experiment design were expounded. The subtractive clustering algorithm was used to improve the RBF neural network, and the data acquired from the specimen by means of the experiment system was used to train the network model. A pair of GMR-based probes were used 25 to increase the detecting performance for the deep defects and the minor surface defects. The experiment results show that the constructed model by the improved algorithm possesses higher accuracy in the inversing of the rail cracks, in the mean time, the training time of the inversing model is decreased, and it satisfies the demand of the inline inspection for the parameter of the rail crack. 30 Key words: rail crack; eddy current nondestructive tes