文献类型：期刊文献

中文题名：基于模糊神经网络的焊缝缺陷识别方法的研究

英文题名：Weld Defects Distinguishing Method Based on Fuzzy Neural Networks

作者：张晓光[1];林家骏[2]

机构：[1]中国矿业大学机电与材料工程学院,江苏徐州221008;[2]华东理工大学信息学院,上海200237

年份：2003

卷号：32

期号：1

起止页码：92

中文期刊名：中国矿业大学学报

外文期刊名：Journal of China University of Mining & Technology

收录：CSTPCD;;Scopus;北大核心:【北大核心2000】；CSCD:【CSCD2011_2012】；

语种：中文

中文关键词：射线底片;焊缝;缺陷识别;模糊神经网络;隶属度

外文关键词：ray negative; weld; defects distinguishing; fuzzy neural network

摘要：通过对射线底片焊缝缺陷特征分析 ,提出了用于焊缝缺陷识别的模糊神经网络模型 ,并介绍了隶属度的构造和 BP网络学习算法 .用 5 2个典型缺陷样本训练该模型后 ,对 8个缺陷样本进行识别试验 .试验结果表明 ,该方法能够提高介于模糊边界模式分类时的识别率 。
The model of fuzzy neural networks for weld defects distinguishing was described, through the analysis of the defect characters in weld of ray inspection. The construction of membership function and learning algorithm of BP networks were introduced. On the basis of the model trained by 52 type defect sample, 8 defect sample were distinguished. The experiment shows the method can improve the distinguishing rate of pattern sort in fuzzy boundary. This method excels the sort distinguishing in weld defects distinguishing.