文献类型：期刊文献

中文题名：声发射辅以模糊神经网络的质量检测

英文题名：Quality evaluation by acoustic emission aided with a fuzzy neural network

作者：刘国光[1];程青蟾[1]

机构：[1]华东理工大学,上海200237

年份：2004

卷号：23

期号：4

起止页码：16

中文期刊名：应用声学

外文期刊名：Journal of Applied Acoustics

收录：CSTPCD;;北大核心:【北大核心2000】；CSCD:【CSCD_E2011_2012】；

语种：中文

中文关键词：声发射;模糊神经网络;质量检测;铸铁工件;BP学习法

外文关键词：Acoustic emission, Fuzzy neural network, Quality evaluation

摘要：利用铸铁工件在加载与卸载时所产生的声发射信号作为神经网络模式识别的输入特征,并对网络输入量与输出量进行模糊化隶属度计算,用最陡下降的BP学习法对连接权重系数进行优化,成功地实现了对铸铁工件样本组的训练和分类。
The acoustic emission signals produced when an iron workpiece is loaded and unloaded are used as input characteristics of pattern recognition with a neural network. The fuzzy subordinate degree is calculated for network input and output, and the connection weight factors are optimized through BP learning of steepest fall. Training and classification of workpiece sample group are successfully fulfilled.