论文部分内容阅读
对于非均匀材料，超声无损检测技术受到能否有效区分有用信号与背景噪声的限制，目前人们大多倾向使用频率分隔与统计算法来提高粗晶材料（一种非均匀材料件相对颗粒散射的缺陷回波比例．文中介绍一种用Ｗｉｇｎｅｒ分布作特征提取、用前馈网络自动识别超声散射回波中的缺陷信号．由于普通的人工神经网络要求输入信号的特征与时间起始点无关，因此采取了一种数学变换方式来实现这一要求，这样训练好的网络就有很强的识别能力．在实验中，正确识别率达到９０％．所述方法对其他非均匀介质的超声检测与评价工作也有益处．
For non-uniform materials, ultrasonic non-destructive testing is limited by whether useful signals and background noise can be effectively distinguished. At present, most people prefer to use frequency separation and statistical algorithms to improve the performance of coarse-grained materials Wave ratio.In this paper, a Wigner distribution is introduced for feature extraction and a feedforward network is used to automatically detect the defect signal in ultrasonic scattering echo.Because the common artificial neural network requires that the characteristics of input signal have nothing to do with the starting point of time, So that the trained network has a strong recognition ability.In the experiment, the correct recognition rate reaches 90% .The method is also beneficial to the ultrasonic testing and evaluation of other inhomogeneous media .