在利用超声信号进行金属基复合材料结构的无损评价时。需要对不同界面脱粘进行识别。利用人工神经网络的方法可以对不同界面脱粘时的实验检测信号进行正确地识别。实验检测结果表明超声回波信号识别方法可用在金属基复合材料结构的无损检测中。
In the ultrasonic non-destructive evaluation of metal matrix composite(MMC) multi- layer structure,it needs to clarify the disbands at different interface . Using artificial neural networks, can succeed in clarifying the experience disbands at different interfaces. This method of feature extraction is expected to be largely applicable to the non - destructive evaluation of MMC multi - layer bonded structure.