基于深度学习的材料超声回波衰减预测方法

材料超声回波衰减是评价材料均匀一致性的常用方法, 针对具有复杂结构的航空发动机盘件难以进行材料底面超声回波衰减评价的问题, 本文提出了利用超声背散射波信号直接预测底面回波衰减的方法。采用10MHz聚焦探头进行超声背散射波数据的采集, 利用深度学习技术构建和训练模型,建立了基于深度学习的材料底面回波衰减预测方法, 同时讨论了采用不同信号形式的超声波信号分类识别模型的准确率差异。研究发现：基于深度学习技术可实现通过超声背散射波预测材料的底面回波衰减, 预测结果和实际底面回波衰减试验结果具有良好的一致性。

Material ultrasonic echo attenuation prediction method based on deep learning

Aiming at the difficulty of evaluating the material bottom ultrasonic echo attenuation of aero-engine disks with complex structures, this paper proposes a method to directly predict the bottom echo attenuation using ultrasonic backscattered wave signals. A 10MHz focusing probe is used to collect the detection data set. Through the construction and training of models by deep learning, an prediction method for bottom echo attenuation of materials based on deep learning is established. At the same time, the accuracy difference of ultrasonic signal classification and recognition models with different signal forms is discussed. The study found that the use of deep learning can realize prediction of the bottom echo attenuation of materials through ultrasonic backscattered waves, and the predicted results are in good agreement with the actual bottom echo attenuation test results.