基于信号融合的镁合金激光焊接熔透状态监测

Monitoring penetration status in magnesium alloy laser welding through signal fusion

针对目前研究中缺乏镁合金激光焊接熔透状态监测的现状,提出了一种基于声图信号融合的镁合金激光焊接熔透状态监测方法;使用麦克风和高速相机监测激光焊接过程,将提取的声音、图像特征进行特征级融合,并输入反向传播神经网络中,实现镁合金熔透状态准确识别,准确率达100%。结果表明,与单独使用声音或图像特征进行监测相比,声音图像信号融合监测能获得更高的识别准确率;将声图融合特征输入经典机器学习方法中,均获得了较高的准确率;该方法准确识别了镁合金激光焊接熔透状态,有助于提高焊接质量。

To address the lack of the penetration status monitoring on magnesium alloy laser welding in the existing study,a monitoring method for the penetration status of magnesium alloy laser welding based on sound-image signal fusion was proposed.A microphone and high-speed camera were used to monitor laser welding process,the extracted sound and image features were fused at the feature level and inputted into a back-propagation(BP)neural network to achieve accurate recognition of magnesium alloy penetration status,with an accuracy of 100%.The fusion monitoring of sound and image signals can achieve higher recognition accuracy as compared to employing either sound or image features for monitoring the magnesium alloy laser welding penetration status.Furthermore,inputting sound-image fusion features into classical machine learning methods also achieves high accuracy.This method accurately identifies the penetration status of magnesium alloy laser welding,which helps to improve welding quality.