研究目的
To develop a data-driven model for monitoring the seam during laser beam welding under the influence of an external magnetic field (LBW-AMF) to improve weld quality.
研究成果
The developed data-driven model provides a reliable method for monitoring the weld bead during LBW-AMF, offering guidance for controlling processing parameters in real time to improve weld quality. The ensemble of neural networks showed better prediction accuracy than single neural networks.
研究不足
The technology used to process images may miss some hidden information and geometries. Future work could explore deep learning algorithms for monitoring weld bead under big data.
1：Experimental Design and Method Selection:
A visible LBW-AMF system was built to track the laser melting pool and keyhole. Image processing techniques were used to extract features of the laser melting pool and keyhole. An ensemble of neural networks (BPNN, RBFNN, GRNN) was proposed to establish correlations between these features and the welding seam.
2：Sample Selection and Data Sources:
AISI 2205 workpieces were used for LBW-AMF experiments. Images of the welding process were recorded for training and testing points.
3：List of Experimental Equipment and Materials:
YLR-4000 ytterbium laser, high-speed CMOS camera, auxiliary pulsed diode laser unit, filter, protect lens, and a special electromagnet system.
4：Experimental Procedures and Operational Workflow:
The welding process was conducted with and without an external magnetic field. Images were processed to extract features of the laser melting pool and keyhole.
5：Data Analysis Methods:
An ensemble of neural networks was used to analyze the data and predict the width of the weld bead.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
