| 注册
首页|期刊导航|航空科学技术|基于同轴视觉传感的激光熔丝直接能量沉积过程稳定性监测

基于同轴视觉传感的激光熔丝直接能量沉积过程稳定性监测

蔡玉华 程新玉 陈勇 熊俊 陈辉

航空科学技术2025,Vol.36Issue(5):67-74,8.
航空科学技术2025,Vol.36Issue(5):67-74,8.DOI:10.19452/j.issn1007-5453.2025.05.008

基于同轴视觉传感的激光熔丝直接能量沉积过程稳定性监测

Monitoring of Deposition Process Stability in Wire-Laser Direct Energy Deposition Based on Coaxial Visual Sensing

蔡玉华 1程新玉 1陈勇 2熊俊 1陈辉1

作者信息

  • 1. 西南交通大学,四川 成都 610031
  • 2. 中航工业成都飞机工业(集团)有限责任公司,四川 成都 610092
  • 折叠

摘要

Abstract

The wire-laser directed energy deposition(DED)technique can balance forming accuracy and deposition efficiency,and has shown broad application prospects in the rapid manufacturing of complex components in the aerospace and defense fields.However,the immaturity of monitoring methods for the deposition process stability and the difficulty in real-time control of forming qualities in wire-laser DED seriously limit the manufacturing process's automation level and component reliability.This paper proposes a monitoring method based on coaxial visual sensing and deep learning to characterize the intersection point of the laser beam and the wire to the top layer distance(IPTD)in wire-laser DED for achieving real-time detection of the deposition process stability.The abilities of three different convolutional neural network models to extract IPTD features from molten pool images are studied by performing classification tasks.An IPTD regression model is designed based on the optimal CNN model determined by classification experiments.The fitting performance of the IPTD regression model is studied and discussed.Compared with other models,the ResNet18 model has the highest training convergence speed and an optimal classification accuracy of 0.996.The prediction results'average absolute error of the IPTD regression model on the testing dataset is less than 0.318mm.The prediction time of the IPTD regression model for a molten pool image is within 4ms.The research results indicate that the proposed detection method of deposition process stability in wire-laser DED has high detection accuracy,providing a theoretical and technical basis for the height stability control in wire-laser DED of complex metal components.

关键词

激光直接能量沉积/增材制造/在线监测/深度学习/成形质量

Key words

laser DED/additive manufacturing/online monitoring/deep learning/forming quality

分类

金属材料

引用本文复制引用

蔡玉华,程新玉,陈勇,熊俊,陈辉..基于同轴视觉传感的激光熔丝直接能量沉积过程稳定性监测[J].航空科学技术,2025,36(5):67-74,8.

基金项目

国家自然科学基金(62173280,51975491) (62173280,51975491)

航空科学基金(2023Z049109001) (2023Z049109001)

四川省科技计划(2023NSFSC1956,2024JDRC0022) (2023NSFSC1956,2024JDRC0022)

中央高校基本科研业务费(2682023ZTPY023) National Natural Science Foundation of China(62173280,51975491) (2682023ZTPY023)

Aeronautical Science Foundation of china(2023Z049109001) (2023Z049109001)

Sichuan Science and Technology Program(2023NSFSC1956,2024JDRC0022) (2023NSFSC1956,2024JDRC0022)

Fundamental Research Funds for the Central Universities(2682023ZTPY023) (2682023ZTPY023)

航空科学技术

1007-5453

访问量1
|
下载量0
段落导航相关论文