基于视觉大模型的激光粉末床熔融铺粉缺陷检测  

Powder⁃Spreading Defect Detection in Laser Powder Bed Fusion Based on Large Vision Model

作　　者：谭鲲鹏 唐甲锋 赵志斌 王晨希 张兴武 何卫锋[2] 陈雪峰 Tan Kunpeng;Tang Jiafeng;Zhao Zhibin;Wang Chenxi;Zhang Xingwu;He Weifeng;Chen Xuefeng(Institute of AeroEngine,School of Mechanical Engineering,Xi’an Jiaotong University,Xi’an 710049,Shaanxi,China;National Key Lab of Aerospace Power System and Plasma Technology,Air Force Engineering University,Xi’an 710038,Shaanxi,China)

机构地区：[1]西安交通大学机械工程学院航空发动机研究所,陕西西安710049 [2]空军工程大学航空动力系统与等离子体技术全国重点实验室,陕西西安710038

出　　处：《中国激光》2024年第10期267-276,共10页Chinese Journal of Lasers

基　　金：国家重点研发计划(2022YFB4600800)。

摘　　要：激光粉末床熔融增材制造面临质量稳定一致性的挑战,铺粉质量是影响成形件质量的重要因素。近年来,计算机视觉在铺粉缺陷监测中的应用表现突出,但其性能却受到标注数据数量不足的限制。针对这一问题,笔者设计了基于视觉大模型分割一切模型(SAM)的铺粉缺陷分割模型(PSAM)。针对SAM预训练参数的知识迁移问题,引入Adapter模块实现参数微调;针对铺粉分割任务中类别信息的需求,改进了SAM中的掩码解码器;针对工业场景中人工提示难的问题,提出了自动提示生成器,实现了视觉提示的自动生成。在训练样本数量仅为50的情况下,PSAM表现出了良好的分割性能,平均交并比(mIoU)可达到65.02%,相较于Deeplab v3和U-Net分别提升了8.52个百分点和5.31个百分点。本研究展示了视觉大模型在增材过程监控中的应用价值和应用潜力。Objective To date,laser powder bed fusion(LPBF)is considered as the most advanced metal additive manufacturing technology.It has been widely adopted for the production of critical metal components in aerospace and healthcare industries.However,realizing quality stability and consistency is challenging because of the coupled effects of various factors during LPBF.The powderspreading quality is a crucial characteristic of LPBF process monitoring.Defects during powder spreading can introduce defects into the formed components.In recent years,the application of computer vision in powderspreading defect detection has shown promising results.However,the limited availability of annotated data constrains its performance.Large vision models,such as the segment anything model(SAM),exhibit remarkable generalization capabilities owing to pretraining on an extremely large dataset.This allows its transfer to various downstream tasks with minimal training data.However,owing to the lack of defect knowledge,absence of category information,and dependence on manual prompts,SAM cannot be directly applied to powderspreading defect segmentation.This study addresses the requirements for powderspreading defect segmentation by improving SAM,achieving excellent defect segmentation performance with minimal training samples,and exploring the potential application of large vision models in monitoring the additive manufacturing process.Methods In this study,the powderspreading defect segment anything model(PSAM),based on SAM,was introduced.The overall structure of PSAM was similar to that of SAM,which consisted of an image encoder,an autoprompt generator,and a mask decoder.Compared to the original SAM,PSAM incorporated the following improvements:To address the issue of knowledge transfer concerning SAM's pretrained parameters,four Adapter modules were introduced into the SAM image encoder structure.These Adapter modules enabled efficient adjustment of image feature encoding.They were inserted behind the multihead attention layer in the transf

关 键 词：激光技术 激光粉末床熔融 过程监测 视觉大模型 缺陷检测 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]