|
| 面向激光选区熔化金属增材制造的检测技术研究进展 |
| Research Progress of Inspection Technology for Addition Manufacturing of SLM Metal |
| Received:June 27, 2019 Revised:July 10, 2019 |
| DOI：10.3969/j.issn.1674-6457.2019.04.006 |
| 中文关键词: 激光选区熔化 金属增材制造 在线检测 离线检测 过程监控 机器学习 |
| 英文关键词: selective laser melting metal additive manufacturing on-line inspection off-line inspection processing control machine learning |
| 基金项目:国家自然科学基金（51775196）；广东省科技计划（2017B090912003，2017A050501058，2014B010131002，2015B010125006，2016B090914002） |
| Author Name | Affiliation | | WU Shi-biao | South China University of Technology, Guangzhou 510641, China | | DOU Wen-hao | South China University of Technology, Guangzhou 510641, China | | YANG Yong-qiang | South China University of Technology, Guangzhou 510641, China | | WANG Di | South China University of Technology, Guangzhou 510641, China | | CHEN Xiao-jun | South China University of Technology, Guangzhou 510641, China | | LIANG Yi-fu | South China University of Technology, Guangzhou 510641, China |
|
| Hits: 3973 |
| Download times: 2377 |
| 中文摘要: |
| 激光选区熔化技术（简称SLM）近些年来发展迅速，已应用于航空航天、医疗、模具等诸多领域。首先综述了近几年SLM领域的在线检测、离线检测技术的进展，重点介绍了SLM的在线检测手段，如利用同轴/旁轴原位架构的高速CCD及红外成像装置获取SLM过程中丰富的可见光和红外信息，介绍了相关描述子提取方法，并研究描述子与SLM成形质量的相关性，另外，少部分学者基于声信号信息源、光电二极管进行了单熔道成形质量的分类识别检测。此外，还介绍了SLM的离线检测手段，除传统的材料测试分析方法外，显微CT和激光诱导击穿光谱学为SLM的缺陷三维表征和成分分析提供了高效新型的工具；在此基础上，进而重点综述了SLM的过程监测及反馈控制策略。其中，介绍了常见的机器学习模型（K均值聚类分析、支持向量机、深度置信网络、卷积神经网络等）及其在SLM过程统计描述子提取中的研究进展。还介绍了统计过程控制方法在SLM的特征量间和特征量与SLM成形质量间的关系分析及控制图生成方面的应用；最后，对SLM在线和离线检测研究进展进行了总结，并对其主要发展方向进行了展望。 |
| 英文摘要: |
| Selective laser melting technology (SLM) has developed rapidly in recent years and has been applied in aerospace, medical treatment, mold and many other fields. Firstly, the progress of online detection and offline detection technology in SLM field was reviewed. The online detection methods of SLM were introduced, especially the high-speed CCD and infrared imaging device using coaxial/parallel in-situ architecture to obtain rich visible and infrared information of SLM process. Extraction method of relevant descriptors was introduced and the correlation between descriptors and SLM forming quality was studied. In addition, a small number of scholars have carried out classification and detection of single melt forming quality based on acoustic signal information sources and photodiodes. In addition, the off-line detection method of SLM was introduced. In addition to traditional material testing and analysis methods, micro-CT and laser inductive breakdown spectroscopy provided a new and efficient tool for three-dimensional characterization and composition analysis of SLM defects. On this basis, the process monitoring and feedback control strategies of SLM were summarized. Among them, the common machine learning models (K-means clustering analysis, support vector machine, deep belief network, convolutional neural network, etc.) and their research progress in SLM process statistical descriptor extraction were introduced. The application of statistical process control method in analysis on relationship between the feature quantity of SLM and that between feature quantity and SLM forming quality as well as generation of control chart was introduced. Finally, the research progress of on-line and off-line SLM detection was summarized, and its main development direction was prospected. |
|
View Full Text
View/Add Comment Download reader |
| Close |
|
|
|
渝公网安备 50010702501719号