Sensor fusion enabled machine learning for quality and material optimization in robotic additive manufacturing

传感器融合支持机器学习，以优化机器人增材制造中的质量和材料

• 批准号: 561049-2020
• 负责人: Qureshi, AhmedJawad
• 金额: $ 3.64万
• 依托单位: University of Alberta
• 依托单位国家: 加拿大
• 项目类别: Alliance Grants
• 财政年份: 2021
• 资助国家: 加拿大
• 起止时间: 2021-01-01 至 2022-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=720703
• 关键词: Sensor; fusion; enabled; machine; learning

Metal additive manufacturing (AM) is an emerging technology that has seen exponential growth in the advanced manufacturing industry. This technology produces complex freeform shapes with precise metal deposition using robotic, or CNC, systems. A core element in quality assurance of metal AM processes is understanding the manufacturing process' effect on the deposited material's microstructure which directly affects the mechanical behavior, geometry, and quality of the printed part. Discovering these effects requires extensive effort and time, and is a laborious process limiting the maturity of technology. However, with the advancement of Industry 4.0, process automation and machine learning (ML) tools, there is an opportunity to accelerate our understanding of metal additive manufacturing processes.The applicants will leverage the cutting edge advances in robotic metal additive manufacturing (RM-AM), machine learning (ML), and deep learning (DL) to develop an automated, scalable, platform agnostic cyber-physical system to achieve two main goals pertaining to robotic metal additive manufacturing (metal-AM): (i) quality optimization of the parts produced through process parameter control, and (ii) predictive modeling of the mechanical properties of parts printed from materials of a given composition. ML techniques circumvent the need for the conventional time, and resource-intensive parameter assessment and optimization process carried out by human experts. This system will consist of a hardware layer to execute metal deposition integrated with sensor-fusion to monitor the geometry of the parts being manufactured. The sensors' data will be processed through an ML trained feedback control system that would send the appropriate control signals to the execution hardware to tune the process parameters to optimize part quality. The tangible deliverables of the project will be in form of an autonomous robotic metal additive manufacturing system, sensor fusion based multi source high frequency sensor suite for robotic metal additive manufacturing systems, and a set of AI based algorithms for model identification and mechanical and geometric property improvement for the additive manufacturing systems.

金属增材制造（AM）是一项新兴技术，在先进制造业中呈指数级增长。该技术使用机器人或CNC系统，通过精确的金属沉积产生复杂的自由形状。金属增材制造过程质量保证的核心要素是了解制造过程对沉积材料微观结构的影响，这直接影响到打印部件的机械性能、几何形状和质量。发现这些效果需要大量的努力和时间，并且是一个限制技术成熟的艰苦过程。然而，随着工业4.0、过程自动化和机器学习（ML）工具的进步，有机会加速我们对金属增材制造工艺的理解。申请人将利用机器人金属增材制造（RM-AM）、机器学习（ML）和深度学习（DL）方面的前沿技术，开发一个自动化、可扩展、与平台无关的网络物理系统，以实现与机器人金属增材制造（metal- am）相关的两个主要目标：(1)通过工艺参数控制生产的零件的质量优化，(2)用给定成分的材料打印零件的机械性能的预测建模。机器学习技术规避了由人类专家进行的传统时间和资源密集型参数评估和优化过程的需要。该系统将由一个硬件层组成，用于执行金属沉积，集成传感器融合，以监测正在制造的部件的几何形状。传感器的数据将通过机器学习训练的反馈控制系统进行处理，该系统将向执行硬件发送适当的控制信号，以调整工艺参数以优化零件质量。该项目的有形交付成果将以自主机器人金属增材制造系统，基于传感器融合的机器人金属增材制造系统多源高频传感器套件，以及一套基于人工智能的增材制造系统模型识别和机械和几何性能改进算法的形式出现。