Large-scale additive manufacturing (LSAM) has a similar mechanism to the fused filament fabrication (FFF) and is capable of fabricating a part in large size. This capability provides LSAM with potentials in a variety of industries, including aerospace and automotive manufacturing. Product quality and production efficiency are two main concerns, as LSAM is implemented. It has been proven that print surface temperature is a major factor that impacts the quality of final products. Therefore, it needs to be controlled throughout the process. As an infrared camera is implemented, the real-time data of surface temperature of parts are available. A dynamic approach is studied in this article to perform real-time layer time control based on the real-time data from the infrared camera to improve both product quality and production efficiency. A regression model is formulated and proved to fit the cooling dynamics. To deal with the layerwise change of cooling dynamics, due to humidity and airflow, the Gaussian process is used to keep the regression model updated. The regression model together with the Gaussian process can predict the surface temperature of a part accurately, even in a dynamic environment. This method to predict surface temperature is then combined into an optimization model for real-time layer time control. Specifically, more than one position on the surface is monitored and considered in the optimization model, and the resulting layer time for each layer by solving the optimization model has the quality requirement satisfied and improves production efficiency. The improved system performance is presented in a case study. This article provides practitioners of LSAM with a useful tool to improve the process through manufacturing automation. Note to Practitioners-Carbon fiber reinforced thermoplastic material is used for large-scale additive manufacturing (LSAM) to fabricate parts in large size. This technology is new compared with other additive manufacturing technologies, and several key problems are to be addressed before it is widely applied in industry. One issue is product quality, which depends largely on print surface temperature. Quality problems caused by improper print surface temperature include cracking, warping, and deformation. Another problem is the operation inefficiency, which results in a high cost. Currently, it takes hours to print a single part. This article provides a framework to improve both quality and efficiency of LSAM by employing the real-time data captured from the infrared thermal camera. Specifically, a regression model is formulated to describe the surface temperature with high accuracy. Then, a layer time control method is proposed to schedule printing operations in real time to guarantee high printing efficiency and quality.

中文翻译：

---

大规模增材制造的打印表面热建模和层时间控制


大规模增材制造 (LSAM) 与熔丝制造 (FFF) 具有类似的机制，能够制造大尺寸的零件。这种能力为 LSAM 提供了在航空航天和汽车制造等多种行业中的潜力。实施 LSAM 时，产品质量和生产效率是两个主要问题。事实证明，印刷表面温度是影响最终产品质量的主要因素。因此，需要在整个过程中进行控制。由于采用了红外摄像头，可以获得零件表面温度的实时数据。本文研究了一种动态方法，根据红外热像仪的实时数据进行实时分层时间控制，以提高产品质量和生产效率。制定并证明回归模型适合冷却动力学。为了处理由于湿度和气流导致的冷却动力学的分层变化，使用高斯过程来保持回归模型的更新。回归模型与高斯过程一起可以准确预测零件的表面温度，即使在动态环境中也是如此。然后将这种预测表面温度的方法结合到实时层时间控制的优化模型中。具体地，在优化模型中监测和考虑表面上的多个位置，通过求解优化模型得到的每一层的层时间满足质量要求并提高生产效率。案例研究中介绍了改进的系统性能。本文为 LSAM 的实践者提供了一个有用的工具，可以通过制造自动化来改进流程。 从业者须知——碳纤维增强热塑性材料用于大规模增材制造（LSAM）来制造大尺寸零件。该技术与其他增材制造技术相比是较新的技术，在工业上广泛应用之前还需要解决几个关键问题。问题之一是产品质量，这在很大程度上取决于印刷表面温度。打印表面温度不当引起的质量问题包括开裂、翘曲、变形等。另一个问题是运营效率低下，导致成本高昂。目前，打印一个零件需要几个小时。本文提供了一个框架，通过利用红外热像仪捕获的实时数据来提高 LSAM 的质量和效率。具体来说，建立回归模型来高精度地描述表面温度。然后，提出了一种层时间控制方法来实时调度打印操作，以保证高打印效率和质量。