Technology development in additive manufacturing is accelerating transition from mass production to mass customization. In this transition, automation in all stages of production including quality control is a key. In this study, a layer-wise framework is proposed to monitor quality of 3D printing parts based on top-view images. The proposed statistical process monitoring method starts with self-start control charts that require only two successful initial prints. Answering the challenges of image processing due to lighting, a Machine Learning (ML) method is adopted to separate each layer from the printing bed. A sample image is compared to the standard image from a good part at each layer. The number of pixels in the difference images is fed into the proposed control charts to monitor printing process at each layer. An Exponentially Weighted Moving Average (EWMA) chart based on the number of pixels is used for process monitoring at each layer. Once enough parts have been printed, homogeneous layers are clustered to reduce the number of control charts needed for process monitoring. Experimental results based on a 3-inch diameter basket part show that the proposed framework based on continuously monitoring of layer-by-layer images is able of detecting small changes in printing process.

增材制造技术的发展正在加速从大规模生产到大规模定制的过渡。在此过渡中，包括质量控制在内的所有生产阶段的自动化都是关键。在这项研究中，提出了一个分层框架来监视基于顶视图图像的3D打印部件的质量。所提出的统计过程监视方法从仅需要两次成功的初始打印的自启动控制图开始。为应对照明带来的图像处理挑战，采用了机器学习（ML）方法将每一层与印刷床分开。将每一层的大部分样本图像与标准图像进行比较。差异图像中的像素数量被馈送到建议的控制图中，以监视每一层的打印过程。基于像素数的指数加权移动平均值（EWMA）图表用于每一层的过程监视。一旦打印了足够多的零件，就将同质层聚在一起以减少过程监控所需的控制图数量。基于直径为3英寸的篮筐部分的实验结果表明，所提出的基于连续监视逐层图像的框架能够检测到打印过程中的细微变化。