No abstract

Real-time and in-situ printing performance diagnostic in vat photopolymerization is critical to control printing quality, improve process reliability, and reduce wasted time and materials. This paper proposed a low-cost smart resin vat to monitor the printing process and detect the printing faults. Built on a conventional vat photopolymerization process, we added equally spaced thermistors along the edges of the resin vat. During printing, polymerization heat transferred to the edges of the resin vat, which increased thermistors’ temperature and enhanced resistances. The heat flux received at each thermistor varied with the distance to the place of photopolymerization. The temperature profiles of all thermistors were determined by the curing image pattern in each layer, and vice versa. Machine learning algorithms were leveraged to infer the printing status from the measured temperatures of these thermistors. Specifically, we proposed a simple and robust Failure Index to detect if the printing was active or terminated. Gaussian process regression was utilized to predict the printing area using the temperature recordings within a layer. The model was trained, validated, and tested using the data set collected by printing six parts. Different printing abnormalities, including printing failures, manual printing pause, and missing features (incorrect printing area), were successfully detected. The proposed approach modified the resin vat only and could be easily applied to all vat photopolymerization processes, including SLA, DLP, and LCD based 3D printing. The limitation and future work are also highlighted.