In-Process Quality Prediction in Additive Manufacturing using Generative AI: An Experimental Study

Fused Filament Fabrication (FFF), a widely adopted additive manufacturing (AM) process, employs an incremental deposition approach to fabricate products layer by layer. Despite its versatility and cost-effectiveness, challenges persist in ensuring desired product quality due to process complexity and layer-to-layer variations. Recent advancements have enabled in-process defect detection and monitoring of printed layers. However, existing methods exhibit limited predictive capabilities in understanding defect propagation across successive layers and its impact on final product quality which represents a critical knowledge gap essential for proactive defect assessment and mitigation. A novel generative AI (GenAI)-based quality prediction method that leverages the image-generative capabilities of generative adversarial network (GAN) models to predict subsequent layer quality based on prior ones is proposed to address this gap. This approach employs a purpose-designed GAN architecture that captures spatial and texture variations across sequential layer images. By integrating printing process parameters alongside these images, the proposed method provides rich information regarding next layer’s texture based on the data collected from the layer images from over prints with various process settings, along with their associated printing parameters. The findings demonstrate the model’s ability to accurately predict next-layer properties, facilitating effective defect tracking and in-process quality prediction.