Abstract: Physics-Informed Neural Networks (PINNs) combine machine learning with physical laws to improve anomaly detection in industrial automation systems. This talk presents a practical PINN framework for automated 3D printing inspection using measurable parameters such as printed dimensions, bead geometry, nozzle temperature, and bed temperature. The model enforces consistency between material deposition, final geometry, and thermal conditions to identify anomalies including under-extrusion, nozzle clogging, warping, overheating, and poor layer adhesion. The session will also briefly cover other computer vision and AI-based industrial automation applications including visual fault detection, OCR-based document verification, and intelligent inspection systems used in manufacturing environments.
