Abstract Scope |
Wire Arc Additive Manufacturing is a promising metal additive manufacturing technology that confers significant advantages over established alternatives. Benefits include high deposition rate, improved material utilization, low equipment costs, and increased part size. However, the mechanical properties of the parts are often sub-standard due to common defects such as porosity, residual stresses, and cracks. Mitigation requires online detection and prediction capabilities. A multi-sensor framework comprising arc current, arc voltage, wire feed speed, robot position and orientation, motor current, thermal imaging, and acoustics has been developed. The data obtained with this framework will be utilized to train a random forest ensemble model to predict defects in real time. Different metrics for classifying defects will be investigated, including visual inspection, dimensional measurements, and pore density. The trained model will provide the basis for a closed loop control system that can adjust the process parameters in response to defect detection. |