Part and process monitoring for 3D printing technologies

• Main Author: Ong, Benjamin Jin Rui
• Other Authors: Moon Seung Ki
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Manufacturing; Engineering::Aeronautical engineering
• Online Access: https://hdl.handle.net/10356/157383
• Institution: Nanyang Technological University

For Direct Energy Deposition (DED) technologies, process monitoring is essential for consistently achieving high quality parts and stabilization of processes. Hence, the appropriate process parameters must be selected for evaluation to enable effective process monitoring. The key contributions of this FYP project are: (1) The development of an in-situ monitoring system to capture optical images of the L-DED process, (2) The implementation of OpenCV image processing pipelines, (3) Provision of 3D and 2D data visualization of the melt pool features. Firstly, a robust and efficient in-situ process monitoring strategy was proposed. A Charged Couple Device (CCD) camera was integrated into a Computer Numerically Controlled (CNC) laser aided additive manufacturing (LAAM) system to identify significant melt pool characteristics. Image processing techniques via various OpenCV libraries were then developed to extract in-situ melt pool features. These libraries include image binarization (Simple Thresholding), image masking (Bitwise functions), image noise removal (morphological transformations) and melt pool shape fitting (Contour generation). The python software infrastructure was tested on pyrometer and Infrared (IR) images from an open-sourced database. Raw pixel data was extracted from the Comma Separated Values (CSV) files and subsequently reorganized and normalized to lie within the OpenCV pixel value limits. Next, data cleaning was carried out to remove irrelevant (laser transition images) before storing the remaining data into a big list as Portable Graphics Format (PNG) files to retain all the image information. Following that, the PNG images were passed through the OpenCV image processing pipelines to extract various melt pool parameters. Some of these parameters include the Heat Affect Zone (HAZ) area, inner melt pool length, width, area, maximum and average temperature. Finally, these parameters were plotted against time and layer number to provide a 2D and 3D visualization of the melt pool. As such, the relative standard deviation, and the coefficient of determination of the melt pool parameters were derived to identify the most reliable parameter to be used for future data-driven control models.