DATA-DRIVEN MONITORING OF MELT POOL, SPATTER AND PLUME IN LASER POWDER-BED FUSION AM PROCESS

Abstract

The thesis proposed an off-axial powder-bed fusion (PBF) process monitoring method with a high-speed camera. The images of three objects, that are melt pool, plume and spatters, were collected and observed. A novel image processing method was proposed for extracting the features from the three objects. The statistical characteristics of the extracted features were studied under different built conditions. To further understand the PBF process, the spattering phenomenon and the interaction of melt pool-plume-spatter were observed by high-speed photography. The laser focal position was introduced for the process dynamic analysis. In addition, the data-driven methods, support vector machine (SVM) and convolutional neural network (CNN) were adopted for process anomalies diagnosis based on the information extracted from melt pool, plume and spatters. The results demonstrate the suitability of the proposed method provide guidelines for advancing PBF process monitoring technique.