Selective laser melting (SLM) is a combined process of melting and stacking three-dimensional products by fusing micro-metal powder using a laser. It has the advantage of manufacturing parts with complex structures with reduced production time. However, in the case of aluminum, the disadvantages of poor laser formability due to its high thermal conductivity, diffusivity, and reflectivity result in process defects such as bowling, pore, and poor surface quality. This study aims to develop a surface defect removal methodology during aluminum melting by laser processing and to enhance process automation capabilities by introducing a sensor monitoring scheme. In the laser experiments, aluminum specimens (AL-7075) with mechanical scratches were used and the level of surface defect removal during processing was classified depending on surface conditions. In addition, a PVDF-type acoustic emission (AE) sensor monitoring system was implemented to collect characteristic signals during the laser polishing of the machined surface(scratch). It was shown that the degree of surface defects removal and surface state could be effectively classified through a convolution neural network (CNN) utilizing the collected signals as input vectors.
|