Computer vision based online monitoring technique: part quality enhancement in the selective laser melting process

This chapter envisages the practicability of integrating computer vision and machine learning approaches in the selective laser melting (SLM) technology for analyzing the end-used parts in real time during melting in a layer-by-layer sequence. Real-time monitoring offers scope and control of the processing parameters and results in high-quality parts by optimizing different SLM processing factors. Therefore the computer vision technique is exploited on the deposited layer after the SLM process. The computer vision edge detection technique is used on the layer to detect the laser melted surface area. The edge detection data for the laser melted surface area details the melted layer conditions, the porosity developed in the layer, and its deviations from the CAD model. Further, the portion left out from melting or overmelting is traced out using the proposed technique. The outermost edge of the laser-deposited layer provides a hint about the outer surface properties, such as surface finish, and thus the obtained data can be employed for the very efficient postprocessing of the part. The Greyscale pixel value (GPV) calculation from the image of a deposited layer over the entire height is captured using a high-resolution camera. This technique is exploited to determine the surface properties of the layer to inspect the laser melted zone and to find any improper melting, defects like process-induced porosity, gas-induced porosity, and many other flaws in the printed part. Further, the GPV is combined with the machine learning clustering model to classify the deposited layer into two segments: perfect and imperfect layers. Hence, combining the proposed two techniques can be an effective methodology to control part quality in SLM in an automated manner instead of employing them individually.