Enhancing Aluminum Additive Manufacturing Quality through Optimized Cycle Boundaries and ANN Modeling of Laser Energy Distribution

Abstract

Metal powder can be used in additive manufacturing, especially with methods such as Selective Laser Melting (SLM), to create intricate metallic components. Optimising process parameters, sometimes referred to as cycle limits, such as laser power, speed, hatching distance, and layer thickness, is crucial to producing high-quality produced products. A lot of the final parts' mechanical qualities come from changing these parameters. An Artificial Neural Network (ANN) model that makes use of the Levenberg-Marquardt learning method has been used to tackle this optimization challenge. The tangent sigmoid function, which is easily implemented using MATLAB, is used as the activation function in the training and testing phases of the ANN. The primary material utilised in this experiment is powdered aluminum metal. The conventional mechanical attributes have been replaced with output metrics including Volumetric Energy Density (VED), Surface Energy Density (SED), and Linear Energy Density (LED). A structural integrity and functional assessment is eventually impacted by these factors, which shed light on energy distribution and fusion characteristics during the SLM process. Measuring the difference between expected and actual outcomes, the Mean Square Error (MSE), must be minimized by optimizing cycle boundaries. Further evaluating the prediction accuracy of the ANN model is the correlation coefficient (R²). This work intends to push quality and control in aluminum additive manufacturing forward more quickly with SLM. In conjunction with ANN-based modelling, this is accomplished by deliberately altering cycle boundaries based on LED, SED, and VED properties. Researchers hope to improve outcomes in aluminum additive manufacturing by better understanding and controlling the SLM process through the integration of various technologies.