Artificial neural network (ANN) based prediction of process parameters in additive manufacturing

Hardik D. Sondagar, Seema Bhadauria, Vin Sharma

2021IOP Conference Series Materials Science and Engineering17 citationsDOI Open Access PDF

Abstract

Abstract In recent years, selective laser melting (SLM), a part of additive manufacturing (AM) is one of the most encouraging ones that permit fabricating metallic parts from metal powder with complex geometry. Diversities in thesecycle boundaries become an imperative system to improve the nature of the outcome.Cycleboundaries, for example, laser power, scan speed, hatch spacing, layer height used as input parameters and have a significant impact on the mechanical property taken as an output parameter of the manufactured part. The Artificial Neural Network (ANN) model includes a multi-layer perceptron (MLP) learning algorithm named as Levenberg-Marquardt and tangent sigmoid function consider as preparing and testing functions respectively utilizing MATLAB toolkit. Ideal cycle boundaries are attained dependent on the mean square error function (MSE) and correlation coefficient(R 2 ).

Topics & Concepts

Sigmoid functionArtificial neural networkSelective laser meltingMean squared errorPerceptronMATLABMultilayer perceptronProcess (computing)Computer scienceActivation functionAlgorithmArtificial intelligenceMathematicsLaserStatisticsOpticsPhysicsOperating systemAdditive Manufacturing Materials and ProcessesAdditive Manufacturing and 3D Printing TechnologiesWelding Techniques and Residual Stresses