Data-driven approaches for predicting mechanical properties and determining processing parameters of selective laser...

Area of Science:

• Materials Science and Engineering
• Additive Manufacturing
• Computational Modeling

Background:

• Conventional material models for selective laser sintering (SLS) are accurate but costly and time-consuming.
• Engineers require efficient methods to optimize laser settings and predict mechanical properties of SLS components.
• Data-driven approaches offer promising alternatives for modeling complex material behaviors.

Purpose of the Study:

• To develop and compare computational data-driven methodologies for selective laser sintering (SLS) nylon-12.
• To establish cross-correlations between processing parameters and mechanical properties.
• To introduce direct (laser settings to properties) and inverse (properties to laser settings) estimation frameworks.

Main Methods:

• Investigated three data-driven methodologies: fuzzy inference system (FIS), artificial neural networks (ANN), and adaptive neural fuzzy inference system (ANFIS).
• Formulated direct and inverse estimation frameworks to link laser settings and material properties.
• Analyzed and compared the accuracy of the proposed computational models.

Main Results:

• All proposed data-driven methodologies provided accurate estimation results for SLS nylon-12.
• The fuzzy inference system (FIS) emerged as the most accurate among the evaluated methods.
• The developed frameworks effectively correlated processing parameters with mechanical properties.

Conclusions:

• Data-driven models, particularly FIS, offer a viable and accurate alternative to conventional methods for SLS nylon-12.
• The proposed direct and inverse frameworks enable informed decision-making for laser settings and property prediction.
• Computational approaches significantly enhance the efficiency of material characterization in additive manufacturing.