Genetic Algorithm-Based Data-Driven Process Selection System for Additive Manufacturing in Industry 4.0

Area of Science:

• Manufacturing Engineering
• Computer Science
• Materials Science

Background:

• Additive Manufacturing (AM) enables complex 3D object creation but lacks automated design rules.
• Selecting appropriate AM processes requires specialized expertise, hindering widespread adoption.
• Industry 4.0 demands data-driven solutions for Design for Additive Manufacturing (DFAM).

Purpose of the Study:

• To develop an automated, data-driven system for AM process selection within the DFAM framework.
• To capture and apply expert knowledge for optimizing AM process selection.
• To address manufacturability challenges in 3D-printed parts.

Main Methods:

• Utilized a Genetic and Evolutionary Feature Weighting technique with 3D CAD data.
• Developed a two-stage predictive model.
• Benchmarked Steady-State Genetic Algorithm (SSGA) against Estimation of Distribution Algorithm (EDA) and Particle Swarm Optimization (PSO).

Main Results:

• The two-stage model achieved average accuracies of 70% (Stage 1) and up to 97.33% (Stage 2).
• SSGA demonstrated superior performance compared to EDA and PSO for AM process selection.
• The system successfully identifies optimal AM processes for specific 3D object geometries.

Conclusions:

• The developed automated system accurately identifies optimal AM processes, supporting DFAM.
• The data-driven approach effectively captures and applies expert knowledge.
• This system offers a scalable solution for improving AM process selection and manufacturability.