Programmable Density of Laser Additive Manufactured Parts by Considering an Inverse Problem.

Area of Science:

• Materials Science
• Manufacturing Engineering
• Artificial Intelligence

Background:

• Increasing demand for tailored process parameters in laser additive manufacturing (LAM) of AlSi10Mg components.
• Potential for reduced production times and costs by decreasing component densities, especially with post-processing like hot isostatic pressing.
• Need for automated methods to predict and control component density.

Purpose of the Study:

• To develop an artificial neural network (ANN) model for predicting the relative density of AlSi10Mg components based on process parameters.
• To create an inverse prediction model using concatenated ANNs to determine optimal process parameters for a target density.
• To enhance the efficiency and reduce costs in laser additive manufacturing through automated parameter generation.

Main Methods:

• Training an ANN model on a synthetic dataset and a statistical test series (256 instances) for density prediction across a 70%-100% range.
• Developing a database approach and supervised training of concatenated ANNs to solve the inverse parameter prediction problem.
• Validating the model's ability to reproduce data distribution and predict relative density with high accuracy (R² = 0.98).

Main Results:

• The concatenated ANN model accurately predicts relative density for synthetic data with an R²-value of 0.98.
• A 12% increase in mean build rate was achieved through backward model training.
• The model demonstrated the ability to generate parameter predictions within a high-dimensional result space.

Conclusions:

• The developed ANN model effectively predicts and enables the inverse prediction of relative density for AlSi10Mg components in laser additive manufacturing.
• The approach offers a pathway to reduce processing time and scrap rates by generating optimal process parameters.
• Future work includes training on larger datasets and incorporating additional properties like surface quality and mechanical strength for broader practical application.