Process Parameter Prediction for Fused Deposition Modeling Using Invertible Neural Networks

Abstract

Additive manufacturing has revolutionized prototyping and small-scale production in the past years. By creating parts layer by layer, a tool-less production technology is established, which allows for rapid adaption of the manufacturing process and customization of the product. However, the geometric freedom of the technologies comes with a large number of process parameters, especially in Fused Deposition Modeling (FDM), all of which influence the resulting part's properties. Since those parameters show interdependencies and non-linearities, choosing a suitable set to create the desired part properties is not trivial. This study demonstrates the use of Invertible Neural Networks (INN) for generating process parameters objectively. By specifying the desired part in the categories of mechanical properties, optical properties and manufacturing time, the demonstrated INN generates process parameters capable of closely replicating the desired part. Validation trials prove the precision of the solution with measured properties achieving the desired properties to up to 99.96% and a mean accuracy of 85.34%.

Keywords: additive manufacturing; fused deposition modeling; neural network; part quality; process parameters; production management.

Figure 1

Triangle of goals for FDM, depicted as radial chart. Not all properties can be reached optimally at the same time.

Figure 2

General architecture of the invertible neural network used for process parameter prediction.

Figure 3

Design of experiments for creating training data. For better illustration, corner points are separated from the star pattern. Combined, the two graphs represent the entire trial plan.

Figure 4

Five tensile test specimens according to DIN EN ISO 527 in lying (left) and standing orientation (right).

Figure 5

Scenario 1 to 3 used to validate the accuracy of the demonstrated INN. Each scenario favors two distinct part qualities while allowing a compromise in the third. The gray area shows the fulfilment of each quality category.

Figure 6

Scenarios 4 and 5 evaluate whether process parameters for medium and above-average part properties can be generated. The gray area shows the fulfilment of each quality category.

Figure 7

Radar chart for input accuracies in scenarios 1 (left) and 2 (right). The gray band shows the interval between best and worst accuracy for each quality characteristic.

Figure 8

Radar chart for input accuracies in scenario 3 (left) and 4 (right). The gray band shows the interval between best and worst accuracy for each quality characteristic.

Figure 9

Radar chart for input accuracies in scenario 5 (left) and the combination of all scenarios (right). The gray band shows the interval between best and worst accuracy for each quality characteristic. The dashed line in the right plot shows the mean value.