Machines as Craftsmen: Localized Parameter Setting Optimization for Fused Filament Fabrication 3D Printing

Abstract

FFF is known to have issues with part quality and consistency, which has limited its use to prototyping and noncritical applications where functional reliability does not affect safety. [2] The occurrence of issues such as poor surface finish, layer delamination, and poor dimensional stability depends on a number of parameter settings, including nozzle temperature, print speed, environmental conditions, geometry, and location (Figure 1a,b). [3] Experienced operators must set these parameters according to the material, part geometry, and 3D printer. It can be difficult for even an expert to select optimum parameter settings (Figure 1c,d). [4] An operator can specify settings for over 100 different printing options using a slicing software such as Cura. Attempting all combinations is not practical, so an operator must rely on their knowledge of 3D printing to adjust parameter settings. This leads to operator-dependent part-to-part variations and uncertainty in the performance of the final part. Additionally, an operator will typically select a set of global parameter settings that are used throughout the part, or at most make some layer-to-layer adjustments. A more objective solution for selecting parameter settings is needed to ensure optimal use of the printer/material performance space and consistency in 3D-printed parts, regardless of part geometry, material, system, or operator.Machine learning offers a potential approach to addressing this problem. Machine learning models are trained on thousands to millions of data points to recognize patterns that are too difficult to identify using deterministic algorithms. [5] Machine learning has been successfully applied in applications such as image processing, text classification, and speech recognition. [5][6][7] Potential uses for machine learning in 3D printing have also been studied in a limited capacity. [8] Examples of their use in both monitoring/feedback applications and predictive models include predicting property outcomes based on parameter settings, predicting global parameter settings for specific outcomes, identifying failures during printing, predicting bead geometry, adjusting geometry to prevent failures, and assessing part manufacturability. [9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27] An example of the utility of machine learning in the established quality control method of visual inspection is demonstrated by the use of a neural network to identify flaws in laser powder bed fusion 3D printing. [28,29] An Quality control and repeatability of 3D printing must be enhanced to fully unlock its utility beyond prototyping and noncritical applications. Machine learning is a potential solution to improving 3D printing performance and is explored for areas including flaw identification and property prediction. However, critical problems must be resolved before machine learning can truly enable 3D printing to reach its potential, including the very large data sets required for training and the inherently local nature of 3D print...

Influence of fused filament fabrication parameters on tensile properties of polylactide/layered silicate nanocomposite using response surface methodology

“…14,15 To date, the impact of printing speed on mechanical properties is not clear, and has been shown to have both positive and negative effects. [14][15][16][17][18][19] In general, it seems that a low printing speed leads to better adhesion between layers. This effect was reported by Deng et al.…”

Machine Learning for Advanced Additive Manufacturing

“…In recent studies, ML approaches have been realized by many researchers who have implemented them in different aspects of AM jobs. For example, Gardner et al 16 applied CNNs to predict print quality under various process parameter settings, including print speed, extrusion multiplier, and fan speed, in the FFF process. In the study, a CNN model was developed to distinguish the print condition (flaw or good quality) based on images captured under different combinations of process parameters.…”

“…14,15 Additionally, machine learning (ML) and computer vision models are developed to study the relationships between process parameters and product quality. 16,17 These techniques have the potential to create the next generation of smart, low-cost, and efficient AM systems.…”

Automated Real‐Time Detection and Prediction of Interlayer Imperfections in Additive Manufacturing Processes Using Artificial Intelligence

“…It turns out that these imperfections are largely dictated by the settings of printing parameters, first‐layer calibration, and model geometry . With recent advances in applying artificial intelligence and machine learning to materials science and engineering problems, researchers have started using machine‐learning algorithms to classify and predict different printing defections including blob, warp, and delamination based on the settings of printing parameters . Another interesting approach in the field adopts a new slicing mechanism which splits prints into spatially locked bricks to reduce warping .…”