Predictive modeling of dimensional accuracies in 3D printing using artificial neural network
Additive manufacturing, particularly Fused Deposition Modeling (FDM) using three-dimensional (3D) printing, has revolutionized the manufacturing industry by offering design flexibility, customization options, affordability, and high printing speed. However, improper selection of process parameters...
Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Taylor's University
2023
|
| Online Access: | http://eprints.utem.edu.my/id/eprint/27419/2/0061718122023.PDF http://eprints.utem.edu.my/id/eprint/27419/ https://jestec.taylors.edu.my/Vol%2018%20Issue%204%20August%202023/18_4_24.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Additive manufacturing, particularly Fused Deposition Modeling (FDM) using three-dimensional (3D) printing, has revolutionized the manufacturing industry by offering design flexibility, customization options, affordability, and high
printing speed. However, improper selection of process parameters in FDM can lead to suboptimal surface efficiency, defective mechanical properties, increased waste, and higher production costs. In this research, an Artificial Neural Network
(ANN) model was developed to optimize dimensional properties in FDM by considering control factors such as layer thickness, orientation, raster angle, raster width, and air gap. Experimental data consisting of 27 sets of control parameters and corresponding dimensional outputs were used to train and validate the ANN model. The ANN model was developed using MATLAB software, employing training functions and learning algorithms to optimize the neural network architecture. The optimized ANN structure comprised 15 neurons and 2 layers, and it demonstrated accurate prediction of dimensional properties with percentage errors ranging from 0.01% to 25.49% for length, less than 10%
for weight, and less than 4% for thickness. The mean absolute percentage error (MAPE) and root mean square error (RMSE) were used to quantify the errors, indicating the effectiveness of the ANN model in predicting dimensional properties. The results highlight the potential of ANN in optimizing FDM process parameters for improved dimensional accuracy. The ANN model provides a reliable tool for manufacturers to predict and optimize the length, weight, and thickness of 3D-printed components, leading to enhanced product quality and reduced production costs. The developed ANN model can be further extended to consider other parameters and optimize various aspects of the additive manufacturing process. |
|---|