The property palette: A rapid printing of performance-tunable blended polymers guided by artificial neural network

Suo, Shuaipeng; Yang, Yaodong; Wang, Ziliang; Rao, Wei-Feng

doi:10.1016/j.apmt.2023.101837

Cited by 3 publications

(3 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Combining the data in Tables 4 and 5, it can be seen that the prediction performance of the BPNN and KNN methods is significantly higher than the other three methods. The neural network method excels at addressing the challenges posed by the non-smooth and non-linear features [46,52] and showed good prediction results in this study. The KNN method hinges on predicting the filament dimensions based on their proximity to processing parameters.…”

Section: Machine Learning Results Analysismentioning

confidence: 70%

“…Three indicators were used for model performance evaluation, namely, root mean square error (RMSE), mean absolute error (MAE), and the coefficient of determination (R 2 ). The formulae for calculating these evaluation metrics are provided below [46,47]: Three indicators were used for model performance evaluation, namely, root mean square error (RMSE), mean absolute error (MAE), and the coefficient of determination (R 2 ). The formulae for calculating these evaluation metrics are provided below [46,47]:…”

Section: Machine Learningmentioning

confidence: 99%

“…The formulae for calculating these evaluation metrics are provided below [46,47]: Three indicators were used for model performance evaluation, namely, root mean square error (RMSE), mean absolute error (MAE), and the coefficient of determination (R 2 ). The formulae for calculating these evaluation metrics are provided below [46,47]:…”

Section: Machine Learningmentioning

confidence: 99%

See 2 more Smart Citations

Material Extrusion Filament Width and Height Prediction via Design of Experiment and Machine Learning

Shi,

Sun,

Tian

et al. 2023

Micromachines

View full text Add to dashboard Cite

The dimensions of material extrusion 3D printing filaments play a pivotal role in determining processing resolution and efficiency and are influenced by processing parameters. This study focuses on four key process parameters, namely, nozzle diameter, nondimensional nozzle height, extrusion pressure, and printing speed. The design of experiment was carried out to determine the impact of various factors and interaction effects on filament width and height through variance analysis. Five machine learning models (support vector regression, backpropagation neural network, decision tree, random forest, and K-nearest neighbor) were built to predict the geometric dimension of filaments. The models exhibited good predictive performance. The coefficients of determination of the backpropagation neural network model for predicting line width and line height were 0.9025 and 0.9604, respectively. The effect of various process parameters on the geometric morphology based on the established prediction model was also studied. The order of influence on line width and height, ranked from highest to lowest, was as follows: nozzle diameter, printing speed, extrusion pressure, and nondimensional nozzle height. Different nondimensional nozzle height settings may cause the extruded material to be stretched or squeezed. The material being in a stretched state leads to a thin filament, and the regularity of processing parameters on the geometric size is not strong. Meanwhile, the nozzle diameter exhibits a significant impact on dimensions when the material is in a squeezing state. Thus, this study can be used to predict the size of printing filament structures, guide the selection of printing parameters, and determine the size of 3D printing layers.

show abstract

Section: Machine Learning Results Analysismentioning

confidence: 70%