Forecasting Warping Deformation Using Multivariate Thermal Time Series and K-Nearest Neighbors in Fused Deposition Modeling

Abstract

Over the past decades, additive manufacturing has rapidly advanced due to its advantages in enabling diverse material usage and complex design production. Nevertheless, the technology has limitations in terms of quality, as printed products are sometimes different from their desired designs or are inconsistent due to defects. Warping deformation, a defect involving layer shrinkage induced by the thermal residual stress generated during manufacturing processes, is a major factor in lowering the quality and raising the cost of printed products. This study utilized a variety of thermal time series data and the K-nearest neighbors (KNN) algorithm with dynamic time warping (DTW) to detect and predict the warping deformation in the printed parts using fused deposition modeling (FDM) printers. Multivariate thermal time series data extracted from thermocouples were trained using DTW-based KNN to classify warping deformation. The results showed that the proposed approach can predict warping deformation with an accuracy of over 80% by only using thermal time series data corresponding to 20% of the whole printing process. Additionally, the classification accuracy exhibited the promising potential of the proposed approach in warping prediction and in actual manufacturing processes, so the additional time and cost resulting from defective processes can be reduced.

A Combination of Vision‐ and Sensor‐Based Defect Classifications in Extrusion‐Based Additive Manufacturing

“…For the effect of extruder temperature and printing speed, the experimental results show that high temperature and too fast printing speed lead to abnormal material cooling and filling defects with regular patterns on the surface of the printed part. Second, low extruder temperatures can lead to uneven cooling rates throughout the printing process, and as the number of deposited layers increases, temperature gradients in the part can create residual thermal stresses that can predispose to defects such as warping [14]. However, the results also show that if the print speed is fast enough, the printed part adheres to the build platform even at a slightly lower extruder temperature, and no warping occurs at the corners of the printed part.…”

“…For example, Song et al used thermocouple sensors to record thermal timing data from a build platform and used the K-nearest neighbors (KNN) algorithm to predict the warping phenomenon in the FDM process. However, the classification accuracy was only 84% and was not compared with other related algorithms [14].…”

Development of a 3D printer optimized for rapid prototyping with continuous fiber fabrication technology

“…In fact, this feature leads to a considerable improvement in performance, as working in a heated environment reduces the effects of warping, an anaesthetic printing defect that causes strains due to uncontrolled cooling and excessive temperature changes to which the newly melted material is subjected. In addition, internal heating improves the quality of adhesion between subsequent layers of the workpiece [18,19].…”

Predicting dimensional accuracy in 3D printed polydimethylsiloxane‐carbon nanotubes composites via machine learning

“…15 The same algorithm was used to forecast warping deformation in FDM-based 3D printing, and 84% accuracy was found in 10-fold cross-validation. 16 Compared to other ML algorithms, the k-NN algorithm has a number of benefits; (a) The k-NN algorithm is much faster than other algorithms because it does not need to tune multiple parameters during training, (b) The algorithm only requires the values of the number of neighbors "k" to be taken into account for prediction rather than making assumptions about the data set. (c) The k-NN algorithm does not need to create a completely new model for each additional training instance.…”

“…Recently, the k ‐nearest neighbors ( k ‐NN) algorithm was implemented to predict surface roughness in plasma arc‐based AM 15 . The same algorithm was used to forecast warping deformation in FDM‐based 3D printing, and 84% accuracy was found in 10‐fold cross‐validation 16 . Compared to other ML algorithms, the k ‐NN algorithm has a number of benefits; (a) The k ‐NN algorithm is much faster than other algorithms because it does not need to tune multiple parameters during training, (b) The algorithm only requires the values of the number of neighbors “ k ” to be taken into account for prediction rather than making assumptions about the data set.…”