Scholarlink

<p>Fused Filament Fabrication (FFF) is a widely utilized additive manufacturing (AM) technique, but nozzle clogging remains a significant challenge, affecting print quality and increasing material wastage. Early and accurate detection of clogging is essential for maintaining process efficiency and reliability. This research explores the use of machine learning (ML) models for nozzle clogging condition detection in FFF printing. A Cartesian 3D printer was equipped with three sensors namely Rotary Encoder, Load Cell, and Thermocouple to measure filament feed rate, extrusion force, and extruder temperature, respectively. Data collection followed the Taguchi L9 Design of Experiments (DoE), ensuring a diverse dataset of 718,200 data points across three nozzle conditions: Healthy, Partially clogged, and Fully clogged. ML models namely Logistic Regression (LR), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Naïve Bayes (NB), Decision Tree (DT), Random Forest (RF), and Gradient Boosting Classifier (GBC) were trained and evaluated based on accuracy, precision, recall, F1 score, and the Areas Under Curve (AUC) values of Receiver-Operating Characteristic (ROC) plots. Among these, GBC emerged as the best-performing model with an accuracy of 99.92%, achieving perfect classification across all classes. RF and DT also exhibited high accuracy of 99.84% and 99.51%, respectively. While models like SVM and KNN demonstrated slightly lower performance, NB had limitations in distinguishing between clogging conditions. The results validate the effectiveness of ML models in nozzle clogging detection, with GBC being particularly suited for real-time deployment in closed-loop systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Nozzle Clogging Detection in FFF 3D Printing Using Machine Learning Approach: A Comparative Investigation

  • 期刊论文

  • 发表日期：

    2025年6月12日

查看全文
  • Anil Ansoliya,
  • Vijay Kumar Gupta

摘要

Fused Filament Fabrication (FFF) is a widely utilized additive manufacturing (AM) technique, but nozzle clogging remains a significant challenge, affecting print quality and increasing material wastage. Early and accurate detection of clogging is essential for maintaining process efficiency and reliability. This research explores the use of machine learning (ML) models for nozzle clogging condition detection in FFF printing. A Cartesian 3D printer was equipped with three sensors namely Rotary Encoder, Load Cell, and Thermocouple to measure filament feed rate, extrusion force, and extruder temperature, respectively. Data collection followed the Taguchi L9 Design of Experiments (DoE), ensuring a diverse dataset of 718,200 data points across three nozzle conditions: Healthy, Partially clogged, and Fully clogged. ML models namely Logistic Regression (LR), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Naïve Bayes (NB), Decision Tree (DT), Random Forest (RF), and Gradient Boosting Classifier (GBC) were trained and evaluated based on accuracy, precision, recall, F1 score, and the Areas Under Curve (AUC) values of Receiver-Operating Characteristic (ROC) plots. Among these, GBC emerged as the best-performing model with an accuracy of 99.92%, achieving perfect classification across all classes. RF and DT also exhibited high accuracy of 99.84% and 99.51%, respectively. While models like SVM and KNN demonstrated slightly lower performance, NB had limitations in distinguishing between clogging conditions. The results validate the effectiveness of ML models in nozzle clogging detection, with GBC being particularly suited for real-time deployment in closed-loop systems.