Assessment and Applicability of Machine Learning Models for Quality Monitoring in Electric Vehicle Connector Manufacturing
Fragment książki (Rozdział monografii pokonferencyjnej)
MNiSW
20
Poziom I
| Status: | |
| Autorzy: | Antosz Katarzyna, Kulisz Monika, Michaluk Justyna, Knapčíková Lucia |
| Dyscypliny: | |
| Aby zobaczyć szczegóły należy się zalogować. | |
| Wersja dokumentu: | Drukowana | Elektroniczna |
| Język: | angielski |
| Strony: | 189 - 204 |
| Scopus® Cytowania: | 0 |
| Bazy: | Scopus |
| Efekt badań statutowych | NIE |
| Materiał konferencyjny: | TAK |
| Nazwa konferencji: | 10th EAI International Conference on Management of Manufacturing Systems |
| Skrócona nazwa konferencji: | 10th EAI MMS 2025 |
| URL serii konferencji: | LINK |
| Termin konferencji: | 8 października 2025 do 10 października 2025 |
| Miasto konferencji: | High Tatras |
| Państwo konferencji: | SŁOWACJA |
| Publikacja OA: | NIE |
| Abstrakty: | angielski |
| This study explores the use of machine learning algorithms to predict product conformity in the production of electric vehicle (EV) charging connectors. The main aim was to evaluate and compare the performance of five classification models—Logistic Regression (LR), Naive Bayes (NB), Decision Tree (DT), support vector machine (SVM), and neural network (NN)—using real-world process data. The dataset included 19 input variables representing technical, environmental and organizational factors, which were collected at various stages of the cable processing workflow. It also included binary output labels indicating whether a product was conforming or non-conforming. Each model was assessed using standard classification metrics, including accuracy, precision, recall, and F1 score. In addition to predictive performance, model interpretability and computational efficiency were considered to determine their practical suitability for industrial implementation. The results showed that the neural network achieved the highest classification accuracy (94.5%), followed by SVM (93.1%) and Decision Tree (91.2%). LR and NB achieved lower accuracy (89.5% and 85.7%, respectively), but offered advantages in terms of simplicity and efficiency. Of all the models, the DT provided the best overall balance between accuracy, speed, and interpretability. These findings demonstrate the potential of machine learning to enhance data-driven decision-making in industrial quality control, emphasizing the importance of selecting appropriate model architectures based on performance requirements and operational constraints. |
