Application of Machine Learning Methods with Dimension Reduction Techniques for Fault Prediction in Molding Process
Bildiriler > Tam Metin Bildiri · 2019
Yazarlar: Fatma DEMİRCAN KESKİN, İnanç KABASAKAL
10th International Symposium on Intelligent Manufacturing and Service Systems, Türkiye, Sakarya, Uluslararası, Sözlü Sunum, Tam Metin Bildiri, 2019-2019, ss 1001-1010
Özet
<p>Significant advances in technology, advanced analytical tools that can create value from big data, have had a substantial impact on the production environment, and laid the foundation for Industry 4.0 and intelligent production concepts. Predictive engineering, one of the key pillars of smart manufacturing, necessitates collecting real-time, accurate data, analyzing the data from an anticipatory point of view with advanced analytical techniques to provide value. </p><p>In the literature, machine learning based methods have received a great deal of attention to extract valuable information from the process data and predict faults. In this study, fault prediction problem in the molding process that includes a number of successive steps is addressed by applying machine learning methods with dimension reduction techniques. The dimension reduction techniques of Principal Component Analysis (PCA), and Isometric Feature Mapping (Isomap) are first utilized for feature selection. Then, machine learning methods of Support Vector Machine (SVM), Neural Network (NN), and Logistic Regression (LR) are applied for predicting faults. The data for the analysis include measurements of faults and possible fault causes taken from the molding process of a wheel rim manufacturer. Prediction performances of the applied methods and the effects of PCA and Isomap on these methods are assessed by employing the criteria of prediction accuracy, area under curve (AUC), Type I, and Type II error.</p>