Improving Process Control Through Decision Tree-Based Pattern Recognition

This paper explores the integration of decision tree classifiers in the assessment of machining process stability using control charts. The inherent variability in manufacturing processes requires a robust system for the early detection and correction of disturbances, which has traditionally relied on operators' experience. Using decision trees, this study presents an automated approach to pattern recognition on control charts that outperforms the accuracy of human operators and neural networks. Experimental research conducted on two datasets from surface finishing processes demonstrates that decision trees can achieve perfect classification under optimal parameters. The results suggest that decision trees offer a transparent and effective tool for quality control, capable of reducing human error, improving decision making, and fostering greater confidence among company employees. These results open up new possibilities for the automation and continuous improvement of machining process control. The contribution of this research to Industry 4.0 is to enable the real-time, data-driven monitoring of machining process stability through decision tree-based pattern recognition, which improves predictive maintenance and quality control. It supports the transition to intelligent manufacturing, where process anomalies are detected and resolved dynamically, reducing downtime and increasing productivity.