Machine learning-based anomaly detection for product quality monitoring with noncontact sensor

Abstract Measurement systems and sensor technologies are essential components of production engineering in advanced manufacturing. On-machine measurements allow for real-time evaluation anomaly detection for product quality monitoring of machining parameters, operational conditions, and outcomes directly on the machine tool. This study presents an anomaly detection framework based on Support Vector Regression (SVR) to enhance monitoring in automated manufacturing systems. By leveraging small to medium dimensional sensor data from laser-based non-contact probing, the SVR model achieves a 99% accuracy rate in anomaly detection. The results demonstrate the model’s capability to capture complex nonlinear patterns and adapt to various abnormal conditions, offering a practical and intelligent solution for real-time production monitoring in Industry 4.0.