Machine Learning Models for Industrial Process Monitoring : A study at IPercept

This study was conducted as part of a master’s thesis at KTH Royal Institute of Technology in collaboration with IPercept in Sweden. The study focuses on the development of an intelligent process monitoring system for a pressing machine of a washer manufacturing, integrating vibration signal analysis with machine learning techniques to enhance quality control, reduce downtime, and enable predictive maintenance, thereby optimizing process efficiency. The data analysis for a specific washer type with labeled data focuses on the Z-axis data, which aligns with the expected vertical pressing motion of the pressing machine. The CRISP-DM method is followed using Python programming with the Pandas Library to develop the model for process monitoring. The most significant features were selected for classification, and the boundary values for each feature were determined for the sub-signals corresponding to individual punches in the pressing machine. The monitoring process enables predictive maintenance, fault detection, and process optimization throughout the manufacturing operation.