AI-Driven Concept for Monitoring Grinding Wheel Conditions

Grinding processes are vital in manufacturing because they impact the quality of components, which is why errors are costly. The condition of the grinding wheels used significantly influences the process/result quality and should therefore always be in the optimum range of the process. This paper describes a study focused on an external cylindrical plunge-cut grinding process that generates data to monitor the grinding wheel condition using machine learning (ML). Various sensors and a vitrified bonded corundum grinding wheel were used. Facilitating principal component analysis (PCA) and autoencoder, a methodology was developed to correctly detect the grinding wheel wear from the sensor data.