Digitalization of a milling machine: development of a condition monitoring application and human-machine interface

The goal of this thesis project was to digitize a milling machine. To achieve this, a data collection system was created using current sensors connected to a PLC external to the machine's CNC control to measure the phase currents of the machine's motors. During a period of observation, in which the machine exhibited only healthy behavior, data on the motor currents were collected by running a diagnostic cycle between the normal production cycles. This data was then processed to develop a condition monitoring PLC application for the motors. The application allows the user to determine how similar the signals recorded are to those recorded during the collection period and in this way determine if any anomalies are present. To determine the presence of anomalies, the signals are processed using three different approaches: first they are compared to a nominal signal in the time domain, then through a classifier obtained by machine learning techniques, their harmonic content is analyzed; finally, an autoencoder is used and the reconstruction error is evaluated. Combining the results of these approaches, the final judgment on the executed diagnostic cycle is then produced. In addition, a mobile human-machine interface has been created to allow the machine operator to visualize the produced diagnostic data, to update the production data and to monitor the machine operation through a mobile device.