Tool wear estimation in CNC machine milling processes: an embedded real-time approach based on power balance

During the last decades, the implementation of digital technologies in industrial plants has led to significant improvements in the efficiency and cost reduction of many processes. Predictive maintenance is one of the emerging fields of this revolution and it consists in repairing or replacing components only when needed by monitoring machinery through different techniques. In many industries, such as aerospace and automotive, milling processes often require working expensive materials and complex parts, thus early detection of tool wear is a key factor to save additional costs. In this scenario, MOREPRO project aims at developing a digital prototype able to monitor in real-time the end-effector tool condition in high precision CNC machine milling processes. Since tool wear estimation is a highly debated topic, the first step of this thesis work was to accurately review the articles in literature, evaluate the main available technologies and compare them. Although many approaches rely on machine learning techniques showing good results, they are only evaluated over a limited range of operating conditions, limiting the scalability of the system. At second, an innovative prediction algorithm based on power balance of the CNC machine was developed according to model-based design approach since the target platform is an embedded hardware. The main idea behind the algorithm is to combine spindle current data acquired from the CNC machine with virtual data coming from a digital twin and obtain a wear condition index in real-time. A parameter identification strategy was included to accomplish different working conditions. Finally, the code was simulated employing data collected from a real case study to demonstrate the feasibility of the approach and C-code was generated from the Simulink Model to lay the foundations for the future implementation.