Tool wear analysis and SoH estimation in CNC machine milling processes: an embedded real-time model-based approach.

Since the advent of Industry 4.0, many companies feel the need to rely on industrial renewal to be more competitive on the market in terms of increasing productivity and reducing time and costs. In the context of predictive maintenance, the estimation of the State of Health (SoH) of CNC machines end-effector tools plays a crucial role and turns out to be an interesting challenge for the scientific community and a real need for cutting-edge companies. Some possible solutions have been proposed in literature to address this topic but most of them rely on machine learning and cloud computing algorithms which are not efficiently suitable to a real-time edge device implementation. Therefore, some limitations relating to the state of the art have been highlighted and to tackle this issue, in this thesis, a low computational power approach has been developed in order to make it compatible with low capabilities edge devices. After having analyzed the dynamical model which represents the plant, an innovative approach is presented, which is responsible for estimating in real time the friction coefficient relative to the interface between the cutting tool and the workpiece through the use of a Batch Least Square estimator. Subsequently, a relationship between tool wear and friction coefficient increment is presented, through which it was possible to extrapolate the State of Health (SoH) of the end-effector tool. Finally, through a model-based approach, an executable C-code was generated to lay solid foundations for a future real implementation on embedded platforms. The research activities presented in this thesis were carried out within the framework of the MorePRO project which aims to the realization of innovative technologies for industrial automation in the field of high precision machining, that sees the company Brain Technologies Srl as one of the main research partners.