Cutting tools with ceramic inserts are often used in the process of machining many types of super alloys, mainly due to their high strength and thermal resistance. Nevertheless, during the cutting process, the plastic flow wear generated in these inserts enhances and propagates cracks due to high temperature and high mechanical stress. This leads to a very variable failure of the cutting tool. Furthermore, in high-speed rough machining of nickel-based super alloys, such as Inconel 718 and Waspalloy, it is recommended to avoid the use of any type of coolant. This in turn, enables the clear visualization of cutting sparks, which in these machining tasks are quite distinctive.
The present doctoral thesis attempts to set the basis of a potential Tool Condition Monitoring (TCM) system that could use vison-based sensing to calculate the amount of tool wear. This TCM system would work around the research hypothesis that states that a relationship exists between the continuous wear that ceramic SiAlON (solid solutions based on the Si3N4 structure) inserts experience during a high-speed machining process, and the evolution of sparks created during the same process. A successful TCM system such as this could be implemented at an industrial level to aid in providing a live status of the cutting tool’s condition, potentially improving the effectiveness of these machining tasks, whilst preventing tool failure and workpiece damage.
During this research, sparks were analyzed through various visual methods in three main experiments. Four studies were developed using the mentioned experiments to support and create a final predictive approach to the TCM system. These studies are described in each thesis chapter and they include a wear assessment of SiAlON ceramics, an analysis of the optimal image acquisition systems and parameters appropriate for this research, a study of the research hypothesis, and finally, an approach to tool wear prediction using Neural Networks (NN). To carry out some of these studies, an overall methodology was structured to perform experiments and to process spark evolution data, as image processing algorithms were built to extract spark area and intensity. Towards the end of this thesis, these spark features were used, along with measured values of tool wear, namely notch, flank and crater wear, to build a Neural Network for tool wear prediction.
