Hoog-Efficiënt Microvonken met een Niet-Invasieve Meettechniek: In-Proces Monitoring en Regeling

As a non-contact material removal technique based on thermoelectric coupling mechanism, micro electrical discharge machining (micro-EDM) is capable of processing any electrically conductive material, including those that are difficult-to-machine, and has thus found widespread application in various industrial sectors. However, the extremely narrow inter-electrode gap and complex gap conditions lead to highly stochastic and dynamic discharge behaviours. During machining, multiple categories of discharge phenomena can occur, affecting material removal and machining performance. To ensure process stability and efficiency, a holistic overview of processing history is required in micro-EDM operation. The mainstream methods for process monitoring in EDM nowadays rely on invasive electrical sensing, i.e., inferring the actual gap state through voltage and current signals, which hinders the process development towards low-cost and interference-free digitalization. To address the aforementioned challenges, this research starts from fundamental research to explore the discharge-state-dependent material removal characteristics in micro-EDM and subsequently proposes a non-invasive in-process monitoring and control scheme based on radio frequency (RF) radiation. First, a novel continuous pulse discharge experiment is designed to study how different types of sparks influence material removal. By linking individual sparks to generated craters, the research explains the material removal characteristics from the perspectives of pulse energy, energy density, and the fraction of discharge energy entering the workpiece. Next, the study explores RF radiation mechanism and potential influencing factors in micro-EDM. It is concluded that RF sources in EDM include high-frequency radiation induced by dielectric breakdown and low-frequency radiation generated by the internal circuit of the power supply. Variations in RF behaviour are evaluated from the perspectives of material properties, process conditions, and monitoring equipment. Finally, RF signals are used to monitor and control the machining process. Machine learning techniques are applied to identify discharge states with high accuracy. By further analysing the machining states at the pulse-train level, an adaptive control strategy is implemented through machining state definition, prediction, and feedback control, and is validated in deep-hole drilling scenario. The research findings will contribute to the intelligent and digital transformation of EDM and provide an accessible and practical solution for small and medium-sized enterprises.