Data-driven technology development in wire EDM (Data-WEDM)

• Applicant: Professor Dr.-Ing. Thomas Bergs
• Subject Area: Metal-Cutting and Abrasive Manufacturing Engineering
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 555289522

Wire electrical discharge machining is a flexible manufacturing process for machining high-temperature materials. To use the process economically, the machine settings must be ideally matched to the respective machining conditions. For this reason, extensive technology studies are conducted for new materials, modified component heights or for machining special geometric fea-tures. In these studies, the machine settings are iteratively adjusted and the effects on manufactur-ing performance are investigated. Particularly in the case of technology development for the trim-cut, where the primary focus is on improving the quality of the workpieces, elaborate analyses of the workpiece surfaces are required after the process. Only based on these investigations can deci-sions be made regarding the adaptation of the machine settings for process optimization. To reduce the effort and time required for this process adaptation and to systematize it, the technology devel-opment of the trim-cut process is to be realized by data-driven models in this project. For this pur-pose, the electrical signals are extracted from the continuous process and the manufacturing quality is determined using artificial intelligence methods. For this purpose, a measurement system is de-veloped that analyzes the high-frequency and low-energy electrical signals of the manufacturing process and characterizes them using process parameters and patterns. To build the models, several process conditions are artificially generated in experiments. Process data is recorded using the measurement system and the machined workpiece surfaces are examined to quantify the influences on the quality of the components. The information collected is used to train and test the data-based models, enabling evaluation of the process based on the electrical signals. The direct feedback on process performance will be used in conjunction with further algorithms for technological optimiza-tion of the first trim-cut. To this end, evolutionary algorithms will be applied and these, in combina-tion with the evaluation models, will generate improved virtual process parameters for the trim-cut. The machine parameters will then be adjusted to approximate the virtual process parameters and thus optimize the trim-cut. Overall, this research project enables systematic and data-based tech-nology development for trim-cut. In particular, knowledge gains from the interrelationships and effects of the physical process variables on the machining quality can be expected.

• DFG Programme: Research Grants