Towards adaptive electrochemical machining via signal-based data-driven modelling

Electrochemical machining remains largely restricted to medium- and large-batch production and to companies with substantial prior experience despite its significant potential, driven by advantages such as contact-free machining and virtually wear-free operation. This limitation arises from the strongly experience-based nature of present process setups. Reliable process monitoring is a prerequisite for the implementation of adaptive and closed-loop control. This paper demonstrates how process dynamics can be monitored using data-driven models trained on historical machining signals, enabling the inference of prevailing process conditions directly from measured data and forming a basis for future closed-loop process control.