Data-Driven Formulation and Monitoring of Process Stability in Precise Electrochemical Machining (PECM)

Precise electrochemical machining (PECM) continues to be implemented in an open-loop approach. The online monitoring of process stability is a promising endeavor. It is well known that the active avoidance of instabilities (e.g. the occurrence of short circuits) and parameter optimization rely on such closed-loop systems. As a result of the open-loop approach, conservative technologies with considerable safety margins are used, which solely provide stable processing. Productivity is often of secondary concern in this approach. The main reason for the current approach is the complexity (multiple physical factors involved) and the high dynamics of the process (transient working gap and pulsed electric signals). However, the availability of process signals with sufficiently high sampling rates can be used to mitigate the complexity of the process and implement online monitoring tools in a first step and closed-loop control in a later step. This work presents a data-driven model based on machine learning methods to evaluate the process stability of non-stationary PECM processes. By using offline process signals of various trials from different cathode geometries and machining parameters, a definition of process stability if formulated. For this, unsupervised machine learning methods are employed. In a next step, supervised machine learning models are trained and tested. The models are validated using additional trials.