Clustering approach for improving model-based analyses of milling operations

Data-based modeling techniques can be used to predict various characteristics of milling operations. In production engineering, however, the constantly increasing demands on the manufactured components also increase the required complexity of the corresponding milling processes. Consequently, these processes can involve complicated cause-and-effect relationships, rendering an adequate representation in a comprehensive global model difficult. This contribution presents an approach for a data-based analysis to group complex milling operations into elementary process segments. By subsequently modeling characteristics, such as process dynamics, for each identified process segment, the overall prediction accuracy can be increased compared to modeling attempts considering the entire process progression.