Signal-based Context Comparative Analysis for Identification of Similar Manufacturing Modules

Abstract Industry 4.0 connects different machines and their modules to each other. Integrating already existing non-modular machines and establishing the required modularization in such a scenario requires a lot of time-consuming analysis. But Industry 4.0 also allows previously unconnected machines to establish a comparative analysis between each other by comparing monitored results of new and old machines. This analysis allows finding behavior that overlaps between machines and allows to identify parts that can be encapsulated in new or substituted by already known modules. In this paper, we propose a method to identify those overlapping parts by exploiting learned behavior models during runtime and combining production paths with sequence search algorithms by using context information of observable event signals.