Towards Pattern Modeling for Asynchronicity in Production Data

Modern Production and Next Generation Manufacturing Systems rely heavily on data from production and production environments. This leads to critical dependency on the quality of said data, where lacks in quality result in limited performance, reduced resilience, and applicability of data-driven models. In complex production setups, multiple sources of data must be aggregated and synchronized accurately to enable correct assignment of sensors to the same location and time during production. The Time Synchronization Problem in manufacturing describes the problem of asynchronous data streams based on the technical limitations of technical clocks. In this paper, we present modeling approaches to the asynchronicity of production data streams in short- and long-term data acquisition. Based on experiments with production machines, we propose a set of typical asynchronicity patterns, which can be used to model the asynchronicity in offline synchronization methods to improve quality of the production data quality for manufacturing systems and models.