Root Cause Analysis for Global Anomalous Events in Self-Organizing Industrial Systems


In self-organizing industrial systems (SOIS) workflows are not defined by engineers in advance, but the system decides by itself at runtime how to route workpieces through the factory, so that the desired output is manufactured as optimal as possible in the present circumstances. As a consequence, the number of possible workflows is not limited to those which were manually predefined, but limited to all possible routes in the factory (state space explosion). Accordingly, analysing anomalies in such a huge solution space becomes more challenging. In this paper, we present a root cause analysis (RCA) approach for finding the root cause of global anomalous events which handles this state space explosion in SOIS. To do so, the dependencies between path usage and external factors like available machines and demanded tasks are subdivided into several sub-dependencies. In addition, we propose for one of these sub-dependencies a heuristical description which avoids the enormous computational effort for modeling the dependency exactly. The operating principle of our RCA method is evaluated based on simulation data of an example factory.

21st IEEE International Conference on Intelligent Engineering Systems (IEEE INES 2017)