The distinctive feature of new generation information systems is not only their complexity in terms of number of elements, number of connections and hierarchy levels, but also their constantly changing structure and behavior. In this situation the problem of receiving actual information about the observed complex Cyber–Physical Systems (CPS) current status becomes a rather difficult task. This information is needed by stakeholders for solving tasks concerning keeping the system operational, improving its efficiency, ensuring security, etc. Known approaches to solving the problem of the complex distributed CPS actual status definition are not enough effective. The authors propose a model based approach to solving the task of monitoring the status of complex CPS. There are a number of known model based approaches to complex distributed CPS monitoring, but their main difference in comparison with the suggested one is that known approaches by the most part use static models which are to be build manually by experts. It takes a lot of human efforts and often results in errors. Our idea is that automata models of structure and behavior of the observed system are used and both of these models are built and kept in actual state in automatic mode on the basis of log file information. The proposed approach is based, on one hand, on the results of the authors researches in the field of automatic synthesis of multi-level automata models of observed systems and, on the other hand, on well known algorithms of process mining. In the paper typical monitoring tasks are described and generalized algorithms for solving them using the proposed system of models are presented. An example of real life systems based on the suggested approach is given. The approach can be recommended to use for building CPS of medium and high complexity, characterized by high structural dynamics and cognitive behavior.
Enabling the human in the loop: Linked data and knowledge in industrial cyber-physical systems
