Considering the human operator cognitive process for the interpretation of diagnostic outcomes related to component failures and cyber security attacks

In this work, we consider diagnostics of cyber attacks in Cyber-Physical Systems (CPSs), based on data analytics. For the first time to authors knowledge, the performance of such diagnosis is quantified considering the possible failure of the human operator cognitive process in interpreting and understanding the diagnosis support tool outcomes. A Non-Parametric CUmulative SUM (NP-CUSUM) approach is used for data-driven diagnostic, and the cognitive process of the human operator who interprets its outputs is modeled by a Bayesian Belief Network (BBN). The overall framework is applied on the digital controller of the Advanced Lead-cooled Fast Reactor European Demonstrator (ALFRED).