Fusion of Data and Expert Knowledge for Fault Tree Reliability Analysis of Cyber-Physical Systems
Parisa Niloofar, Sanja Lazarova‐Molnar · 2021
Reliability analysis of cyber-physical systems have benefitted substantially from the introduction of a range of technology enablers. Internet of things (IoT), advanced computing architectures and digital platforms are among the new technologies that are enhancing the data collection and analytics perceptions in the era of Industry 4.0. Fault tree modelling and failure analysis of systems have been traditionally performed using exhaustively expert knowledge. However, nowadays cyber-physical systems are equipped with sensors and meters, enabling reliability analysis to become more automated and less human-dependent. There have been approaches that fully depend on data that utilized these new developments. However, completely ignoring human cognitive capabilities and expert knowledge causes a great loss of information, which might only be compensated by collecting large amounts of data that is costly in many aspects, and sometimes even impossible. In this paper we discuss how and to what extend expert knowledge can be fused or combined with data to learn fault trees of cyber-physical systems. We, furthermore, point out the gap in availability of systematic methods for fusing data with expert knowledge for the purpose of reliability analysis of cyber-physical systems. Results of an initial simulation study indicate that hybrid reliability analysis of a system increases the accuracy and is less tedious.