Probabilistic Relational Models for Reliability Centered Asset Management of Active Distribution Management Systems

This paper presents an approach for reliability centred asset management of active distribution management systems. The application of Probabilistic Relational Models (PRMs) together with a set of defined components and their attribute allows both the ability to capture system configurations in architectural models and the ability of probabilistic reasoning offered by Bayesian networks. The approach is based on two key concepts; first, it addresses both the reliability of primary system components and the supporting secondary, ICT-based systems. Secondly, it enables representation of architecture of the ICT systems, including for instance redundancy of hardware and allocation of software functions to several hardware devices. The increasing number of software dependent systems for controlling and supervising the power grid enhances the risk of software-caused failures. Thus, for reliable operation it is of high importance to not only concern primary component, but also the software and hardware of the secondary systems controlling it. The approach is demonstrated using an example composing both primary equipment and secondary systems.