Power network reliability computations using multi-agent simulation

Abstract

The reliability of a power network as a whole is limited by the reliability of the weakest element in the network. Ensuring the reliability just with nothing more than a readiness to repair or replace broken parts is not feasible economically, and planned element replacements in the middle of their working life is often a waste of working resources. This article presents a multi-agent simulation system designed for modelling all aspects of power network elements and power network processes. The modelling of this system is able to predict the weak points of a simulated power network over an extended time span (decades). This mode of prediction takes into account the network topology, which means that each element is modelled in a context that is close to the real one. To show the system capabilities, we present two models of computations based on real power networks of 10 kV. Both models make use of the main feature of the system, which is the ability to use local interactions even for global analytic computations, such as the computation of a power network steady state or locating the least reliable point in the network. © 2012 Springer-Verlag.