IFAC PapersOnLine 51-2 (2018) 260–265 ScienceDirect Available online at www.sciencedirect.com 2405-8963 © 2018, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. Peer review under responsibility of International Federation of Automatic Control. 10.1016/j.ifacol.2018.03.045 © 2018, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. Keywords: Colored Petri nets, electrical networks, fault detection, static model, interval-valued uncertainty 1. INTRODUCTION Electrical networks are thoroughly investigated and prac- tically important objects of model-based diagnosis. Both of their dynamic and steady-state models are used for di- agnostic purposes of different kind. Their classical models are based on linear and/or nonlinear ordinary differential equations in the dynamic, and on linear and/or nonlinear algebraic equations in the static case (see e.g. in Bird (2010)). In practical cases, however, one needs to consider the effect of uncertainties and to cope with the situation when relatively rare measurements are available with possible measurement errors, and the available measurements are not uniformly placed over the network. The uncertainties can be handled either by using stochastic models or by considering interval-valued variables in the model. Colored Petri nets (see Jensen (1996) for an informal introduction) offer a good modeling opportunity in interval-valued un- certainty description case. Different variants of Petri nets can be used for modeling electrical networks. In Ning (2002) a multi layer Petri net is developed to investigate the bidding strategy of the power network. Physical components of the network are modeled by basic Petri nets while colored Petri nets are used to model the participants of the energy market. In Dey et al. (2011) generalized stochastic Petri net model of a smart grid is developed to analyze its behavior. Electrical network diagnosis usually concentrates on the failures reported by protective relays and circuit breakers. Fuzzy Petri nets (Sun et al. (2004)) and basic Petri nets (Lo et al. (1997) , Bi et al. (2005), Calderaro et al. (2011)) can be used to detect faults in the protection system and the distribution network. Other approaches use Petri nets together with coding theory (Ren and Mi (2006)) to diagnose and identify faults in power networks. The aim of our paper is to propose a simple and effective colored Petri net model for a special type of electrical networks that describe local transformer areas, in order to detect and localize illegal loads that may be present. 2. MODELING PROBLEM STATEMENT The modeling problem statement, i.e. the system descrip- tion together with the modeling assumptions of electrical network models for diagnosing non-technical losses are described here. The modeling approach and methodology described in Hangos and Cameron (2001) was adopted here to the case of electrical networks. 2.1 Electrical networks and their operation The conventional electrical grids consist of supplies (power plants including the ones with renewable energies), trans- formers and transmission lines. They transfer electrical energy generated by the supplies to the loads that are connected to the transmission lines. In classical grids the flow of energy comes from the big power plants to the customers. Nowadays a lot of small domestic power plants which are powered by Renewable Energy Sources (RES) are connected to the grid. These power plants realize the Distributed Generation (DG) of the electrical energy (DG/RES). The topology, the diameter of the electrical grid is known, it is public information that can be seen in the utility maps. If the position of the customers and its loading profile are known, the currents and the loads of the network elements are computable. Abstract: A simple and effective colored Petri net (CPN) model is proposed in this paper for a special type of electrical networks that describe local transformer areas, in order to detect and localize illegal loads that may be present. The model is decomposed to characteristic structural elements called feeders with one or two sources the computation of which can be performed independently and in parallel. The CPN model allows to handle the interval-type uncertainties in the model in a transparent and effective way. * Department of Electrical Engineering and Information Systems, University of Pannonia, Veszpr´ em, H-8201 Hungary (e-mail: {pozna.anna, fodor.attila, gerzson.miklos}@ virt.uni-pannon.hu) ** Institute for Computer Science and Control, Hungarian Academy of Sciences, Budapest, H-1518 Hungary (email: hangos@scl.sztaki.hu) A. I. P´ ozna *,** A. Fodor * M. Gerzson * K. M. Hangos **,* Colored Petri net model of electrical networks for diagnostic purposes