Advances in Electrical and Computer Engineering Volume 12, Number 1, 2012 Cyber Physical Systems: A New Approach to Power Electronics Simulation, Control and Testing Nikola L. ČELANOVIĆ 1 , Ivan L. ČELANOVIĆ 2 , Zoran R. IVANOVIĆ 1 1 Faculty of Technical Sciences, University of Novi Sad, Trg Dositeja Obradovića 6, 21000 Novi Sad, Serbia 2 Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, USA zorani@uns.ac.rs 1 Abstract—This paper presents a Cyber Physical Systems approach to power electronics simulation, control and testing. We present a new framework based on generalized hybrid automaton and application specific ultra-low latency high- speed processor architecture that enables high fidelity real- time power electronics model computation. To illustrate the performance of this approach we experimentally demonstrate two extremely computationally demanding power electronics applications: real-time emulation for Hardware-in-the-Loop (HIL) testing, and hybrid system observers for fault detection and isolation. Index Terms—power electronics, real-time systems, hybrid intelligent systems, computational modeling, observers. I. INTRODUCTION Cyber physical systems (CPS) represent tight integration and coordination of computation and physical processes [1]. Potential applications of CPS range from process control, advanced automotive control systems, traffic control, and robotics all the way to power electronics and smart grid, to name a few. The intimate coupling between computational layer and physical layers has a potential to bring significant benefits in terms of overall system performance, functionality, reliability, and fault-tolerance. Arguably, the application of CPS to energy conversion and in particular to control and coordination of smart grids and power electronics, as one of smart grid’s key physical layers, is one of the key CPS applications, as it is poised to bring innumerable benefits to our society in terms of energy efficiency, sustainability, and overall environmental impact[2], [3]. CPS is not a completely new concept. Embedded systems are considered as its predecessor since in embedded systems digital processors monitor and control physical processes, most often in a feedback loop configuration, in such a way that the computation affects physical layer and vice versa. While embedded systems have been widely successful, CPS present the next revolutionary step in the development towards high-performance, highly networked, ultra reliable embedded systems. In order to achieve the full potential of CPS approach fundamental rethinking of real-time modeling, computation, networking, simulation, testing, and control is needed [1]. The key motivation for revisiting and re-inventing some of the prevailing computational paradigms stem from the fact that physical processes are intrinsically time driven and concurrent while most of the computational and software platforms were developed and optimized for data centric systems and hence lack proper abstractions and specific implementations to deal with hard real-time requirements. Indeed, this limitation manifests trough the lack of time predictability both in terms of computation and communication capabilities. This work was supported by the Ministry of Education and Science of Republic of Serbia (project no. III 42004) In this paper we focus on the application of CPS to power electronics modeling, simulation, control, and testing. We adopt the hybrid system modeling approach to power electronics which naturally lends itself to efficient and time predictable computation. Furthermore, we present a new processor architecture that is tailored for ultra-fast real-time computation - featuring time predictable execution. To demonstrate the new capabilities of proposed CPS approach, based on hybrid system modeling and new digital processor architecture, we examine two critical high performance power electronics applications: real-time emulation for Hardware-in-the-Loop (HIL) [4], and hybrid system observers for fault detection and isolation [5]. The paper is organized in five sections. Section II gives a brief overview of the hybrid system modeling framework and how it pertains to power electronics. Section III presents an application specific processor architecture that is tailored for real-time emulation of hybrid systems with ultra-low latency and high-speed. Two illustrative examples, together with experimental results, are given in Section IV. Finally, Section V summarizes the paper and outlines potential new applications for the ultra-low latency processor. II. HYBRID DYNAMICAL SYSTEM MODELING Power electronics (PE) systems are inherently non-linear, switched circuits where the control of power flow is achieved with precisely timed switching events [6]. Generic block diagram of PE converter is shown in Fig. 1.a). The combination of continuous time dynamics (continuous-time state-space) and discrete events (finite automaton) that PE exhibits lends itself naturally to hybrid system modeling approach. This motivated us to adopt the modeling framework based on Generalized Hybrid Automaton (GHA) with the piecewise linear continuous dynamics [7]. 33 1582-7445 © 2012 AECE Digital Object Identifier 10.4316/AECE.2012.01006 [Downloaded from www.aece.ro on Monday, April 02, 2012 at 09:20:53 (UTC) by 188.2.59.193. Redistribution subject to AECE license or copyright. Online distribution is expressly prohibited.]