High Performance Repetitive Control of an Active Filter under Varying Network Frequency ⋆ Ramon Costa-Castell´ o, Shane Malo, Robert Gri˜ n´ o Institut d’Organitzaci´ o i Control de Sistemes Industrials (IOC), Universitat Polit` ecnica de Catalunya (UPC), Barcelona, Spain, (e-mail: {ramon.costa, shane.malo, roberto.grino}@upc.edu). Abstract: Shunt active power filters are power electronics devices that are connected in parallel with nonlinear and reactive loads to compensate these characteristics in order to assure the quality of service of the electrical distribution network. This work proposes and designs a controller, based on combined feedforward and feedback actions, the last using repetitive control, to obtain a good closed-loop performance (power factor close to 1 and, load current harmonics and reactive power compensation) in spite of the possible frequency variations that may occur in the electrical network. It is known that these variations clearly affect the performance of the usual discrete-time implementations of the repetitive based controllers. This work analyzes the effect of these variations and describes the architecture of the controller, its design, and the mechanism to compensate the network frequency variations. Some experimental results that show the good performance of the closed-loop system are also included. Keywords: Active power filters, current harmonics compensation, reactive power compensation, repetitive control, digital control implementation 1. INTRODUCTION Active filters are devices which allow to coexist nonlinear loads and good energy quality in distribution networks. A main effort in the design and control of these devices has been carried out in the past years. One research line deals with topologies and architectures, see for example (Akagi [1996], El-Habrouk et al. [2000], Salo and Tuusa [2005]): several types of topologies have been proposed including parallel (shunt active filters), serial and hybrid serial-parallel connections, mixed passive-active devices and converter based active filters with voltage or current dc bus. Another important research line is the control of active filters, where many approaches have been proposed (Wu and Jou [1996], Choi [2005], Buso et al. [1998], Mattavelli [2001], Singh et al. [1998]). Most of them are based on two hierarchical control loops, an inner one in charge of assuring the desired current and an outer one in charge of determining the required shape as well as the appropriate power balance. The current control loop needs to be fast and precise in order to assure the desired energy flow quality. In this sense, an approach which has proved to be specially efficient is repetitive control. This control technique is based on the Internal Model Principle (Francis and Wonham [1976]) which allows the design of a controller capable of rejecting or tracking periodic signals in steady state (Costa-Castell´o et al. [2004], Costa- Castello et al. [2005]). However, repetitive controllers are ⋆ This work was supported in part by the Spanish Ministerio de Educaci´on y Ciencia under project DPI2007-62582. designed for a predefined frequency 1 and, unfortunately, when this frequency slightly changes the tracking/rejecting capabilities decay dramatically. In order to overcome this problem several approaches, that can be grouped in two main areas, have been proposed: to preserve the sampling time (Steinbuch [2002], Cao and Ledwich [2002]) or to change it adaptively (Liu and Yang [2004], Manayathara et al. [1996], Hillerstr¨om and Sternby [1994]). For the first approach there are also two main ideas: improving robustness by using large memory ele- ments (Steinbuch [2002]) or introducing a fictitious sam- pler operating at a variable sampling rate and later using a fixed frequency internal model (Cao and Ledwich [2002]). These two ideas improve the performance of the system for small frequency variations but increase the computational burden. An alternative approach is to adapt the controller sampling rate according to the disturbance/reference pe- riod (Hillerstr¨om and Sternby [1994], Manayathara et al. [1996], Liu and Yang [2004]). This allows to preserve the steady-state performance while maintaining a low compu- tational cost but, on the other hand, it implies structural changes in the system behavior which may destabilize the closed-loop system. The controller designed in this work uses the traditional two control loops decomposition. The current controller is composed by a feedforward action in charge of assuring very fast transient response and a feedback control law in charge of assuring closed-loop stability and a very good harmonic correction performance. The feedback control 1 Once the sampling period is fixed this frequency is structurally embedded in the control algorithm. Proceedings of the 17th World Congress The International Federation of Automatic Control Seoul, Korea, July 6-11, 2008 978-1-1234-7890-2/08/$20.00 © 2008 IFAC 3344 10.3182/20080706-5-KR-1001.2179