1 Abstract— This paper presents a Model Predictive Control (MPC) design for the Unified Power Quality Conditioner (UPQC), an integration of series and shunt active filters to improve power quality in a power distribution system. The control strategy aims to regulate the load voltage and the source current to the desired references in spite of the existence of harmonic components in the supply voltage and the load current, possible sag or swell in the supply voltage, and non- unity power factor of the supply side. Kalman filters are used to extract the fundamental as well as the harmonic components abovementioned, which will then be used to formulate the desired references and regarded as measurable disturbances, respectively. Based on a state-space model developed for the UPQC, an MPC controller is designed. Simulation studies on a single-phase power distribution system are also presented to verify the performance. Index Terms— UPQC, MPC, Kalman filters, power quality. I. INTRODUCTION he po increasing use of power electronics degrades the wer quality in power distribution system. Nonlinear loads such as the adjustable speed drives or the rectifier introduce periodic distortions to the supply current. Voltage distortion may occur at the Point of Common Coupling (PCC) as a result of this distorted current. This causes malfunctioning of devices, overheating of power factor correction capacitors, motors, transformers and cables, and thermal tripping of protective devices installed there. In addition, the operation of manufacturing plants like petrochemical plants and semiconductor plants is highly dependant on a stable supply voltage. The supply voltage has to be free from any voltage sag or swell, while concurrently, be maintained at a normally fixed frequency. These power quality problems have become a great concern for both utilities and customers [1], [2]. Therefore, many researches were carried out on compensating devices to improve power quality in powersystems, while at the same time, maintain a stable supply voltage. The use of Dynamic Voltage Restorer (DVR) This research work is supported by the School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore. All the authors are with the School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore. K. H. Kwan and P. L. So are also with the Centre for Advanced Power Electronics, School of EEE, Nanyang Technological University. *Corresponding author’s postal address: K. H. Kwan, School of Electrical and Electronic Engineering, Power Electronics and Drives Lab., Nanyang Technological University, Block S2, Nanyang Avenue, Singapore 639798 (email: kwan0019@ntu.edu.sg). Model Predictive Control of Unified Power Quality Conditioner for Power Quality Improvement *K. H. Kwan, Y. S. Png, Y. C. Chu, Senior Member, IEEE, and P. L. So, Senior Member, IEEE [3], [4], Uninterruptible Power Supply (UPS) [5], and Unified Power Quality Conditioner (UPQC) [6]-[8] are some of the ways to overcome the above power quality problems. The UPQC, which is an integration of series and shunt active filters, has the capability of improving power quality at the point of installation in power or industrial distribution systems. The series active filter is employed to compensate for any distortions in the supply voltage, while the shunt active filter is used to compensate for current harmonics in the load current, as well as to correct the power factor of the supply side near to unity. Over the past few years, major research works have been carried out on control circuit designs for UPQCs. Modern control theories are considered and adopted to design control circuits for power quality devices [9]-[11]. This paper introduces the application of Model Predictive Control with Kalman filter to the UPQC to improve the power quality in a power distribution system. The Kalman filter also functions like a harmonics extraction circuit. Model Predictive Control (MPC) is an approach to controller design that involves online optimization. The online optimization takes into account the system dynamics, constraints and control objectives. The key motivations of using MPC for UPQC are:- 1) It can fully utilize the state-space model developed for the UPQC presented in Section II below. Furthermore, this state-space model is linear, time-invariant with a low order, which is particularly suitable for the online optimization required by the MPC. 2) MPC is able to handle hard constraints of the process variables explicitly [12], [13]. In UPQC, the manipulated variable is implemented with a Pulse Width Modulation (PWM) technique, which has to be limited to 1 ± , so that the active filters will operate in the linear modulation region. Conventional linear controllers need to avoid such limits, otherwise instability might occur. This results in a more conservative design, thus limiting the compensation capability of the UPQC. In this paper, various functional modules of the UPQC and the MPC controller are discussed in subsequent sections. Simulation results for different operating conditions in a 2kVA single phase power distribution system are then presented to verify the performance of the control design. II. MODELLING OF THE PLANT The equivalent single-phase representation of the UPQC topology considered [14, pp. 380-406] is shown in Fig. 1. T 16th IEEE International Conference on Control Applications Part of IEEE Multi-conference on Systems and Control Singapore, 1-3 October 2007 TuC04.2 1-4244-0443-6/07/$20.00 ©2007 IEEE. 916