RECURSIVE LEAST SQUARES HARMONIC IDENTIFICATION IN ACTIVE POWER FILTERS. A. El Zawawi, K. H. Youssef, and O. A. Sebakhy

Similar documents
FUNDAMENTALS OF POWER SYSTEMS

Question 1 Equivalent Circuits

ISSN: [Basnet* et al., 6(3): March, 2017] Impact Factor: 4.116

EE/ME/AE324: Dynamical Systems. Chapter 8: Transfer Function Analysis

ECE Linear Circuit Analysis II

Section Induction motor drives

POWER QUALITY AND RELIABILITY SUPPLY IMPROVEMENT USING A POWER CONDITIONING SYSTEM WITH ENERGY STORAGE CAPABILITY

ECE 3510 Root Locus Design Examples. PI To eliminate steady-state error (for constant inputs) & perfect rejection of constant disturbances

MAE140 Linear Circuits Fall 2012 Final, December 13th

Lecture 10 Filtering: Applied Concepts

Digital Control System

SERIES COMPENSATION: VOLTAGE COMPENSATION USING DVR (Lectures 41-48)

R. W. Erickson. Department of Electrical, Computer, and Energy Engineering University of Colorado, Boulder

A Simple Approach to Synthesizing Naïve Quantized Control for Reference Tracking

Improving Power System Transient Stability with Static Synchronous Series Compensator

Introduction to Laplace Transform Techniques in Circuit Analysis

What lies between Δx E, which represents the steam valve, and ΔP M, which is the mechanical power into the synchronous machine?

Design By Emulation (Indirect Method)

SIMON FRASER UNIVERSITY School of Engineering Science ENSC 320 Electric Circuits II. Solutions to Assignment 3 February 2005.

Gain and Phase Margins Based Delay Dependent Stability Analysis of Two- Area LFC System with Communication Delays

376 CHAPTER 6. THE FREQUENCY-RESPONSE DESIGN METHOD. D(s) = we get the compensated system with :

Determination of the local contrast of interference fringe patterns using continuous wavelet transform

SIMON FRASER UNIVERSITY School of Engineering Science ENSC 320 Electric Circuits II. R 4 := 100 kohm

Comparison of Hardware Tests with SIMULINK Models of UW Microgrid

Chapter 2 Sampling and Quantization. In order to investigate sampling and quantization, the difference between analog

into a discrete time function. Recall that the table of Laplace/z-transforms is constructed by (i) selecting to get

Improvement of Transient Stability of Power System by Thyristor Controlled Phase Shifter Transformer

Tuning of High-Power Antenna Resonances by Appropriately Reactive Sources

Lecture 4. Chapter 11 Nise. Controller Design via Frequency Response. G. Hovland 2004

Digital Control System

CHAPTER 4 DESIGN OF STATE FEEDBACK CONTROLLERS AND STATE OBSERVERS USING REDUCED ORDER MODEL

The Influence of the Load Condition upon the Radial Distribution of Electromagnetic Vibration and Noise in a Three-Phase Squirrel-Cage Induction Motor

Massachusetts Institute of Technology Dynamics and Control II

A Novel Direct Torque Control Scheme for Induction Machines With Space Vector Modulation

R. W. Erickson. Department of Electrical, Computer, and Energy Engineering University of Colorado, Boulder

Finding the location of switched capacitor banks in distribution systems based on wavelet transform

Induction Motor Drive

Sampling and the Discrete Fourier Transform

Dynamic Simulation of a Three-Phase Induction Motor Using Matlab Simulink

Delhi Noida Bhopal Hyderabad Jaipur Lucknow Indore Pune Bhubaneswar Kolkata Patna Web: Ph:

BASIC INDUCTION MOTOR CONCEPTS

Basic parts of an AC motor : rotor, stator, The stator and the rotor are electrical

Robust Decentralized Design of H -based Frequency Stabilizer of SMES

ABSTRACT- In this paper, a Shunt active power filter (SAPF) is developed without considering any harmonic detection

POWER SYSTEM SMALL SIGNAL STABILITY ANALYSIS BASED ON TEST SIGNAL

ME 375 FINAL EXAM Wednesday, May 6, 2009

Control Systems Engineering ( Chapter 7. Steady-State Errors ) Prof. Kwang-Chun Ho Tel: Fax:

Bogoliubov Transformation in Classical Mechanics

DIRECT TORQUE CONTROL OF THREE PHASE INDUCTION MOTOR USING FUZZY LOGIC SPEED CONTROLLER FOR STEADY/DYNAMIC STATE RESPONSE

A Simplified Methodology for the Synthesis of Adaptive Flight Control Systems

ME 375 FINAL EXAM SOLUTIONS Friday December 17, 2004

Direct Torque Control using Matrix Converters

HOMEWORK ASSIGNMENT #2

Design of Digital Filters

SMALL-SIGNAL STABILITY ASSESSMENT OF THE EUROPEAN POWER SYSTEM BASED ON ADVANCED NEURAL NETWORK METHOD

Module 4: Time Response of discrete time systems Lecture Note 1

Direct Torque Control of Saturated Induction Machine with and without speed sensor

MATHEMATICAL MODELING OF INDUCTION MOTORS

ECEN620: Network Theory Broadband Circuit Design Fall 2018

15 Problem 1. 3 a Draw the equivalent circuit diagram of the synchronous machine. 2 b What is the expected synchronous speed of the machine?

An estimation approach for autotuning of event-based PI control systems

Direct Torque Control for Induction Motor Using Fuzzy Logic

Control Systems Analysis and Design by the Root-Locus Method

A Parallel Power Conditioning System with Energy Storage Capability for Power Quality Improvement in Industrial Plants

No-load And Blocked Rotor Test On An Induction Machine

Chapter 2: Problem Solutions

Chapter 13. Root Locus Introduction

S_LOOP: SINGLE-LOOP FEEDBACK CONTROL SYSTEM ANALYSIS

III.9. THE HYSTERESIS CYCLE OF FERROELECTRIC SUBSTANCES

Experimental Direct Torque Control Induction Motor Drive with Modified Flux Estimation and Speed control Algorithm.

The Measurement of DC Voltage Signal Using the UTI

Advanced Digital Signal Processing. Stationary/nonstationary signals. Time-Frequency Analysis... Some nonstationary signals. Time-Frequency Analysis

EE 508 Lecture 16. Filter Transformations. Lowpass to Bandpass Lowpass to Highpass Lowpass to Band-reject

Root Locus Contents. Root locus, sketching algorithm. Root locus, examples. Root locus, proofs. Root locus, control examples

Performance Improvement of Direct Torque Controlled Interior Permanent Magnet Synchronous Motor Drive by Considering Magnetic Saturation

MODERN CONTROL SYSTEMS

THE EXPERIMENTAL PERFORMANCE OF A NONLINEAR DYNAMIC VIBRATION ABSORBER

Properties of Z-transform Transform 1 Linearity a

Simulation and Analysis of Linear Permanent Magnet Vernier Motors for Direct Drive Systems

Quantifying And Specifying The Dynamic Response Of Flowmeters

EE Control Systems LECTURE 14

Lecture 8. PID control. Industrial process control ( today) PID control. Insights about PID actions

Lecture 12 - Non-isolated DC-DC Buck Converter

Source slideplayer.com/fundamentals of Analytical Chemistry, F.J. Holler, S.R.Crouch. Chapter 6: Random Errors in Chemical Analysis

March 18, 2014 Academic Year 2013/14

Mathematical modeling of control systems. Laith Batarseh. Mathematical modeling of control systems

Reference:W:\Lib\MathCAD\Default\defaults.mcd

Homework 12 Solution - AME30315, Spring 2013

A Method for Assessing Customer Harmonic Emission Level Based on the Iterative Algorithm for Least Square Estimation *

ECE-202 FINAL December 13, 2016 CIRCLE YOUR DIVISION

Chapter 7. Root Locus Analysis

Reliability Analysis of Embedded System with Different Modes of Failure Emphasizing Reboot Delay

Evolutionary Algorithms Based Fixed Order Robust Controller Design and Robustness Performance Analysis

LOAD FREQUENCY CONTROL OF MULTI AREA INTERCONNECTED SYSTEM WITH TCPS AND DIVERSE SOURCES OF POWER GENERATION

Root Locus Diagram. Root loci: The portion of root locus when k assume positive values: that is 0

CONTROL OF INTEGRATING PROCESS WITH DEAD TIME USING AUTO-TUNING APPROACH

Department of Mechanical Engineering Massachusetts Institute of Technology Modeling, Dynamics and Control III Spring 2002

Sensorless speed control including zero speed of non salient PM synchronous drives

ECE 325 Electric Energy System Components 6- Three-Phase Induction Motors. Instructor: Kai Sun Fall 2015

A PLC BASED MIMO PID CONTROLLER FOR MULTIVARIABLE INDUSTRIAL PROCESSES

Transcription:

RECURSIVE LEAST SQUARES HARMONIC IDENTIFICATION IN ACTIVE POWER FILTERS A. El Zawawi, K. H. Youef, and O. A. Sebakhy Department of Electrical Engineering, Alexandria Univerity, Alexandria 21544, Egypt.P.O. BOX 21544. Tel: 203 5925557 Fax: 203 5921853. Email: omarebakhy @ hotmail.com and karim_hy24 @ hotmail.com) Abtract: A new leat quare Harmonic identification technique i developed. Thi technique i ued baically to etimate the fundamental component of the drawn by nonlinear load in order to calculate the reference of the active power filter on line. The effectivene of the propoed technique i teted under a udden change of the nonlinear load. The performance of the propoed technique i compared with the conventional harmonic extraction technique baed on the ynchronou reference frame method. Simulation reult are given for both the propoed and the conventional technique. Keyword: Harmonic compenation, Parameter etimation, Sytem identification, Active power filter. 1. INTRODUCTION Due to the advancement in power electronic technology and due to the intenive ue of power electronic equipment in all branche of indutry, the harmonic pollution ha increaed everely in electrical network. Thi affect the operation of motor, control equipment and protection device. Harmonic alo caue reonance and they increae neutral and ytem loe (Singh, et al., 1999). Conventionally paive (L-C) filter were ued to olve thi problem but they are characterized by fixed compenation, large ize and complex deign. They can alo caue reonance in the power ytem and their performance i affected by aging of the paive component or change of power frequency (Moran, et al., 1995). A a reult, the active power filter ha become the bet way for harmonic and reactive power compenation. A three phae hunt active power filter i hown in Fig. 1. where: ia, ib and ic are the three phae upply, ila, ilb and ilc are the three phae load, ica, icb and icc are the three phae filter, va, vb and vc are the three phae voltage and 1 : 6 are the inverter witche. va vb vc L L L ia ib ic ica L L L ila ilb ilc icb icc 1 3 5 2 4 6 Fig. 1- Three phae hunt active power filter The load can be expreed in the form: il( t) = ao a1 in ωt a3 in 3 ωt LL am in Mωt b1 co ωt b3 co3 ωt LL bm co Mωt (1) where M i the order of the maximum harmonic of interet. The role of the active power filter i to upply the harmonic and reactive power component of the load o that the ource upplie only the fundamental component in-phae with the upply voltage ( a1 in ωt ). Many technique have been developed for eparating the R L C

fundamental and harmonic component of the load and for calculating reference for active power filter, for example: the p-q theory, the ynchronou reference frame theory (Soare, et al., 2000) and the correlation technique (Harmelen and Enlin, 1993; Jung, et al., 2000). The reference can be etimated uing capacitor voltage regulation technique (Jain, et al., 2002; Chatterjee, et al., 1999;Torrey and Al-Zamel, 1995). There i alo the fat Fourier tranform (FFT) technique which perform the harmonic extraction in the frequency domain. It i found that the capacitor voltage regulation and the p-q technique are not efficient under non-inuoidal upply voltage condition. On the other hand, the Fat Fourier Tranform and the correlation technique method are dependent on the ampling frequency and require extenive calculation. They alo require toring of ample during one period at any time. The ynchronou reference frame method require le calculation burden and it i le dependent on upply voltage harmonic but it performance i affected by the filter ued to extract the reference. Recurive parameter identification technique can be ued to etimate the fundamental and harmonic component of the load in order to etimate the reference of active power filter. Thi paper applie the leat quare identification technique to generate the reference for an active power filter. Simulation reult for the propoed technique are given and compared with the reult of the ynchronou reference frame technique. 2. THE SYNCHRONOUS REFERENCE FRAME THEORY In thi method the nonlinear load il a,il b and il c are tranformed into the ynchronou reference frame baed on Park tranformation a follow: ila ilα 2 1 1/ 2 1/ 2 = (2). ilb ilβ 3 0 3 / 2 3 / 2 ilc va vα 2 1 1/ 2 1/ 2 = (3). vb vβ 3 0 3 / 2 3 / 2 vc il d co θ in θ ilα vβ θ = il =., tan 1 (4) q in θ co θ ilβ vα Intantaneou active and reactive load can be decompoed into average and ocillatory term: ~ ~ il d = Il d i l d and il q = Il q i l q.under balanced and inuoidal upply voltage condition, the firt harmonic of poitive equence i tranformed into a dc quantity and thi contitute the average component (Soare, et al.,2000). All higher order harmonic including the firt harmonic of negative equence are tranformed to non-dc quantitie and o contitute the ocillatory component. ila ilb ilc Vdc* Vdc θ LPF abc dq ild ilq LPF LPF PI 0 icd * dq icq* Fig. 2- Reference generation uing ynchronou frame method Extracting the average uing a low pa filter (LPF) and ubtracting it from the active and reactive component the ocillatory component can be obtained and the required reference compenating are then calculated a follow: ~ ic α 1 vα vβ i ld (5) = ~. icβ 2 2 β α α v v v v i lq β ila 2 1 1/ 2 1/ 2 T ilα (6) ilb =. 3 0 3 / 2 3 / 2 ilβ ilc A Butterworth type filter i choen in order to obtain magnitude and phae characteritic a cloe a poible to an ideal filter ince it magnitude repone i maximally flat in the pa-band (Soare, et al., 2000). The filter conidered i a fourthorder filter with a cut-off frequency 25 Hz. It hould be noted that the average component, being dc quantitie, do not uffer from phae hift and thi i the reaon for not extracting the ocillatory component directly uing high pa filter. The block diagram of harmonic extraction and reference generation i hown in Fig. 2. A proportional-integral (PI) controller i ued for the dc voltage regulation of the inverter and to control the active power flow for compenating the inverter loe. 3.THE RECURSIVE HARMONIC IDENTIFICATION TECHNIQUE. θ abc ica * icb* icc * Equation (1) can be written in the tandard form: yk il ( tk) = u k a (7) Where t k i the ampling time, u k = [ 1,in ωtk, K,in Mωtk, co ωtk, K, co Mωtk] i the regreor vector, and a = [ ao, a1, K, am, b1, K, bm ] i the unknown parameter vector to be etimated. In the teady-tate cae, the off-line etimate of the parameter a that minimize the weighted um of the quare of the error, namely J = E WE (8) where E = Y Ua, Y = [ y 1, K y N],

W = diag { λ N i, i = 1, K, N}, 0 < λ <1i a forgetting factor to peed up the etimation (uually λ 0.95-0.99) and U = [ u1, L, u N ] i given by: a ˆ U 1 = ( WU ) U WY (9) In the literature, there are many variant of the recurive leat quare algorithm (RLS) to etimate the unknown parameter vector a (Goodwin and Sin, 1984). We hall preent here the mot baic algorithm, correponding to the above mentioned off-line etimate. The recurive algorithm take the form: aˆ k = aˆ k 1 Pk u k ( y k u k aˆ k 1) (10) P = k 1u k uk Pk 1 1 P P (11) k k 1 λ uk Pk 1u k λ where P k i the covariance matrix. It can be hown that if the initial condition are: aˆ(0) = 0 P(0) = ρ I where I i the identity matrix and ρ i a very large contant, then at teady tate, the on-line etimate of aˆ k (10) tend to the off line etimation â a k ( Södertröm and Stoica, 1989). Thi technique can be ued for the on-line etimation of the fundamental component which i then ubtracted from the load to obtain the reference of the active power filter a hown in Fig. 3. A proportional-integral (PI) controller i ued for the dc voltage regulation of the inverter and to control the active power flow in the inverter. The etimated fundamental component i then ubtracted from the load to obtain the reference for the active power filter. Load Sampling harmonic Recurive uing etimation DSP Fundamenta component l Vdc * PI Vdc V/Vmax Reference filter Fig.3 Reference generation uing recurive identification 4. DESIGN OF THE PI CONTROLLER A PI controller i ued to regulate the capacitor voltage by adding an active component I cp to the reference. The deign of the PI controller i done by linearizing the DC voltage control ytem around the teady tate operating point (Soare, et al., 2000; Chatterjee, et al., 1999). Uing the average power balance principle, the open loop tranfer function model of the compenator about a particular operating pointv dco, i obtained a: V V G dc ( ) 3 1 ( ) = = = * I cp ( ) CV dco where; 3V A = CV dco A (12) V i the r.m. phae voltage V dco i the required teady tate capacitor voltage and C i the capacitance at the DC ide. Fig. 4 how the cloed loop DC voltage control ytem including the PI controller with a proportional and integral gain of k P and k I repectively and a firt-order low-pa filter F() with a gain of k I / k P and a pole at = k I / k P to eliminate the influence of the zero introduced by the PI controller (Soare, et al.,2000). V dc* k F() I k P G() V dc Fig.4- The cloed loop control ytem The overall cloed loop tranfer function i given by: V dc ( ) Ak = I (13) V * dc ( ) 2 Ak P Ak I Equation (13) repreent a econd order ytem and the deign of PI controller gain i baed on the choen damping ratio ζ and undamped natural frequencyω n. Thi dc voltage controller will give a zero teady tate error due to the introduction of a pole at = 0 by the PI controller, making a type 1 ytem. When the DC voltage control i done in the d-q ynchronou reference frame, the obtained value for k p and k I hould be multiplied by 3 due to tranformation. 5. THE CURRENT CONTROLLER The obtained reference for each * phaei c and it actual i c are compared uing a two level hyterei controller with hyterei band ± 0.25 A a hown in Fig. 5. The lower witch of each leg i ued to increae the (if the actual i le than the reference ) and the upper witch i ued to decreae the (if the actual i greater than the reference ). Reference Actual Error 1 0 h 0 h Fig. 5 - Hyterei control for one inverter leg Upper witch Lower witch

Fig. 6 Simulation reult of the conventional technique. Fig. 7 Simulation reult of the propoed technique

. Fig. 8. The etimated fundamental component.. Fig. 9. Spectrum of upply uing both technique. 6. SIMULATION RESULTS The ytem parameter are choen a follow: Supply rm phae voltage: 50 V at 50 Hz. Filter inductance and capacitance: L = 2.2 mh., C = 2000µF. Dc voltage operating point = 175 V. The PI controller i deigned to have ζ = 2 / 2 and ω n = 314 rad/ec uing a linearization method around the teady tate operating point and thi give: k P = 1. 796, k I = 398. 88 for the cae of the conventional controller and k P = 1. 036 k I = 223.29 for the new technique. The nonlinear load i a three phae bridge rectifier with a contant DC of 20 A which i uddenly decreaed to 10 A at t = 0.15 ec. Fig. 6 how the performance of the filter uing the conventional technique and Fig. 7, 8 how the performance of the filter uing the propoed technique. Fig.9 how the upply pectrum uing both technique. From Fig. 6, 7 and 9 it i obviou that the performance of the propoed technique i better than that of the conventional technique. The harmonic content of the upply uing the propoed technique i better than the one obtained uing the conventional technique. The reaon i that in the ynchronou reference frame, harmonic frequencie are reduced due to tranformation and they may pa through the low pa filter which i practically not ideal. Fig. 8 how the etimated fundamental component. The etimation proce ha reached teady tate in nearly half cycle time and it ha tracked the change in load effectively within the ame time period. Although both technique ue the ame DC voltage regulator, but the DC voltage profile in the conventional technique i better. Thi i due to the delay aociated with the etimation proce becaue of the relatively low ampling frequency (4 khz) and o, the DC voltage repone of the propoed technique will be much better at higher ampling frequencie. 7.CONCLUSION A new leat quare harmonic identification technique ha been propoed and it wa applied to generate the reference of a three phae active power filter. Simulation reult of the propoed technique have been given and they are found better than thoe of the ynchronou

reference frame technique. The on line recurive parameter etimation technique ue only two conecutive ample at any intant for the harmonic extraction proce and it i not affected by the harmonic in the upply voltage. REFERENCES Chatterjee, K., B. G. Fernande and G. K. Dubey, (1999) "An intantaneou reactive volt-ampere compenator and harmonic uppreor ytem," IEEE Tranaction on Power Electronic, Vol. 14, No.2, pp.381-392. Goodwin, G. C., and K.S. Sin, (1984) "Adaptive filtering prediction and control," Printice-Hall, Inc., New Jery, U.S.A. Harmelen, G. L. and J. H. R. Enlin, (1993), "Realtime dynamic control of dynamic power filter in upplie with high contamination," IEEE Tranaction on Power Electronic, Vol. 8, No.3,, pp.301-308. Jain, S. K., P. Agrawal and H. O. Gupta, (2002), "Fuzzy logic controlled hunt active power filter for power quality improvement," IEE Proc.-Electr. Power Appl., Vol. 149, No.5, pp.317-328. Jung, Y. G., Y. C. Lim and S. H. Yang, (2000), "Single-phae active power filter baed on three-dimenional coordinate," IEE Proc.-Electr. Power Appl., Vol. 147, No.6, pp.572-578. [Moran, L. A., J. W. Dixon and R. R. Wallace, (1995), "A three phae active power filter operating with fixed witching frequency for reactive power and harmonic compenation," IEEE Tranaction on Indutrial Electronic, Vol. 42, No.4, pp.402-408. Singh, B., K. Al-Haddad and A. Chandra, (1999), "A review of active filter for power quality improvement," IEEE Tranaction on Indutrial Electronic, Vol. 46, No.5, pp.960-967. Soare, V., P. Verdelho and G. D. Marque, (2000), "An intantaneou active and reactive component method for active filter," IEEE Tranaction on Power Electronic, Vol. 15, No.4, pp.660-669. Södertröm, T., and P. G. Stocia, (1989), "Sytem identification," Prentice-Hall,. Torrey, D. A., and A. M. Al-Zamel, (1995), "Single-phae active power filter for multiple nonlinear load," IEEE Tranaction on Power Electronic, Vol. 10, No.3, pp. 263-272..