Shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions

Introduction. The widespread adoption of power-electronic loads has made harmonic distortion a critical power-quality issue. shunt active power filters (SAPFs) remain the most versatile solution. Problem. The conventional harmonic compensation algorithms suffer from degraded filtering performance un...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Datum:2026
Hauptverfasser: Belhadj Mostefa, M. M., Boussaid, A., Khezzar, A.
Format: Artikel
Sprache:Englisch
Veröffentlicht: National Technical University "Kharkiv Polytechnic Institute" and Аnatolii Pidhornyi Institute of Power Machines and Systems of NAS of Ukraine 2026
Schlagworte:
Online Zugang:https://eie.khpi.edu.ua/article/view/353265
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Назва журналу:Electrical Engineering & Electromechanics
Завантажити файл: Pdf

Institution

Electrical Engineering & Electromechanics
_version_ 1869562799232909312
author Belhadj Mostefa, M. M.
Boussaid, A.
Khezzar, A.
author_facet Belhadj Mostefa, M. M.
Boussaid, A.
Khezzar, A.
author_institution_txt_mv [ { "author": "M. M. Belhadj Mostefa", "institution": "University Freres Mentouri Constantine 1" }, { "author": "A. Boussaid", "institution": "University Freres Mentouri Constantine 1" }, { "author": "A. Khezzar", "institution": "University Freres Mentouri Constantine 1" } ]
author_sort Belhadj Mostefa, M. M.
baseUrl_str http://eie.khpi.edu.ua/oai
collection OJS
datestamp_date 2026-07-01T21:42:56Z
description Introduction. The widespread adoption of power-electronic loads has made harmonic distortion a critical power-quality issue. shunt active power filters (SAPFs) remain the most versatile solution. Problem. The conventional harmonic compensation algorithms suffer from degraded filtering performance under non-ideal grid condition, while the conventional low-pass filters (LPFs) in instantaneous reactive power theory (p-q theory) create an unavoidable trade-off between transient speed and harmonic rejection. Goal. To develop an adaptive and efficient harmonic current compensation algorithm that can generate reference currents with rapid convergence and high accuracy under non-ideal grid conditions. Methodology. The proposed method combines a multivariable filter phase-locked loop (MVF-PLL) for precise extraction of instantaneous components normalized to unit amplitude (i.e., sinq, cosq) with a variable leaky least mean squares (VLLMS) adaptive filter for DC component extraction. The algorithm was tested in MATLAB/Simulink across five scenarios, including balanced and unbalanced voltages, variable loads, and voltage distortions. Experimental validation was conducted on a field-programmable gate array (FPGA) using real-time co-simulation and hardware implementation. Results. MATLAB/Simulink simulations and real-time FPGA implementation on a low-cost Spartan-6 board show that the proposed method reduces the 2–98 % rise time of the extracted DC active power from ≈ 12 ms (4th-order Butterworth LPF) to 0.6– 0.8 ms (93–95 % improvement) while maintaining source current total harmonic distortion below 4.92 % in the worst case fully compliant with IEEE 519. The extremely low computational cost makes the solution ideal for industrial controllers. Scientific novelty. This paper proposes a novel control strategy that replaces the traditional LPF with a single-coefficient VLLMS adaptive filter while ensuring robust positive-sequence synchronisation via an MVF-PLL. Practical value. The algorithm improves SAPF performance, reduces response time, and ensures stable operation across diverse grid scenarios, offering a reliable solution for industrial applications. References 24, tables 2, figures 15.
doi_str_mv 10.20998/2074-272X.2026.4.06
first_indexed 2026-07-02T01:00:24Z
format Article
fulltext Industrial Electronics 40 Electrical Engineering & Electromechanics, 2026, no. 4 © M.M. Belhadj Mostefa, A. Boussaid, A. Khezzar UDC 621.314 https://doi.org/10.20998/2074-272X.2026.4.06 M.M. Belhadj Mostefa, A. Boussaid, A. Khezzar Shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions Introduction. The widespread adoption of power-electronic loads has made harmonic distortion a critical power-quality issue. shunt active power filters (SAPFs) remain the most versatile solution. Problem. The conventional harmonic compensation algorithms suffer from degraded filtering performance under non-ideal grid condition, while the conventional low-pass filters (LPFs) in instantaneous reactive power theory (p-q theory) create an unavoidable trade-off between transient speed and harmonic rejection. Goal. To develop an adaptive and efficient harmonic current compensation algorithm that can generate reference currents with rapid convergence and high accuracy under non-ideal grid conditions. Methodology. The proposed method combines a multivariable filter phase-locked loop (MVF-PLL) for precise extraction of instantaneous components normalized to unit amplitude (i.e., sin, cos) with a variable leaky least mean squares (VLLMS) adaptive filter for DC component extraction. The algorithm was tested in MATLAB/Simulink across five scenarios, including balanced and unbalanced voltages, variable loads, and voltage distortions. Experimental validation was conducted on a field-programmable gate array (FPGA) using real-time co-simulation and hardware implementation. Results. MATLAB/Simulink simulations and real-time FPGA implementation on a low-cost Spartan-6 board show that the proposed method reduces the 2–98 % rise time of the extracted DC active power from ≈ 12 ms (4th-order Butterworth LPF) to 0.6– 0.8 ms (93–95 % improvement) while maintaining source current total harmonic distortion below 4.92 % in the worst case fully compliant with IEEE 519. The extremely low computational cost makes the solution ideal for industrial controllers. Scientific novelty. This paper proposes a novel control strategy that replaces the traditional LPF with a single-coefficient VLLMS adaptive filter while ensuring robust positive-sequence synchronisation via an MVF-PLL. Practical value. The algorithm improves SAPF performance, reduces response time, and ensures stable operation across diverse grid scenarios, offering a reliable solution for industrial applications. References 24, tables 2, figures 15. Key words: shunt active power filter, harmonic compensation, adaptive filtering, non-ideal grid, power quality. Вступ. Широке впровадження силових електронних навантажень зумовило перетворення гармонічних спотворень на одну з ключових проблем якості електроенергії. Шунтуючі активні силові фільтри (SAPFs) залишаються найбільш універсальним засобом її розв’язання. Проблема. Традиційні алгоритми компенсації гармонік характеризуються погіршенням ефективності фільтрації за неідеальних умов електричної мережі, тоді як класичні фільтри низьких частот (LPF), що застосовуються в теорії миттєвої реактивної потужності (p-q теорії), створюють компроміс між швидкодією перехідного процесу та ступенем пригнічення гармонік. Мета. Розроблення адаптивного та ефективного алгоритму компенсації гармонічних струмів, здатного формувати опорні струми з високою точністю та швидкою збіжністю за неідеальних умов мережі. Методика. Запропонований метод поєднує контур фазового автопідлаштування частоти на основі багатозмінного фільтра (MVF-PLL) для точного виділення миттєвих компонент, нормованих до одиничної амплітуди (sinθ, cosθ), з адаптивним фільтром Variable Leaky Least Mean Squares (VLLMS) для виділення постійної складової. Алгоритм досліджено в MATLAB/Simulink у п’яти режимах роботи, зокрема за симетричних та несиметричних напруг, змінного навантаження та спотворень напруги. Експериментальну перевірку виконано на програмованій вентильній матриці (FPGA) із використанням співмоделювання в реальному часі та апаратної реалізації. Результати. Моделювання в MATLAB/Simulink та реалізація в реальному часі на недорогій платі Spartan-6 FPGA показали, що запропонований метод зменшує час наростання (2–98 %) виділеної постійної складової активної потужності приблизно з 12 мс (LPF Баттерворта 4-го порядку) до 0,6–0,8 мс (покращення на 93–95 %) при одночасному забезпеченні коефіцієнта гармонічних спотворень струму мережі нижче 4,92 % у найгіршому випадку, що повністю відповідає вимогам стандарту IEEE 519. Надзвичайно низька обчислювальна складність робить запропоноване рішення придатним для промислових контролерів. Наукова новизна. У роботі запропоновано нову стратегію керування, у якій традиційний LPF замінено однокоефіцієнтним адаптивним фільтром VLLMS із забезпеченням робастної синхронізації за додатною послідовністю за допомогою MVF-PLL. Практична значимість. Запропонований алгоритм підвищує ефективність роботи SAPF, зменшує час реакції та забезпечує стабільне функціонування в різних режимах електричної мережі, що робить його надійним рішенням для промислових застосувань. Бібл. 24, табл. 2, рис. 15. Ключові слова: шунтуючий активний фільтр потужності, компенсація гармонік, адаптивна фільтрація, неідеальна електромережа, якість електроенергії. Introduction. The ever-increasing penetration of power-electronic-based loads has made harmonic distortion one of the most critical power-quality issues in modern distribution networks [1–3]. Shunt active power filters (SAPFs) remain the most widely adopted and versatile solution for compensating harmonic currents generated by nonlinear loads [4–6]. Conventional SAPF control strategies based on instantaneous reactive power p-q theory [7–9] or synchronous reference frame methods [10–12] require accurate extraction of the DC component of the instantaneous active power )( p . This task is traditionally performed using fixed low-pass filters (LPFs), typically 4th-order Butterworth filters with 20–40 Hz cutoff frequency. However, the inherent bandwidth-speed trade-off of such filters results in slow transient response (10–20 ms settling time), which is unacceptable for fast-varying industrial loads [10]. Advanced phase-locked loop (PLL) techniques, such as the decoupled double synchronous reference frame PLL [13, 14], double second order generalized integrator PLL [15–17], and especially the multivariable filter PLL (MVF-PLL) [18, 19] have successfully solved the synchronization problem under unbalanced and distorted grid conditions. Similarly, several adaptive filtering algorithms, including recursive least squares, Kalman filters and variable-step least mean square have been proposed for undesired component extraction and signal component estimation, including the extraction of the DC component of the instantaneous active power, but at the cost of significantly higher computational complexity and difficult parameter tuning [20–24]. The variable leaky least mean square (VLLMS) algorithm offers an outstanding compromise, it guarantees stability, provides extremely fast convergence [20–23], and requires negligible computational effort when configured with a single coefficient to extract only a DC value [24]. Although VLLMS has already shown excellent Electrical Engineering & Electromechanics, 2026, no. 4 41 performance under ideal balanced grid conditions [22], its behavior under simultaneous voltage unbalance, harmonic distortion, and sudden load changes particularly when combined with robust positive-sequence synchronization has not yet been investigated in the literature. This paper fills this gap by proposing a novel SAPF control architecture that integrates a MVF-PLL for accurate extraction of positive-sequence components even under severe grid disturbances [18, 19], and a single-coefficient VLLMS (VLLMS with x(k) = 1) that completely replaces the conventional LPF for ultra-fast and stable extraction of the DC component of active power [24]. The goal of the paper is to develop an adaptive and efficient harmonic current compensation algorithm that can generate reference currents with rapid convergence and high accuracy under non-ideal grid conditions. The main contributions of this work are:  The first reported integration of MVF-PLL and single-coefficient VLLMS in a p-q theory-based SAPF, fully eliminating the classic LPF bandwidth-speed trade- off under non-ideal grid conditions.  Rigorous performance evaluation under extreme scenarios: balanced/unbalanced/distorted grids and sudden load variations.  Comprehensive quantitative comparison with conventional Butterworth LPF, VLLMS algorithms in terms of transient response, total harmonic distortion (THD).  Fixed-point real-time implementation on a low-cost Xilinx Spartan-6 field-programmable gate array (FPGA) with complete resource utilization report. System description and conventional p-q theory control. Figure 1 illustrates the test system. A balanced 3-phase 400 V (line-to-line), 50 Hz supply feeds a 6-pulse diode rectifier, initially operating with a load resistor of RL = 40 . At t = 0.1 s, an additional resistor is connected in parallel to produce a sudden load increase. The SAPF is based on a conventional 2-level voltage-source inverter rated at 2.5 kVA, interfaced with the point of common coupling through inductors Lf = 2 mH. The DC-link voltage is regulated at 250 V using a 1100 µF capacitor, controlled by PI regulator. The inverter operates with a switching frequency of 10 kHz. The classical p-q theory control transforms grid voltages and load currents into the stationary  reference frame using the power-invariant Clarke transformation:                               c b a v v v v v 2 3 2 3 0 2 1 2 1 1 3 2   ; (1)                               c b a i i i i i 2 3 2 3 0 2 1 2 1 1 3 2   , (2) where v, v are the grid voltages; i, i are the load currents in  coordinates . Instantaneous active p and reactive q powers are:      . ;   ivivq ivivp (3) The DC component is extracted using a 4th-order Butterworth LPF with 20 Hz cutoff. Reference compensating currents are then computed as:                          q p vv vv vvi i c c ~ ~1 22     , (4) where p~ , q~ are the AC components of the instantaneous active and reactive powers, respectively. The reference compensating currents in the abc frame can be determined by applying the inverse Clark transformation:                                        c c cc cb ca i i i i i 2321 2321 01 32 . (5) Fig. 1. SAPF structure Despite its widespread use, this approach suffers from 2 fundamental limitations:  Slow transient response: the fixed LPF bandwidth yields a 2–98 % rise time of the extracted DC component of the instantaneous active power p is longer than 12 ms when using the conventional 4th-order Butterworth LPF.  Poor robustness under non-ideal grids: raw voltages are used without proper synchronization, transferring negative-sequence components and harmonics directly into the current references. Proposed MVF-PLL-VLLMS control scheme. The proposed control strategy (Fig. 2) eliminates both limitations by integrating 2 complementary blocks:  An MVF-PLL that delivers clean positive-sequence components normalized to unit amplitude. These components are defined as: u = sin, u = cos, where  is the instantaneous phase of the fundamental grid voltage. These signals represent the normalized fundamental positive-sequence components of the grid voltages in the αβ-frame, from which negative-sequence components, harmonics and voltage unbalance have been effectively removed, even under severe grid disturbances.  A single-coefficient VLLMS adaptive filter (x(k) = 1) that replaces the conventional LPF for ultra-fast extraction of the DC active power. Fig. 2. Block diagram of the proposed control scheme combining MVF-PLL and single-coefficient VLLMS 42 Electrical Engineering & Electromechanics, 2026, no. 4 Positive-sequence synchronization using MVF- PLL. The MVF-PLL extracts the positive-sequence fundamental component in the  frame using the continuous-time state equations [18, 19]:         ,ˆˆ ;ˆˆ     xxxkx xxxkx c c   (6) where  Txx  ˆˆˆ x is the MVF-PLL state vector, corresponding to the filtered output signal of x; x = [x x] T is the input electrical signal (current or voltage) in the  reference frame; k is the positive constant; c = 250 rad/s is the nominal grid angular frequency; x is the time derivative of the state vector x̂ . After Tustin discretization at 20 kHz, the normalized components are:                    , ˆ ; ˆ ˆ 22 22       xx x u xx x u    (7) where u, u are the normalized positive‑sequence unit components, derived from the MVF‑PLL. These components are perfectly synchronized with the fundamental positive‑sequence component of the grid voltage, free from negative‑sequence and harmonic distortions. Figure 3 shows the architecture of the MVF- PLL model used in the control algorithm. Fig. 3. Architecture of the MVF-PLL model used in the control algorithm DC active-power extraction using VLLMS. The DC component is extracted by modelling the problem as an adaptive linear combiner with constant input signal x(k) = 1: )()()()( kpkkky  xw , (8) where x(k) is the input signal, which serves as the reference signal in the filtering process; w(k) is the single adaptive weight [24]; y(k) is the output signal. The objective is to adjust w(k) in order to minimize the error: )()()( kpkdke  , (9) where d(k) is the active power p. The single adaptive weight w(k) is updated using the VLLMS algorithm [20, 24]:   )()(2)()()(21)1( kekkkkk   ww , (10) where the time varying step size (k) and leakage factor (k) are automatically adjusted according to [20]: 2)()()()1( kRkkk   ; (11) )1()()(2)()1(  kwkekkk  , (12) where  is the step size adaptation parameter; R(k) is the autocorrelation of the error:   )1()(1)1()(  kekekkR  , (13) where  is the positive adaptation parameter controlling the update of the leakage factor;  is the exponential weighting parameter. To avoid possible divergence when the DC component varies significantly, the step size (k) is limited between min and max (parameters:  = 0.97,  = 0.997,  = 4.210–5, min = 510–5, max = 0.5). The scheme (Fig. 4) shows adaptive linear filtering. This structure requires only 5–6 arithmetic operations per sample and achieves a 2–98 % rise time of 0.6–0.8 ms. Fig. 4. Single-coefficient VLLMS adaptive filter used for DC active-power extraction Let ic * and ic * denote the reference compensating currents in the  frame. They are computed as:                             u u p i i i i c c * * . (14) Subsequently, the reference currents in the abc coordinate system are obtained through the inverse Clarke transformation. Compared with the conventional method, the proposed combination uses the clean components normalized to unit amplitude (u, u) from the MVF-PLL together with an adaptive filter without fixed bandwidth, completely eliminating the classic speed-rejection trade- off while ensuring excellent robustness. Simulation and experiment. All simulations were carried out in MATLAB/Simulink using the complete 3-phase SAPF model. Five representative operating scenarios were considered in order to evaluate the behavior of the proposed MVF-PLL combined with the single- coefficient VLLMS algorithm. In each scenario, results are directly compared with the conventional solution based on a 4th-order Butterworth LPF tuned at 20 Hz. Unless otherwise stated, each figure presents (from top to bottom): a) the 3-phase grid voltages vabc; b) the phase A load current iLa; c) the instantaneous active power p (gray), the extracted DC component using the LPF LPFp (blue) and using the proposed VLLMS VLLMSp (red); d, e) the reference and compensated source currents (blue with LPF, red with VLLMS). These waveforms allow a direct visual comparison of the convergence speed, tracking capability, harmonic compensation performance and robustness under different grid conditions. Scenario 1: balanced sinusoidal grid with constant nonlinear load. Figure 5 shows that the proposed method extracts the DC component of the active Electrical Engineering & Electromechanics, 2026, no. 4 43 power extremely rapidly. The VLLMS reaches its steady value in approximately 0.65 ms (2–98 % rise time), compared to about 12 ms for the LPF. This demonstrates the intrinsic advantage of the adaptive estimator, which reacts almost instantly to the stationary nonlinear load. Scenario 2: balanced grid with sudden load variation. The load power is doubled at t = 0.05 s and restored to its nominal value at t = 0.1 s. The LPF exhibits a clear delay after each load transition, whereas the VLLMS adapts almost instantly and maintains accurate tracking of the new power value as shown in Fig. 6. As a consequence, the source current compensation remains stable and free from transient distortions. Scenario 3: unbalanced grid with constant load. Under voltage unbalance, the instantaneous active power contains oscillations that may degrade the performance of conventional filtering (Fig. 7). The LPF attenuates these oscillations but introduces a significant delay. By contrast, the MVF-PLL ensures clean extraction of the positive-sequence fundamental components, enabling the VLLMS to operate with the same speed and accuracy as in the balanced case. Scenario 4: unbalanced grid with load steps. This scenario confirms the robustness of the proposed system under simultaneous disturbances: voltage unbalance and abrupt variations of the load (Fig. 8). The VLLMS maintains its rapid convergence, ensuring smooth compensation even during large transitions. t, s t, s p , W i L a, A v ab c, A i sa -f il te re d, A i s a- re f, A a b c e d Fig. 5. Scenario 1 – balanced grid with constant nonlinear load t, s t, s p , W i L a, A v ab c, A i sa -f il te re d, A i s a- re f, A a b c e d Fig. 6. Scenario 2 – sudden load steps under balanced grid t, s t, s p , W i L a, A v ab c, A i sa -f il te re d, A i s a- re f, A a b c e d Fig. 7. Scenario 3–10 % voltage unbalance with constant load 44 Electrical Engineering & Electromechanics, 2026, no. 4 t, s t, s p , W i L a, A v ab c, A i sa -f il te re d, A i s a- re f, A a b c e d Fig. 8. Scenario 4 – voltage unbalance with load steps at t = 0.1 s and t = 0.2 s Scenario 5: severe grid distortion. Voltage distortion is generated by inserting series resistors in phases a and c between t = 0.1 s and t = 0.2 s (Fig. 9). This produces approximately 20 % THDv (THD of the voltage). Even under heavy distortion, the MVF-PLL successfully isolates the clean positive-sequence components, enabling the VLLMS to extract the DC component with sub- millisecond response (Fig. 10). The LPF, on the other hand, exhibits attenuation but poor reactivity. Performance summary. The proposed method consistently provides a 13–19 times improvement in convergence speed with respect to the conventional LPF, while maintaining comparable current THD after compensation. Table 1 presents the system performance under different grid voltage conditions. Fig. 9. Circuit used to generate controlled voltage distortion in phases A and C t, s t, s p , W i L a, A v ab c, A i sa -f il te re d, A i s a- re f, A a b c e d Fig. 10. Scenario 5 – severe grid distortion (20 % THDv) FPGA implementation and experimental validation. The algorithm was deployed on a Xilinx Spartan-6 XC6SLX75 FPGA. Hardware co-simulation was first carried out through a Joint Test Action Group (JTAG)-based verification setup. Table 1 Performance comparison under different grid conditions Grid condition Balanced Unbalanced Distorted THD load iLa, % 27.8 25.1 38.66 THD source-LPF, % 1.65 1.72 4.46 THD source-VLLMS, % 1.76 2.45 4.92 Rise time-LPF, ms 12.0 12.0 10.5 Rise time-VLLMS, ms 0.65 0.78 0.62 Real-time validation was then performed on a 2.5 kVA laboratory prototype following the workflow illustrated in Fig. 11. The corresponding experimental schematic configuration is shown in Fig. 12. Fig. 11. Flow diagram of real-time FPGA implementation Experimental results under balanced grid and constant load are given in Fig. 13. The proposed method reproduces the fast convergence observed in simulations, resulting in clean and sinusoidal compensated source currents. Electrical Engineering & Electromechanics, 2026, no. 4 45 Fig. 12. Experimental schematic configuration t, s i s a- fi lt er ed , A i s a- re f, A Fig. 13. Experimental results under balanced grid conditions using the proposed method Additional tests under unbalanced and distorted grids are shown in Fig. 14, 15. In all cases, the measured rise time of remained below 0.8 ms, confirming the robustness and real-time suitability of the method. t, s i s a- fi lt er ed , A i s a- re f, A Fig. 14. Experimental results under unbalanced grid conditions using the proposed method t, s i s a- fi lt er ed , A i s a- re f, A Fig. 15. Experimental results under distorted grid conditions using the proposed method The real-time FPGA resource utilization of the proposed method is summarized in Table 2. The “Available” column gives the total resources of the device, and the “Used” column indicates the number of primitives consumed by the implementation. The low usage of look- up tables (LUTs), registers (flip-flops), and digital signal processing (DSP) resources confirms the computational efficiency and real-time feasibility of the proposed method. Table 2 Post-synthesis resource utilization of the proposed control algorithm on Xilinx Spartan-6 XC6SLX75 FPGA Resource Available Used LUTs 46656 3867 Flip flops 93312 339 DSP 132 51 Conclusions. A novel SAPF control strategy combining MVF-PLL with single-coefficient VLLMS adaptive filtering has been proposed and rigorously validated through detailed simulation and real-time FPGA implementation. The method completely eliminates the inherent bandwidth speed trade-off of conventional LPFs, achieving sub-millisecond transient response (0.6–0.8 ms rise time) while keeping source-current THD well below 5 % even under severe simultaneous voltage unbalance and harmonic distortion fully compliant with IEEE 519 requirements. Thanks to its extremely low computational complexity, the proposed solution is ideally suited for efficient fixed-point implementation on inexpensive industrial FPGAs, making it highly attractive for next- generation power quality improvement systems. Conflict of interest. The authors declare that they have no conflicts of interest. REFERENCES 1. Ishaq M., Langella R. Aggregated Load Modelling Approach to Study Impact of Heat Pumps Harmonic Distortion on the Low Voltage Distribution Network. Journal Européen Des Systèmes Automatisés, 2024, vol. 57, no. 5, pp. 1497-1502. doi: https://doi.org/10.18280/jesa.570525. 2. Sai Thrinath B.V., Prabhu S., Meghya Nayak B. Power quality improvement by using photovoltaic based shunt active harmonic filter with Z-source inverter converter. Electrical Engineering & Electromechanics, 2022, no. 6, pp. 35-41. doi: https://doi.org/10.20998/2074-272X.2022.6.06. 3. Baros J., Sotola V., Bilik P., Martinek R., Jaros R., Danys L., Simonik P. Review of Fundamental Active Current Extraction Techniques for SAPF. Sensors, 2022, vol. 22, no. 20, art. no. 7985. doi: https://doi.org/10.3390/s22207985. 4. Dwinanto B., Setiyono S., Thalib F., Siswono H. Single-phase power shunt active filter design using photovoltaic as reactive power compensator. Electrical Engineering & Electromechanics, 2025, no. 3, pp. 59-64. doi: https://doi.org/10.20998/2074-272X.2025.3.09. 5. Mishra N., Gawande S.P., Deshmukh S., Tan K.T., Ansari K., Khan T.M.Y., Almakayeel N., Khan W.A. A modified instantaneous reactive 46 Electrical Engineering & Electromechanics, 2026, no. 4 power algorithm for shunt compensation. Scientific Reports, 2025, vol. 15, no. 1, art. no. 26666. doi: https://doi.org/10.1038/s41598-025-12315-w. 6. Jampana B., Askani J., Veramalla R. DC Component Extraction of Notch Filter Algorithm for Active Power Filters. Journal Européen Des Systèmes Automatisés, 2022, vol. 55, no. 2, pp. 207-212. doi: https://doi.org/10.18280/jesa.550207. 7. Boopathi R., Indragandhi V. Enhancement of power quality in grid- connected systems using a predictive direct power controlled based PV- interfaced with multilevel inverter shunt active power filter. Scientific Reports, 2025, vol. 15, no. 1, art. no. 7967. doi: https://doi.org/10.1038/s41598-025-92693-3. 8. Zorig A., Babes B., Hamouda N., Mouassa S. Improving the efficiency of a non-ideal grid coupled to a photovoltaic system with a shunt active power filter using a self-tuning filter and a predictive current controller. Electrical Engineering & Electromechanics, 2024, no. 6, pp. 33-43. doi: https://doi.org/10.20998/2074-272X.2024.6.05. 9. Govind A., Jayaswal K., Tayal V.K., Kumar P. Simulation and real time implementation of shunt active power filter for power quality enhancement using adaptive neural network topology. Electric Power Systems Research, 2024, vol. 228, art. no. 110042. doi: https://doi.org/10.1016/j.epsr.2023.110042. 10. Popescu M., Bitoleanu A., Suru C.V., Linca M., Alboteanu L. Shunt Active Power Filters in Three-Phase, Three-Wire Systems: A Topical Review. Energies, 2024, vol. 17, no. 12, art. no. 2867. doi: https://doi.org/10.3390/en17122867. 11. Asadi Y., Eskandari M., Mansouri M., Chaharmahali S., Moradi M.H., Tahriri M.S. Adaptive Neural Network for a Stabilizing Shunt Active Power Filter in Distorted Weak Grids. Applied Sciences, 2022, vol. 12, no. 16, art. no. 8060. doi: https://doi.org/10.3390/app12168060. 12. El Ghaly A., Tarnini M., Chahine K. A robust reference extraction method for shunt active power filters in the presence of noise and source voltage distortion. Measurement, 2025, vol. 242, art. no. 116243. doi: https://doi.org/10.1016/j.measurement.2024.116243. 13. Lu Y., Li B., Teng G., Zhang Z., Xu X. A harmonic current detection algorithm for aviation active power filter based on generalized delayed signal superposition. Scientific Reports, 2025, vol. 15, no. 1, art. no. 10435. doi: https://doi.org/10.1038/s41598-025-94829-x. 14. Rajak M.K., Pudur R. Adaptive hybrid PSO-GD optimized phase- locked loop for robust grid synchronization in renewable energy systems. COMPEL - The International Journal for Computation and Mathematics in Electrical and Electronic Engineering, 2025, vol. 44, no. 4, pp. 566- 596. doi: https://doi.org/10.1108/COMPEL-11-2024-0463. 15. Chemidi A., Benhabib M.C., Bourouis M.A. Performance improvement of shunt active power filter based on indirect control with a new robust phase-locked loop. Electrical Engineering & Electromechanics, 2022, no. 4, pp. 51-56. https://doi.org/10.20998/2074- 272X.2022.4.07. 16. Srilakshmi R., Chayapathy V., Anitha G.S. Digital Phase-Locked Loops with Backward Euler Approximation for Harmonic Reduction in Solar PV Integration. Smart Grids and Sustainable Energy, 2025, vol. 10, no. 2, art. no. 57. doi: https://doi.org/10.1007/s40866-025-00285-x. 17. Nkambule M., Hasan A., Ali A., Shongwe T. A Novel Control Strategy in Grid-Integrated Photovoltaic System for Power Quality Enhancement. Energies, 2022, vol. 15, no. 15, art. no. 5645. doi: https://doi.org/10.3390/en15155645. 18. Boussaid A., Nemmour A.L., Louze L., Khezzar A. A novel strategy for shunt active filter control. Electric Power Systems Research, 2015, vol. 123, pp. 154-163. doi: https://doi.org/10.1016/j.epsr.2015.02.008. 19. Boussaid A., Chelli S.E.I., Nemmour A.L., Khezzar A. An effective control algorithm for dynamic voltage restorer under symmetrical and asymmetrical grid voltage conditions. Electrical Engineering & Electromechanics, 2021, no. 4, pp. 53-63. doi: https://doi.org/10.20998/2074-272X.2021.4.07. 20. Subudhi B., Ray P.K., Ghosh S. Variable leaky least mean-square algorithm-based power system frequency estimation. IET Science, Measurement & Technology, 2012, vol. 6, no. 4, pp. 288-297. doi: https://doi.org/10.1049/iet-smt.2011.0103. 21. Ray P.K., Puhan P.S., Panda G. Real time harmonics estimation of distorted power system signal. International Journal of Electrical Power & Energy Systems, 2016, vol. 75, pp. 91-98. doi: https://doi.org/10.1016/j.ijepes.2015.08.017. 22. Bag A., Subudhi B., Ray P.K. An Adaptive Variable Leaky Least Mean Square Control Scheme for Grid Integration of a PV System. IEEE Transactions on Sustainable Energy, 2020, vol. 11, no. 3, pp. 1508-1515. doi: https://doi.org/10.1109/TSTE.2019.2929551. 23. Ray P., Ray P.K. Design and Control of PV-UPQC Using Variable Leaky LMS Based Algorithm for Power Quality Enhancement. 2020 3rd International Conference on Energy, Power and Environment: Towards Clean Energy Technologies, 2021, pp. 1-6. doi: https://doi.org/10.1109/ICEPE50861.2021.9404514. 24. Zoghbi A., Berkani D. Performance improvement of the shunt active power filter using a novel adaptive filtering approach. Turkish Journal of Electrical Engineering & Computer Sciences, 2021, vol. 29, no. 1, pp. 203-222. doi: https://doi.org/10.3906/elk-2005-193. Received 28.01.2026 Accepted 11.03.2026 Published 02.07.2026 M.M. Belhadj Mostefa1, PhD Student, A. Boussaid 1,2, Doctor of Electrical Engineering, Professor, A. Khezzar1, Doctor of Electrical Engineering, Professor, 1 Laboratoire d’electrotechnique de Constantine (LEC), University Freres Mentouri Constantine 1, Algeria, e-mail: mohammed.belhadj-mostefa@lec-umc.org (Corresponding Author) 2 Institute des Sciences et des Techniques Appliquees, University Freres Mentouri Constantine 1, Algeria. How to cite this article: Belhadj Mostefa M.M., Boussaid A., Khezzar A. Shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions. Electrical Engineering & Electromechanics, 2026, no. 4, pp. 40-46. doi: https://doi.org/10.20998/2074-272X.2026.4.06
id eiekhpieduua-article-353265
institution Electrical Engineering & Electromechanics
keywords_txt_mv keywords
language English
last_indexed 2026-07-02T01:00:24Z
publishDate 2026
publisher National Technical University "Kharkiv Polytechnic Institute" and Аnatolii Pidhornyi Institute of Power Machines and Systems of NAS of Ukraine
record_format ojs
resource_txt_mv eiekhpieduua/89/5600aa6c68a5d008ce249c9f215f0389.pdf
spelling eiekhpieduua-article-3532652026-07-01T21:42:56Z Shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions Shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions Belhadj Mostefa, M. M. Boussaid, A. Khezzar, A. shunt active power filter harmonic compensation adaptive filtering non-ideal grid power quality шунтуючий активний фільтр потужності компенсація гармонік адаптивна фільтрація неідеальна електромережа якість електроенергії Introduction. The widespread adoption of power-electronic loads has made harmonic distortion a critical power-quality issue. shunt active power filters (SAPFs) remain the most versatile solution. Problem. The conventional harmonic compensation algorithms suffer from degraded filtering performance under non-ideal grid condition, while the conventional low-pass filters (LPFs) in instantaneous reactive power theory (p-q theory) create an unavoidable trade-off between transient speed and harmonic rejection. Goal. To develop an adaptive and efficient harmonic current compensation algorithm that can generate reference currents with rapid convergence and high accuracy under non-ideal grid conditions. Methodology. The proposed method combines a multivariable filter phase-locked loop (MVF-PLL) for precise extraction of instantaneous components normalized to unit amplitude (i.e., sinq, cosq) with a variable leaky least mean squares (VLLMS) adaptive filter for DC component extraction. The algorithm was tested in MATLAB/Simulink across five scenarios, including balanced and unbalanced voltages, variable loads, and voltage distortions. Experimental validation was conducted on a field-programmable gate array (FPGA) using real-time co-simulation and hardware implementation. Results. MATLAB/Simulink simulations and real-time FPGA implementation on a low-cost Spartan-6 board show that the proposed method reduces the 2–98 % rise time of the extracted DC active power from ≈ 12 ms (4th-order Butterworth LPF) to 0.6– 0.8 ms (93–95 % improvement) while maintaining source current total harmonic distortion below 4.92 % in the worst case fully compliant with IEEE 519. The extremely low computational cost makes the solution ideal for industrial controllers. Scientific novelty. This paper proposes a novel control strategy that replaces the traditional LPF with a single-coefficient VLLMS adaptive filter while ensuring robust positive-sequence synchronisation via an MVF-PLL. Practical value. The algorithm improves SAPF performance, reduces response time, and ensures stable operation across diverse grid scenarios, offering a reliable solution for industrial applications. References 24, tables 2, figures 15. Вступ. Широке впровадження силових електронних навантажень зумовило перетворення гармонічних спотворень на одну з ключових проблем якості електроенергії. Шунтуючі активні силові фільтри (SAPFs) залишаються найбільш універсальним засобом її розв’язання. Проблема. Традиційні алгоритми компенсації гармонік характеризуються погіршенням ефективності фільтрації за неідеальних умов електричної мережі, тоді як класичні фільтри низьких частот (LPF), що застосовуються в теорії миттєвої реактивної потужності (p-q теорії), створюють компроміс між швидкодією перехідного процесу та ступенем пригнічення гармонік. Мета. Розроблення адаптивного та ефективного алгоритму компенсації гармонічних струмів, здатного формувати опорні струми з високою точністю та швидкою збіжністю за неідеальних умов мережі. Методика. Запропонований метод поєднує контур фазового автопідлаштування частоти на основі багатозмінного фільтра (MVF-PLL) для точного виділення миттєвих компонент, нормованих до одиничної амплітуди (sinθ, cosθ), з адаптивним фільтром Variable Leaky Least Mean Squares (VLLMS) для виділення постійної складової. Алгоритм досліджено в MATLAB/Simulink у п’яти режимах роботи, зокрема за симетричних та несиметричних напруг, змінного навантаження та спотворень напруги. Експериментальну перевірку виконано на програмованій вентильній матриці (FPGA) із використанням співмоделювання в реальному часі та апаратної реалізації. Результати. Моделювання в MATLAB/Simulink та реалізація в реальному часі на недорогій платі Spartan-6 FPGA показали, що запропонований метод зменшує час наростання (2–98 %) виділеної постійної складової активної потужності приблизно з 12 мс (LPF Баттерворта 4-го порядку) до 0,6–0,8 мс (покращення на 93–95 %) при одночасному забезпеченні коефіцієнта гармонічних спотворень струму мережі нижче 4,92 % у найгіршому випадку, що повністю відповідає вимогам стандарту IEEE 519. Надзвичайно низька обчислювальна складність робить запропоноване рішення придатним для промислових контролерів. Наукова новизна. У роботі запропоновано нову стратегію керування, у якій традиційний LPF замінено однокоефіцієнтним адаптивним фільтром VLLMS із забезпеченням робастної синхронізації за додатною послідовністю за допомогою MVF-PLL. Практична значимість. Запропонований алгоритм підвищує ефективність роботи SAPF, зменшує час реакції та забезпечує стабільне функціонування в різних режимах електричної мережі, що робить його надійним рішенням для промислових застосувань. Бібл. 24, табл. 2, рис. 15. National Technical University "Kharkiv Polytechnic Institute" and Аnatolii Pidhornyi Institute of Power Machines and Systems of NAS of Ukraine 2026-07-02 Article Article application/pdf https://eie.khpi.edu.ua/article/view/353265 10.20998/2074-272X.2026.4.06 Electrical Engineering & Electromechanics; No. 4 (2026); 40-46 Электротехника и Электромеханика; № 4 (2026); 40-46 Електротехніка і Електромеханіка; № 4 (2026); 40-46 2309-3404 2074-272X en https://eie.khpi.edu.ua/article/view/353265/351643 Copyright (c) 2026 M. M. Belhadj Mostefa, A. Boussaid, A. Khezzar http://creativecommons.org/licenses/by-nc/4.0
spellingShingle shunt active power filter
harmonic compensation
adaptive filtering
non-ideal grid
power quality
Belhadj Mostefa, M. M.
Boussaid, A.
Khezzar, A.
Shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions
title Shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions
title_alt Shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions
title_full Shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions
title_fullStr Shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions
title_full_unstemmed Shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions
title_short Shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions
title_sort shunt active power filter with variable leaky least mean squares and multivariable filter phase-locked loop for fast harmonic compensation under non-ideal grid conditions
topic shunt active power filter
harmonic compensation
adaptive filtering
non-ideal grid
power quality
topic_facet shunt active power filter
harmonic compensation
adaptive filtering
non-ideal grid
power quality
шунтуючий активний фільтр потужності
компенсація гармонік
адаптивна фільтрація
неідеальна електромережа
якість електроенергії
url https://eie.khpi.edu.ua/article/view/353265
work_keys_str_mv AT belhadjmostefamm shuntactivepowerfilterwithvariableleakyleastmeansquaresandmultivariablefilterphaselockedloopforfastharmoniccompensationundernonidealgridconditions
AT boussaida shuntactivepowerfilterwithvariableleakyleastmeansquaresandmultivariablefilterphaselockedloopforfastharmoniccompensationundernonidealgridconditions
AT khezzara shuntactivepowerfilterwithvariableleakyleastmeansquaresandmultivariablefilterphaselockedloopforfastharmoniccompensationundernonidealgridconditions