60 PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 95 NR 7/2019 Samira DAHMANI 1 , Abdelhafid SEMMAH 1 , Mohamed KADEM 1 , Patrice WIRA 2 Djillali Liabes University of Sidi Bel Abes, Algeria (1), Haute Alsace University of Mulhouse, France (2) doi:10.15199/48.2019.07.13 Genetic algorithm optimization of a SAPF based on the fuzzy- DPC concept Abstract. This article presents a study on the use of the concept of direct power control (DPC) based on intelligent techniques in the control of a shunt active power filter (SAPF). In order to improve harmonic mitigation and reactive power compensation capabilities, the conventional switching table is replaced by a fuzzy inference system to generate the switching sequences of the shunt active power filter. To ensure an active power exchange stable and efficient, the DC voltage of the SAPF in controlled using an integrated proportional controller (PI) optimized by a heuristic optimization technique based on genetic algorithms (GA). The combination of two intelligent techniques in this proposed control strategy makes it possible to reduce ripples in different variables of the SAPF, to maintain the direct voltage at their reference value and to improve the THD of the grid current. The numerical simulation results obtained under Matlab / Simulink confirm the importance of the SAPF's proposed control technique. Streszczenie. W artykule opisano wykorzystanie metody DPC (direct power control) do poprawy parametrów bocznikowego filtru aktywnego SAPF. Konwencjonalna tabela przełączeń jest zastąpiona przez system logiki rozmytej. Do optymalizacji filtru wykorzystano też algorytm genetyczny. Optymalizacja aktywnego filtru bocznikowego SAPF z wykorzystaniem algorytmu genetycznego i logiki rozmytej Keywords: Shunt active power filter, direct power control, fuzzy inference system, genetic algorithm. Słowa kluczowe: aktywny filtr bocznikowy SAPF, logika rozmyta, algorytm genetyczny Introduction Harmonic distortion is generated by non-linear loads connected to the grid that absorb non-sinusoidal currents [1]. These current harmonics will successively generate harmonic voltages at the different grid connection points [2]. Many solutions for cleaning up electricity networks have already been proposed in the literature [3, 4]. Active power filters (APF) are to date the most appropriate advanced pollution control solutions [5, 6], APF has several advantages such as the elimination of harmonic currents, reactive energy compensation, rebalancing of non-linear load currents [7]. The principle of parallel active filters was introduced by [8]. The principle is based on the injection of harmonic currents or voltages opposite to those generated by the nonlinear load so as to have a current or voltage resulting quasi-sinusoidal. The performance of the active filter depends on the control strategy adopted. Several control strategies have been proposed in the literature [9]. The DPC concept is equivalent to direct torque control (DTC) one for electrical machines [6]. The principle is based on the selection of a voltage vector using a switching table, direct self-control, or space vector modulation [10]. Among these DPC techniques, the strategy of voltage vector selection using a switching table is widely studied and marketed. This is due to its concept which is simple to implement. The selection of the voltage vector is based on the active power error, reactive power and therefore the position of the voltage vector. The most disadvantages in conventional DPC are: the system doesn’t differentiate between very small and relatively large errors of the active and reactive powers which can affect the stability of the system response, the occurrence of ripples at different SAPF quantities, additionally the utilization of conventional analysis methods to select the parameters of the PI controller need determining a transfer function including the controller and the system to be controlled. This needs the adoption of simplifying assumptions which will take us away from studying the important behavior of the system. Iteratively, it is identified that the PI parameters strongly have an effect on the THD of the system input current [11]. To remedy this, a new direct power control for the SAPF is proposed in this article, based on the use of fuzzy logic, replacing the conventional switching table with a fuzzy inference system, for the generation of the inverter switching times, and on the other hand, it is necessary to select the PI controller parameters in an optimal way to control the DC voltage, thus using a heuristic optimization technique based on genetic algorithms (GA). Shunt active power filter Fig. 1 illustrates the structure and principle of an SAPF based on voltage source converter.It consists of a bridge of six power transistors with anti-parallel diodes. Let us consider the voltages on the AC side of the active filter V fa , V fb , V fc and the input voltages of the converter V AN , V BN , V CN . S a , S b , S c represent the logical control variables Fig.1. Structure of the SAPF based on voltage source converter The operating principle of the SAPF on the AC side is given by: (1) NO fa f fa f fa ON AO AN V dt dI L I R V V V V       (2) NO fb f fb f fb ON BO BN V dt dI L I R V V V V       (3) NO fc f fc f fc ON CO CN V dt dI L I R V V V V       The phase-to-phase voltages V AN , V BN , V CN are created at the switch terminals by the commands S a , S b , S c applied to