Finite control set model predictive control to balance DC-link capacitor voltage for TType NPC inverter of grid-connected photovoltaic systems

TAÏP CHÍ PHAÙT TRIEÅN KH&CN, TAÄP 18, SOÁ K6- 2015 Finite control set model predictive control to balance DC-link capacitor voltage for T- Type NPC inverter of grid-connected photovoltaic systems . Phan Quoc Dzung . Nguyen Dinh Tuyen . Nguyen Minh Nhat Ho Chi Minh city University of Technology, VNU-HCM, Vietnam (Manuscript Received on July 15, 2015, Manuscript Revised August 30, 2015) ABSTRACT This paper proposes the Finite control set function of the proposed FCS-MPC uses the Model Predictive Control (FCS-MPC) with 27 possible switching states generated by T- delay compensation for three-phase three- Type NPC, the optimal switching state is level T-Type NPC inverter (T-Type NPC) of selected in each sampling time that grid-connected photovoltaic systems (PV). minimizes the cost function. The proposed The proposed FCS-MPC controls the FCS-MPC has also proposed the delay objectives: current tracking control, DC-link compensation with two-step prediction capacitor voltage balance, the reduction of horizon at time k+2 to reduce the total switching frequency to ensure issues of the harmonic distortion (THD) of the grid current. power quality and improve the efficiency of The proposed FCS-MPC is verified by using grid-connected of PV system. The cost Matlab/Simulink. Keywords: Finite control set Model predictive control, reduction of switching frequency, DC- link Capacitor voltage balance, T-Type NPC, PV system. 1. INTRODUCTION In the recent decades, photovoltaic system these disadvantages, the three-phase three-level (PV system) quickly has been developed about the T-Type Neutral Point Clamp inverter (T-Type installed power, as well as the penetration into the NPC), shown in Figure 1, is proposed, which has electrical grid. Photovoltaic is the distributed the advantages about the low THD of output generation, thus the stability when connecting the current, high efficiency with switching frequency grid with the photovoltaic is a problem that worth in 4kHz-30kHz, small size, low investment cost to concern about at this moment. Due to the [1], [2], [3]. restrict of the two-level inverter is the output The structure of a T-Type NPC includes two current with high THD distortion, the large size DC-link capacitors connected in series, thus the filters is chosen, however which causes some small voltage vectors and medium voltage vectors problems: reduction of the efficiency of PV of T-Type NPC produce the voltage oscillations system, high cost, bulky system [1], [2]. To solve Trang 5 SCIENCE & TECHNOLOGY DEVELOPMENT, Vol.18, No.K6 - 2015 of the mid-point “O” of dc-link capacitor, making FCS-MPC has presented with the limited the unbalanced DC-link capacitors [5], [6]. The switching states of the converter for solving the voltage oscillations of the mid-point of dc-link optimization problem from a discrete model of the capacitor voltage causes the voltage stress on the system [8], [10], [11], [14], [15]. Many researches capacitors and the semiconductor devices [4], [5]. with FCS-MPC have been presented for current Namely the failure in switching and the increase tracking control and DC-link capacitor voltage of THD of output current, due to the low-order balance [10], [13], [21], [14]; current tracking harmonics of output voltage [3], [5], [6]. Pulse control and reduce switching frequency [7]. The width modulation (PWM) is used to balance FCS-MPC with delay compensation (two-step prediction horizon) has presented to reduce the capacitor voltage with three-level NPC inverter THD of current [8], [11], [14], [15]. The FCS- [4]; three-level T-Type NPC [2], [3], [5]. MPC with delay compensation will cause a big However, the PWM algorithm has the complex quantity of switching states due to two-step calculation and is not easy to implement for the prediction horizon: k+1, k+2, for three-level T- controlled objects that are nonlinearities and have Type NPC, number of the switching states can many constraints [7], [8]. correspond to 272 = 729, the computation amount to select optimal switching state is massive. This paper focuses on the control for the objectives of grid-connected of PV system: current tracking control, balancing dc-link capacitor voltage, the reduction of switching frequency by using the proposed FCS-MPC with delay compensation. In order to reduce the computation amount, this paper proposes that the values of current and dc-link capacitor voltages at Figure 1. Three-level T-Type NPC inverter for three- time k+1, will be estimated based on the optimal phase grid-connected of PV system switching state at time k-1, then use these Model predictive control (MPC) bases on the estimated values to predict the values at time k+2. good mathematical model of system to support the Therefore, the proposed FCS-MPC only chooses exact and complete the prediction about the future 27 switching states of T-Type NPC for two-step behaviors of system in control. In addition, the prediction horizon for the reduction of the development of the microprocessor’s speed computation amount in comparison [8], [15]. The allows calculating quickly the MPC [9], [10]. This optimal switching state is selected that minimizes makes MPC more and more popular in many the cost function of the proposed FCS-MPC. The fields of the power electronics. MPC have distinct change of weighting factor in the interrelationship advantages: simple, easy to implement, well between the objectives of cost function, will prove control with the systems that are non-linear and the flexibility of the proposed FCS-MPC for have many constraints, fast responds with solving the problems of power quality and changing of the system, sustainable when enhancing efficiency of grid-connected for PV compared to the classic control uses the system via the reduction of switching frequency. proportional - integral controller (PI) and the 2. MODEL OF T-TYPE NPC FOR GRID- PWM modulators [8], [9], [11], [12]. CONNECTED OF PV SYSTEM 2.1 Model of the system Trang 6 TAÏP CHÍ PHAÙT TRIEÅN KH&CN, TAÄP 18, SOÁ K6- 2015 T-Type NPC for grid-connected of PV Grid current and grid voltage vector are system using L-filter is illustrated in Figure 1, the defined [14], [17]: dc-link voltage is assumed constant, we have the 2 2 i() ia  ai b  a i c (5) relation between voltage and current of grid- 3 connected of PV system in (1) [11], [14], [17]: 2 2 e() ea  ae b  a e c (6) di 3 v La  Ri  e  v aodt a a No With, 2 2 di (v av  a2 v )  v (1  a  a 2 )  0 (7) v Lb  Ri  e  v (1) 3No No No 3 No bodt b b No From (4) (5) (6) (7), mathematical model of di v Lc  Ri  e  v the grid-connected for PV system is re-written: codt c c No di v L  Ri  e (8) ia, ib, ic : grid current dt Equation (8) demonstrates the mathematical ea, eb, ec : grid voltage model of the grid-connected for PV system with L,R : filter inductance and filter resistance of T-Type NPC inverter. With the voltage vector is L-filter defined by (2), there are totally 19 separate vao, vbo, vco : output voltage of T-Type NPC voltage vectors for corresponding to 27 switching The voltage vector of T-Type NPC is defined: combinations of T-Type NPC as Figure 2. 2 2 v() Vao  aV bo  a V co (2) 2.2 Grid-connected control for PV system 3j2π/3 Where, a = e Voltage oriented control (VOC) method is Considering a constant dc-link voltage, and proposed to control grid-connected for PV balanced capacitor voltages, the voltages system. According to [16], the instantaneous generated by the T-Type NPC at the inverter active and reactive powers in the rotating dq frame terminals are obtained in (3) and Table 1 [14]. injected into the grid by PV system are given in V VS dc (3) (9): xo x 2 Table 1. Switching state of T-Type NPC, with x = a, b, c [1] Switching status Output Switching states voltage (Sx) Sx1 Sx2 Sx3 Sx4 (Vxo) 1 1 1 0 0 Vdc/2 0 0 1 1 0 0 -1 0 0 1 1 -Vdc/2 Figure 2. Voltage vectors and switching states can be generated by a T-Type NPC From (2), equation (1) is re-written: 3 P() e i  e i (9) d 2 2 2 d d q q vL( ( iaiai  2 ))  R ( ( iaiai   2 )) a b c a b c 3 dt 3 3 (4) Q() e i  e i 2 2 2 q d d q  ()()e  ae  a2 e  v  av  a 2 v 3a b c 3 No No No Trang 7 SCIENCE & TECHNOLOGY DEVELOPMENT, Vol.18, No.K6 - 2015 ed, eq: grid voltage in the rotating dq frame The proposed FCS-MPC for T-Type NPC of id, iq : grid current in the rotating dq frame the grid-connected of PV system is illustrated in Figure 4. The proposed FCS-MPC for T-Type Assuming that the d axis is perfectly aligned NPC of the grid-connected of PV system uses 27 with the grid voltage, thus eq = 0. The active switching states corresponding to 19 separate power and reactive power are rewritten: voltage vectors as Figure 2. The optimal switching 3 P e i state is selected that minimizes the cost function. 2 d d (10) The proposed FCS-MPC for grid-connected of PV 3 Q  e i system produces the optimized switching state 2 d q From (10), we can independently control directly to the T-Type NPC, without using the modulation techniques such as CB-PWM or between active and reactive powers with id and iq currents. We assume that: the dc-link voltage is Space Vector Modulation (SVM). constant, delivering symmetrical and balanced ac The proposed FCS-MPC for grid-connected currents to the grid, so we can skip the calculation of PV system has two main assumptions: keeping * of the reference grid current (i dq) [13]. In this the stable DC-link voltage, symmetrical and * * paper, we only choose i dq as an external reference balanced grid voltages, so we treat only i dq(k) for for implementing the proposed FCS-MPC, as controlling the grid-connected of PV system, not shown in Figure 4. operating in islanded case with grid systems. 3. THE PROPOSED FCS-MPC From Figure 4, grid voltage and grid current are measured at time k, which are used for predicting 3.1 Finite control set model predictive the grid current at time k+2 and the predicted grid control (FCS-MPC) with delay current shows in stationary αβ frame. The compensation reference grid current in the dq frame have to FCS-MPC for the current tracking control show in stationary αβ frame for implementing the presented in [11], [14]. x*(k) represents the proposed FCS-MPC, so we use Phase-Locked- reference values of the desired control object, x(k) Loop block (PLL) to synchronize the phase angle is the measured value at time k, x(k+2) is the (θ) between the grid system and the PV systems predicted value for possible switching states of T- [16]. We use Park and Clark transformations for Type NPC inverter at time k+2 (two-step the conversion between the abc, αβ, dq frames prediction horizon). The error between x(k+2) and with phase angle (θ) calculated by PLL blocks. x*(k) is obtained to minimize the cost function The cost-function of the proposed FCS-MPC and the switching state that minimizes the cost produces the optimal switching state based on the function will be chosen. FCS-MPC block diagram reference grid current and the predicted grid is illustrated in Figure 3 [9]. current, or other controlled objectives. In this paper, x(k) presents for the desired control objectives: grid current, DC-link voltage, switching frequency. Figure 3. FCS-MPC block diagram Trang 8 TAÏP CHÍ PHAÙT TRIEÅN KH&CN, TAÄP 18, SOÁ K6- 2015 Figure 4. The proposed FCS-MPC applies for T-Type NPC of grid-connected of PV System 3.2 Current tracking control According to the proposed FCS-MPC with Applying a sampling period Ts, the derivative delay compensation, the grid current at k+1 time of grid current in (8) will be approximated by is estimated based on the grid current and voltage Euler forward method [14], [17], as follows: current measured at time k, thus the estimated grid current has the value as (12), meaning ip(k+1) = di i( k  1)  i(k) (11)  i^(k+1), with i^(k+1) is the estimated value at time dt Ts k + 1 and v(k) is the optimal voltage vector which i(k+1) is the predicted grid current at time was calculated at time k-1. The estimation value k+1, i(k) is the value of grid current measured at i^(k+1) will be used for predicting the grid current time k. From (8), (11) and grid voltage e(k) ip(k+2) at time k+2, considering all possible measured at time k, the equation of the predicted switching states of T-Type NPC [14]. The grid current at time k+1 is rewritten: prediction of grid current at time (k+2) is defined: RT T p RTs T s ip ( k 1)(1  s )() i k  s (() v k  e ()) k (12) i( k 2)  (1  ) i ( k  1)  ( v ( k  1)  e ( k  1))(13) LL LL v(k+1) is voltage vector of T-Type NPC for v(k) is voltage vector of inverter and considering all possible switching states depending on the switching states of T-Type NPC generated by the inverter. e(k+1) is the grid inverter as (2) and (3), so the predicted grid voltage at time k+1. The value e(k+1) can be current i(k+1) at each time will vary and depends determined by Lagrange extrapolation formula as on the switching states. follows [14]: e( k 1)  3() e k  3( e k  1)  e (k2)  (14) Trang 9 SCIENCE & TECHNOLOGY DEVELOPMENT, Vol.18, No.K6 - 2015 For a small enough Ts, we can approximate time k-1 to thereby calculate switching instants for e(k+1) ≈ e(k), equation (13) is rewritten: each switching state. The cost function for RT T ip ( k 2)  (1 s ) i ( k  1)  s ( v ( k  1)  e ( k )) (15) reducing the switching frequency is defined: LL g  . N From equation (15) shows that the different 2s w k  2 (22) switching states of T-Type NPC inverter will λsw : weighting factor of the reduction of produce the different values of predicted grid switching frequency current at time k+2. The cost function for current Nk+2 : the total switching instants is calculated tracking control is defined with the predicted grid by the change between of each expected switching current and the reference grid current at time k+2, state of the 27 possible switching states generated gives in (16): by T-Type NPC with optimal switching state was *p 2 * p 2 (16) chosen at time k-1. g1 ( i ( k  2)  i  ( k  2))  ( i  ( k  2)  i  ( k  2)) p p p (17) 3.4 Balancing DC-link capacitor voltages i( k 2)  i ( k  2)  ji  ( k  2) *** (18) Assuming the DC-link capacitor voltage is i( k 2)  i ( k  2)  ji  ( k  2) p p i α(k+2), i β(k+2) of the predicted grid current constant. From Figure 1, the prediction for the * * voltage of each capacitor at the time k+1 is [14]: and i α(k+2), i β(k+2) of the reference grid current TT show in αβ frame at time k+2. v( k  1)  v (k) s i ( k )  v (k)  s i ( k ) C1 C1CC C 1 C1 2 o The i*(k+2) at time k+2 can be determined by (23) TT Lagrange extrapolation formula as follows [14]: v( k 1)  v (k) s i ( k )  v (k)  s i ( k ) C2 C2CC C 2 C2 2 o **** i( k 2)  6() i k  8( i k  1)  3i(k  2) (19) vC1, vC2 : DC-link capacitor voltage For a small enough Ts, we can approximate C : capacitance of each DC-link capacitor i*(k+2) ≈ i*(k) [14], [17]. The cost function for io : current at the mid-point of dc-link current tracking control is rewritten: capacitor “O” *p 2 * p 2 g1 (() i k  i  ( k  2))  (() i  k  i  ( k  2)) (20) The current at the mid-point of dc-link 3.3 Reduction of switching frequency capacitor “O” can be defined from currents ia, ib, The switching frequency for MPC is limited ic on each phase at the current time and switching to half the sampling frequency (fs) [14], [17]. The state depended function Sxo , with switching state switching frequency is the average switching Sx in Table 1 [14]: frequency value of power semiconductors of (24) iko()()()()()()() Skik ao a  Skik bo b  Skik co c inverter. Switching frequency for FCS-MPC for Sxo = 1 if Sx=0; Sxo = 0 if Sx = 1 or -1; (25) T-Type NPC with 12 power semiconductors with x = a,b,c (IGBTs) is calculated [11], [17]: N (21) The proposed FCS-MPC will delay 01 fsw  12*t sampling period (Ts), thus we will predict the N : the total switching instants of 12 IGBTs future value k+2 based on the estimated value at during period of t the time k+1. DC-link capacitor voltage is t : time of simulation or experiment predicted at time k+2 [14]: The total switching instants N should be TTs s decreased to reduce the switching frequency. vC1( k 2)  v C 1 ( k  1)  i C 1 ( k  1)  v C 1 ( k  1)  i o ( k  1) CC2 (26) Comparing the 27 switching states of T-Type TTs s vC2( k 2)  v C 2 ( k  1)  i C 2 ( k  1)  v C 2 ( k  1)  i o ( k  1) NPC inverter with optimum switching state at CC2 Trang 10 TAÏP CHÍ PHAÙT TRIEÅN KH&CN, TAÄP 18, SOÁ K6- 2015 v(k+1) is the voltage vector corresponding to link=Vdc=700V; fgrid = 50Hz; C1 = C2 = 5mF. To 27 the possible switching states in the proposed demonstrate the flexibility for controlling grid- FCS-MPC. The current at the mid-point of dc-link connected of PV systems, the reference grid capacitor “O” at time k+1 is written: current of d-axis of the rotating dq frame will have (27) the different values at different times, and the iko( 1)  Skik ao (  1)( a  1)  Skik bo (  1)( b  1)  Skik co (  1)( c  1) reference grid current of q-axis with iqref = 0A; io(k+1) is defined based on the estimated grid sample time Ts = 25µs with sampling frequency current i(k+1) at the time k+1, and Sao(k+1), fs = 1/Ts = 40kHz applies the proposed FCS-MPC: Sbo(k+1), Sco(k+1) correspond to all of the switching states of inverter as (25). + t = 0s – 0.2s; idref = 4A, iqref = 0A The differentiation of DC-link capacitor + t = 0.2s – 0.3s; idref = 10A, iqref = 0A voltage is defined: + t = 0.3s – 0.5s; idref = 6A, iqref = 0A VDC( k  2)  v C1 (k  2)  V C2 (k  2) (28) In this paper, weighting factor of the From the (26) and (28), we have: capacitor voltage balance with λDC = 8 and Ts weighting factor of the reduction of switching VDC( k  2)  v C1 (k  1)  V C2 (k  1)  i o ( k  1) (29) C frequency with λsw = 0.1 và λsw = 1.5 are used to Cost function for DC-link capacitor voltage compare the responses of the grid-connected of balance is defined in (34): PV system. g (  V ( k  2))2 3 DC DC (30) The proposed FCS–MPC for T-Type NPC of λDC: weighting factor of capacitor voltage grid-connected of PV system provides the fast balance. response to the power change of system with 3.5 Cost function of the proposed FCS – decoupled control active power (P) and reactive MPC power (Q) in the rotating dq frame via the current From (20), (22), (30), the proposed FCS – tracking control. At the time t = 0.2s, the reference MPC for controlling many objectives: current grid current is ordered to change from idref = 4A tracking control, balancing the DC-link capacitor to idref = 10A, and at the time t = 0.3s, the reference voltage, reduction of switching frequency with the grid current is ordered to change from idref = 10A cost function is defined: to idref = 6A, system’s response is fast, as shown in Figure 5 and Figure 6. Corresponding to the g g  g  g 1 2 3 change of power supply, the grid current in abc *p 2 * p 2 2 (ikik ( )   (  2))  ( ikik  ( )   (  2)) DC .(  Vk DC (  2))   sw . N k 2 frame in Figure 8, at the time of t = 0.2s and t = Depending on the priority of the controlled 0.3s, has a fast response and a good current objectives, we can adjust the weighting factors tracking control. (λDC, λsw) represented for objectives to obtain the For the proposed FCS – MPC, the reduction optimal switching state. Some guidelines on of switching frequency shows the interrelation choosing weighting factor for the desired between the weighting factor of the reduction of objectives may refer in [18]. switching frequency and weighting factor of The optimal switching state is selected by the capacitor voltage balance. The switching minimized of cost function g. frequency of PWM is constant. FCS–MPC 4. SIMULATION RESULTS chooses the optimal voltage vector, minimizes the cost function, hence there is a independence The specifications for the system in Figure 1 between sampling frequency and switching are: R=0.5Ω; L=5mH; Vgrid = 220V; Vdc- Trang 11 SCIENCE & TECHNOLOGY DEVELOPMENT, Vol.18, No.K6 - 2015 frequency [7], [14]. We can see that the switching the THD of grid current is higher than frequency depends on the weighting factors λsw in implemented by λDC = 8, λsw = 0.1, shown in the cost function. With the same capacitor balance Figure 5(a) and Figure 5(b). weighting factor λDC = 8, and λsw = 1.5 produces (A) (A) abc abc i i - - Grid current current Grid Grid current current Grid Time Time (s) (A) (A) ia abc abc i i - - ib ic Grid current current Grid Grid current current Grid Time (s) Time THD = 11.20% THD = 2.81% (a) Figure 5. Waveform and THD of grid current in abc frame: (a) λDC = 8, λsw = 1.5; (b) λDC = 8, λsw = 0.1 Trang 12 TAÏP CHÍ PHAÙT TRIEÅN KH&CN, TAÄP 18, SOÁ K6- 2015 (A) dq dq i i - - Grid current current Grid Grid current current Grid Time (s) id,P iq,Q Time (s) (a) (b) Figure 6. Waveform of grid current in the rotating dq frame: (a) λDC = 8, λsw = 1.5; (b) λDC = 8, λsw = 0.1 (V) (V) C2 C2 V V - - C1 C1 = V = V DC DC ΔV ΔV (V) (V) C2 C2 V V - - C1 C1 = V = = V DC DC ΔV ΔV C2 (V) C2 (V) C2 Voltage of C1, C1, of Voltage C1, of Voltage Time (s) Time (s) VC1 VC2 (a) (b) Figure 7. The differentiation of DC-link capacitor voltage ΔVDC, capacitor voltage of C1 and C2 capacitor: (a) λDC = 8, λsw = 1.5 ; (b) λDC = 8, λsw = 0.1 Trang 13 SCIENCE & TECHNOLOGY DEVELOPMENT, Vol.18, No.K6 - 2015 (V) AB AB V - V - Line to line line voltage to Line Line to line line voltage to Line Time (s) Time (s) (a) (b) Figure 8. Wave form of output voltage of T-Type NPC inveter - VAB (Line to line voltage - VAB) and VAN (Voltage of phase A): (a) λDC = 8, λsw = 1.5; (b) λDC = 8, λsw = 0.1 The increasing weighting factor of the reduction of switching frequency will make the reduction of switching frequency and the increase of THD of grid current, as shown in Figure 5, Figure 9 and Table 2. Similarly, the increasing weighting factor of the reduction of switching frequency, the priority of DC-link capacitor voltage balance will decrease, so the voltage oscillations of the mid- point of dc-link capacitor “O” will increase, as shown in Table 2. With λDC = 8 and λsw = 1.5 produces ΔVDC = 1.2V, as shown in Figure 7(a); with λDC = 8 and λsw = 0.1 only produces ΔVDC = Figure 9. The effects of weighting factor of the 0.27V as shown in Figure 7(b). From Table 2 and reduction switching frequency to: The DC-link Figure 9 (using Microsoft Excel to sketch), with capacitor voltage balance and THD of grid current the λDC = 8, the increasing weighting factor of reduction of switching frequency corresponds to Table 2 shows the independence between the reduce the switching frequency and increase the sampling frequency and the switching frequency voltage oscillations of the mid-point of dc-link via the control of the interrelation between capacitor, and the correlation between the voltage objectives: grid current, DC-link capacitor oscillations of the mid-point of dc-link capacitor voltage, switching frequency by the cost function and weighting factor of reduction of switching of the proposed FCS-MPC. With λDC = 8, λsw = frequency is almost linear. Waveforms of line-to- 0,1, the THD of grid current is only 2,81%, the line voltage of T-Type NPC is illustrated in Figure DC-link capacitor voltage is the best value with 8, which comply with wave forms of output ΔVDC = 0,27V and the switching frequency is voltage of T-Type NPC. 4.99kHz, complying with the switching frequency for T-Type NPC with high efficiency Trang 14 TAÏP CHÍ PHAÙT TRIEÅN KH&CN, TAÄP 18, SOÁ K6- 2015 [1]. The proposed FCS-MPC provides the THD of 5. CONCLUSION grid current to meet the requirements for grid The proposed FCS-MPC provides the connected of PV system as IEEE 1547 and IEC excellent of grid-connected of PV system with the 61727. changes of injected power into grid, the low THD of grid current and the good DC-link capacitor Table 2. The quantitative of the desired objectives, voltage balance. Using the proposed FCS-MPC with Ts = 25µs, fs = 40kHz can achieve flexible to choose the switching DC-link Weighting Weighting Switching THD capacitor frequency for the three-level T-Type NPC to factor of factor of frequency (%) voltage - enhance the efficiency of system while still the λDC λsw fsw (kHz) ΔVDC (V) guaranteeing about the problems concerned with 8 0 6.961 3.07 0.35 8 0.1 4.99 2.81 0.27 the power quality as: DC-link capacitor voltage 8 0.3 3.428 3.53 0.4 balance and THD of grid current. The proposed 8 0.5 2.477 4.54 0.55 FCS–MPC only uses 27 switching states to 8 0.7 1.77 5.86 0.8 calculate the cost function, considerably reduces 8 0.9 1.414 7.07 1 8 1.1 1.151 8.95 1.1 the computation amount to enhance the efficiency 8 1.3 0.973 9.82 1.15 of the proposed FCS-MPC. In addition, the 8 1.5 0.871 11.2 1.2 proposed FCS–MPC is simple and easy to control 8 1.7 0.781 13.47 1.3 8 1.9 0.712 14.12 1.45 in comparison with the traditional PWM for the control of DC-link capacitor voltage balance with the constant switching frequency [3]. Giải thuật finite control set model predictive control để cân bằng điện áp tụ DC-link cho nghịch lưu T-Type NPC của hệ thống năng lượng mặt trời kết nối lưới . Phan Quốc Dũng . Nguyễn Đình Tuyên . Nguyễn Minh Nhật Trường Đại học Bách Khoa, ĐHQG-HCM, Việt Nam TÓM TẮT: Bài báo này đề xuất giải thuật điều khiển Control (FCS-MPC) có bù trễ cho nghịch lưu dự báo Finite control set Model Predictive 3 pha 3 bậc T-Type NPC (T-Type NPC) của Trang 15 SCIENCE & TECHNOLOGY DEVELOPMENT, Vol.18, No.K6 - 2015 hệ thống năng lượng mặt trời kết nối lưới. cắt tối ưu được chọn trong mỗi thời gian lấy FCS-MPC được đề xuất cho điều khiển nhiều mẫu là trạng thái mà ở đó hàm mục tiêu có đối tượng: điều khiển bám dòng điện, cân giá trị bé nhất. Giải thuật FCS-MPC được đề bằng điện áp tụ DC-link, giảm tần số đóng cắt xuất là có bù trễ với hai thời khoảng dự báo để qua đó đảm bảo các vấn đề chất lượng tại thời điểm k+2 để qua đó giúp giảm méo điện năng và hiệu suất của hệ thống năng dạng sóng hài của dòng điện lên lưới. FCS- lượng mặt trời kết nối lưới. Hàm mục tiêu của MPC được đề xuất sử dụng phần mềm FCS-MPC sử dụng 27 trạng thái đóng cắt Matlab/Simulink để kiểm chứng. khác nhau của T-Type NPC, trạng thái đóng Từ khóa: Finite control set Model predictive control, giảm tần số đóng cắt, cân bằng điện áp tụ DC-link, nghịch lưu T-Type NPC, hệ thống năng lượng mặt trời. REFERENCES [1]. Schweizer, M.; Kolar, J.W., Design and connected Three-Level Inverters using a Implementation of a Highly Effiient Three- Discontinuous Pulse Width Modulation, Level T-Type Converter for Low-Voltage Vehicle Power and Propulsion Conference Applications, Power Electronics, IEEE (VPPC), IEEE Transactions on, pp. 638 - Transactions on, pp. 899 - 907, June 2012. 642, Oct. 2012. [2]. June-Seok Lee; Kyo-Beum Lee, New [6]. Bin Wu, High-power converters and AC Modulation Techniques for a Leakage drivers, IEEE press and John Wiley & Sons Current Reduction and a Neutral-Point Ltd, 2006. Voltage Balance in Transformerless [7]. Cortes, P.; Vattuone, L.; Rodriguez, J., A Photovoltaic Systems Using a Three-Level comparative study of Predictive Current Inverter, Power Electronics, IEEE Control for Three-Phase Voltage Source Transactions on, pp. 1720 - 1732, June 2013. Inverters based on switching frequency and [3]. Ui-Min Choi; Blaabjerg, F.; Kyo-Beum Lee, current error, Power Electronics and Method to Minimize the Low-Frequency Applications, IEEE Transactions on, pp. 1 - Neutral-Point Voltage Oscillations with 8, Aug. 30 2011-Sept. 1 2011. Time-Offset Injection for Neutral-Point- [8]. Yaramasu,V; Wu, B.; Rivera,M.; Clamped Inverters, Industry Applications, Rodriguez,J.; Wilson,A., Cost-Function IEEE Transactions on, pp. 1678 - 1691, based Predictive Voltage Control of Two- August 2014. Level Four-Leg Inverters using Two Step [4]. June-Seok Lee; Kyo-Beum Lee, A Carrier- Prediction Horizon for Standalone Power Based PWM Method for Neutral-Point Systems, Applied Power Electronics Ripple Reduction of a 3-Level Inverter, Conference and Exposition (APEC), 2012 Energy Conversion Congress and Exposition Twenty-Seventh Annual IEEE, IEEE (ECCE), IEEE Transactions on, pp. 2095 - Transactions on, pp. 128 - 135, Feb. 2012. 2100, Sept. 2014. [9]. Vazquez,S.; Leon,J.I.; Franquelo,L.G. ; [5]. Hyun-Hee Lee, Ui-Min Choi, Kyo-Beum Rodriguez,J.; Young,H.A.; Marquez,A.; Lee, Neutral-Point Voltage Control for Grid Zanchetta,P., Model Predictive Control A Trang 16 TAÏP CHÍ PHAÙT TRIEÅN KH&CN, TAÄP 18, SOÁ K6- 2015 Review of Its Applications in Power [14]. Calle-Prado,A.; Alepuz,S.; Bordonau, J.; Electronics, Industrial Electronics Magazine, Nicolas-Apruzzese, J. ; Cortes, P.; IEEE Transactions on, pp. 16 - 31, March Rodriguez, J., Model Predictive Current 2014. Control of GridConnected Neutral-Point [10]. Almaktoof, A.M.; Raji, A.K.; Kahn, M.T.E., Clamped Converters to Meet Low Voltage Finite-Set Model Predictive Control and DC- Ride-Through Requirements, Industrial link Capacitor Voltages Balancing for Three- Electronics, IEEE Transactions on, pp. 1503 Level NPC Inverters, Power Electronics and - 1514, October 2014. Motion Control Conference and Exposition [15]. Uddin, Muslem ; Mekhilef, Saad; Nakaoka, (PEMC), IEEE Transactions on, pp. 224 - Mutsuo; Rivera, Marco, Model Predictive 229, Sept. 2014. Control of Induction Motor with Delay Time [11]. Jiefeng Hu; Jianguo Zhu; Dorrell,D.G., Compensation: An Experimental Model Predictive Control of Grid-Connected Assessment, Applied Power Electronics Inverters for PV Systems With Flexible Conference and Exposition (APEC), IEEE Power Regulation and Switching Frequency Transactions on, pp. 543 - 548, March 2015. Reduction, Energy Conversion Congress and [16]. Remus Teodorescu, Marco Liserre and Pedro Exposition (ECCE), IEEE Transactions on, Rodríguez, Grid converters for photovoltaic pp. 540 - 546, Sept. 2013. and wind power systems, IEEE press & John [12]. Karamanakos,P.; Geyer, T.; Oikonomou,N.; Wiley & Sons, 2011. Kieferndorf,F.D.; Manias,S., Direct Model [17]. Jose Rodriguez; Patricio Cortes, Predictive Predictive Control: A Review of Strategies control of power converters and electrical That Achieve Long Prediction Intervals for drivers, IEEE press and John Wiley & Sons Power Electronics, Industrial Electronics Ltd, 2012. Magazine, IEEE Transactions on, pp. 32 - [18]. Cortes, P.; Kouro, S.; La Rocca, B.; Vargas, 43, March 2014. R.; Rodriguez, J.; Leon, J.I.; Vazquez, S.; [13]. Scoltock, J.; Geyer, T.; Madawala,U.K., A Franquelo, L.G., Guidelines for Weighting Model Predictive Direct Current Control Factors Design in Model Predictive Control Strategy with Predictive References for MV of Power Converters and Drives, Industrial Grid-Connected Converters with LCL- Technology, IEEE International Conference Filters, Power Electronics, IEEE on , pp. 1 - 7, Feb. 2009. Transactions on, pp. 1, December 2014. Trang 17