Decentralized Finite Control Set Model Predictive Control Strategy of Microgrids for Unbalanced and Harmonic Power Management

In this article, a modified decentralized finite control set model predictive control (FCS-MPC) scheme for the distributed energy resources (DERs) is proposed to improve the power management quality of the prosumers integrated microgrids under the condition of harmonic and unbalance loads. The proposed control strategy for the microgrids mainly consists of the power droop controller, the model predictive controller delay compensation, feedback correction and the unbalance compensation mechanism. The feedback correction method is used to correct the delay compensation, which effectively reduces the average switching frequency (ASF) and voltage total harmonic distortion (THD). By sharing the negative sequence reactive power of the microgrid, power distribution between the prosumers’ DERs is improved. The DERs in the prosumers can be integrated without any communication wire. The transient response and robustness to parameter changes are far superior to hierarchical cascaded control. Moreover, the proposed control strategy can better suppress harmonic and reduce and share fundamental negative sequence reactive power under microgrid unbalance and nonlinear load conditions. Finally, the effectiveness of the proposed FCS-MPC control strategy is validated by time-domain simulation results and real-time tests with RT-Lab under the condition of unbalanced and nonlinear loads in the microgrids.


I. INTRODUCTION
With the development of the global economy and society, the use of fossil energy has caused a series of environmental problems. The growing demand for renewable energy (RES) has led to the increasing penetration of distributed generation (DG) units in the power systems, such as solar, wind, tidal and fuel cells [1]- [5]. Generally, various types of DG systems can be connected to the microgrid to improve the The associate editor coordinating the review of this manuscript and approving it for publication was Abdullah Iliyasu . flexibility and reliability of the distribution network system. [6]- [8]. Microgrids are an effective method to solve the high penetration rate of renewable energy in the future smart grid [9]- [12].
In general, the ac microgrid facilitates deployment and compatibility with existing equipment. For the centralized control structure, the control of all units relies on a dedicated central controller, which requires extensive communication between the central controller and the controlled units. The high reliance of data acquisition and supervisory control on communication not only makes deploying new DGs cumber-some but also results in the security issues of the communication failure [13]- [15].
On the other hand, decentralized control structure eliminates the need for complicated communication network for data transmission within the microgrids. In this case, droop control strategies and their variants are the universal choices [16], [17]. In the islanded microgrid, the active power and reactive power sharing among the DG units can be achieved by adopting the P-f/Q-E droop control method with only local information [18], [19]. However, the traditional droop control approach cannot effectively cope with the harmonic voltage and negative sequence voltage when nonlinear loads or unbalance loads connected to the microgrids [20], [21]. In order to overcome these issues and address the harmonic and unbalanced power sharing in microgrids, improved droop control approaches are proposed in [22]- [24]. In [22], a dynamic consensus algorithm is adopted to realize the negative sequence current sharing to compensate the unbalance effectively. [23] proposes an adaptive virtual impedance control method based on injecting a very small ac signal (SACS) into the output voltage of each inverter. The droop method is used to adjust the virtual impedance, thereby adjusting the distribution of unbalance and harmonic power in the microgrid system. In [24], an auxiliary controller is supplemented to the cascaded hierarchical control scheme of an islanded microgrid to enhance the unbalance and harmonics control and obtain accurate power sharing. In [25] and [26], the enhancement of the voltage quality of critical load bus in microgrids are studied. In [25], low-bandwidth communication (LBC) technology is used to send appropriate signals of the secondary control from the microgrid control center to the primary control; thus, the voltage unbalanced compensation can be achieved. In [26], a complementary part is added to the secondary control loop to limit the voltage unbalance factor at local buses and DG terminals.
However, the conventional cascaded control structure presented in [20]- [26] has inherent drawbacks such as slow response speed, which will dramatically degrade the effect of the compensated and improved control schemes in hierarchical control architecture. Therefore, the main issues associated with the nonlinearity of the loads and dc dynamics will deteriorate the power quality of the voltage source converter (VSC) based microgrids.
Model predictive control (MPC) which was originally applied to industrial process control has many attractive features: 1) dealing with multivariate control problems; 2) easy to adjust; 3) explicitly considering constraints [27], [28]. With the development of digital processor technology, model predictive control is gradually applied to power converters and drivers [29], [30], such as motor driver [31], [32], uninterrupted power supply (UPS) [33], static compensator [34], [35]. Recently, MPC has been applied to microgrids, providing a new solution for the control of grid-connected inverters. In [36], a continuous control set model predictive control (CCS-MPC) method is introduced for grid-connected inverters. In [16], MPC is combined with the droop control method, which is used to minimize voltage unbalance, improve current limiting and prevent overload of active power. In [37], a cost function-based CCS-MPC is proposed to compensate the voltage imbalance and different harmonics of the power grid. However, when CCS-MPC is applied to the grid-connected inverters, the compensation for harmonics and unbalance will bring a substantial calculation burden. Moreover, the non-ideality characteristics of the inverters will bring more challenges to the modeling and control process.
On the other hand, the finite control set model predictive control (FCS-MPC) is easy to implement and has low computational cost. The FCS-MPC method can predict the future behavior of system control variables under finite switching conditions. The preset cost function defines the weight of different control variables (such as voltage, current, switching frequency) in the control target. This feature provides new options for improving power quality.
Nevertheless, the variable switching frequency of this control method will lead to more harmonics for the converters. In [38]- [45], the finite control model predictive control is applied to the grid-connected inverter, and the results show that MPC is superior to traditional controllers in terms of steady-state and transient response. In [40], a FCS-MPC is developed that does not require tuning and has low computational burden. [41] proposes a hybrid control strategy of FCS-MPC and deadbeat control to achieve a fixed frequency and robustness. In [42], a FCS-MPC is applied to photovoltaic power generation, which realizes maximum power tracking and voltage and frequency control on the ac bus. In [43], a decentralized FCS-MPC achieves excellent dynamic characteristics and robustness. [44] provides a method to test the stability of FCS-MPC. In [45], a FCS-MPC based microgrid control scheme is presented considering limitation of fault current and smooth transition between islanded and grid-connected modes.
It should be noted that nonlinear and unbalance loads are typical loads in residential microgrids. In [46], a cost function-based finite control set model predictive control method is used to implement unbalance compensation for grid mode operation. However, the studies mentioned above have not considered the harmonic compensation and unbalance control in islanded microgrids when adopting FCS-MPC. Therefore, there is a research gap in developing the FCS-MPC control strategy for islanded microgrids under unbalanced and harmonic load conditions.
In this article, a decentralized FCS-MPC control strategy for the VSC-based islanded microgrids in the case of unbalanced and nonlinear loads is proposed. Feedback correction is used to reduce prediction errors and switching times, thereby reducing voltage distortion. An unbalance compensation is proposed, which reduce and share fundamental negative sequence reactive power. The numerical simulation was implemented under the condition of unbalanced and nonlinear loads using MATLAB/Simulink tools, and real-time verification testing was performed with RT-Lab. The effectiveness of the proposed control strategy was verified.
The key contributions of this article are highlighted as follows: 1) A modified decentralized FCS-MPC control framework is proposed to reduce the harmonics of the microgrid and share negative sequence reactive power between distributed energy resources (DERs).
2) A feedback correction method for FCS-MPC is proposed to improve delay compensation and output vector prediction, thereby improving voltage tracking and control of voltage harmonics.
3) Compared with conventional FCS-MPC, the proposed modified decentralized FCS-MPC strategy improves the robustness of circuit parameter changes, thereby reducing the dependence on model matching.
The rest of this article is organized as follows. In Section II, the discrete dynamic model of VSC DG units is presented. Section III proposes the decentralized FCS-MPC strategy. In Section IV, the numerical simulation is implemented using MATLAB/ Simulink tools. In Section V, real-time verification testing is performed with RT-Lab. Section VI gives the conclusion.

II. DISCRETE DYNAMIC MODELING OF VSC DG UNITS
As shown in Fig. 1, the studied microgrid in this article consists of several parallel VSC DG units, which are connected to the point of common coupling (PCC) through corresponding feeders respectively. The DG is composed of renewable energy source (RES), a three-phase inverter and an LC filter. Note that the studied microgrid in Fig. 1 is a three-phase three-wire system. Therefore, zero-sequence current is not considered in this article.

A. VSC DG MODEL
As shown in Fig. 2, the inverter is connected to the microgrid via LC filter. The output voltage vector of the three-phase  inverter in the αβ stationary orthogonal reference frame is given in Table 1. In Table 1, S a , S b and S c are the switch configurations of phases A, B and C, respectively. The LC filter is connected to the outlet of the inverter and is three-phase symmetrical. The dynamic characteristics of the filter inductance L f in complex αβ frame can be described as: where i l is the current of the filter inductance; r f is the resistance of the filter inductance; v i is the output voltage of the inverter; v o is the voltage of the filter capacitor C f . The dynamic characteristics of the filter capacitor C f in complex αβ frame can be described as: where i o is output current of the LC filter. Thus, the state space equations of the VSC with an LC filter can be obtained by combining (1)-(4), which is expressed as follows: where

DISCRETIZATION OF THE DG DYNAMIC MODEL
Discretization of continuous state space equation is necessary for the digital controller in practical applications. In this article, because of the advantage of the accurate approximation between continuous and discrete systems, the zero-order hold (ZOH) discretization method is adopted for the discretization of the DG dynamic model. The discretization equation is used to predict the output of the inverter. Hence, Equation (5) can be discretized as follows: where x(k + 1) are the predicted current of the filter inductance i f (k + 1) and the voltage of the filter capacitor v f (k + 1) at the time of (k +1)T, respectively; x(k) are the current of the filter inductance i f (k) and the voltage of the filter capacitor v f (k) at the time of kT; y(k) are the output voltage of inverter and output current of LC filter at the time of kT. A d and B d can be calculated by: and where T s is the sampling time of the controller.

III. PROPOSED MODIFIED DECENTRALIZED FCS-MPC CONTROL STRATEGY FOR MICROGRIDS
As shown in Fig. 3, the proposed control strategy is comprised of two parts: the harmonic compensation and the unbalance compensation based on FCS-MPC. The harmonic compensation function is achieved by a supplement of feedback correction in the FCS-MPC cost function. The unbalance compensation is implemented by the auxiliary loop, which uses the fundamental negative sequence reactive power to change the output reference voltage. The voltage and current signals of the VSC DG are sampled and feed to the FCS-MPC strategy, and then provide switching signals to the inverter.

A. POSITIVE/NEGATIVE SEQUENCE POWER CALCULATION
The fundamental positive sequence and fundamental negative sequence current of the DG unit are separated by using the second-order generalized integrator, which are detailed in Section III-F. Then, the active power P + , reactive power Q + VOLUME 8, 2020 and unbalance power Q − of the DG unit can be expressed as follows [47]: where I + is the fundamental positive sequence current of the DG unit; I − is the fundamental negative sequence current of the DG unit; E * is the nominal voltage.

B. DROOP CONTROL
Positive sequence active power-frequency and positive sequence reactive power-voltage droop control method are used for power sharing among the DG units, while avoid using the communication wire.
, ω nom is the system nominal frequency.P andQ are the active power and reactive power of the DG unit, which can be calculated byP where ω c is the cut-off frequency of the low pass filter.

C. MODIFIED FINITE CONTROL SET MODEL PREDICTIVE CONTROL SCHEME
In order to further reduce the computational burden and to simplify the analysis, this article considers a one-step prediction in the FCS-MPC strategy. At the beginning of each control cycle, the controller simultaneously samples the system state variables and switches the system. In actual application, considering the calculation time and sampling time of the controller cannot be neglected, the FCS-MPC strategy is performed in the following steps to compensate for the delay time. Firstly, the current of the filter inductor i l , the voltage of the filter capacitor V c and the output current I o at kT are measured. And then, the state variables at (k + 1)T with the compensation of the controller delay are predicted according to the output voltage vector V i and Equation (6). Secondly, the output of the LC filter for all candidate voltage vectors in the FCS at the (k + 2)T sampling instant are predicted. Finally, using the proposed cost-function, the predicted results of each voltage vector are evaluated. Then the corresponding switch state will be selected from all candidate voltage vectors, which can minimize the cost function J .
It should be noted that the cost function value represents the deviations between the control result and the object.
In addition, the weights of different control targets in the control process are defined in the cost function, so that users can intuitively optimize the control targets [28]. In this work, the proposed cost function is expressed as follow: (16) where J v is the predicted voltage tracking error, J der is the voltage differential term and J sw is the penalty for switching effort. λ der and λ sw are the corresponding associated weighting factors. It should be noted that simplifying the weighting factors into λ der and λ sw can facilitate analysis and tuning while reducing the amount of calculations in the processor.
In (16), the predicted voltage tracking error term can be defined as follows: In the second-order systems of VSC with LC filter, the tracking of voltage and voltage derivatives allows the system to be effectively controlled, which is more satisfactory than the tracking control effect only by voltage. The capacitor reference voltage is defined as follows: The derivative of the capacitor reference voltage is: The derivative of the capacitor voltage through the dynamic characteristics of the capacitor is: The tracking error of the capacitor voltage differential can be obtained by subtracting Equation (20) from Equation (19). In order to facilitate computer operation, the capacitor C f is multiplied on both sides.
The penalty item for switching J sw can reduce switch frequency which can be calculated by: where S i is the switch configuration of phase A, B and C. 202302 VOLUME 8, 2020

D. FEEDBACK CORRECTION
The measures to reduce the tracking error can be achieved not only by the constraints on the cost function, but also by reducing the prediction error. As shown in Fig. 4, in delay compensation, prediction result with small differences may make a different choice of optimal switching state, such as V pre1 (k + 1) and V pre2 (k + 1). Therefore, reducing the prediction error of delay compensation will effectively reduce the tracking error and the resulting oscillations and harmonics. Compared with the control period, the fundamental periodic error changes slowly, which can be approximated as: where v sam (k) is the sampling value of capacitor voltage at kT; v pre (k) is the predicted value of the capacitor voltage at kT from the last control cycle. Therefore, the correction value of the predicted voltage used for delay compensation can be expressed as: where v pre (k + 1) is the predicted voltage calculated by the discrete model, λ c is the feedback correction coefficient.

E. UNBALANCE COMPENSATION METHOD
When the load is unbalanced, the voltage unbalance can be compensated by reducing the negative sequence voltage. As shown in Fig. 3, the compensation reference can be obtained by the product of the negative reactive power Q − , an unbalance compensation gain (UCG), and the fundamental negative-sequence voltage v − o . With the compensation proceeds, Q − gradually decreases. Also, the compensation efforts will be reduced, helping microgrids to distribute compensation efforts properly. The adjustment of the constant UCG needs to consider the compensation effect of unbalance.

F. SEQUENCE EXTRACTION
For the calculation of the DG output power, the output current of the DG is extracted by using a dual second-order generalized integrator quadrature signal generator (DSOGI-QSG) and the positive-negative-sequence calculation (PNSC) block, which has fast dynamic response and high accuracy. The implementation of sequence extraction is shown in Fig. 5. In Fig. 5, k is the coefficient of the bandwidth of the SOGI-QSG; ω is the resonant frequency. The output I + and I − is the positive and negative sequence current in αβ frame.

IV. SIMULATION RESULTS
The proposed FCS-MPC control strategy is implemented in an islanded ac microgrid consisting of two VSC DG units, which helps simplify the analysis in this study. The microgrid is established for time-domain simulation in MATLAB/Simulink environment. As shown in Fig. 1, an unbalanced load and a nonlinear load are connected to the microgrid via the switches sw1 and sw2, respectively. Thus, the proposed control strategy can be verified under complex load conditions, including nonlinear and unbalanced loads. The main parameters of the islanded ac microgrid are given in Table 2.

A. PERFORMANCE UNDER THE CONDITION OF NONLINEAR LOAD
As shown in Fig. 1, the nonlinear load is a three phases diode rectifier with a parallel RC load connected to the PCC. The comparison of the microgrid performance of the conventional cascaded PI method, the conventional proportional resonant (PR) control method and the proposed modified   In this operating case, two fundamental periods of the capacitor voltage are sampled with a sampling time of 1 µs. Then, in order to identify individual harmonic amplitudes, the signal is processed using the Fast Fourier Transform (FFT) algorithm provided in the MATLAB.
It can be seen from voltage and current waveforms of the DG unit in Fig. 6 that, under the condition of nonlinear load, the microgrid using the conventional cascaded PI control method appears a severe distortion of the output voltage waveform. Moreover, the output current of the DG unit cannot be effectively controlled. The microgrid system loses stability and diverges gradually.
The voltage and current waveforms of the DG unit when using the conventional PR control method and the proposed modified decentralized FCS-MPC control strategy are shown in Fig. 7 and Fig. 8, respectively. Obviously, the output voltage of the DG unit which uses the proposed control strategy presents better sinusoidal. Due to the nonlinear characteristics of the V/I relationship of the diode, the output voltage of the DG based on conventional PR control has significant distortion, and the corresponding total harmonic distortion (THD) is 5.63%. At the same time, relatively good harmonic currents are shared between the DG units. It can be observed from the diagram of the FFT analysis that the output voltage distortion of DG using traditional cascaded PR control is mainly concentrated on odd harmonics.
As shown in Fig. 8, due to the fast response of FCS-MPC and the reduction of the prediction error by the feedback correction method, the proposed modified decentralized FCS-MPC strategy is well capable of achieving excellent suppression effect on each order harmonic. It can be seen from Fig. 8 (c) that when using the proposed modified decentralized FCS-MPC control strategy, the output voltage THD of the DG is 1.63%. Compared with the conventional PR control method shown in Fig. 7 (c), the harmonics can be significantly reduced with the proposed control strategy. The feature in Fig. 8 shows excellent performance of the proposed method under nonlinear load condition.

B. PERFORMANCE UNDER THE CONDITION OF UNBALANCED AND NONLINEAR LOADS
In order to test the validity of the proposed modified decentralized FCS-MPC strategy in a VSC based microgrid with 202304 VOLUME 8, 2020 generalized loads, a nonlinear load and an unbalanced load are connected to the PCC at t = 0 s and t = 1 s, respectively. Thus, sw2 and sw1 switch to the closed states.
The system performance of the microgrid adopting only the conventional PR control method is presented in Fig. 9. As can be observed from Fig. 9, the microgrid system operates stably using the conventional PR control method under only nonlinear load. After an unbalanced load is added at t = 1 s, the voltage and current waveforms of the DG unit start to fluctuate, and the microgrid system is gradually unstable.
In order to improve the system performance under unbalanced and nonlinear loads conditions, the proposed modified decentralized FCS-MPC strategy is adopted, and the response of the voltage, current, fundamental negative sequence active power, reactive power and negative power are shown in Fig. 10 and Fig. 11. The unbalance compensation of the proposed method is activated at t = 2 s. Note that the unbalanced load is 80 . It is obvious that the voltage can still maintain a good sinusoidal shape even after the unbalanced load is connected at t = 1 s. After the unbalance compensation is implemented at t = 2 s, the unbalance current gradually decreases, and the voltage can still maintain an excellent sinusoidal during this period. Furthermore, as shown in Fig. 11, the negative-sequence reactive power is effectively weakened, and it is well shared among different DG units. Compared with the conventional PR control method in Fig. 9, it can be obviously demonstrated that the proposed modified decentralized FCS-MPC strategy provides robust performance for the microgrid under unbalanced and nonlinear loads conditions.

C. PERFORMANCE UNDER THE CONDITION OF STEP LOADS
In order to test the transient performance of the proposed modified decentralized FCS-MPC strategy in a VSC based microgrid with generalized loads, a step load (R = 40 ) are connected to the PCC at t = 2 s. The system response comparisons between the conventional PR control method and the proposed modified decentralized FCS-MPC control strategy are illustrated in Fig. 12 and Fig. 13, respectively. Obviously, even after a step load is connected at t = 2 s, the voltage can still maintain a good sinusoidal shape, and at the same time, the current increases rapidly in response to load changes. However, some distortions generated by the voltage waveform under the PR control method indicate that the transient response of the proposed modified decentralized FCS-MPC strategy is better than the PR control.

D. CONTROL PARAMETERS AND STEADY-STATE ANALYSIS
To evaluate the switching frequency of FCS-MPC, the average switching frequency (ASF) is defined as follows: where T = 0.2 s is the calculation period of the average switching frequency. sw a , sw b , sw c are the switching times of phase A, B and C, respectively in T . In order to determine the optimal values of the key control parameters, the influence of the weighting factors λ der , λ sw and the feedback correction coefficient λ c on the output voltage THD and ASF are studied with simulation results in Figs. 14 and 15. And the value of λ der is varied from 0.1 to 0.6, and λ sw is varied from 1 to 5, when λ c is 0, 0.5, 1, 1.5. As shown in Fig. 14, the output voltage THD decreases as λ der increases, and increases as λ sw increases. The increase of feedback correction coefficient λ c makes the control effect improve under different parameters. Furthermore, it can be seen from Fig. 15 that, the average switching frequency of the inverter increases with the increase of λ der and decreases with the increase of λ sw . Combined with the the simulation results shown in Figs. 14 and 15, the optimal values of the key  control parameters λ der , λ sw and λ c are selected, as presented in Table 2.

E. SENSITIVITY OF THE MODEL PARAMETER VARIATION
In order to evaluate the sensitivity of the model parameters, the performance of the decentralized modified FCS-MPC in the case of capacitance or inductance mismatch was tested. The system output voltage THD with respect to parameter matching was tested when the capacitance or inductance mismatch is ±5%, ±10%, and ±20%. The simulation results are shown in Fig. 16. The decentralized modified FCS-MPC shows the robustness of matching model parameters. When the model parameters are highly mismatched, it will not significantly reduce the control effect. At the same  time, the decentralized modified FCS-MPC has better control effect than conventional FCS-MPC. When the actual circuit parameters are slightly larger than the model parameters, a smaller voltage THD may be achieved. When the actual circuit parameters are slightly smaller than the model parameters, the voltage THD only slightly increases. Thus, it can be seen that the precise match of the model has no significant influence on the control effect.

F. DISCUSSION
The above simulation reflects the excellent control performance of the proposed decentralized modified FCS-MPC strategy under various measures working conditions (including harmonics, unbalance and harmonics, transient step load change and parameter mismatch). The proposed decentralized modified FCS-MPC strategy mainly has the following characteristics: (1) Under harmonic and unbalanced load conditions, the proposed decentralized modified FCS-MPC strategy has a lower THD than conventional PR and conventional FCS-MPC.
(2) The proposed modified decentralized FCS-MPC shows more robustness of modeling parameter changes than conventional FCS-MPC. VOLUME 8, 2020 (3) Under the unbalanced load working state, the proposed modified decentralized FCS-MPC strategy can effectively share and reduce the unbalanced power.

V. REAL-TIME VERIFICATION WITH RT-LAB
In order to further verify the effectiveness of the proposed modified decentralized FCS-MPC strategy, the proposed control strategy implemented in the microgrid is verified with the real-time digital platform OP5600-OPALRT, as shown in Fig. 17. The model which is established with MATLAB/Simulink and OPAL-RT libraries is added to facilitate the real-time implementation. The implemented model in MATLAB/ Simulink is compiled with RT-LAB thus to convert the model in C-language. The loaded model in OPAL-RT is run with a sampling time of 10 us. The input/output ports of OPAL-RT are connected to the fourchannel Tektronix oscilloscope MDO3024. The parameters of the microgrid system are selected to be the same with simulation, as presented in Table 2. The analog signals are shown in Fig. 18 and Fig. 19.   18 shows the voltage and current performance of the DG units using the proposed control strategy under nonlinear load. Only slight voltage distortion is observed due to the active harmonic control capability introduced by the feedback correction in the modified decentralized FCS-MPC approach. In spite of the severe nonlinear characteristics caused by the diode, the modified decentralized FCS-MPC strategy is well able to obtain a remarkable harmonic control effect. This is consistent with the simulated results in Fig. 8.
The performance of the proposed modified decentralized FCS-MPC strategy is also tested under the complex conditions of unbalanced and nonlinear loads, and the results of the output voltage, current and the negative-sequence reactive power are depicted in Fig. 19. It can be clearly observed that the voltage waveform can still maintain a sinusoidal shape. Moreover, the introduced unbalance compensation method is able to effectively suppress the negative-sequence reactive power under unbalanced loads, as shown in Fig. 19 (b). Also, these characteristics agree with the results shown in Fig. 10.

VI. CONCLUSION
This article proposes a modified decentralized finite control set model predictive control strategy of prosumers integrated microgrids for unbalanced and harmonic power management. The proposed strategy is developed to enhance the voltage quality of the microgrids with reduced the total harmonic distortion and the load unbalance conditions. Based on the FCS-MPC strategy of the two-stage three-phase VSC, the control strategy is modified by feedback correction to suppress the harmonics of the output voltage using only one-step prediction. Compared with conventional FCS-MPC, the average switching frequency, parameter adaptability and voltage harmonics are improved. In addition, an unbalanced compensation scheme was introduced in the decentralized FCS-MPC control architecture to mitigate the impact of unbalanced loads on the DG in the microgrid. Negative sequence reactive power is well shared, and power distribution between DERs is improved. Comprehensive simulation and real-time verification with RT-Lab from a VSC microgrid have been presented to verify the effectiveness of the proposed control strategy. The obtained results show that the unbalance and harmonic are well compensated and controlled by using the proposed control technique, and the compensation effort is properly shared between the prosumers' DERs. In comparison with the conventional control method, it can be demonstrated that the proposed modified decentralized FCS-MPC strategy provides satisfactory and robust performance for practical microgrids under unbalanced and nonlinear loads conditions.