Comparison study between NPWM and NSVPWM strategy in FSMC control of stator reactive and active powers control of a DFIG-based wind turbine system

Received May 28, 2019 Revised Jan 22, 2020 Accepted Mar 10, 2020 In this work, we present a comparative study between neural space vector pulse width modulation (NSVPWM) and neural pulse width modulation (NPWM) technique in fuzzy-sliding mode control (FSMC) of stator active and stator reactive power control of a doubly fed induction generator (DFIG) for wind energy conversion systems (WECSs). Two strategies approach using FSMC-NSVPWM and FSMC-NPWM are proposed and compared. The validity of the proposed strategies is verified by simulation tests of a DFIG (1.5MW). The reactive power, electromagnetic torque, rotor current and stator active power is determined and compared in the above strategies. The obtained results showed that the proposed FSMC with NSVPWM strategy has stator reactive and active power with low powers ripples and low rotor current harmonic distortion than NPWM strategy.


INTRODUCTION
Traditionally, pulse width modulation (PWM) is most popular technique used in the AC machine drives. The PWM strategy is simple and easy to implement [1]. But, this technique gives more total harmonic distortion (THD) of current and voltage output [2]. This strategy is unable to fully utilize the available DC bus supply voltage to the VSI [3]. In order to overcome the drawbacks of the traditional PWM technique, space vector pulse width modulation (SVPWM) strategy has been presented [4][5][6]. This strategy based on the principles of space vectors and need to calculate of angle and sector [7]. This strategy gives 15% more voltage output compare to the conventional PWM method, thereby increasing the DC bus utilization [8]. On the other hand, this strategy reduces the THD value of current/voltage compared to the PWM strategy.
In [9], the author has proposed a new SVPWM strategy, this strategy based on calculating minimum (Min) and maximum (Max) of three-phase voltages. SVPWM and artificial neural networks (ANNs) controller are combined to control DFIG-based wind turbine [10]. Fuzzy SVPWM (FSVPWM) is proposed to regulate the stator reactive and active power of the DFIG [11]. In [12], a reactive and active power proportional-integral (PI) controllers and three-level FSVPWM strategy were combined to regulate the electromagnetic torque and current of the DFIG. In [13], a three-level neural SVPWM strategy of a DFIG was presented. In [14], four-level SVPWM based on FLC control to regulate the active and reactive power of the DFIG.
Field oriented control (FOC) using PI controllers are the traditional strategy used for DFIG. In [15], FOC control is the most popular technique used in the DFIG-based wind energy conversion system.

NEURAL PULSE WIDTH MODULATION
Using the traditional PWM strategy provides many advantages, such as simple modulation and easy implementation. On the other hand, the major problem of the traditional PWM strategy is the harmonic distortion of voltage and current caused by the hysteresis comparators. The traditional PWM strategy designed to control the two-level inverter is illustrated in Figure 1. To reduce the harmonic distortion of stator current and voltages of DFIG, we have applied the NPWM strategy. The principle of the NPWM strategy is similar to traditional PWM strategy. The difference results in using a neural networks controller to replace the hysteresis comparators. As shown in Figure 2. This strategy is simple modulation scheme and easy to implement. This proposed strategy reduces the harmonic distortion of stator current and gives minimum power ripples of the DFIG-based wind turbine. The training used is that of the algorithm, gradient descent with momentum and adaptive LR (ALR). The convergence of the network in summer obtained by using the value of the parameters grouped in Table 1. The block diagram of the ANN controllers is shown in Figure 3. The construction of layer 1 and layer 2 is shown in Figures 4 and 5 respectively.    Figure 6 show the principle of the two-level SVPWM strategy. This strategy based on calculating minimum and maximum of three-phase voltages [29]. This strategy not needed to calculated of the angle and sector compared with traditional SVPWM technique. This strategy is detailed in [30]. On the other hand, this strategy is simple and easy to implement compared to conventional SVPWM technique. The block diagram of the hysteresis comparators is shown in Figure 7. The principle disadvantages of the SVPWM strategy is that the harmonic distortion of stator current and voltages. In order to overcome the drawbacks of the SVPWM strategy a complimentary use of the ANN controller is proposed. However, the ANN controller contains a hidden layer, input layer and output layer [31]. Recent years, ANN controller has found many applications such as control AC machines. . The principle of the two-level NSVPWM technique is similar to two-level SVPWM strategy. The difference results in using the ANN controllers to replace the hysteresis comparators [29]. As shown in Figure 8. This proposed strategy is simple control and easy to implement. This proposed strategy reduces the harmonic distortion of stator current and voltages. The training used is that of the retro propagation of Levenberg-Marquardt (LM). The parameters of the LM algorithm are shown in Table 2, the block diagram of the neural hysteresis comparators is shown in Figure 9, and the block diagram of the ANN controllers is shown in Figure 10.

FUZZY SLIDING MODE CONTROL
Sliding mode control is one of the most popular control in use nowadays. This strategy control based on theory of variable structure systems [33]. This strategy was proposed by Utkin in 1977 [34]. However, the SMC is a strategy to adjust feedback by previously defining a surface. Since the robustness is the best advantage of the SMC control, it has been widely employed to control nonlinear systems that have model uncertainty and external disturbance [35]. On the other hand, the principle disadvantages of the SMC technique is that the chattering effect created by the discontinuous part of control [36].
Fuzzy logic is a technique based on engineering experience and observations. In FLC, an exact mathematical model is not necessary because linguistic variables are used to divine system behavior rapidly [37]. On way to improve SMC performance is to combine it with FLC controller to form a FSMC strategy. The design of a SMC incorporating FLC control helps in achieving minimized chattering, reduces the harmonic distortion of stator current, simple rule base, and robustness against disturbances and nonlinearities. The FSMC strategy is a modification of the SMC technique, where the switching controller term Sat(S(x)), has been replaced by a FLC control input as given by (6).
The proposed FSMC control with NSVPWM strategy, which is designed to control the stator reactive and stator active powers of the DFIG-based WTSs, is shown in Figure 11. The internal structure of FSMC control is shown in Figure 12. Membership functions in triangular shape are shown in Figure 13. The rule bases of the FSMC control are shown in the Table 3. The properties of our controller are given in the Table 4. The block diagram of the FLC is illustrated in Figure 14.   Figure 14. Block diagram of the FLC

RESULTS AND DISCUSSION
Simulations of the proposed control techniques for a DFIG-based wind turbine are conducted by using the MATLAB/Simulink software. The DFIG is rated at 1.5 MW and its parameters are listed in Table 5. The parameters of the proposed FSMC control are selected to provide optimum performance as K1=0.0003, K2=0.00005, and K3=0.02. The proposed FSMC control with two-level NPWM technique, which is designed to control the DFIG-based WTSs, is shown in Figure 15. The proposed strategies will be tested and compared in two different configurations: robustness against parameters variations and reference tracking.

CONCLUSION
This paper presents simulation of FSMC strategy for reactive and active power control of a DFIGbased wind turbine systems, using the modulation technique of the two-level NPWM and NSVPWM strategy. With results obtained from simulation, it is clear that for the same operation condition, the 1.5MW DFIG with FSMC control using two-level NSVPWM strategy had better performance than the two-level NPWM strategy and that is clear in the ripples of active and reactive powers which the use of the two-level NSVPWM strategy, it is minimized of powers ripples more than two-level NPWM strategy.

APPENDIX Design of SMC control
The SMC control does not need accurate mathematical models like classical controllers. We choose the error between the reference stator energies and measured as sliding mode surfaces, so we can write the following expression: The equivalent control vector V eq can express by: To obtain good performances, dynamic and a commutation around the surface, the control vector is imposed as follows: V n dq is the saturation function defined by: where K determine the ability of overcoming the chattering. The SMC will exist only if the following condition is met: The block diagram of the conventional SMC strategy is shown in Figure 30.