A Genetic Algorithm-Based Low Voltage Ride-Through ... - IEEE Xplore

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO. 3, MAY 2014

1325

A Genetic Algorithm-Based Low Voltage Ride-Through Control Strategy for Grid Connected Doubly Fed Induction Wind Generators Theodoros D. Vrionis, Xanthi I. Koutiva, and Nicholas. A. Vovos, Senior Member, IEEE

Abstract—This paper proposes a new computational intelligence-based control strategy, to enhance the low voltage ride-through capability of grid-connected wind turbines (WTs) with doubly fed induction generators (DFIGs). Grid codes world-wide require that WTs should supply reactive power to the grid during and after the fault, in order to support the grid voltage. The conventional crowbar-based systems that were initially applied in order to protect the rotor-side converter at the occurrence of grid faults, do not fulfill this requirement, as during the connection of the crowbar, the DFIG behaves as a squirrel cage machine, absorbing reactive power from the grid. This drawback led to the design of control systems that eliminate or even avoid the use of the crowbar. In order to conform to the above-mentioned requirement, this paper proposes a coordinated control strategy of the DFIG converters during a grid fault, managing to ride-through the fault without the use of any auxiliary hardware. The coordination of the two controllers is achieved via a fuzzy controller which is properly tuned using genetic algorithms. To validate the proposed control strategy, a case study of a 1.5-MW DFIG supplying a relatively weak electrical system is carried out by simulation. Index Terms—Doubly fed induction generator (DFIG), fuzzy control, genetic algorithms (GAs), low voltage ride through (LVRT), power systems faults, wind power generation.

I. INTRODUCTION

O

VER the last few years, doubly fed induction generators (DFIGs) have dominated the largest world market share of wind turbines (WTs), as an alternative concept to conventional variable speed generators [1]. A DFIG consists of a wound rotor induction generator with its stator windings directly connected to the grid and its rotor windings connected to the grid via an arrangement of two ac/dc back-to-back converters [2]. As the converters are sized for only the one third of the rated power of the turbine, this topology accomplishes a cost-effective, decoupled control of the active and reactive power [3], [4]. However, DFIGs present a major drawback concerning their operation during grid faults. The voltage drop at the stator windings, due to grid faults, results in a sudden change

Manuscript received June 03, 2013; revised September 30, 2013; accepted November 03, 2013. Date of publication December 11, 2013; date of current version April 16, 2014. Paper no. TPWRS-00713-2013. The authors are with the Department of Electrical and Computer Engineering, University of Patras, Patras, Greece (e-mail: [email protected]; [email protected]; [email protected]). Digital Object Identifier 10.1109/TPWRS.2013.2290622

in the stator flux of the DFIG, which finally leads to an overcurrent to the rotor windings, due to the magnetic coupling. This overcurrent may cause severe damage to the semiconductors of the rotor side converter and large fluctuations of the dc-link voltage [5], [6]. When DFIGs started being used in WTs, the penetration of WTs to the power system was relatively low, so their control concerning faults focused on the protection of the DFIGs themselves and no special actions were taken in order to provide the DFIGs with the capability of contributing to network support. Under this concept, in order to protect the generator, its rotor windings were protected with a crowbar circuit [7]–[9]. This device consists of a bank of resistors, which is connected to the rotor windings through power electronic devices. When a fault is detected, the rotor windings are connected to the crowbar resistors, whereas the rotor-side converter is temporary disabled. Thus, the short-circuit current flows through the crowbar instead of the rotor-side converter. With this solution, the machine is effectively protected, but due to the fact that the blocking of the rotor-side converter leads to the partial loss of power control during the crowbar action, large transients are generated after the fault, which may lead to the disconnection of the machines from the grid. In addition, during the activation of the crowbar, the DFIG is converted to a conventional squirrel-cage induction generator, absorbing a large amount of reactive power from the grid [18], [19]. Nowadays, due to the fact that WTs represent a significant part of the total generation in electrical systems, system operators worldwide have revised their grid codes (GCs), making the requirements concerning the fault ride-through (FRT) capability of WTs more stringent. The majority of the GCs require that the WTs should provide low voltage ride-through (LVRT) capability for grid faults resulting in a 85% voltage drop or even more. This means that they should stay connected to the grid during and after grid faults, contributing to the system stability. Moreover, they should supply reactive power to the grid in order to support the voltage recovery [10]–[13]. This requirement cannot be fulfilled by DFIGs protected with a conventional crowbar circuit, as they cannot generate reactive power during the activation of the crowbar. For this reason, researchers started addressing the issue of the FRT of the DFIGs from several other points of view. For instance, [14] proposes an improved version of the crowbar circuit, eliminating the duration of the crowbar action. Reference [15] investigates the application of a STATCOM to achieve the uninterrupted operation of a wind turbine equipped

0885-8950 © 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

1326

with a DFIG during grid faults. In [16], a control strategy using a series grid-side converter is proposed. Although these last two arrangements are promising in some cases, the complexity and the additional cost impair their applicability. Other papers propose some methods for FRT without any auxiliary external devices [17]–[27]. They show that if the DFIG controllers are suitably designed, it is possible to limit the rotor overcurrent. Xiang et al. in [17] propose a control strategy that orients the rotor current to counteract the dc and negative sequence components of stator flux linkage. Experimental results show that under certain circumstances, the rotor current is limited; therefore, the use of the crowbar can be avoided. However, this method has high dependence on the estimation of stator flux linkage and the exact knowledge of the parameters of the DFIG. In [18] is proposed a flux linkage tracking-based control strategy, to suppress the fault rotor current. Its control objective is to keep the rotor flux linkage close to the stator flux linkage instead of maintaining a certain value accurately. In this way, there is no need to decouple the dc and negative sequence components, which means the control scheme does not depend on the exact knowledge of the system parameters. Although the proposed control method is promising, the paper could have studied a bigger voltage dip for the case of the three-phase fault, as it is imposed by international grid codes. In [19], Hu et al. propose the use of a virtual resistance in combination with demagnetization control to limit the rotor side overcurrents. This method manages to enlarge the control range in relation to the control method proposed in [17], but the drawbacks that were met in [17] still cannot be avoided. The study described in [20] suggests that a properly designed fuzzy controller (FC) presents a better performance in presence of variations of parameters and external disturbances than a traditional proportional integral (PI) controller. Comparative results between the two controllers showed that the FC manages to limit the generator currents during the fault, avoiding the use of the crowbars. However, it would be interesting to see which would be the response of this control system and what changes should possibly be done, for the case where the DFIG would be connected to a weaker bus, instead of an infinite bus. Reference [21] proposes a control strategy based on genetic algorithms (GA) for the acquisition of the optimal gains of the PI controllers to the rotor-side converter of the DFIG. The GA fitness function is defined with the objective to reduce the over-current in the rotor circuit, in order to maintain the converter in operation during the fault period. Simulation results show that, in some cases, if the GA gain adjustment is used, the use of the crowbar protection-scheme can be avoided. However, in this work, no analytical information is given about the electrical grid connected to the DFIG. In addition, it is tested for a small voltage dip in relation to the voltage dip described in most grid codes. Papers [22]–[27] propose some methods for FRT without any auxiliary external hardware for the case of asymmetrical grid faults. Based on the experience gained from the studies published so far, this paper tries to extend the concept of the protection of the DFIG without the use of additional hardware, in more demanding situations, such as the connection of a relatively weak electrical system with the DFIG in the case of a fault resulting

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO. 3, MAY 2014

Fig. 1. Configuration of the DFIG WT.

in a bigger voltage dip, as it is required from the GCs. The proposed control scheme contributes to the optimal coordination of the two converters, aiming to attenuate the disturbances to the system caused by the fault and ensure system stability. In order to encounter the difficulties met due to the uncertainties of the system modeling and considering the nonlinearity of the system, the controllers were designed based on fuzzy logic (FL) and GAs, which are more efficient in such cases. By this concept, the overcurrents at the rotor windings and the dc side overvoltages are effectively eliminated. In addition, the FRT requirement concerning the reactive power supply is fulfilled. In this point, it would be necessary to point out that many previous papers have tested their proposed systems using small scale test equipment that had a rotor voltage rating far too high, when compared with the megawatt range WT generators that have stator to rotor winding turns ratio around 2.5 to 3 [5], [28]. Thus, the results have been unrealistically good. In this paper, the system is simulated choosing a ratio within the above-mentioned range, in order to simulate the electrical system more properly. The paper is organized as follows: In Section II, the dynamic model of the DFIG and the description of its control system are analyzed. The validation of the proposed scheme is performed by simulations in Section III. Finally, the conclusions are summarized in Section IV. II. MODELING AND CONTROL OF THE DFIG WT SYSTEM The schematic diagram of a grid connected DFIG WT is shown in Fig. 1. The WT is connected to the DFIG through a mechanical shaft system, which consists of a low speed turbine shaft connected to the high speed generator shaft via a gearbox. The DFIG consists of a wound rotor induction generator with its stator windings directly connected to the grid and its rotor windings connected to the grid via an arrangement of two ac/dc back-to-back converters, as depicted in Fig. 1. The rotor side converter (RSC) and the grid side converter (GSC) are pulse width modulated (PWM), IGBT voltage-source converters (VSCs). A. Modeling of the DFIG A synchronously rotating d-q reference frame has been selected to model the dynamic behavior of the DFIG. Considering the generator convention, the stator and rotor voltages are given by the following equations: (1)

VRIONIS et al.: GENETIC ALGORITHM-BASED LOW VOLTAGE RIDE-THROUGH CONTROL STRATEGY

1327

(2) (3) (4) where and are the stator and rotor flux linkages; and are the stator voltages and currents; and are the rotor voltages and currents; and are the stator and rotor resistances; and are the stator and rotor angular frequencies, respectively. The indices and indicate the direct and quadrature axis components of the reference frame. The flux linkage in (1)–(4) is defined as (5) (6)

Fig. 2. RSC control system.

(7) (8) where and are the stator, rotor, and magnetizing inductances, respectively. In the following paragraph is cited a brief description of a conventional control system, where no special provisions are taken for the FRT of the DFIG. This system would normally be combined with an auxiliary hardware such as a crowbar or a STATCOM, presenting the drawbacks mentioned in Section I. The reason why it is presented here is to highlight its main differences from the proposed control system, which is presented just afterwards. B. Description of the Conventional Control System The conventional DFIG control system, presented in several papers [28]–[30], is divided to RSC control and GSC control. Its basic principles are described in this paragraph. The control system of RSC is shown in Fig. 2. Its objective is to independently regulate the stator active and reactive power, and , respectively. In order to achieve decoupled control of and , the rotor current is transformed to components, and using a reference frame oriented to the stator-flux. The -axis current component is used to control the stator active power . The reference value of the active power is obtained using a maximum power tracking (MPPT) technique [31], [32]. The measured value of is subtracted from and the error is driven to the power controller. The output of this controller is the reference value of the -axis rotor current . This signal is compared to the actual value of and the error is passed through the current controller whose output is the reference voltage for the -axis component . The reactive-power control of RSC can be tuned to keep the stator voltage within the desired range, when the DFIG feeds into a weak power system without any local reactive compensation. When the DFIG feeds into a strong power system, the command of can be simply set to zero. In this paper, the case of the DFIG which feeds a weak ac grid is studied; therefore, ac voltage control is used instead of reactive power control. The actual voltage at the generator terminals is compared to its reference value and the error is passed through the ac

Fig. 3. GSC control system.

voltage controller to generate the reference signal for the -axis current . This signal is compared to the -axis current value and the error is sent to the current controller, which determines the reference voltage for the -axis component . The signals and are transformed back to abc quantities which are used by the PWM module to generate the IGBT gate control signals to drive the RSC. The objective of the GSC control system is mainly to keep the dc-link voltage constant. In this paper, it is designed to be reactive neutral by setting . This setting is chosen considering that the converter is rated for only 30% of the DFIG rating, and that it is primarily used in order to supply active power to the grid. The control system of GSC is shown in Fig. 3. For the transformation of the measured instantaneous signals to quantities, a reference frame oriented to the stator voltage is used. As shown in Fig. 3, the dc voltage is controlled through the signal and the reactive power is controlled through the signal . C. Description of the Optimized Control System In the above-described control system, no provisions are taken for the FRT of the DFIG. In this paragraph, a modification of this system is attempted, in order to ride-through the fault

1328

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO. 3, MAY 2014

TABLE I RULES

Fig. 5. Structure of a chromosome.

Fig. 4. Optimized control system.

without any additional hardware. The proposed control scheme, achieving an optimal coordination between the two converters, manages to attenuate the system disturbances caused by the fault, even in the case where the WT feeds a relatively weak ac grid. As the control system has to act efficiently in a very short period of time, it should be insensitive to the measurement noise and to the lack of accurate information concerning the machine parameters. In order to encounter these difficulties and considering the nonlinearity of the system, the controllers were designed based on computational intelligence (CI), which is more efficient in such cases. More precisely, all the controllers used are FCs whereas the tuning of the FC which achieves the FRT, , is realized using GAs. The reason why a is that the GA-based approach is used for the tuning of derivation of its rules is quite complicated and it cannot come up from simple fuzzy reasoning. As shown in Fig. 4, the optimized control system is a modification of the RSC control system, accomplished by adding the block “Fault Detection and Confrontation System” (FDCS). This block is active only when the ac voltage, , deviates more than 10% from its reference value. The control of the GSC remains unchanged. The concept of the control strategy is analyzed below: in order to successfully protect the DFIG, there are two major issues that should be addressed properly: the rotor overcurrent and the dc-link overvoltage [19]. Neither the dc voltage nor the rotor current should exceed their acceptable limits during the restoring period. The amount of the extra energy that is induced to the rotor during the transient must be properly “pumped” through the converters to the grid, in order to bring the values of the rotor current and the dc voltage back to their normal values. The problem that rises when trying to do this is the following: if the value of the rotor current is sharply dropped by quickly “pumping” the stored energy in the rotor to the grid, the value of the dc voltage will rise suddenly, risking exceeding its limits. Respectively, if the value of rotor current

is slowly reduced in order to avoid the dc overvoltage, there is a high risk that it reaches unacceptable values. Therefore, the correction signals of the rotor current should also take into account the respective values of the dc voltage, in order to achieve a successful FRT. As depicted in Fig. 4, the output of the current controller is corrected by a quantity derived by a FC, . The inputs of , , and are given by (9) (10) Equations (9)–(10), where the indicator ss stands for the steady state value just before the dip and the indicator mv stands for the maximum acceptable value, imposed by the manufacturer. is the rotor rms current (the maximum value of the three phases). In order to equally participate to the modulation of the deviations of the two quantities from their steady state values are divided by their maximum acceptable deviations. It should also be mentioned that only the positive deviations are taken into account to the modulation of . Negative deviations are taken as zeros. This contributes to a smoother transition from the FDCS to the steady state control system. The rules of are given in Table I. Three subsets are used for both inputs: Small (S), Medium (M), and Big (B), whereas five subsets are used for the output : OK, Small Positive (SP), Big Positive (BP), Small Negative (SN), and Big Negative (BN). Triangular membership functions (MFs) are selected for all fuzzy sets. As it was mentioned above, a GA-based approach is used to tune the MFs of . This approach is described above. Before the GA is applied, the optimization problem is converted to a suitably described function, called fitness function, which represents the performance of the solution to the problem. Each set of variables for the given problem is encoded into a binary bit string, called chromosome. Such a string is made up of sub-strings, called genes, which correspond to each different variable. Several chromosomes representing different solutions comprise the population. In this case, 20 chromosomes each consisting of 18 genes are generated as the initial population, Fig. 5. The first 16 genes are parameters related to the shape and the position of the membership functions whereas the rest

VRIONIS et al.: GENETIC ALGORITHM-BASED LOW VOLTAGE RIDE-THROUGH CONTROL STRATEGY

Fig. 9. Fig. 6. Tuning of the input

output of

1329

after the GA optimization.

MFs.

Fig. 10. Electrical system studied.

Fig. 7. Tuning of the input

MFs.

The selected fitness function, which is to be minimized, is given by (11), where the indicator max denotes the maximum value of the signal during the whole restoring period: (11)

Fig. 8. Tuning of output

MFs.

of the genes are related to the range of the output. In Figs. 6–8 is shown the impact of the selected genes to the MFs. The GA starts the evaluation of each chromosome’s fitness. The repopulation of the next generation is done using three operators: reproduction, crossover, and mutation. Through reproduction, strings with high fitness (i.e., low value of the fitness function) receive multiple copies in the next generation while strings with low fitness receive fewer or even none at all. Crossover refers to taking a string, splitting it into two parts at a randomly generated crossover point and recombining it with another string which has also been split at the same crossover point. This procedure serves to promote change in the best strings which could give them even higher fitness. Mutation is the random alteration of a bit in the string which assists in keeping diversity in the population. Finally, the new population replaces the old (initial) one. This procedure continues until a specified termination condition is reached.

The reasons why (11) is chosen as the fitness function to the problem of the FRT of the DFIG are the following: 1) Usually, an integral function is chosen as a fitness function in a wide range of problems to be solved. When an integral is used, the target is the overall behavior of the system in a time interval. On the contrary, in this case, the target is to limit the instantaneous values of and in order to avoid the tripping of the DFIG. Therefore, the selected function does not include any integral, but a sum of the squared values of the quantities that have to be minimized. 2) The target is not just a minimization of the two quantities but the specific “balance” between them in order to retain both quantities below the acceptable values. This is the reason why the squares of the two quantities are used. 3) The maximum deviations of and from their steady state values during the whole restoring period are divided by their maximum acceptable deviations from their steady state values, in order to use their normalized quantities in the fitness function. A 3-D graph for the output as a function of and , after the GA optimization, is shown in Fig. 9. The concept of the proposed control system could be used in several sizes of machines. The rules of the fuzzy controller, , and the process of their optimization would be the same. The only variation would be in the maximum values imposed by the manufacturer, used in the fitness function.

1330

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO. 3, MAY 2014

Fig. 11. Response of the system without the “fault detection and confrontation system”-wind speed 12 m/s. (a) Three-phase voltage at the PCC. (b) dc-link voltage. . (c) Rotor current. (d) Stator current. (e) WT output active power. (f) WT output reactive power. (g) Rotor speed. (h) Signal

III. SIMULATION RESULTS To validate the effectiveness of the proposed control strategy, a case study of a 1.5-MW DFIG supplying a relatively weak electrical system (with short circuit ratio equal to 2.23) is carried out by simulation. The layout of the electrical system studied is depicted in Fig. 10. Apart from the response of the proposed control system, the response of the conventional control system is also cited in this paragraph. This is not done in order to compare

the effectiveness of the two systems, as a conventional control system would also require an auxiliary system to ride-through the fault, presenting the drawbacks mentioned in Section I. The reason why the responses of the two systems are cited together is to show that with the proposed modification of the conventional control system, the DFIG manages to ameliorate its overall response during the fault and post-fault periods and to successfully ride through the fault, without the use of any auxiliary hardware. The paper concentrates on three-phase symmetrical grid

VRIONIS et al.: GENETIC ALGORITHM-BASED LOW VOLTAGE RIDE-THROUGH CONTROL STRATEGY

1331

Fig. 12. Response of the system using the “fault detection and confrontation system”-wind speed 12 m/s. (a) Three-phase stator voltage. (b) dc-link voltage. . (j) Signal . (c) Rotor current. (d) Stator current. (e) WT output active power. (f) WT output reactive power. (g) Rotor speed. (h) Pitch angle. (i) Signal

faults, since the term “fault-ride-through capability” of national grid codes refer to this type of faults.

In the case study a three-phase fault takes place at , resulting in a 85% depth of voltage dip at the PCC, as imposed

1332

by the Greek Grid Code and several other grid codes around the world. After 80 ms, the reclosers open to disconnect the faulted line and reclose after 200 ms. The technical characteristics of these reclosers are those of the custom reclosers used in the Greek national grid. The parameters of the electrical system are cited in the Appendix. The system has been modeled and simulated in the Simulink toolbox extension of MATLAB. The modeling of the FCs and their tuning via GA were realized using MATLAB programming. A detailed converter model is used to model the PWM. The response of the system using the conventional control system is shown in Fig. 11(a)–(h), whereas its response using the proposed control system is depicted in Fig. 12(a)–(j). As illustrated in Fig. 11(a)–(h), when the conventional system is used, the dc voltage fluctuates intensively above its maximum acceptable limit of 125% of its rated value, endangering to destroy the dc capacitor. Furthermore, the rotor current exceeds its maximum acceptable rating (100% higher than the continuous current rating) which is imposed by the IGBTs of the RSC. Finally, the overall response of the system is quite fluctuating, submitting the drive train to great stress. Consequently, this system needs a proper modification, in order to ride-through the fault and ameliorate its performance during the fault and post-fault period. On the contrary, as depicted in Fig. 12(a)–(j), when the “Fault Detection and Confrontation System” is used, all the above fluctuations are alleviated and the system reaches sooner its steady state. Through the optimal coordination of the two converters, the FRT of the DFIG is achieved, as the overcurrents at the rotor windings and the dc-link overvoltages are effectively constrained below their maximum acceptable values. In addition, the dc link fluctuations are effectively attenuated, so the stressing and the possible destruction of the dc capacitor are avoided. Fig. 12(f) shows the WT output reactive power. As mentioned in the introduction, using the proposed method, the rotor side converter is not disabled during the fault, so it is possible to supply reactive power to the grid in order to support the voltage recovery, as imposed by several grid codes. In the case studied, the grid voltage, being supported by the proposed control system, recovers quickly, so there is no need for large amounts of reactive power from the DFIG after the fault. However, it is obvious that while the ac voltage remains below its reference value, the DFIG delivers the required amount of reactive power to the grid, in order to support the grid voltage. During the fault, the voltage drops, so the power from the wind that is not dissipated to the grid is stored as kinetic energy to the rotor. The rotor of the WT, having a big mass, has a high capability to store energy. Due to this fact, its speed rises. This can be depicted by Fig. 12(g), where the rotor speed is illustrated. When the fault duration ends, this energy is sent to the grid as electrical power, so the rotor speed gradually turns back to its pre-fault value. The pitch angle response is also cited, Fig. 12(h). As it was expected, due to the large inertia of the WT, the pitch angle control cannot react in the limited time interval of the fault duration. The pitch angle increases by less than 1 degree. So, it could be

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO. 3, MAY 2014

Fig. 13. Response of the system using the “fault detection and confrontation system”- wind speed 10 m/s. (a) Three-phase stator voltage. (b) dc-link voltage. (c) Rotor current. (d) WT output active power. (e) WT output reactive power.

considered that it practically does not participate at all to the FRT process.

VRIONIS et al.: GENETIC ALGORITHM-BASED LOW VOLTAGE RIDE-THROUGH CONTROL STRATEGY

TABLE II PARAMETERS OF THE 1.5 MW DFIG SYSTEM

Finally, in Figs. 11(h) and 12(i) is shown the q-component of the rotor voltage, , whereas in Fig. 12(j) is illustrated the signal , which is the signal modified by the “Fault Detection and Confrontation System”. At , a transient behavior can be observed. This transient is caused by the steady state control system, not by the Fault Detection and Confrontation System. In this work, no effort was made in order to optimize the steady state control system, as the subject of the paper emphasizes only on the confrontation of the faults. In fact, the steady state controller induces this transient at the moment where the ac voltage, which has started dropping gradually after the post fault period, goes below its reference value. This controller takes a very quick action in order to bring the ac voltage back to its reference value, causing the above-mentioned transient. Probably this transient would be avoided, if the controller had been optimized. However, this optimization gets off the subject of the proposed control system. The fitness function described in (11) is not optimized just for the case shown in Fig. 12, but for a variety of fault cases leading to 85% voltage dip. As mentioned above, this limit is imposed by the Greek Grid Code and coincides with the voltage limit of several other grid codes around the world. For the case of a bigger voltage dip, the DFIG is allowed to disconnect. This is why we have only studied cases that lead to 85% voltage dip. This is the worst-case scenario while being connected to the grid. Although, in order to prove that the system is also effective under other operating conditions, we cite a case study where a fault takes place under 10 m/s wind speed (which corresponds to 1 MW output power) and leads to 85% voltage dip. Apparently, the selected fault resistance which is used by the simulation is now different from the previous case, in order to cause the same dip. As shown in Fig. 13, in this case too, the proposed control system manages to effectively ride-through the fault. IV. CONCLUSIONS This paper proposes a control strategy to improve the LVRT capability of grid connected DFIG WTs without the need of any auxiliary hardware. Its basic idea is the optimal coordination of the DFIC converters through a fuzzy controller, which is designed using genetic algorithms. The simulation results show

1333

TABLE III PARAMETERS OF THE AC GRID

that using the proposed control system, the DFIG can successfully ride-through the fault, even in the case where the WT feeds a relatively weak ac grid. The overcurrents at the rotor windings and the dc side overvoltages are effectively eliminated and the DFIG can continuously supply the electrical system with reactive power during and after the fault, contributing to the support of the ac voltage. APPENDIX Tables II and III show the parameters of the DFIG system and the ac grid system used in the simulation tests, respectively. REFERENCES [1] A. D. Hansen, F. Iov, F. Blaabjerg, and L. H. Hansen, “Review of contemporary wind turbine concepts and their market penetration,” Wind Eng., vol. 28, no. 3, pp. 247–263, Jul. 2009. [2] T. Ackermann, Wind Power in Power Systems. New York, NY, USA: Wiley, 2005. [3] J. A. Baroudi, V. Dinavahi, and A. Knight, “A review of power converter topologies for wind generators,” in Renewable Energy. New York, NY, USA: Elsevier, 2007, pp. 2269–2385. [4] A. Peterson, “Analysis, modeling and control of doubly fed induction generators for wind turbines,” Ph.D. dissertation, Chalmers Univ. Technol., Goteborg, Sweden, 2005. [5] J. Lopez, P. Sanchis, X. Roboam, and L. Marroyo, “Dynamic behavior of the doubly fed induction generator during three-phase voltage dips,” IEEE Trans. Energy Convers., vol. 22, no. 3, pp. 709–717, Sep. 2007. [6] F. Lima, A. Luna, P. Rodriguez, E. Watanabe, and F. Blaabjerg, “Rotor voltage dynamics in the doubly fed induction generator during grid faults,” IEEE Trans. Power Electron., vol. 25, no. 1, pp. 118–130, Jan. 2010. [7] V. Akhmatov, Induction Generators for Wind Power. Brentwood, CA, USA: Multi-science, 2005. [8] S. Seman, J. Niiranen, S. Kanerva, A. Arkkio, and J. Saitz, “Performance study of a doubly fed wind-power induction generator under network disturbances,” IEEE Trans. Energy Convers, vol. 21, no. 4, pp. 883–890, Dec. 2006. [9] M. Kayikci and J. V. Milanovic, “Assessing transient response of DFIG based wind plants the influence of model simplifications and parameters,” IEEE Trans. Power Syst., vol. 23, no. 2, pp. 545–554, May 2008.

1334

[10] E.ON Netz GmbH, Bayreuth, Requirements for Offshore Grid Connections in the E.ON Netz Network, Updated: Apr. 1, 2008 [Online]. Available: http://www.eon-netz.com. [11] J. Kabouris and F. D. Kanellos, “Impacts of Large Scale Wind Penetration on Energy Supply Industry, Energies 2009, 2, 1031-1041; doi:10. 3390/en20401031,” [Online]. Available: available: http://www.mdpi. com/1996-1073/2/4/1031. [12] M. Tsili and S. Papathanassiou, “A review of grid code technical requirements for wind farms,” IET Renew. Power Gen., vol. 3, no. 3, pp. 308–332, 2009. [13] W. Christiansen and D. T. Johnsen, Analysis of Requirements in Selected Grid Codes, Technical Report, Section of Electric Power Engineering, Technical Univ. Denmark (DTU), 2006. [14] J. Lopez, E. Gubia, E. Olea, J. Ruiz, and L. Marroyo, “Ride through of wind turbines with doubly fed induction generator under symmetrical voltage dips,” IEEE Trans. Ind. Electron., vol. 56, no. 10, pp. 4246–4254, Oct. 2009. [15] W. Qiao, G. K. Venayagamoorthy, and R. G. Harley, “Real-time implementation of a STATCOM on a wind farm equipped with doubly fed induction generators,” IEEE Trans. Ind. Appl., vol. 45, no. 1, pp. 98–107, Jan./Feb. 2009. [16] P. S. Flannery and G. Venkataramanan, “Fault tolerant doubly fed induction generator wind turbine using a parallel grid side rectifier and series grid side converter,” IEEE Trans. Power Electron., vol. 23, no. 3, pp. 1126–1135, May 2008. [17] D. Xiang, L. Ran, P. J. Tavner, and S. Yang, “Control of a doubly fed induction generator in a wind turbine during grid fault ride-through,” IEEE Trans. Energy Convers., vol. 21, no. 3, pp. 652–662, Sep. 2006. [18] S. Xiao, G. Yang, H. Zhou, and H. Geng, “An LVRT control strategy based on flux linkage tracking for DFIG-based wecs,” IEEE Trans. Ind. Electron., vol. 60, no. 7, pp. 2820–2832, Jul. 2013. [19] S. Hu, X. Lin, Y. Kang, and X. Zou, “An improved low-voltage ridethrough control strategy of doubly fed induction generator during grid faults,” IEEE Trans. Power Electron., vol. 26, no. 12, pp. 3653–3665, Dec. 2011. [20] R. G. Almeida, J. A. P. Lopes, and J. A. L. Barreiros, “Improving power system dynamic behavior through doubly fed induction machines controlled by static converter using fuzzy control,” IEEE Trans. Power Syst., vol. 19, no. 4, pp. 1942–1950, Nov. 2004. [21] J. P. A. Vieira, M. V. A. Nunes, U. H. Bezerra, and A. C. d. Nascimento, “Designing optimal controllers for doubly fed induction generators using a genetic algorithm,” IET Gen. Transm. Distrib., vol. 3, no. 5, pp. 472–484, Oct. 2009. [22] Y. Zhou, P. Bauer, J. A. Ferreira, and J. Pierik, “Operation of grid connected DFIG under unbalanced grid voltage,” IEEE Trans. Energy Convers., vol. 24, no. 1, pp. 240–246, Mar. 2009. [23] Y. Wang, L. Xu, and B. W. Williams, “Compensation of network voltage unbalance using doubly fed induction generator-based wind farms,” IET Renew. Power Gen., vol. 3, no. 1, pp. 12–22, 2009. [24] L. Xu and Y. Wang, “Dynamic modeling and control of DFIG-based wind turbines under unbalanced network conditions,” IEEE Trans. Power Syst., vol. 22, no. 1, pp. 314–323, Feb. 2007. [25] L. Xu, “Coordinated control of DFIG’s rotor and grid-side converters during network unbalance,” IEEE Trans. Power Electron., vol. 23, no. 3, pp. 1041–1049, May 2008. [26] J. Hu, Y. He, L. Xu, and B. W. Williams, “Improved control of DFIG systems during network unbalance using PI-R current regulators,” IEEE Trans. Ind. Electron., vol. 56, no. 2, pp. 439–451, Feb. 2009. [27] J. Santos-Martin, L. Rodriguez-Amenedo, and S. Arnaltes, “Providing ride-through capability to a doubly fed induction generator under unbalanced voltage dips,” IEEE Trans. Power Electron., vol. 24, no. 7, pp. 1747–1775, Jul. 2009.

IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 29, NO. 3, MAY 2014

[28] R. Pena, J. C. Clare, and G. M. Asher, “Doubly fed induction generator using back-to-back PWM converters and its application to variable speed wind-energy generation,” Proc. Inst. Elect. Eng., Elect. Power Appl., vol. 143, no. 3, pp. 231–241, May 1996. [29] S. Tao, “Power quality of grid-connected wind turbines with DFIG and their interaction with grid,” Ph.D. dissertation, Aalborg Univ., Aalborg, Denmark, 2004. [30] O. Noureldeen, “A novel crowbar protection technique for DFIG wind farm during fault ride through,” Int. J. Eng.Technol. IJET-IJENS, vol. 12, no. 3, pp. 85–95, Jun. 2012. [31] B. Shen, B. Mwinviwiwa, Y. Zhang, and B. T. Ooi, “Sensorless maximum power point tracking of wind by DFIG using rotor position phase lock loop (PLL),” IEEE Trans. Power Electron., vol. 24, no. 4, pp. 942–951, Apr. 2009. [32] Y. Zou, M. Elbuluk, and Y. Sozer, “A novel maximum power points tracking (MPPT) operation of doubly-fed induction generator (DFIG) wind power system,” in Proc. IEEE Industry Applicat. Soc. Annual Meeting (IAS), 2012, pp. 1–6.

Theodoros D. Vrionis was born in Athens, Greece, in 1976. He received the diploma in electrical engineering from the University of Patras, Greece, in 2001, where he is currently pursuing the Ph.D. degree. His main research interests include HVdc transmission based on VSCs, renewable energy sources, power quality, computer applications in power systems analysis, and intelligent control.

Xanthi I. Koutiva was born in Tripolis, Greece, in 1976. She received the diploma and Ph.D. degree in electrical engineering from the University of Patras, Greece, in 2000 and 2007, respectively. She is mainly specializing in HVdc transmission based on VSCs, renewable energy sources, and computational intelligence applications in power systems.

Nicholas A. Vovos (M’76–SM’95) was born in Thessaloniki, Greece, in 1951. He received the diploma and Ph.D. degrees from the University of Patras, Greece, and the M.Sc. degree from the University of Manchester Institute of Science and Technology (UMIST), U.K., in 1974, 1978, and 1975, respectively. He is a Professor in the Electrical and Computer Engineering Department of the University of Patras, and his main fields of interest are the transient stability study of integrated AC/DC systems, FACTS, power quality, renewable energy sources, and smart grids.