IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 58, NO.

1, JANUARY 2011

29

A Novel Algorithm for Fast and Efcient Speed-Sensorless Maximum Power Point Tracking in Wind Energy Conversion Systems
Syed Muhammad Raza Kazmi, Hiroki Goto, Member, IEEE, Hai-Jiao Guo, Member, IEEE, and Osamu Ichinokura, Member, IEEE

AbstractThis paper proposes a novel solution to the problems that exist in the conventional hill climb searching (HCS) maximum power point tracking (MPPT) algorithm for the wind energy conversion system. The presented solution not only solves the tracking speed versus control efciency tradeoff problem of HCS but also makes sure that the changing wind conditions do not lead HCS in the wrong direction. It intelligently adapts the variable step size to keep up with the rapid changes in the wind and seizes the perturbation at the maxima to yield 100% control efciency. For this purpose, a novel peak detection capability has been devised which, in contrast with conventional peak detection, can work robustly under changing wind conditions. The proposed MPPT performs self-tuning to cope with the nonconstant efciencies of the generatorconverter subsystemsa phenomenon quite rarely discussed in research papers so far. In addition, a smart speed-sensorless scheme has been developed to avoid the use of mechanical sensors. The experimental results conrm that the proposed algorithm is remarkably faster and more efcient than the conventional HCS. Index TermsHill climb searching (HCS), maximum power point (MPP) tracking (MPPT), speed sensorless, wind energy conversion system (WECS).

I. I NTRODUCTION S BRIEFLY reviewed in [1], several maximum power point (MPP) tracking (MPPT) algorithms for the wind energy conversion system (WECS) can be found in the literature. Among these algorithms, the hill climb searching (HCS) control happens to be the one that does not require any prior knowledge of the system and is absolutely independent of the turbine, generator, and wind characteristics. Also, it can work independently in conjunction with variable-pitch control, whereas the rest of the methods have to be reprogrammed for every change in pitch angle. Therefore, the HCS control bears the potential to be the most appropriate choice for MPPT, but there are two serious problems with HCS which signicantly deteriorate its performance under rapidly changing
Manuscript received April 2, 2009; revised July 14, 2009 and November 10, 2009; accepted December 16, 2009. Date of publication March 8, 2010; date of current version December 10, 2010. S. M. R. Kazmi, H. Goto, and O. Ichinokura are with the Department of Electrical and Communication Engineering, Graduate School of Engineering, Tohoku University, Sendai 980-8579, Japan (e-mail: syed. raza.kazmi@gmail.com). H.-J. Guo is with the Department of Electrical Engineering and Information Technology, Tohoku Gakuin University, Sendai 980-8511, Japan. Color versions of one or more of the gures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identier 10.1109/TIE.2010.2044732

wind conditions. These two problems are the speedefciency tradeoff and the wrong directionality under rapid wind change, as explained in Section II. Only very few attempts have been made so far to address these problems, and mostly, they belong to the MPPT in a photovoltaic (PV) system for a very similar method called perturb and observe [2][5], [7]. However, none of these papers provide a complete and easily implementable solution to both problems. For instance, Femia et al. [2] focus on solving the issue of wrong directionality and propose a formula for the optimal perturbation step size. However, the solution is very complicated and requires the knowledge of various system parameters. References [3][5] are on the speedefciency tradeoff problem, proposing a variable-step hill climbing governed by the magnitude of the derivative of power P with respect to a control variable such as the duty ratio D of the MPPT converter. Such variable-step-size schemes are impractical under uctuating environmental conditions because, then, dP/dD does not provide the true sense of distance from the maxima. Reference [6] also uses the slopes of power P and generator speed with respect to a variable whose calculation requires the knowledge of the turbines mechanical power which is practically not available in most systems. Even if the mechanical power is assumed equal to the electrical power measured at the output, still, the methodology of Agarwal et al. [6] is most likely to fail in uctuating wind conditions because, then, the signs of dP/d and d/d cannot indicate the true operating sector (dened in [6]). Reference [7] suggests an additional measurement in the middle of the sampling period through which the change in power solely caused by the MPPT can be calculated. However it requires a constant rate of change of irradiation over a sampling period; but unlike the PV system where the output current is a linear function of irradiance, the output torque in WECS is a square function of wind velocity [8]. Therefore, a linear change in wind velocity over a sampling period does not result in a linear change in power, and hence, the method of Sera et al. [7] cannot be extended to WECS. Moreover, generally, the inertia of a WECS is much higher than that of a PV system, and hence, WECS has a rather lethargic settling time. The sampling period is usually set equal to the settling time; hence, an intermediate measurement will not be a true value. Owing to the inadequacies of these techniques in solving the intrinsic problems of HCS, WECS designers tend to use other MPPT techniques which employ the prerecorded

0278-0046/$26.00 2011 IEEE

30

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 58, NO. 1, JANUARY 2011

Fig. 1. Size of perturbation in (a) is larger than that in (b). Hence, the tracking speed in (a) is faster than that in (b) but at the cost of larger oscillations around the MPP.

characteristics of the optimal curve [1]. Such techniques are indeed very fast and efcient but are also system specic and cannot be used together with variable-pitch control. Even if the pitch angle is unchanged, still, the nonconstant efciencies of generatorconverter subsystems can result in the deviation of the actual optimum points from the stored ones. This phenomenon has never been catered for before, and quite always, in lookup-table-like techniques, a unique optimal curve is assumed [8][10]. The experimental results in this paper prove that the optimal curves are actually nonunique for the output electrical power of a WECS at different wind velocities. This paper presents a very simple yet very effective solution to the two problems of HCS by making use of the general characteristic of the optimal power curve. The nonuniqueness of the optimal curve is catered for by incorporating a self-tuning capability into the system. This research paper extends the work of Raza et al. [1] with an ingenious scheme to make the proposed system completely mechanical sensorless. Moreover, the experimental evaluation, which, in [1], was based on a simulated wind prole, has been carried out here under a real wind prole measured on a windy day in Sendai. II. P ROBLEMS IN THE C ONVENTIONAL HCS C ONTROL A. Perturbation Step Size and SpeedEfciency Tradeoff As shown in Fig. 1, a larger perturbation step size increases the speed of convergence but deteriorates the efciency of the MPPT [2] by amplifying the oscillations Pmpp around the MPP Pmpp . This is due to the fact that the HCS control does not halt at MPP because it does not possess a peak detection capability, and hence, the oscillations are an unavoidable attribute of the HCS control. A smaller step size boosts the efciency, but then, the convergence speed becomes slower; therefore, the controller may become incapable of tracking MPP under rapidly varying wind conditions. Hence, in the conventional HCS control, a tradeoff always exists between the tracking speed and the control efciency. B. Perturbation Direction and Trackability Under Wind Change In the normal HCS control method, the direction, i.e., the sign of the next perturbation, is decided by the increase or decrease in power due to the previous perturbation. Since the HCS control is blind to the atmospheric changes, this rule can

Fig. 2. HCS control losing its trackability under changing wind conditions and traveling downhill instead of climbing uphill.

be misleading, as the sign might be dictated by the change in wind rather than the applied perturbation. This wrong decision leads to the failure in keeping track of MPP, and the HCS control moves downhill, as shown in Fig. 2. Of the two problems discussed previously, the latter one is of major concern because it signicantly deteriorates the energy extraction from the renewable source and hence greatly affects the overall inputoutput efciency of the system. III. P ROPOSED MPPT A LGORITHM A. Principle The novel MPPT algorithm proposed in this paper exploits the fact that a constant-pitch variable-speed wind turbines mechanical power Pm has a unique optimal power curve Popt which exhibits a cubic function of the generator speed [8]. Therefore, the optimal curve of a wind turbines mechanical power is characterized by a unique constant kopt , as given in Popt = kopt 3 . (1)

During the normal hill climbing, if we happen to reach a maximum and detect it successfully, then we can extract kopt simply by measuring the corresponding power and rotational speed. A point to be noted here is that the MPPT techniques found so far in literature do not have the peak detection capability under varying wind conditions. This research paper proposes a set of checks, shown later in (3), to formulate a robust yet very simple peak detection under varying wind conditions. Once the peak is detected and kopt is extracted, the latter can then serve as an accurate reference for the size and the direction

KAZMI et al.: NOVEL ALGORITHM FOR FAST AND EFFICIENT SPEED-SENSORLESS MPPT IN WECSs

31

Fig. 3. Pm curves for various wind speeds and the optimal curve that passes through the MPPs. The arrowheads on the operating points A and B show the desired direction of perturbation.

of the next perturbation. Fig. 3 shows an illustration of this ideawith a known optimal curve, if the operating point lies on its right, like A, the next perturbation would be in the direction of decreasing in order to move the operating point closer to the optimal curve. Also, the size of the step can be determined through the measure of how far the operating point lies from the optimal curve. This idea constitutes the very core of the algorithm presented in this paper. It gives a very effective solution to both the problems stated in Section II because now, regardless of the wind change, the operating point lying anywhere, like A or B in Fig. 3, will always move toward the optimal curve. This eradicates the loss of trackability under changing wind conditions. Also, it should be mentioned that the farther the operating point from the optimal curve is, the larger the perturbation size for faster tracking will be. However, as it converges into the optimal curve, the perturbation size will automatically approach zero, thus eliminating the speedefciency compromise of HCS. B. Optimal Po Versus Optimal Pm The principle presented previously is actually not straightforward to implement. The existence of a unique kopt , on which the proposed method fully depends, is associated with the captured mechanical power Pm , whereas the power being actually measured and subjected to maximization is the output electrical power Po supplied to the load. In order to extend the proposed MPPT principle to Po , it is necessary that Po should also have a unique optimal curve for all wind speeds. This can be investigated by considering the relationship of the optimal curve constants (kopt )m and (kopt )o of Pm and Po , respectively, via generator efciency g and converter efciency c , given in steady state as [1] Po = g c Pm (kopt )o = g c (kopt )m . (2)

[10], which can most likely be erroneous because the generator efciency generally gets worse on the increase in phase current due to the copper losses, whereas the increase in rotational speed boosts the eddy current losses in the generator core. Reference [11] has a graph that shows such varying efciency of a switched reluctance generator with respect to its torque and speed. Similarly, on the converter side, the copper losses in the inductor and the losses in the power electronic devices may vary the converter efciency with the change in load current. Therefore, the overall efciency of a WECS is not constant under wind and load variations. With the nonconstant efciency of WECS, we can infer two very important conclusions. 1) There does not exist a unique optimal curve constant kopt for Po (for ease of working, the subscripts have been dropped, and from now on, kopt will be associated with Po only). 2) The maximum of the Po curve does not coincide with the peak of the Pm curve. These two inferences have been conrmed by the experimental results presented in a later section. C. Methodology To cater for the nonunique kopt of Po , a self-tuning strategy has been developed in the new intelligent MPPT algorithm. The owchart in Fig. 4 shows the three modes of operation. Mode 0 searches for a kopt with HCS via a novel peak detection capability. Mode 1 retains the system at the detected maximum, unless there is a change observed in wind velocity Vw . Mode 2 gets into action under changing wind conditions and implements the novel adaptive hill climbing via the earlier found kopt . This may not yield the true MPP, owing to the nonuniqueness of kopt for a different wind velocity, but still, it drives the operating point in the close vicinity of the true peak power. Therefore, this strategy is quite useful for fast tracking. Note that, here, the control input is the duty ratio D of the converter which is driven by the MPPT controller, as also shown in Fig. 8 in a later section. Each cycle of the algorithm ends up with updating D by a perturbation step d. All the states of the systems thus become the function of D and Vw . The salient features of the three modes are as follows. 1) Mode 0: This mode runs the HCS control equipped with the new peak detection capability to search for an MPP. The contrasting feature of this HCS is the checks in (3) which ensure that the peak detection should not be misled by the change in wind. MPP is detected when all the following checks are true: P (k ) < 0 Vw (k ) = 0 Vw (k 1) = 0.

(3)

Equation (2) implies that the key to the existence of a unique optimal Po curve for various wind speeds lies in a very important question whether the generatorconverter efciencies are constant or they vary with wind velocity, rotor speed, and load variations. So far, there is no signicant research work formulating these efciencies. Usually, they are regarded as constants [8]

The rst check detects the zero crossing of P that happens when the hill climbing crosses over the MPP. In the algorithms published so far, this check is the only provision for peak detection which surely fails when the decrease in power is due to the wind change instead of the MPP crossing. Therefore, a check on Vw (k ) has been proposed here.

32

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 58, NO. 1, JANUARY 2011

Fig. 4. Flowchart of the new intelligent adaptive MPPT.

Fig. 5. For the detection of the correct maxima, both Vw (k) and Vw (k 1) should be checked for the wind change.

Fig. 6.

Adaptive tracking in Mode 2 with variable step size.

Another important check has been proposed on the previous sample Vw (k 1), which is necessary even if the current Vw says no wind change. Fig. 5 shows an example where just the check on Vw (k ) would result in announcing the k th sample as the MPP, which obviously is not true, whereas augmenting a check on Vw (k 1) will retrace the correct MPP. Once (3) becomes true, kopt is calculated through (1) by measuring the power and rotational speed, and then, the algorithm switches to Mode 1. 2) Mode 1: This mode sets the perturbation to zero to retain the maxima. Thus, it yields 100% control efciency due to no oscillations about the maxima. The system is shifted to Mode 2 when the wind change is detected through the change in rotor speed or the drift in output power. 3) Mode 2: This mode implements the central theme of this research paperthe adaptive hill climbing according to the stored kopt . As outlined earlier, the perturbation size is decided by the distance of the operating point from the kopt 3 curve, whereas the sign of this distance vector determines the sign of

the perturbation. Mathematically the control law of mode 2 can be formulated as: d(k + 1) = ( ) (4)

where is the abscissa of the optimal curve corresponding to the current power P and is a positive denite gain. Fig. 6 shows a graphical explanation on the working of the adaptive MPPT during Mode 2. Note that is not opt , but eventually, it converges to opt as the algorithm proceeds in time. Another important attribute here is that the direction in which the operating point travels is opposite to the direction of . This is due to the fact that the sign of is negative to the sign of d for no wind change. A mathematical proof to this statement can be seen in [12]. Mode 2 is kept in operation until the operating point reaches sufciently closer to the curve characterized by kopt . Once reached in the close proximity of the pseudo-optimal curve, if there is no signicant wind change, then the algorithm switches back again to Mode 0 to search for the correct optimum corresponding to the current wind level. Hence, this way, the

KAZMI et al.: NOVEL ALGORITHM FOR FAST AND EFFICIENT SPEED-SENSORLESS MPPT IN WECSs

33

algorithm possesses a self-tuning capability which caters for the nonuniqueness of kopt under changing wind conditions and for the parametric changes in the system. Note the rst check block of Mode 0 in Fig. 4. Its utility is during the self-tuning when the system switches to Mode 0 from Mode 2 in search for the new kopt . During this search, if a wind change happens, through this check, the system immediately resumes the adaptive tracking of Mode 2. Hence, the trackability of the algorithm is never compromised. D. Detecting Wind Change Worthy to mention here is that Vw is not measured through an anemometer; instead, a novel sensorless scheme is devised. Its working principle is based on the three fundamental properties of WECS, stating that, for no wind change, the following conditions should be satised. 1) The change in rotor speed should be bounded by a limit if the perturbation is of a constant magnitude. 2) The change in rotor speed will bear the sign opposite to the change in duty ratio [12]. 3) The system cannot go downhill on the power curve for two consecutive samples. These three rules can be formulated as (5) and (6). A wind change will be signaled either if (5) is not true or if any one of the two checks in (6) is true | (k ) (k 1)| sgn(d) = sgn( ) P (k ) < 0 P (k 1) < 0. IV. G ENERATOR S PEED -S ENSORLESS S CHEME The measurement of the generator speed is essential not only to implement the adaptive hill climbing proposed in (4) but also to detect the wind changes. Using a rotary encoder for this purpose will not only increase the cost of the system but will also make the system more prone to sensor noise as the mechanical sensors always have maintenance issues and cannot be used for a long run. This paper proposes a very smart sensorless scheme that simply takes into account the cyclic nature of the generator phase current whose frequency (and, hence, the number of zero crossings) is directly proportional to the speed of the generator. This empirical knowledge can be converted into the estimate of speed simply by knowing the number of rotor poles p and measuring the time lapse t between two rising zero crossings (one complete cycle) of the generator phase current. The rotor displacement between these two zero crossings would be 2/p. Hence, the speed can be estimated as = 2/p . t (7) (5.a) (5.b) (6)

Fig. 7. Speed estimator for the speed-sensorless scheme.

Fig. 8. Schematic of the designed WECS.

Fig. 9. Components of the experimental setup.

The extreme simplicity and ease of implementation of the proposed speed-sensorless scheme makes it quite outstanding when compared with the techniques published so far for salientpole machines. As evident from the literature review [13] [19], these speed-sensorless techniques require various machine parameters, model equations, or lookup tables. In contrast, the proposed method only needs a single current sensor. Even the information of the number of rotor poles is not a necessity because the proposed MPPT algorithm, shown in Fig. 4, does not require the true measure of generator speed. Therefore, for an unknown number of rotor poles, any positive integer can be assigned to p in (7). Hence, this speed-sensorless scheme is very attractive for the wide class of salient-pole generators in which the zero crossings of the cyclic phase current depend on the rotor speed and the rotor poles. V. E XPERIMENTAL R ESULTS A. Experimental Setup The schematic of the designed WECS system is shown in Fig. 8, and the respective hardware in the loop is shown in Fig. 9. As shown in Fig. 9(a), an emulator of the wind turbine is developed via a servo-driver-based dc motor for the fast prototyping of the system. It takes a torque command from a computer running the dynamic simulation model of the wind turbine in MatlabdSpace, as developed in [20]. The control hardware driven by a eld programmable gate array (FPGA) is shown in Fig. 9(b).

Fig. 7 shows the proposed speed estimator as implemented in MatlabdSpace for a four-pole generator. To avoid detecting the false zero crossing due to the sensor noise, a rst-order lowpass lter with hysteresis band has been put into service.

34

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 58, NO. 1, JANUARY 2011

B. Conguring Control Systems Parameters The conguration of the control parameters of Fig. 4 is enumerated as follows. 1) The initial step size d(0) for Mode 0 is set as 0.01. 2) The bound of (5) should be set low enough to detect any change in the wind prole. However, it must be higher than the of constant wind. For the designed WECS, it has been found by a test run with the duty step d = 0.01, that = 1 is a good candidate for detecting the wind change. 3) The gain for Mode 2 is set as 5/1000. Note that the adaptive step control of (4) is like a proportional control, and therefore, a higher may result in oscillations. 4) The perturbation frequency is determined by observing the settling time of the system to a step input [9]. C. Results and Discussion In [1], the experimental evaluation of the sensor-based system was carried out on a simulated wind prole. Here, the performance of the novel sensorless MPPT system is validated via both simulated and real wind proles. The new intelligent adaptive HCS algorithm has been compared with the conventional HCS operating at the same step size as that in Mode 0 and at the same perturbation frequency. Fig. 10 shows the tracking results of the two algorithms subjected to the simulated wind prole of varied slopes shown in Fig. 10(a). Fig. 10(b) shows the cyclic nature of the generator phase current, and Fig. 10(c) conrms the excellent performance of the speed estimator in comparison with the actual speed. Fig. 10(d) clearly shows the remarkable boost in MPPT with the novel algorithm. Note that, during the drop in wind speed, such as from 5 to 3.5 m/s, the new algorithm tends to slow down the resulting decrease in power, and therefore, it exhibits a slower gradient of the output power curve as compared with the power tracked by the normal MPPT. On the other hand, during the increase in wind energy, the new algorithm exhibits a very steep slope as compared with the old one to instantly catch up with the added available energy. This can be seen in Fig. 10(d) when the wind speed suddenly changes from 3.5 to 6.5 m/s. In addition, note that, during such sharp wind bursts, the normal MPPT consumes too much time to reach the maximum and hence become very inefcient. A similar conclusion can be drawn from Fig. 10(e), which corroborates that the new algorithm is quite fast in reaching the optimal duty ratio, whereas the normal HCS control gets confused during the wind change. A very detrimental feature of the conventional HCS that resembles ringing or chattering happens when the wind energy decreases continuously, as shown in Fig. 10(e) during the continuous drop in wind speed from 5 to 3.5 m/s. The controller remains hanging in a confused state because the next perturbation is overruled by the change in wind energy rather than the change in control variable. The proposed algorithm eliminates this effect and gives a clear mind to the controller no matter how fast the wind changes.

Fig. 10. Experimental results with the simulated wind prole. (a) Simulated wind prole, (b) generator phase current, (c) estimated generator speed in comparison with the actual one, (d) output power tracked by the new and the conventional MPPT, and (e) duty ratio driven by the new and the conventional MPPT.

KAZMI et al.: NOVEL ALGORITHM FOR FAST AND EFFICIENT SPEED-SENSORLESS MPPT IN WECSs

35

Fig. 11. Characteristic curves. (a) Po curves for various wind speeds and the corresponding kopt 3 curves. A unique optimal curve does not exist for Po contrary to Pm which has a unique optimal curve. (b) Po drawn along with Pm , showing that their maxima do not coincide.

Another important point to note from Fig. 10(e) is the 100% control efciency that is achieved by seizing the perturbation when the MPP is reached. The perturbation is reactivated as soon as the change in wind is detected. It should also be noted here that the new algorithm provides the user with control over the performance. The parameter can further improve the tracking speed, but a larger value may result in oscillations, as mentioned earlier. Fig. 11 conrms the conclusions that were inferred in Section III-B. Fig. 11(a) shows the Po curves at the wind velocities used in Fig. 10. Different optimal curves for each wind speed clearly imply the nonuniqueness of kopt . Furthermore, Fig. 11(b) shows the Po curve in comparison with the Pm curve, and it can be witnessed that their maxima do not coincide. Hence, the conventional lookup-tablebased techniques cannot yield the true optimum under varying wind conditions. Yet, the new intelligent algorithm successfully ensures the tracking of the true maximum through its self-tuning. Fig. 12 shows the performance comparison of the conventional versus new MPPT under the real wind prole shown in Fig. 12(a), as measured on a windy day in Sendai. The comparison has been established by observing the power coefcient Cp curves tracked by both algorithms. As evident from Fig. 12(b), the conventional method fails to maintain the optimum Cp and results in large uctuations under rapidly varying wind conditions, whereas the proposed method, as shown in Fig. 12(c), is very fast with quite-less uctuations that last for a considerably lesser amount of time. Hence, these results conrm that the proposed method is very fast and efcient as compared with the conventional one.

Fig. 12. Experimental results with the real wind prole. (a) Measured wind prole. (b) Power coefcient tracked by the conventional MPPT. (c) Power coefcient tracked by the new intelligent MPPT.

VI. C ONCLUSION A novel MPPT algorithm that eradicates the problems that exist in the conventional hill-climbing MPPT technique has been established in this research paper. The strategy is based on the idea of detecting the cubic optimal power curve online and then employing this curve as a reference for the future perturbation size and direction. With such reference, the algorithm adapts a larger perturbation step during the wind change, which gradually approaches to zero as the peak gets nearer. It always drives the system toward MPP despite the wind change and keeps the control efciency to maximum by seizing the perturbation at MPP. Due to the inconsistent efciencies of generatorconverter subsystems, the algorithm performs selftuning to update the optimal curve for different wind speeds. For the rst time, this research paper has introduced HCS with peak detection capability. Another novelty of this research paper is the smart and simple speed-sensorless scheme that is devised for the class of salient-pole generators in which the zero crossings of the cyclic phase current depend on the speed and the rotor poles of the generator. Hence, the proposed MPPT is a sensorless intelligent adaptive algorithm promising higher control efciency and faster tracking under rapid wind

36

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 58, NO. 1, JANUARY 2011

change without requiring any prior knowledge of the systems characteristics. R EFERENCES
[1] K. S. M. Raza, H. Goto, H.-J. Guo, and O. Ichinokura, A novel algorithm for fast and efcient maximum power point tracking of wind energy conversion systems, in Proc. 18th Int. Conf. Elect. Mach., Vilamoura, Portugal, Sep. 2008, pp. 16. [2] N. Femia, G. Petrone, G. Spagnuolo, and M. Vitelli, Optimization of perturb and observe maximum power point tracking method, IEEE Trans. Power Electron., vol. 20, no. 4, pp. 963973, Jul. 2005. [3] P. Huynh and B. H. Cho, Design and analysis of a microprocessor controlled peak-power-tracking system, IEEE Trans. Aerosp. Electron. Syst., vol. 32, no. 1, pp. 182190, Jan. 1996. [4] J. Yaoqin, Y. Zhongqing, and C. Binggang, A new maximum power point tracking control scheme for wind generation, in Proc. PowerCon, 2002, pp. 144148. [5] F. Liu, S. Duan, F. Liu, B. Liu, and Y. Kang, A variable step size INC MPPT method for PV systems, IEEE Trans. Ind. Electron., vol. 55, no. 7, pp. 26222628, Jul. 2008. [6] V. Agarwal, R. K. Aggarwal, P. Patidar, and C. Patki, A novel scheme for rapid tracking of maximum power point in wind energy generation systems, IEEE Trans. Energy Convers., vol. 25, no. 1, pp. 228236, Mar. 2010. [7] D. Sera, R. Teodorescu, J. Hantschel, and M. Knoll, Optimized maximum power point tracker for fast-changing environmental conditions, IEEE Trans. Ind. Electron., vol. 55, no. 7, pp. 26292637, Jul. 2008. [8] R. Cardenas, R. Pena, M. Perez, J. Clare, G. Asher, and P. Wheeler, Control of a switched reluctance generator for variable-speed wind energy applications, IEEE Trans. Energy Convers., vol. 20, no. 4, pp. 781791, Dec. 2005. [9] Q. Wang and L. Chang, An intelligent maximum power extraction algorithm for inverter-based variable speed wind turbine systems, IEEE Trans. Power Electron., vol. 19, no. 5, pp. 12421249, Sep. 2004. [10] C.-T. Pan and Y.-L. Juan, A novel sensorless MPPT controller for a high-efciency microscale wind power generation system, IEEE Trans. Energy Convers., vol. 25, no. 1, pp. 207216, Mar. 2010. [11] D. A. Torrey, Switched reluctance generators and their control, IEEE Trans. Ind. Electron., vol. 49, no. 1, pp. 314, Feb. 2002. [12] E. Koutroulis and K. Kalaitzakis, Design of a maximum power tracking system for wind-energy-conversion applications, IEEE Trans. Ind. Electron., vol. 53, no. 2, pp. 486494, Apr. 2006. [13] J. Brahmi, L. Krichen, and A. Ouali, A comparative study between three sensorless control strategies for PMSG in wind energy conversion system, Appl. Energy, vol. 86, no. 9, pp. 15651573, Sep. 2009. [14] S. M. Barakati, M. Kazerani, and J. D. Aplevich, Maximum power tracking control for a wind turbine system including a matrix converter, IEEE Trans. Energy Convers., vol. 24, no. 3, pp. 705713, Sep. 2009. [15] B. Shen, B. Mwinyiwiwa, 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. 942951, Apr. 2009. [16] R. Pena, R. Cardenas, J. Proboste, G. Asher, and J. Clare, Sensorless control of doubly-fed induction generators using a rotor-current-based MRAS observer, IEEE Trans. Ind. Electron., vol. 55, no. 1, pp. 330339, Jan. 2008. [17] S. Ichikawa, M. Tomita, S. Doki, and S. Okuma, Sensorless control of permanent-magnet synchronous motors using online parameter identication based on system identication theory, IEEE Trans. Ind. Electron., vol. 53, no. 2, pp. 363372, Apr. 2006. [18] R. Esmaili and L. Xu, Sensorless control of permanent magnet generator in wind turbine application, in Conf. Rec. 41st IAS Annu. Meeting, 2006, pp. 20702075. [19] B. K. Bose, Modern Power Electronics and AC Drives. Englewood Cliffs, NJ: Prentice-Hall, Oct. 2001. [20] K. S. M. Raza, H. Goto, H.-J. Guo, and O. Ichinokura, Maximum power point tracking control and voltage regulation of a DC grid-tied wind energy conversion system based on a novel permanent magnet reluctance generator, in Proc. ICEMS, 2007, pp. 15331538.

Syed Muhammad Raza Kazmi was born in Lahore, Pakistan, in 1977. He received the B.Sc. degree from Ghulam Ishaq Khan Institute of Engineering Sciences and Technology, Topi, Pakistan, in 1999, the M.Eng.Sc. degree (with high distinction) from The University of New South Wales, Sydney, Australia, in 2001, and the Ph.D. degree from Tohoku University, Sendai, Japan, in 2010. From 2001 to 2005, he was a Lecturer and an Assistant Professor with the National University of Sciences and Technology, Islamabad, Pakistan. He is currently a Postdoctoral Researcher with the New Energy and Industrial Technology Development Organization, Kawasaki, Japan. His research interests involve efcient control and power electronics for renewable energy conversion systems.

Hiroki Goto (M06) was born in Shizuoka, Japan, on June 27, 1979. He received the M.S. degree in electrical engineering from Tohoku University, Sendai, Japan, in 2004. He is currently a Research Associate with the Department of Electrical and Communication Engineering, Graduate School of Engineering, Tohoku University. He has a wide spectrum of research interests ranging from the analysis and design of new kinds of special machines to the control and power electronics of electric vehicles and renewable energy systems. Mr. Goto is a member of the IEEE Magnetics Society of Japan.

Hai-Jiao Guo (M03) was born in Shanghai, China. He received the B.E. and Ph.D. degrees in electrical engineering from Tohoku University, Sendai, Japan, in 1988 and 1993, respectively. From 1990 to 1996, he was an Assistant Professor with the Department of Electrical and Communication Engineering, Graduate School of Engineering, Tohoku University, where he was an Associate Professor from 1996 to 2004. He is currently a Professor with the Department of Electrical Engineering and Information Technology, Tohoku Gakuin University, Sendai. His research interests are in the design of robust digital servo systems and advanced motor control. Prof. Guo is a member of The Society of Instrument and Control Engineers of Japan.

Osamu Ichinokura (S77M80) was born in Morioka, Japan, on August 28, 1951. He received the B.S., M.S., and Ph.D. degrees in electrical engineering from Tohoku University, Sendai, Japan, in 1975, 1977, and 1980, respectively. Since 1980, he has been with the Department of Electrical and Communication Engineering, Graduate School of Engineering, Tohoku University, where he is currently a Professor. His current research interests are in the areas of power electronics and power magnetics. Prof. Ichinokura is a member of the Magnetics Society of Japan, The Society of Instrument and Control Engineers, and the Institute of Electrical Installation Engineers of Japan.