Implementation of a new maximum power point tracking control strategy for small wind energy conversion systems without mechanical sensors

Energy Conversion and Management 97 (2015) 298–306

Contents lists available at ScienceDirect

Energy Conversion and Management journal homepage: www.elsevier.com/locate/enconman

Implementation of a new maximum power point tracking control strategy for small wind energy conversion systems without mechanical sensors Yacine Daili a,⇑, Jean-Paul Gaubert b, Lazhar Rahmani a a b

Automatic Laboratory of Setif, University of Ferhat Abbas – Setif 1, Algeria Laboratory of Computer Science and Automatic Control for Systems, University of Poitiers, France

a r t i c l e

i n f o

Article history: Received 7 December 2014 Accepted 16 March 2015

Keywords: Small wind energy conversion systems (SWECS) Maximum power point tracking (MPPT) Perturbation and observation (P&O) Permanent magnet synchronous generator (PMSG) Sensorless

a b s t r a c t This paper proposes a modified perturbation and observation maximum power point tracking algorithm for small wind energy conversion systems to overcome the problems of the conventional perturbation and observation technique, namely rapidity/efficiency trade-off and the divergence from peak power under a fast variation of the wind speed. Two modes of operation are used by this algorithm, the normal perturbation and observation mode and the predictive mode. The normal perturbation and observation mode with small step-size is switched under a slow wind speed variation to track the true maximum power point with fewer fluctuations in steady state. When a rapid change of wind speed is detected, the algorithm tracks the new maximum power point in two phases: in the first stage, the algorithm switches to the predictive mode in which the step-size is auto-adjusted according to the distance between the operating point and the estimated optimum point to move the operating point near to the maximum power point rapidly, and then the normal perturbation and observation mode is used to track the true peak power in the second stage. The dc-link voltage variation is used to detect rapid wind changes. The proposed algorithm does not require either knowledge of system parameters or of mechanical sensors. The experimental results confirm that the proposed algorithm has a better performance in terms of dynamic response and efficiency compared with the conventional perturbation and observation algorithm. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Wind energy is one of the most promising renewable energy sources that can serve as an alternative to traditional sources based on fossil fuels. This is not only for high power, but also for the small power range (1–100 kW) [1]. Small wind energy conversion systems (SWECS) equipped with permanent magnet synchronous generator (PMSG) have several benefits compared to those using induction generators. First, PMSG can provide high-reliability power generation since there is no need for external magnetization. Second, the high torquedensity of the PMSG allows a reduction in the cost of the system [2]. Moreover, a wind turbine with a multi-polar PMSG removes the need for a gearbox. Thus, the system requires less servicing [3].

⇑ Corresponding author. Tel.: +213 778065901. E-mail addresses: [email protected] (Y. Daili), jean.paul.gaubert@ univ-poitiers.fr (J.-P. Gaubert), [email protected] (L. Rahmani). http://dx.doi.org/10.1016/j.enconman.2015.03.062 0196-8904/Ó 2015 Elsevier Ltd. All rights reserved.

To realize a maximum power point tracking (MPPT) operation, a power electronic interface is necessary. The converter topology based on a three-phase diode rectifier cascaded with a boost converter is more suitable for SWECS applications due to its low cost and high reliability [4]. The control objective of the power electronic interface is to extract the maximum available power from the incident wind for different wind speed values. Typically, two MPPT methods are commonly used for the control of SWECS; namely, the perturbation and observation (P&O) technique, and the technique based on optimum relationship [5]. With the first MPPT technique, the control variable allowing the system to operate at the maximum power point (MPP) is obtained, based on a pre-established optimum relationship. Numerous optimum relationships have been presented in the literature, which are all suitable for MPPT control. For example, the power versus rotor speed relationship has been used in [6]. The optimal rotational speed is generated from the measured power by [7]. In Ref. [8], the desired optimum torque is obtained from the wind turbine’s optimum torque curve. The control algorithm using such

Y. Daili et al. / Energy Conversion and Management 97 (2015) 298–306

relationships to achieve the MPPT control requires rotor speed information, which is obtained from a mechanical sensor placed on the shaft of the PMSG, increasing the cost and complexity of the overall system [9]. Some authors have reported the possibility of using only the dc-variables for wind generation systems equipped with diode rectifiers to implement the MPPT without mechanical sensors, making the system more reliable. The output power and the dc-voltage were taken as input and output of the lookup table in [10]. Instead of a look up table, the authors [11] have expressed the optimal dc-voltage as a function of the dc output power with one coefficient. The dc-voltage versus dc-current optimum relationship was calculated analytically based on the system parameters in [5]. In [12], an optimum relation between the dc-voltage and the dc-current was used to track the MPP. The authors of [13] further simplify this relationship to a linear equation between the dc-current and the square of the dc-voltage. The method based on the optimum relationship has the advantage of a good dynamic response. Nevertheless, to implement the MPPT successfully, a priori relationship needs to be known, which is not easy to determine for real wind energy systems. Moreover, the optimum relationship does not remain constant throughout the wind generation system’s operational lifetime due to changes in the wind generation system itself (the aging of the system, debris built-up on the blades. . .) and/or in its surrounding environment (air density changes). Consequently, the MPPT controller fails to track the MPP [14]. The P&O control algorithm can overcome the problem associated with the method based on the optimum relationship. The MPP tracking can be achieved without any knowledge of the wind energy system’s characteristics. This method consists of perturbing a control variable in fixed step-sizes and observing the resulting changes in the electrical output power, the sign of the next perturbation will be decided according to the comparison result between successive output power. If the power increases, the sign of the perturbation is maintained, it is inversed when the output power decreases. The authors of [15] have suggested to perturb the rotor speed and observe the output power. The measured rotor speed is controlled to track its reference provided by the P&O algorithm. Instead of measuring the rotor speed, an estimator was employed in [16] to implement the MPPT based P&O algorithm without mechanical sensor. Other control variables could be used to avoid the rotor speed measurement/estimation, such as the dclink voltage, the dc-current or by directly adjusting the boost duty cycle [4]. Although this method is very simple and does not need any knowledge of the wind turbine system parameters, it suffers from some limitations. The efficiency/rapidity trade-off such as the large step-size means that the system reaches the MPP quickly, but exhibits large oscillations around the MPP when it is reached. Hence, the efficiency of the system is low. By using a small stepsize, the efficiency of the system is improved, but the time taken by the system to reach the MPP is large, which renders the P&O algorithm incapable of tracking the MPP under rapidly varying wind conditions. Furthermore, the power variation caused by the wind changes can be misinterpreted by the P&O mechanism, resulting in a divergence of the P&O algorithm from the peak power [17]. The authors of [18] review the most proposed MPPT methods and conclude that the P&O method has not satisfactory performance under fluctuating wind speed. Aiming to improve the efficiency and rapidity trade-off of the conventional P&O method, several researchers propose replacing the fixed step-size with a variable step-size. The perturbing stepsize is obtained using the power slope with respect to the control variable. Thus, if the system is working far from the optimal point, due to the large value of the slope, a large step-size is applied, while a small step-size is imposed for an operating point close to the MPP. In one study [19], the up-dated duty cycle was selected

299

based on the scaled measured of the slope of the power with respect to the duty cycle. In another study [20], dc-current step size was modified depending on the output power slope. The rectifier voltage variable step size has been defined based on Newton–Raphson method by the authors of [21]. Similarly, a fuzzy logic controller has been used by the authors of [22] to adjust the step-size, where the input variables of this controller are the power variation and the rotor speed variation. With the adaptive P&O technique proposed in [23], the fixed perturbing step-size used in the conventional P&O was multiplied by an adaptive coefficient, which is increased automatically during the tracking phase, so that the algorithm converges rapidly to the peak power. Once the MPP is reached, which is detected through observing the oscillations of the control variable, the algorithm gradually reduces the multiplicative coefficient to avoid the large oscillations of the system in the steady state. The P&O which is based on a line search optimization method was proposed in [24]. In this approach, an adequate perturbing direction and step-size are computed for each iteration to satisfier the Wolfe criteria. Generally speaking, the aforementioned P&O methods permit a high dynamic in the tracking of the MPP with fewer oscillations in the steady state. Nevertheless, the problem of the divergence from the MPP under rapid wind speed variations was not resolved by these techniques, and the system might still fail to track the MPP under these conditions. To deal with the divergence problem of the P&O method, the authors of [20] have proposed a dual- modes operation method, such as under slow wind speed fluctuations, the step-size and direction of the perturbation are defined based on the power slope, whereas they are computed from the rectifier voltage slope when the wind speed changes suddenly to avoid the divergence of the system from the MPP. Reference [14] introduced an advanced P&O torque control to track the MPP under a fluctuating wind profile, in this approach the P&O mechanism was used to search for the optimum torque relationship. In this paper, a modified P&O MPPT algorithm is proposed to accomplish a fast tracking of the MPP under a rapidly changing of the wind speed. The proposed MPPT is easily implemented; it requires neither the knowledge of wind system parameters nor mechanical sensors. This algorithm used in two modes of operation to deal with the aforementioned limitations associated with the traditional P&O technique. It switches from one operational mode to another according to the nature of the wind speed variation: the normal P&O mode with a small step-size is unable under a slow wind speed variation to track the MPP with fewer fluctuations in steady state, while the predictive mode is switched when a fast wind speed is detected in order to move the operating point rapidly to near the MPP .The dc-link voltage variation is employed to detect a fast wind speed variation. Experimental work has been done to demonstrate the validity and performance improvement of the proposed MPPT control algorithm. The rest of this paper is organized as follows: Section 2 is dedicated to the presentation and modeling of the studied system. The limitations of the conventional P&O technique are illustrated in Section 3. In Section 4, the proposed MPPT algorithm is detailed, and then the experimental prototype, as well as the validation results, are discussed in Section 5. In Section 6, we conclude this paper.

2. System configuration The synoptic scheme of the SWECS configuration studied in this work is shown in Fig. 1. The PMSG is coupled directly to a

300

Y. Daili et al. / Energy Conversion and Management 97 (2015) 298–306

Fig. 1. System configuration under study.

three-blade horizontal axis wind turbine, a three-phase diode bridge is employed to rectify the generated output voltage. Due to the uncontrollability of the diode rectifier, a boost converter is used to realize the MPPT operation; only one active power switch device S is needed, which both reduces the cost and simplifies the control of the system. For simplicity, a resistance load RL is connected directly to the output of the boost converter to consume the power generated by the SWECS, it can be substituted by dc– ac inverter with a unity power factor connected to the utility grid or supplying a local load in case of off-grid applications. The control algorithm uses only the dc-link voltage Vdc(t) and the dc-current Idc(t) measurements to adjust directly the duty cycle d(t). The proposed MPPT algorithm will be presented in detail in Section 4. 2.1. Wind turbine model The mechanical power extracted from wind has the following expression [25]:

PW ¼

1 qpR2 V 3W C P 2

ð1Þ

R is the turbine radius (m), q is the air density (kg/m3), VW is the wind speed (m/s) and Cp is the power coefficient of the turbine, usually it is provided by the wind turbine manufacturer. The torque on the wind turbine shaft can be calculated from the power expression as:

TW

1 V3 ¼ ¼ qpR2 W C P Xm 2 Xm PW

ð2Þ

Om is the rotation speed of the turbine (rad/s). The power coefficient used in this study has the following expression [26]:

  21 116 C P ¼ 0:5176  0:4b  5 e ki þ 0:0068k ki

ð3Þ

1 1 0:055 ¼  ki k þ 0:08b b3 þ 1

ð4Þ

RXm VW

PWmax ¼ K X3m

ð6Þ

opt

where Xm opt is an optimum rotation speed corresponding to a specific wind speed and K is an optimal power gain given by: 1

K¼2

qpR5 C Pmax

ð7Þ

k3opt

The relationship (6) can be used to realize the MPPT control. The optimum rotational speed is calculated from the output power, by controlling the wind turbine to operate at the optimum rotor speed, this guarantees that the maximum power is extracted from the incident wind [6]. 2.2. Permanent magnet synchronous generator model The PMSG dynamic equations are expressed in the d–q reference frame, which eliminates all time-varying inductance, giving rise to the following equations [15].



V sd V sq



 ¼

Rs  Ld s xe L d xe Lq Rs  Lq s



isd isq



 þ

0



xe Um

ð8Þ

Rs: stator resistance per phase (O), Ld, Lq: stator inductances in direct and quadrature axis (H), Vsd, Vsq: two axis machine voltages (V), isd, isq: two axis machine currents (A), s: Laplace operator, xe: electric speed of the stator voltage (rad/s), Um: amplitude of the flux linkage established by permanent magnet (Wb). The relationship between the electric speed and the mechanical speed can be expressed as:

P 2

xe ¼ Xm

where b is the angle of the blades (rad), in this work set to zero, and k is the tip speed ratio (TSR),it is represented by:

k¼

If the system operates at the optimal point, the maximum power expression can be found by substituting (5) in (1) as follows:

P is the poles number of the PMSG. The expression of electromagnetic torque is given by [27]:

ð5Þ

Fig. 2 shows the power coefficient versus TSR curve given by (3). It can be seen that there is an optimum TSR kopt at which the power coefficient is maximum CP max. Therefore, the mechanical power extracted from the wind is also maximum.

ð9Þ

Te ¼

    3 P  Ld  Lq isd isq þ Um isq 2 2

ð10Þ

The stator inductances in d and q axis are approximately equal for a non-salient-pole PMSG. Thus, the electromagnetic torque expression (10) can be simplified to:

Y. Daili et al. / Energy Conversion and Management 97 (2015) 298–306

301

Fig. 2. Power coefficient versus Tip Speed Ratio.

Te ¼

   3 P  Um isq 2 2

ð11Þ

Finally, the mechanical dynamic equation has the following expression:

dXm 1 ¼ ðT m  T e Þ J dt

ð12Þ

J is the total moment inertia (wind turbine and machine) (kg m2).

Fig. 3. P&O algorithm using (a) large step-size and (b) small step-size.

3. Conventional P&O algorithm For a wind system with the topology of Fig. 1, the peak power tracking can be achieved by adjusting the dc-link voltage [10]. Based on the conventional P&O principle, the MPPT algorithm could be implemented by perturbing the operating dc-link voltage around an initial value, the effect of the perturbation being observed on the output power in order to decide the direction of the next perturbation; if the current dc-power is greater than the previous one, the algorithm continues to perturb the dc-link voltage in the same direction (the sign of the next perturbation is maintained), and it is inversed otherwise. Theoretically, this algorithm has the advantage of being simple and easy for implementation since it does not require either a prior knowledge of the system parameters or mechanical sensors. However, it has some limitations; namely, the rapidity/efficiency trade-off and the divergence from the MPP under fast changes of wind speed. As shown in Fig. 3, a large step-size means that the system reaches the MPP rapidly, but exhibits large fluctuations around the peak power when it is reached, and thus the efficiency of the system is lower. By using a small step-size, the efficiency of the system is improved, but the dynamic of the system to reach the MPP is slow, which renders the P&O algorithm incapable of tracking the peak power under fluctuating wind speed conditions [17]. Another drawback of the P&O algorithm is the divergence from the MPP in case of rapid changes of wind speed. Since the P&O algorithm is not able to differentiate between power changes caused by the wind speed variations and those resulting from the previous perturbation, the algorithm could diverge from the MPP during a rapid change in wind speed. Fig. 4 illustrates how such a change in wind speed might affect the P&O algorithm decision in determining the direction of the next perturbation, considering the MPP has been reachedand the P&O algorithm oscillates the operating point around the MPP from 1–2–3 and from 3–2–1. In these circumstances, if the wind speed increases rapidly from V1 to V2, and the operating point has been perturbed from 2 to 3, the operating point will be moved to 4 instead of to 3 as shown

Fig. 4. Divergence of P&O algorithm from the MPP under rapidly increasing wind speed.

in Fig. 4, which represents an increase in power and the P&O algorithm will continue perturbing in the same direction, toward point 5. As long as the wind speed continues to increase rapidly, the operating point will be maintained diverging from the MPP. 4. Proposed P&O algorithm 4.1. Principle of the algorithm To overcome the aforementioned problems of the conventional P&O technique, a simple MPPT algorithm is proposed in this work. The idea behind this MPPT technique is to use two distinct operations modes. The first mode (normal P&O mode) is employed when the wind speed varies slowly, in which the conventional P&O with small step-size is sufficient to converge the system toward the MPP. Under a rapid speed variation, the direction and amplitude of the perturbation step are defined based on an auto-adjusted curve (the predictive mode) rather than the output power variation to avoid the divergence of the algorithm from the MPP. This allows to bring the system rapidly to the vicinity of the peak power. The details of this algorithm will be presented below.

302

Y. Daili et al. / Energy Conversion and Management 97 (2015) 298–306

As mentioned in the introduction, the relationship Vdc = f(Idc) with one coefficient has been proved to be suitable for the MPPT control in [13], it can expressed as: 1=2 V dc ðtÞ ¼ K opt Idc ðtÞ

ð13Þ

where Kopt is an optimal gain. If the optimal gain Kopt in (13) is known, the MPPT control can be achieved by estimating the reference dc-link voltage V⁄dc from the measured current Idc, then the duty cycle might be adjusted to force the dc-link voltage to follow its reference V⁄dc. Such a control technique is similar to the other techniques based on optimum curves, it has a fast dynamic in tracking of the MPP. Nevertheless, it is not easy to get a precise value of Kopt for a real wind system, this coefficient is obtained by testing the wind system off-line either experimentally or by simulation. Moreover, the value of Kopt does not remain constant throughout the wind generation system’s operational lifetime due to a possible drift in the system parameters. In the present study, the optimum relationship (13) is used as a reference to help the algorithm to predict the direction and the adequate step-size of the perturbation regardless of variations in the power when the wind speed changes rapidly to drive the system quickly close to the MPP. The perturbing step-size in the predictive mode is defined as the distance between the optimum dc voltage V⁄dc(k) calculated from (13) and the measured voltage Vdc(k) at the sampling time k expressed by:

Ddðk þ 1Þ ¼ aðV dc ðkÞ  V dc ðkÞÞ

ð14Þ

where a is a positive gain, it should be chosen carefully, as a large value of a allows a fast dynamic in tracking of the reference voltage but might result in oscillations around the reference voltage. Fig. 5 illustrates the operating principle of the proposed technique under the predictive mode. If the wind speed changes suddenly from V1 to V2, the operating point moves from 1 to 2. As can be seen from Eq. (14), the next perturbation will be in the direction of increasing the dc-link voltage Vdc(k) to move the operating point toward the optimum curve; the opposite is true when the wind speed decreases from V2 to V1. Also, the perturbing stepsize is auto-adjusted according to the distance between the operating point and the optimum curve. A large step-size is imposed when the operating point is far from the optimum curve for fast convergence to the MPP, whereas a small step-size is applied for an operating point near the optimum curve. The coefficient Kopt is auto-adjusted by the algorithm when the normal P&O mode is switched and the peak power is reached, simply by using the corresponding measurements of dc-link voltage Vdc(k) and dc-current Idc(k) (Eq. (13)). The new value of Kopt(k) will be taken into account in the next use of the predictive mode. To detect the MPP when it is reached, we exploit the fact that the system continues to oscillate around the peak power even

Fig. 5. Operating principle of the proposed technique under the predictive mode.

though it has been reached, due to the change of the perturbation sign. The MPP could be detected by analyzing the variable e(k) obtained from the multiplication of the old sign of the perturbing step and the new one as follows:

eðkÞ ¼ signðDdðk  1ÞÞsignðDdðkÞ

ð15Þ

where k and k  1 correspond to the current and the previous value of the perturbing step, respectively and signðDdðkÞÞ ¼ 1. Note that the variable e(k) can take either +1 or 1. It changes according to the P&O operation phases, such as:  In the tracking phase the sign(Dd(k)) does not change from one iteration k to another in order to move the operating point toward the peak power point, resulting e(k) = +1.  Once the MPP is reached, the sign(Dd(k)) changes for two successive iterations, consequently e(k) = 1. 4.2. Initialization The initial value of the coefficient Kopt(0) should be as close as possible to the optimum value in order to help the algorithm to converge the operating point rapidly in the vicinity of the MPP when a rapid change in wind speed is noticed at the start-up of the system. By assuming that for a specific wind speed the optimum point is close to the generator nominal point [14], thus, the initial coefficient value of the optimum curve can be expressed as:

K opt ð0Þ ¼

V dcrated

ð16Þ

1=2 Idcrated

4.3. Detection of the wind speed changes The wind speed information is necessary for the operation of the proposed algorithm to differentiate between the two modes of operation. Instead of using an anemometer to detect the rapid wind speed variation, the dc-link voltage information is employed. For the synchronous generator, the relationship between the dc-link voltage and the rotational speed can be approximated by a linear equation as follows [13]:

V dc ¼ K u xe

ð17Þ

with Ku is a positive constant. Substituting (2), (9) and (17) into (12), the time derivative of the dc-link voltage is given by:

dV dc P P V3 ¼ qpR2 K u W C P  T e 2K u J 4 dt V dc

! ð18Þ

As can be seen from (18), the dc-link voltage derivative is zero if the wind turbine torque is equal to the torque produced by the generator, which corresponds to an equilibrium point. Any difference between the two torques will be projected into a time derivative of the dc-link voltage. This difference can be caused either by the variation in wind speed or by the changing of the generator torque due to the perturbation. For a correct operation of the proposed MPPT algorithm, it is mandatory to differentiate between the two cases. Taking into account that the detection of a rapid change of wind speed is used only to switch the algorithm from normal P&O mode to predictive mode, this means that when the predictive mode is activated the decision of the algorithm to switch the normal P&O mode is not influenced by detecting a fast wind speed variation. From the other side, under the normal P&O mode the step-size is small and the corresponding dc-link voltage derivative is also

Y. Daili et al. / Energy Conversion and Management 97 (2015) 298–306

small. Consequently, the derivative of the dc-link voltage can be employed to detect any variation of the wind speed when the normal P&O mode is switched. Also, it should be mentioned that the faster the variation of the wind speed is, the larger the Vdc derivative will be, while in the case of a slow variation of the wind speed, the derivative of Vdc is small. In order to detect a rapid speed change, the difference between two successive values of the dc-link voltage DVdc(k) is used, and then the absolute value of DVdc(k) is compared with a certain threshold l, if it is found higher than l, then the variation of the wind speed is rapid

303

This mode remains active until detecting that the step-size calculated according to (14) is lower than the fixed step-size C used in the normal P&O mode. 2. If the step-size calculated in predictive mode is lower than the fixed step-size C, and the wind speed changes slowly (DVdc < l), the algorithm switches the conventional P&O to search for the true MPP. Once the peak power is detected (e(k) = 1), the coefficient of the optimum relationship Kopt(k) is updated.

5. Experimental system and results 4.4. Methodology The algorithm is illustrated by the flow chart shown in Fig. 6, and its operation is summarized as follows.  Firstly, initialize the coefficient Kopt(k) by Eq. (16).  Read the dc-link voltage Vdc(k) and dc-current Idc(k) measurements for each sampling instant Ts.  Calculate the dc-power Pdc(k) from Vdc(k) and Idc(k),and the variation in Pdc(k) and Vdc(k) for two consecutive sampling instants.  Monitor continuously the variation of dc-link voltage DVdc(k) and the perturbing step-size Dd(k). 1. If the absolute value of DVdc (k) exceeds a certain threshold l this means that a rapid change of the wind speed is observed, the algorithm switches to the predictive mode.

Fig. 6. Flowchart of the proposed MPPT algorithm.

To demonstrate the validity of the proposed P&O control algorithm, experimental tests have been carried out. The photograph of the experimental prototype shown in Fig. 7 has been developed at the LIAS-Laboratory, France. It consists of a permanent magnet synchronous motor (PMSM) driven by a commercial inverter unit operating in torque control mode to emulate wind turbine behavior, where the reference torque signal is calculated employing a wind turbine characteristic and the rotor speed measurement is provided by an incremental encoder. The PMSG is coupled directly to the motor and a three-phase diode bridge is employed to rectify the generated AC voltage. A single phase IGBT-based inverter (SEMIKRON) with anti-paralleling diodes (one module SKM 50 GB 12V) and one two-channel driver (SKHI 23/12) plays the role of boost converter to achieve a MPPT control. One Hall-effect CT LEM (PR30) and an isolation amplifier HAMEG (HZ64) are used to measure the dc-current and the dc-link voltage, respectively. A load resistor is connected to the output of the boost to consume the energy produced in the system. Finally, two oscilloscopes are used to visualize the results of the tests. The system is controlled by software running on a dSPACE DS1005 PowerPC board, which offers sixteen 16-bit input channels and four external trigger inputs (DS 2004 A/D board), six 16-bit output channels (DS 2102 D/A board) and sixteen digital outputs with digital pulse patterns (DS5101 Digital Waveform Output board). The entire range of dSPACE I/O boards are accessible via Real-Time Interface Simulink™ blocks. The wind turbine model

Fig. 7. Photograph of the experimental prototype.

304

Y. Daili et al. / Energy Conversion and Management 97 (2015) 298–306

(given by Eqs. (2)–(5)) and the proposed P&O algorithm given in Fig. 6 are both programmed with Matlab/Simulink™ and downloaded in the DS1005 Power PC board. The outputs signals from the DS1005 are the reference torque to the AC driver unit and the PWM signal to control the boost converter. The inputs are the measured dc-voltage and dc-current at the output of the rectifier and the rotor speed provided by the encoder through an incremental encoder interface board (DS 3001). The proposed MPPT algorithm presented in the previous section has been tested experimentally for steps and sinusoidal wind speed profiles. The performances of the MPPT algorithm have been compared with the conventional P&O technique under rapid variations of the wind speed as shown in Fig. 8. The comparison has been accomplished by observing the power coefficient, dc-power and boost duty cycle waveforms for the three cases, namely; the conventional P&O with a small and a large step-size, and the proposed MPPT algorithm. It can be seen from Fig. 9, with a small step-size, the oscillation amplitude around the MPP is lower and the time taken by the system to reach the new MPP when the wind speed change rapidly is quite large. By using a large step-size, the system reaches the MPP with a high dynamic and large power oscillations are visible when the peak power is reached (see Fig. 10). Also, note that the conventional P&O algorithm diverges from the MPP when wind speed changes rapidly from 5 to 8 m/s so that instead of decreasing the duty cycle to move the power coefficient toward the optimum value, the algorithm increases the duty cycle, which results in a diverging of the system from the MPP. Fig. 11 clearly shows the improvement in performance of the proposed algorithm. During rapid wind changes, the duty cycle is adjusted rapidly to an optimum value, which allows the system to track the MPP much faster. The power oscillations when the steady state is established are lower. Furthermore, when the system is subjected to fast wind changes, the power coefficient drop is small compared with the conventional P&O method. This means that the amount of energy captured by the wind energy system under a rapidly fluctuating wind profile using the proposed control scheme is increased over the traditional P&O control.Hence, these results confirm that the proposed P&O control scheme is very fast and efficient compared with the conventional P&O technique. Fig. 12 shows the performance of the proposed P&O algorithm under a mixed wind speed profile with slow and fast wind variations. The algorithm detects the fast wind variations through computing the variation of the dc-link voltage and switches to the

Fig. 9. Conventional P&O algorithm under a small step-size (C = 0.0025).

Fig. 10. Conventional P&O algorithm under a large step-size(C = 0.01).

Fig. 11. Proposed MPPT algorithm.

Fig. 8. Wind speed steps profile.

predictive mode, where the step-size is adapted according to the distance between the operating and optimal point. The system continues operating in this mode till the operating point is near to the optimal one. At that moment, the system switches to normal

Y. Daili et al. / Energy Conversion and Management 97 (2015) 298–306

Fig. 12. Experimental results under mixed wind speed profile using the proposed P&O algorithm (mode equal to zero) means that the normal P&O mode is activated otherwise the predictive mode is switched.

305

Fig. 13. Experimental results under sinusoidal wind speed profile using the proposed P&O algorithm.

P&O mode in which a conventional P&O with small step-size is used to track the true MPP. For a slow change in the wind speed, the normal P&O mode is activated to track the MPP.A sinusoidal wind profile is employed to validate the proposed MPPT control scheme properly. The wind profile used in this test is simulated as a sum of several sinusoidal signals to represent the fluctuating nature of the wind speed as mentioned by authors in the two Refs. [28] and [29]:

V W ðtÞ ¼ 6:3 þ 0:2 sinð3:6645tÞ þ 0:5 sinð1:2930tÞ þ 1:5 sinð0:2665tÞ þ 0:2 sinð0:1047tÞ

ð19Þ

The waveform of the wind profile is shown in Fig. 13. It is clear from Fig. 13, the predictive mode is activated only when the wind speed changes quickly, which allows the system to track the MPP rapidly, while the normal P&O mode is switched when the wind speed varies slowly. Moreover, it is easy to see that the fluctuation of the power coefficient around the maximum value (0.47) is small. This proves that the captured power is optimized. In order to quantify the benefit offered by using the proposed P&O technique in terms of the quantity of the extracted energy, the captured energy is evaluated for the both algorithms under the same wind speed profile. Fig. 14 shows the captured energy for both methods under the same fluctuating wind speed profile of Fig. 13 for a test time duration of 60 s. The extracted electrical energy with the proposed P&O method is 12.82% higher than that of the conventional P&O technique.

Fig. 14. Captured energy with conventional and proposed P&O algorithm under fluctuating wind speed.

6. Conclusion A modified perturbation and observation MPPT technique for small wind generation systems has been proposed in this paper to overcome the limitations of the conventional P&O. The idea behind this strategy is to combine the conventional P&O with a technique based on the optimum relationship. Two modes of operation are used by this algorithm, the normal P&O mode and the predictive mode. The normal P&O mode with small step-size is employed under a slow wind speed variation to track the real MPP with fewer

306

Y. Daili et al. / Energy Conversion and Management 97 (2015) 298–306

oscillations around the peak power when it is reached. For rapid changes in the wind speed, the algorithm tracks the new MPP in two stages: in the first stage, the algorithm switches to the predictive mode in which the step-size is auto-adjusted, in the second stage, the normal P&O mode is used to track the real MPP. The dclink voltage information is used to detect rapid changes in the wind speed and differentiate between the two modes. The proposed MPPT control algorithm has been tested experimentally under various wind speed profiles. The obtained results conclude that, compared with the traditional P&O controller, the proposed algorithm has improved the tracking speed and reduced the power fluctuations in the steady state: the MPP has been tracked successfully even in very fluctuating wind speed conditions. All of these analyses and experiments have demonstrated the effectiveness of the proposed MPPT algorithm. Acknowledgments The authors would like to thank the Ministry of Higher Education and Scientific Research of Algeria for providing financial support. Appendix A. System parameters A.1. Wind turbine parameters Air density q = 1.205 kg/m3, rotor radius R = 1.74 m, optimal TSR kopt = 8.1, maximum power coefficient CP max = 0.47. A.2. PMSG parameters Rated power 6.4 kW, rated speed 3600 rpm, pole pairs number P = 4, stator inductance Ld = Lq = 5.5 mH, stator resistance Rs = 0.57 O, inertia J = 0.01645 kg m2. A.3. Control system parameters Sampling time for MPPT algorithm Ts = 0.1 s, computing sampling time Te = 80 ls, fixed step used in normal P&O mode C = 0.0015, threshold for detecting wind speed changes l = 3 V, gain used to determine the perturbing step-size in the predictive mode a = 0.001. Appendix B. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.enconman.2015. 03.062. References [1] Gsänger S, Pitteloud J. Small wind world report 2014. Published by World Wind Energy Association (WWEA); March 2014, Germany. . [2] Zhipeng Q, Keliang Z, Yingtao L. Modeling and control of diode rectifier fed PMSG based wind turbine. In: IEEE 2011 electric utility deregulation and restructuring and power technologies (DRPT). Weihai, Shandong: IEEE; 6–9 July 2011. p. 1384–8. [3] Ganjefar S, Ghassemi AA, Ahmadi MM. Improving efficiency of two-type maximum power point tracking methods of tip speed ratio and optimum torque in wind turbine system using a quantum neural network. Energy 2014;67:444–56.

[4] Abdullah MA, Yatim AHM, Tan CW, Saidur R. A review of maximum power point tracking algorithms for wind energy systems. Renew Sust Energy Rev 2012;16:3220–7. [5] Urtasun A, Sanchis P, Martín IS, López J, Marroyo L. Modeling of small wind turbines based on PMSG with diode bridge for sensorless maximum power tracking. Renew Energy 2013;55:138–49. [6] Shen B, Mwinyiwiwa B, Yongzheng Z, Ooi BT. Sensorless maximum power point tracking of wind by DFIG using rotor position phase lock loop (PLL). IEEE T Power Electr 2009;24:942–51. [7] Hung LC, Hsu YY. Effect of rotor excitation voltage on steady-state stability and maximum output power of a doubly fed induction generator. IEEE T Ind Electron 2011;58:1096–109. [8] Haque ME, Negnevitsky M, Muttaqi KM. A novel control strategy for a variable speed wind turbine with a permanent magnet synchronous generator. In: IEEE 2008 industry applications society annual meeting (IAS). Edmonton: IEEE; 5-9 October 2008. p. 1–8. [9] Thongam JS, Bouchard P, Ezzaidi H, Ouhrouche M. Wind speed sensorless maximum power point tracking control of variable speed wind energy conversion systems. In: In: IEEE 2009 electric machines and drives conference (IEMDC). Miami: IEEE; 3–6 May 2009. p. 1832–7. [10] Tan K, Islam S. Optimum control strategies in energy conversion of PMSG wind turbine system without mechanical sensors. IEEE T Energy Convers 2004;19:392–9. [11] Chen J, Chen J, Gong C. New overall power control strategy for variable-speed fixed-pitch wind turbines within the whole wind velocity range. IEEE T Ind Electron 2013;60:2652–60. [12] Zhang HB, Fletcher J, Finney SJ, Williams BW. One-power-point operation for variable speed wind/tidal stream turbines with synchronous generators. IET Renew Power Gen 2011;5:99–108. [13] Yuanye X, Ahmed KH, Williams BW. Wind turbine power coefficient analysis of a new maximum power point tracking technique. IEEE T Ind Electron 2013;60:1122–32. [14] Kortabarria I, Andreu J, Alegría IM, Jiménez J, Gárate JI, Robles E. A novel adaptative maximum power point tracking algorithm for small wind turbines. Renew Energy 2014;63:785–96. [15] Datta R, Ranganathan V. A method of tracking the peak power points for a variable speed wind energy conversion system. IEEE T Energy Conver 2003;18:163–8. [16] González LG, Figueres E, Garcerá G, Carranza O. Maximum power point tracking with reduced mechanical stress applied to wind energy conversion systems. Appl Energy 2010;8:2304–12. [17] Kazmi SMR, Goto H, Hai-Jiao G, Ichinokura O. Review and critical analysis of the research papers published till date on maximum power point tracking in wind energy conversion system. In: In: IEEE 2010 energy conversion congress and exposition (ECCE). Atlanta, GA: IEEE; 12–16 September 2010. p. 4075–82. [18] Cheng M, Zhu Y. The state of the art of wind energy conversion systems and technologies: a review. Energy Convers Manage 2014;88:332–47. [19] Koutroulis E, Kalaitzakis K. Design of a maximum power tracking system for wind energy conversion applications. IEEE T Ind Electron 2006;53:486–94. [20] Dalala ZM, Zahid ZU, Wensong Y, Younghoon C, Jih-Sheng L. Design and analysis of an MPPT technique for small-scale wind energy conversion systems. IEEE T Energy Conver 2013;28:756–67. [21] Kesraoui M, Korichi N, Belkadi A. Maximum power point tracker of wind energy conversion system. Renew Energy 2011;36:2655–62. [22] Eltamaly AM, Farh HM. Maximum power extraction from wind energy system based on fuzzy logic control. Electr Pow Syst Res 2013;97:144–50. [23] Belhadji L, Bacha S, Munteanu I, Rumeau A, Roye D. Adaptive MPPT applied to variable-speed micro-hydropower plant. IEEE T Energy Conver 2013;28:34–43. [24] Elnaggar M, Abdel-Fattah AL, Elshafei HA. Maximum power tracking in WECS (Wind energy conversion systems) via numerical and stochastic approaches. Energy 2014;74:651–61. [25] Abdeddaim S, Betka A, Drid S, Becherif M. Implementation of MRAC controller of a DFIG based variable speed grid connected wind turbine. Energy Convers Manage 2014;79:281–8. [26] Nasiri M, Milimonfared J, Fathi SH. Modeling, analysis and comparison of TSR and OTC methods for MPPT and power smoothing in permanent magnet synchronous generator based wind turbines. Energ Convers Manage 2014;86:892–900. [27] Hong CM, Chen CH, Tu CS. Maximum power point tracking-based control algorithm for PMSG wind generation system without mechanical sensors. Energ Convers Manage 2013;69:58–67. [28] Dahbi A, Hachemi M, Nait-Said N, Nait-Said MS. Realization and control of a wind turbine connected to the grid by using PMSG. Energ Convers Manage 2014;84:346–53. [29] Mirecki A, Roboam X, Richardeau F. Architecture complexity and energy efficiency of small wind turbines. IEEE T Ind Electron 2007;54:660–70.