Energy Conversion and Management 157 (2018) 587–599

Contents lists available at ScienceDirect

Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman

Maximum power extraction for wind turbines through a novel yaw control T
solution using predicted wind directions
⁎
Dongran Songa, Jian Yanga, Xinyu Fanb, , Yao Liuc, Anfeng Liua, Guo Chend, Young Hoon Jooe
a
School of Information Science and Engineering, Central South University, Changsha, China
b
School of Automation, Beijing Institute of Technology, Beijing, China
c
Guangdong Power Grid Corp, Zhuhai Power Supply Bur, Zhuhai, China
d
School of Electrical Engineering and Computing, University of Newcastle, Callaghan, Australia
e
Department of Control and Robotics Engineering, Kunsan National University, Kunsan, Republic of Korea

A R T I C L E I N F O A B S T R A C T

Keywords: For modern horizontal axis wind turbines (WTs), a yaw drive mechanism is utilized to adjust the nacelle position
Maximum power extraction to face the wind direction. Depending on historical signals from wind direction sensors, conventional yaw
Yaw control control methods could not provide sufficient performance in tracking winds, and thus result in a reduction of
Wind direction prediction wind power extraction. This issue needs to be tackled using advanced control solutions. Taking advantage of
ARIMA-KF
predicted wind directions, a novel control solution is proposed in this study. Specifically, the proposed solution
Model predictive control
refers to a novel control structure that consists of a wind direction predictive model and a novel yaw control
method. Under the proposed control structure, a hybrid autoregressive integrated moving average method-based
Kalman filter (ARIMA-KF) model is used to predict the wind direction, and two novel yaw control methods are
proposed: one created by using the predicted wind direction as the tracking reference, and the other based on a
model predictive control (MPC) using a finite control set. To demonstrate the feasibility and the superiority of
the proposed solution, two novel yaw controllers are developed and tested through some simulation tests using
industrial data. Their performance is compared to the one of two industrial yaw controllers. Comparison results
show that the two novel yaw controllers are capable of reducing yaw error, and thus increase wind power
extraction for the WTs. Meanwhile, it is noticeable that the MPC-based controller has an advantage in the aspect
of reducing yaw actuator usage.

1. Introduction The operation of the yaw system may affect performance of the WT.
On the one hand, a yaw misalignment leads to a decreased wind power
As the increasing demands of wind energy, the focus of research capture. Theoretically, the wind power captured by a horizontal axis
today in wind turbines (WTs) lies in maximizing the power production WT is decreased by the cube of the yaw error [4]. Although empirical
per unit investment. To make wind energy more competitive with other data have shown that the relationship could be cosine-squared instead
sources of renewable energy, optimal solutions have been developed of cosine-cubed [5], it is obvious that the yaw error results in the power
constantly for WTs [1], where the control technology plays an indis- reduction of the WT. On the other hand, a yaw misalignment may bring
pensable role that directly affects performance of the WTs in the both about an increment of component loads. The impact of yaw misalign-
aspects of power production [2] and component loads [3]. Modern WTs ment on loads of the WTs has been investigated and validated by re-
with horizontal axis have three control actuators: pitch actuator, torque searchers using calculation and measurement methods. For instance,
actuator, and yaw actuator. The former two actuators are considered as Schepers conducted a comparison investigation between calculations
the two dominating ones, since they can provide a fast response that and measurements on a small WT with 10 m rotor diameter in yaw,
answers to the rapid variation of wind force. Accordingly, there are which revealed that the yaw misalignment had effects on blade root and
large quantities of literature that focus on control methods for the pitch shaft loads on a sectional level [6]. Boorsma presented a report of
and torque actuators. By comparison, the literature about the yaw power and loads for a 2.5 MW WT in yawed flow conditions, in which
system control is limited. Nevertheless, the function of the yaw system the edgewise fatigue equivalent loads were found to be increased along
should not be neglected. with the increasing yaw error [7]. Kragh et al. [8] showed the potential

⁎
Corresponding author.
E-mail address: [email protected] (X. Fan).

https://doi.org/10.1016/j.enconman.2017.12.019
Received 18 September 2017; Received in revised form 15 November 2017; Accepted 6 December 2017
0196-8904/ © 2017 Elsevier Ltd. All rights reserved.
D. Song et al. Energy Conversion and Management 157 (2018) 587–599

Nomenclature N number of distribution zones of yaw error

θyej averaged yaw error at the j th zone
Abbreviations f j ∈ [0,1] distribution probability of θyej in the j th zone
(cos(θye ))eq equivalent cosine of yaw error
WT wind turbine Pa wind power extracted by a horizontal axis WT
MPP maximum power point Pred,Pideal reduced power extraction and ideal power extraction
HCM hill climbing method k the kth sampling period
ARIMA autoregressive integrated moving average Ts sampling period
KF Kalman filter Tc control period
MPC model predictive control Ah1,Ah2,Ah3 amplitude thresholds predefined in yaw control al-
MY Ming Yang gorithm
NREL National Renewable Energy Laboratory Th1,Th2,Th3 time thresholds predefined in yaw control algorithm
MAE mean absolute error θnp (k ) nacelle position measured at kth sampling period
RMSE root mean squared error θnṗ (j ) permissible yaw speed
MAPE mean absolute percentage error ̇ (k )
θnp yaw speed during the kth control period
QF quality function w1,w2,w3 weighting factors in the quality function
SCADA Supervisory Control and Data Acquisition System θwd (k ) measured wind direction sampled at the kth sampling
period
Symbols θwd (k + 1|k ) predicted wind direction at the kth sampling period
θye (k + 1|k )
ρ air density θye (k + 1|k ) predicted yaw error at the kth sampling period
Ar rotor area 10s 30s 60s
θye ,θye ,θye mean wind directions averaged at the sampling periods
Cp aerodynamic power coefficient of 10 s,30 s,60 s
V0 free stream wind speed θwd (k + 1|k )Ts = 10s,θwd (k + 1|k )Ts = 30s,θwd (k + 1|k )Ts = 60s predicted
θye yaw misalignment error mean values of wind direction at the kth sampling period,
θwd wind direction Ts = 10 s,30 s,60 s
θnp nacelle position θye (k + 1|k )Ts = 10s,θye (k + 1|k )Ts = 30s,θye (k + 1|k )Ts = 60s predicted mean
tyaw yaw action time values of yaw error at the kth sampling period,
Cyaw yaw action count Ts = 10 s,30 s,60 s
ξ reduction factor of wind power extraction caused by yaw
error

of alleviating blade load variations induced by the wind shear through Table 1. The employed techniques are broadly categorized into four
yaw misalignment for wind speeds above rated wind speed. From their types: free of measurement, normal measurement, advanced measure-
studies, it is observed that the operation of the yaw system significantly ment and indirect measurement. Accordingly, relevant control methods
affects the performance of the WTs. A study of operating WTs revealed a can be also categorized into four types and they have the following
fact that there was a static yaw error of 10 degree for wind speeds features:
below 20 m/s and 5 degree for wind speeds above 20 m/s, which un-
avoidably reduced wind power extraction of the WTs [9]. Besides, an • Controls without wind direction measurement, which originates
early survey of failures in wind power systems showed that the portion from early WTs limited by the wind measurement technology.
of downtime caused by yaw failure comprised 13.3% of the total Because the main objective of yaw control system is to maximize
downtime, and the yaw system failure rate comprised 6.7% [10]; and a wind power extraction, the mechanism for controls without wind
recent analysis for wind turbine reliability concluded that the failure direction measurement is to directly search the maximum power
rate of wind turbines was increased up to 12.5% [11]. Thus, the con- point (MPP). Hill climbing method (HCM) was proposed to find the
trols for the yaw system deserve more attention than they received. desired yaw angle corresponding to the MPP [12], and bisecting-
In the literature, the control methods for yaw systems are directly plane algorithm was presented to enhance the efficiency and accu-
relevant to the measurement techniques which can be seen from rateness of conventional HCM [13]. Besides, a combined maximum

Table 1
Summary of the yaw control methods recorded in the literature.

Measurement Control objective Method WT capacity Refs.

Free Searching optimal power HCM < 50 kW [12]

Searching optimal power Modified HCM 1.5 MW [13]
Tracking optimal rotor speed PI 1.1 kW/2.5 MW [14]

Normal Tracking wind direction Fuzzy-PID Unclear [15]

Tracking wind direction Logic control 2 kW [16]
Tracking wind direction Logic control 600 kW [17]
Tracking wind direction Logic control 1.5 MW [18]

Advanced Tracking wind direction Logic control 600 kW [19–21]

Tracking wind direction Conventional MPC 5 MW [22]

Indirect Maximizing power production/minimizing structural loads Conventional MPC 1 MW [23]

Searching optimal power Logic control 1.1 kW [24]

588
D. Song et al. Energy Conversion and Management 157 (2018) 587–599

power point tracking and yaw control technique aiming at tracking direction, which aims at extracting maximum wind power for the WTs
the optimal rotor speed was presented in [14]. In theory, this type of with a moderate yaw actuator usage. With regards to the literature, the
controls may provide better performance than normal measure- major contributions of this paper are threefold.
ment-based one which may suffer from the inaccurate measure-
ments disturbed by operation of the WTs. However, the MPP of WTs • This paper proposes a novel control structure that consists of a wind
is changing following the variation of wind speed, besides the wind predictive model and a novel control method. To do this, a hybrid
direction. As a result, the real difficulty for controls without mea- autoregressive integrated moving average method-based Kalman
surements consists in locating the MPP, which remains an open issue filter (ARIMA-KF) model is used to predict the wind direction. Then,
in wind energy research community. two novel yaw control strategies are proposed: one created by using
• Controls with normal measurements, which are currently widely the predicted wind direction as the tracking reference, and the other
employed by modern WTs. This type of controls employs active yaw based on a model predictive control (MPC).
control strategies to face the turbine into the wind by acquiring • This paper introduces the novel MPC with a finite control set for
signals from wind vanes installed at the rear of the nacelle. Although controlling the yaw system. The predicted wind directions are dis-
a fuzzy-PID strategy is introduced to track the wind direction, the crete data obtained at every sampling period, and thus the pre-
motivation is unclear [15]. By comparison, most of control strate- dictive model for yaw control system including predicted wind di-
gies employ some predefined logic controls [16–18], where the yaw rections are internally discrete model. Thus, compared with the
actuators are activated when the yaw error measured by wind vanes conventional MPC with a continuous control set, the proposed MPC
exceeds some thresholds. Although the strategies are simple, the strategy is more suitable for controlling yaw systems using predicted
difficulty consists in obtaining a proper reference to adjust the na- wind directions. On one hand, the yaw command sets during each
celle position. The measurements from the wind vanes are always sampling period can be categorized into a finite control set rather
mixed with disturbing noises and outliers. Meanwhile, the wind than a continuous control set. On the other hand, the algorithm
direction constantly changing is different from the future wind di- solution is directly selected from available control sets, and thus
rection. Consequently, the controls with normal measurements reduces the computational burden.
could not provide sufficient performance [9–11]. • This paper discusses how wind power extraction of the WTs can
• Controls with advanced measurements, which have been recently benefit from the predicted wind direction-based control structure
proposed in some advanced wind energy projects. To obtain accu- and introduces four performance indexes to evaluate the yaw con-
rate wind direction measurement, remote sensing instruments based trol system performance, namely yaw error, yaw action time, yaw
on Lidar and hypersonic (Sodar) technologies have been employed. action count, and power reduction factor.
With the powerful remote measurement, performance of the yaw
control system can be potentially enhanced with simple logic con- The remainder of this work is organized as follows: the wind di-
trols [19–21]. Under the assumption that the wind direction preview rection measurement, two industrial yaw control methods, and per-
provided full information about wind direction over the future 60 s, formance indexes of yaw system are discussed in Section 2; and Section
a conventional model predictive control (MPC) can provide an in- 3 describes the novel yaw control solution. It is followed by simulation
crement of 8% wind power extraction and some fatigue load re- tests and result discussions in Section 4. Finally, conclusions are drawn
ductions during an extreme direction change [22]. Nevertheless, the in Section 5.
solutions are expensive and thus, affordable only for high power
WTs [14]. 2. Yaw control system of industrial WTS
• Controls with indirect measurements, which have gradually gained
attentions by researchers. The short-term prediction of wind direc- 2.1. Wind direction measurement
tion is incorporated into a conventional MPC for the yaw control
system, which aims at achieving structural loads minimization and 2.1.1. Wind direction sensor
power production maximization simultaneously [23]. Besides, wind For current industrial WTs, the wind direction measurement is
direction is estimated by an inverted function of wind power and normally provided by one or two wind direction sensors which are
wind speed, and then is employed into the yaw control system with installed on the rear of the nacelle. A typical wind direction sensor is
logic controls [24]. The controls with indirect measurements may be shown in Fig. 1, which is a product of Kriwan with number INT30 [33].
potential for improving performance yaw system, but the presented Its basic specification is given in Table 2.
control solution ignoring prediction algorithms is incomplete and
needs to be further investigated. 2.1.2. Wind direction measurement
Fig. 2 shows the principle of wind direction measurements. Since
From the above, it is concluded that controls with advanced mea- the wind direction sensor rotates along with the WT’s nacelle, it mea-
surements and with indirect measurements may improve performance sures a yaw error rather than the wind direction. Besides the wind
for the yaw system, because the wind direction information in the fu-
ture can be utilized. By comparison to the advanced measurements, the
indirect measurements are normally cost-effective. Until now, some
developed forecasting approaches have been proposed, such as the
wind-power prediction by Azimi et al. [25], Yesilbudak et al. [26], and
Mohammadi et al. [27]; wind-speed prediction by Zameer et al. [28],
Zhang et al. [29], and Noorollahi et al. [30]; and wind direction pre-
diction by Ouyang et al. [31] and Song et al. [32]. These studies ad-
dressed the prediction issues relevant to the wind source, but none of
them tried to employ the predicted wind data into the control appli-
cation of the WTs.
Motivated by the aforementioned observations, this study proposes
a novel control solution by taking advantage of the predicted wind
Fig. 1. The outlook of a typical wind direction sensor [33].

589
D. Song et al. Energy Conversion and Management 157 (2018) 587–599

Table 2 to reduce θye to zero. Since θye is normally disturbed by operation of the
Specifications of a typical wind direction sensor [33]. WT operation, it is usually filtered before being employed as the con-
troller reference.
Parameter Value

Measuring range 0–360° 2.2. Yaw control methods with normal measurements
Resolution <1°
Accuracy ± 2.5°
For megawatt WTs, the yaw speed is normally designed in a range of
Start-up wind speed <0.4 m/s
Permitted ambient temperature −40 to + 70° [0.2deg/s,0.8deg/s], because a fast movement of the yaw system may
Permissible relative humility 0–100% . r . h. induce high loads to the WT. Meanwhile, to avoid over-usage of yaw
actuator, the yaw system is always activated at discrete intervals. The
yaw control methods with normal measurements for modern WTs
normally employ the logic controls, where the yaw actuators are acti-
North vated when the yaw error measured by wind vanes exceeds some
thresholds. In this study, two yaw control algorithms with normal
measurements are given and used as the baseline control algorithms.
Wi
nd 2.2.1. Yaw control algorithm I
dir
ect The yaw control algorithm I is taken from the MY (Ming Yang)
ion
1.5 MW WTs manufactured by China Ming Yang Wind Power, which is
Blade rotor illustrated in Fig. 3 and includes following four steps [18]:

(1) Yaw error filtering. In this step, the yaw error is averaged by three
wd averaged units with different averaged periods: 10 s, 30 s and 60 s;
10s 30s 60s
and thus, three averaged yaw errors (denoted as θye ,θye ,θye ) are
ye obtained.
np
Nacelle 10s 30s 60s
(2) Yaw error judgment. In this step, θye ,θye ,θye are compared to three
predefined amplitude thresholds (denoted as Ah1,Ah2,Ah3) re-
spectively. When any of the three comparisons is satisfied, and the
sustaining time is longer than the corresponding predefined time
thresholds (denoted as Th1,Th2,Th3), the control loop goes on;
otherwise, the control loop is ended for this cycle.
(3) Yaw time calculation. In this step, the yaw action time is calculated
Wind wane
using the corresponding averaged yaw error divided by the yaw
Fig. 2. Schematic diagram of wind direction measurement [32]. speed (Yawspeed ).
(4) Yaw movement. In this step, yaw movement is activated during the
activation time. As a consequence, the nacelle moves toward facing
the wind.
direction sensor, there is another transducer used in yaw control
system, namely, nacelle position transducer. The yaw error is the dif- In Fig. 3, the parameters utilized are summarized in Table 3.
ference between the wind direction and nacelle position, which can be
calculated by 2.2.2. Yaw control algorithm II
θye = θwd−θnp (1) The yaw control algorithm II is taken from the NREL CART3
(Controls Advanced Research Turbine 3-Bladed) turbine, which is il-
Since the wind direction varies along with the time, the yaw control lustrated in Fig. 4. The control logic is comparably simple and detailed
system is needed to adjust the nacelle position to track the wind di- in [16,19,20]. The yaw error is filtered by two low-pass filters, one with
rection. From Eq. (1), it is obtained that tracking the wind direction is a time constant of 1 s, and the other 60 s, producing a more quickly

Step 1 Step 2 Step 3 Fig. 3. Schematic of the yaw control algorithm

for the MY 1.5 MW WTs [18].
10 s
10s 10 s
Yes ye
T( Ah1) Th1
ye ye
Yaw _ speed Step 4
No
Start yaw
30s Yes 30 s movement
ye T( 30 s
Ah2) Th2 ye
ye ye
Yaw _ speed
No

60 s
60s 60 s
Yes ye
ye T( ye Ah3) Th3
Yaw _ speed
No

Do nothing this cycle

590
D. Song et al. Energy Conversion and Management 157 (2018) 587–599

Table 3
1
Yaw control parameters used in Fig. 3. wd (k 1| k ) Wind direction wd (k )
Yaw system
Parameter Ah1 Ah2 Ah3 Th1 Th2 Th3
prediction

Values 13° 10° 8° 10 s 5s 5s

2
Novel yaw control Yaw Command

changing measurement of error and a more slowly changing measure- Fig. 5. Block diagram of the novel yaw control solution.
ment. The quickly changing measurement error is integrated and
monitored. When the integrated error (notated AccErr in Fig. 4) reaches
a value such that it has been off by 10 degrees for 10 min, the yaw angle
of the turbine is moved to the location given by the slowly changing Pa = ρAr Cp V03 (cos(θye ))2 /2 (5)
measurement of the error.
To better check the influence of θye with different yaw control al-
gorithm on Pa , we introduce the power reduction factor (denoted as ξ )
2.3. Performance indexes for evaluating yaw control system which is calculated based on the following formula

To evaluate performance of the yaw control system, two aspects ξ = Pred/ Pideal = 1−((cos(θye ))eq)2 (6)
should be taken into account, one to measure the accuracy of the wind where (cos(θye ))eq is calculated by
direction tracking, and the other to measure the usage of the yaw ac-
N
tuators. The accuracy of the wind direction tracking can be evaluated (cos(θye ))eq = ∑ j=1 (cos(θyej))2 f j , θyej ∈ [−180°,180°]
by checking the statistical value of yaw error, and the usage of the yaw (7)
actuators by calculating the yaw action time and action count.
The yaw error is the difference between the wind direction and 3. Novel yaw control solution
nacelle position. In this paper, mean absolute error (MAE) and root
mean squared error (RMSE) of θye are calculated to evaluate the accu- From the above, it is clear that the available yaw control algorithms
racy of the wind direction tracking. Their calculations are expressed as depend entirely on the historical data of the measured yaw error. To
N handle the issue of the measured yaw error disturbed by operation of
⎧ 1 the WT, averaging or low-pass filters are normally utilized to provide
⎪ MAE (θye ) = N ∑ |θye (t )|
t=1 the reference for the yawing movement, but they may bring about a
⎨
⎪ RMSE (θye ) = 1 N
∑t = 1 (θye (t ))2 time delay. Besides, the filtered measurements only reflect the past yaw
⎩ N (2) error and may be different from the future wind directions. Therefore, it
is reasonable to use prediction values, which may improve the perfor-
The yaw action time (denoted as tyaw ) is the activation time of the
̇ (t ) ), tyaw is mance of the yaw control system. Hence, this study proposes a novel
yaw actuators. When there is a yaw speed (denoted as θnp
solution. Fig. 5 shows its structure.Compared to the methods shown in
cumulated and it can be expressed as
Figs. 3 and 4, the proposed method comprises two parts: a wind di-
N rection prediction module (depicted as 1) and a novel yaw control
tyaw = ∑ ̇ (t )| > 0)
(|θnp module (depicted as 2). The module 1 is designed to estimate the future
t=1 (3)
wind direction θwd (k + 1|k ) at the kth sampling period (Ts ), and the
The yaw action count (denoted as Cyaw ) is the activation count of the module 2 uses θwd (k + 1|k ) to make a decision for the yaw system ac-
yaw actuators. When the current yaw speed is different from the last tion in the next control period. These two modules are explained as
one, Cyaw is increased by one and it can be expressed as follows.

⎧Cyaw (t −1), ̇ (t ) = θnp

∃ θnp ̇ (t −1) 3.1. Wind direction prediction
Cyaw (t ) =
̇ (t ) ≠ θnp
⎨Cyaw (t −1) + 1, ∃ θnp ̇ (t −1)
⎩ (4)
Wind direction prediction approaches have been recently addressed
Since θye has a direct influence on wind power extraction of the WT, by some researchers, such as the data mining algorithm-based pre-
the wind power Pa extracted by a horizontal axis WT should be also dicting approach by Ouyang et al. [31], and the ARIMA-KF based hy-
evaluated and it can be expressed by [5] brid approach by Song et al. [32]. The ARIMA model is suitable for

Fig. 4. Schematic of the CART3 yaw controller [16].

ye Lowpass erfast 2 Do nothing
TC=1s
AccErr sign(erfast )* erfast this cycle

Lowpass
erslow
TC=60s
Yes No
Yaw position AccErr
yaw setpoint Threshold

Yes Yaw setpoint No Stop the

Yaw to yaw
setpoint in cone? turbine

591
D. Song et al. Energy Conversion and Management 157 (2018) 587–599

capturing short-range correlations and has been widely in a variety of errors rather than the filtered measurements. These differences may
forecasting applications [34], but it has a difficulty of adjusting the change the performance of the yaw control system, which will be dis-
model’s parameters when the time series contains new information. cussed in Section 4.
Thus, the ARIMA model in combination with a Kalman filter is used to
solve this problem [35]. In this study, the ARIMA-KF approach is em-
3.2.2. Method II
ployed because of its capability of providing accurate short-term pre-
The MPC enables a novel solution of a control problem while hon-
diction results. Here, it is worthy noticing that the ARIMA-KF methods
oring the constraints imposed upon by the designers of the WT [38].
is not the only one that can be utilized here, and other machine
Meanwhile, it has a potential advantage that it is able to predict the
learning-based prediction approaches may provide more accurate re-
behavior of a plant in future using its model. There are two types of
sults [36,37].
MPC [39]: the MPC with a continuous control set, which requires sol-
To build an ARIMA-KF predictive model for predicting the wind
ving a quadratic programming problem on line and thus has a com-
direction, the relevant procedures refer to three steps, which are de-
putation burden, and the MPC with a finite control set, which has re-
monstrated in Fig. 6. Its details are given as follows:
cently been proposed to control pitch angle and generator torque for
the WTs [40]. The control command for the yaw system can be cate-
(1) Choosing the original wind direction series. Since the yaw con-
gorized into a finite control set; therefore, the MPC with a finite control
troller uses the mean values of wind direction data, three types of
set is proposed and employed in this study. As suggested in [39,40], the
wind direction data with different averaged periods: 10 s, 30 s and
MPC with a finite control set can be developed by following four steps:
60 s, are used to form the original data series.
(2) Defining the ARIMA initial model. In this step, the three types of
(1) Define a quality function (QF).
original data series are used to train the ARIMA predictive models,
(2) Build a model of the target system and its possible control set.
in which a three iterative step is included: model identification,
(3) Build a model of the controlled variables for prediction.
parameter estimation, and diagnostic checking.
(4) Evaluate the predicted value for each control set and select the one
(3) Designing the KF-based predictive model. The KF is generally de-
with minimal value of QF.
scribed as an approach consisting of a state equation and a mea-
surement equation. The system state equation is created by re-
The block diagram of the proposed MPC-based yaw control method
formulating the ARIMA model, and the measurement equation is
is shown in Fig. 8, which mainly refers to the construction of the pre-
defined according to the relation of the measurement variables and
dictive model and the QF. In this study, the primary control objective is
system state variables. Finally, the newly predicted values are ob-
to track the predicted wind direction, which in other words is to de-
tained by using the KF iteration algorithm. KF algorithm also in-
crease the yaw error. Thus, the predictive model is needed to predict
cludes three steps: system modeling, measurement update and time
the yaw error. And, the secondary control objective is to avoid the yaw
update filter gain.
actuator over-usage. In the QF, these two control objectives can be
(4) Outputting one-step prediction values. The ARIMA-KF predictive
evaluated using a convex function. In this way, both objectives of wind
models may provide multiple-step predictions, but multiple-step
direction tracking and actuator usage can be taken care by adjusting the
predictions are with the considerable prediction error. Thus, only
corresponding weighting factors. The proposed method is detailed as
the one-step prediction values are used in this study.
follows.

3.2. Two novel yaw control methods a. Finite control set for yaw actuator

In this section, the two novel yaw control methods are proposed and For modern WTs, the yaw system is driven by yaw motors. When the
discussed: one is created by using wind direction predictions as the yaw system is activated, the nacelle will be moved by the yaw motors at
tracking references (Method I), and the other is based on the MPC using a certain speed; otherwise, the nacelle position remains unchanged. The
a finite control set (Method II). permissible actions for the yaw actuators comprise three elements and
they can be categorized into a finite set as follows:
3.2.1. Method I
The schematic diagram of Method I is illustrated in Fig. 7, where ⎧− Yawspeed, j = 0
̇ (j ) = Yawspeed, j = 1
θnp
three types of predicted wind directions with different sampling per- ⎨
iods: 10 s, 30 s, and 60 s, are used to provide the tracking references. As ⎩ 0, j = 2 (8)
a consequence, the resulted controller has three control periods. In
other words, the controller updates its output when the inputs are up-
b. Predictive model
dated at each 10 s, 30 s, and 60 s.
When comparing with the yaw control method shown in Fig. 3, the
Since the control objective is to decrease the yaw error, the yaw
proposed method shown in Fig. 7 has a main distinction that the update
error is selected as the predictive variable, which can be expressed as
period of its output depends on Ts of the predictive model. Besides, it
follows:
differs in following details: in Step 1, its inputs uses three types of
predicted yaw errors instead of filtered measurements; in Step 2, it ̇ (j ) Tc
θye (k + 1|k ) = θwd (k + 1|k )−θnp (k )−θnp (9)
employs only three amplitude thresholds ( Ah1,Ah2,Ah3), without time
thresholds; and in Step 3, its tracking references use the prediction yaw where the control period Tc is equal to the sampling period Ts .

Fig. 6. Development of the ARIMA-KF wind-di-

1 2 3 4
rection prediction model.
Original wind ARIMA initial One-step
KF model
direction series model prediction

592
D. Song et al. Energy Conversion and Management 157 (2018) 587–599

Step 1 Step 2 Step 3

(k 1| k )Ts (k 1| k )Ts (k 1| k )Ts

wd 10 s ye 10 s Yes ye 10 s
ye (k 1| k )Ts 10 s Ah1
wd (k 1| k ) ye (k ) Yaw _ speed
Step 4
No
Start yaw
(k 1| k )Ts (k 1| k )Ts (k 1| k )Ts movement
wd 30 s ye 30 s Yes ye 30 s
ye (k 1| k )Ts 30 s Ah2
wd (k 1| k ) ye (k ) Yaw _ speed
No

(k 1| k )Ts (k 1| k )Ts (k 1| k )Ts

wd 60 s ye 60 s
(k 1| k )Ts Ah3 Yes ye 60 s
ye 60 s
wd ( k 1| k ) ye ( k )
Yaw _ speed
No

Do nothing this cycle

Fig. 7. Schematic diagram of method I.

The yaw system Fig. 8. Block diagram of method II.

w1 , w2 , w3

wd ( k )
np ( k ) Wind
ye ( k 1| k )
Minization
direciton
np ( j) of QF
prediction
Predictive
model
1 2

np ( k )
wd ( k 1| k )

c. Quality function model. The first two elements are linear functions which are the pe-
nalties on the increment of yaw error and yaw actuator action distance,
The quality function is a representation of the control objectives whereas the third element is established to punish the action count,
which are usually related to make the variables follow the references. which uses a logical function where the value is w3 when the current
Hence, the quadratic value of the error, or its absolute value, is com- yaw action is different from the last one.
monly employed to find the minimum value of the quality function. In
this study, minimization of yaw error and yaw actuator usage should be
achieved simultaneously, so they will be combined in a form of sum. Apply np (op)
Since the yaw error has a direct influence on the power production
of the WT in a cosine-squared fashion, the first part (QF (1) ) of the
Obtaining np (k ), wd (k 1| k )
quality function can be chosen as follows:

QF (1) = w1 (θye (k + 1|k ))2 (10)

QF (op)
where w1 is the weighting factor to be adjusted.
The yaw actuator usage can be evaluated by two indexes: actuator for end
action distance and action count. These two indexes are combined to for(j=0:2)
form the second part (QF (2) ) of the quality function, which is given as
follows:
ye (k 1| k ) wd (k 1| k ) np (k ) np ( j)Tc
̇ (j )|Tc + w3 (θnp
QF (2) = w2 |θnp ̇ (k + 1) ≠ θnp
̇ (k )) (11)

Combining Eq. (11) with Eq. (10), the final QF is obtained as fol-
lows:
̇ (j )|Tc + w3 (θnp
QF = w1 (θye (k + 1|k ))2 + w2 |θnp ̇ (k + 1) ≠ θnp
̇ (k ))
If(QF(j)<QF(op))
(12) QF(op) = QF(j); op=j;
End
As shown in Eq. (12), it is clear that one of the aspects where the
MPC shows its flexibility is the inclusion of different elements in a
Fig. 9. Flowchart of the implemented control algorithm.

593
D. Song et al. Energy Conversion and Management 157 (2018) 587–599

250 Fig. 10. Original wind data in a specific day: (a)

Original wind direction wind direction; (b) wind speed.
Wind direction [degree]

Averaged wind direction/10s

200 Averaged wind direction/30s
Averaged wind direction/60s
150

100

50

0
0 4 8 12 16 20 24
Time [h]
(a)
20

15
Wind speed [m/s]

10

0
0 4 8 12 16 20 24
Time [h]
(b)

d. Control algorithm data shown in Fig. 10(a), where the first 20 h are used to establish the
models, and the leaving 4 h to check the model validity. Since the re-
The control algorithm is detailed in Fig. 9 as a flow diagram. sults have been detailed in [32], only the wind direction prediction
As shown in Fig. 9, the minimization of the quality function is im- results in the form of time series are chosen to show and illustrated in
plemented as a “for” cycle predicting for three permissible control ac- Fig. 11. Besides, three types of performance indexes (including MAE,
tion, evaluating the quality function, and storing the minimum value RMSE, and MAPE) are computed and summarized in Table 4. These
and the index value of the corresponding control action. By the flow- results show that the three types of the prediction results are in good
chart, the proposed control algorithm can be easily implemented in the agreements with the original data. Therefore, these results will be
wind turbine systems. employed in the proposed novel yaw methods.

4. Method validations 4.2. Validation of novel yaw control methods

The proposed methods are tested using wind data obtained from a After obtaining the wind direction prediction data, two novel yaw
wind farm located in Guangdong Province of China, which was saved in controllers were developed using the Matlab software: the first one
a SCADAS (Supervisory Control and Data Acquisition System) per (denoted as C1) followed the schematic illustrated in Fig. 7 and took the
second. In this study, the data used are referred to a specific day, in- parameters given in Table 3, whereas the second one (denoted as C2)
cluding 86, 400 data points. Fig. 10(a) and (b) shows the original wind employed the flow chart illustrated in Fig. 9. For the C2, the control
direction and wind speed series, respectively. Besides, the averaged period and the weighting factors in the QF have to be determined in
wind direction series are also illustrated in Fig. 10(a), which refer to advance. In this study, the control period is set to Tc = 10 s , which is
mean values averaged in 10 s, 30 s and 60 s periods, respectively. As chosen based on the fact that the predicted wind direction can provide
shown in Fig. 10, both the original data of wind direction and wind data per 10 s. By following the trial and error procedures introduced in
speed are mixed with high-frequency noises. After averaging, the wind [41], the three weighting factors are chosen as: w1 = 1.0 , w2 = 14.0 and
direction data become much smooth, and thus they can be utilized as w3 = 50.0 , respectively. Besides, the yaw speed is set to 0.5 deg/s, which
inputs for the conventional yaw controllers. is the designing parameter of the yaw system for the MY 1.5 MW WT.

4.1. Wind direction prediction results 4.2.1. Results of time series

By using the wind direction data shown in Fig. 10(a), the two
The ARIMA-KF predictive model is trained by using wind direction proposed yaw controllers are simultaneously simulated in parallel with

594
D. Song et al. Energy Conversion and Management 157 (2018) 587–599

Fig. 11. Wind direction prediction results of (a) 10 s;

(b) 30 s; (c) 60 s.

Table 4 the performance of a yaw controller can be evaluated in terms of three

Indexes of the three prediction results. performance indexes: yaw error, yaw action time and yaw action count,
these data were calculated during the simulation. The simulation results
Time series of 10 s Time series of 30 s Time series of 60 s
are shown in Fig. 12, which are obtained from the four controllers and
MAE 0.9218 0.9573 1.1954 refer to nacelle position, yaw error, yaw action time and yaw action
RMSE 1.3038 1.2839 1.6431 count, respectively.
MAPE 15.74% 12.91% 12.03%
In Fig. 12, all curves of nacelle position, yaw error, yaw action time
and yaw action count under the four controllers presented similar
each other. Also, for the sake of comparison, the results of the two trends, which prove that all the four controllers were in a normal work
normal yaw controllers were collected together, where the ones from condition. However, there were obvious differences among yaw error,
MY and NREL are denoted as C4_MY and C3_NREL, respectively. Since yaw action time and yaw action count under different controllers.
Fig. 12(b) shows that the variations of yaw error from different

595
D. Song et al. Energy Conversion and Management 157 (2018) 587–599

200 Fig. 12. Simulation results of the four controllers

C3_NREL for: (a) nacelle position; (b) yaw error; (c) yaw
Nacelle position [degree]

C4_MY time; (d) yaw action count.

C1
150 C2

100

50
0 4 8 12 16 20 24
Time [h]

(a)
50
C3_NREL
C4_MY
C1
25
Yaw error [degree]

C2

-25

-50
0 4 8 12 16 20 24
Time [h]
(b)
600

500 C3_NREL
C4_MY
Yaw time [min]

400 C1
C2
300

200

100

0
0 4 8 12 16 20 24
Time [h]
(c)
1400

1200 C3_NREL
C4_MY
1000
Yaw count [-]

C1
800 C2

600

400

200

0
0 4 8 12 16 20 24
Time [h]

(d)

controllers. It can be found out that among the four controllers, error. Fig. 12(c) and (d) shows the yaw time and yaw count, respec-
C3_NREL has the biggest variation of the yaw error, C4_MY comes tively. It is very clear that the trends of yaw time for different con-
secondly, and C1 and C2 have similar and smaller variations of the yaw trollers are similar as the ones of yaw count. Among the four

596
D. Song et al. Energy Conversion and Management 157 (2018) 587–599

Table 5 3.41%. By comparison to the ones of the former two controllers, C1 and
Statistical data comparisons among the four controllers. C2 have less power reductions, being 2.43% and 2.24%, respectively.
These numerical data well justify the developed novel controllers in the
C3_NREL C4_MY C1 C2
aspect of maximizing wind power extraction.
MAE (θ ye ) 9.69 7.96 6.73 6.23 The above observations prove that C2 noticeably outperforms the
RMSE (θ ye ) 12.88 10.57 8.87 8.18 other three controllers in term of performance of wind power extraction
tyaw [min] 85.4 359.4 523.3 328.9 with a moderate yaw actuator usage, whereas C1 outperforms the two
Cyaw [−] 168 876 1377 622 industrial controllers in term of wind power extraction at the expense of
ξ [%] 4.53 3.41 2.43 2.24 yaw actuator usage.

5. Conclusions and future work

controllers, C1 and C4_MY take the first and second places of spending
the most yaw time and yaw count, respectively, C2 comes next, and To operate the yaw system of the WTs effectively, a novel yaw
C3_NREL is with the last place. These observations indicate a fact that control solution using predicted wind direction data has been proposed
the four controllers have different performance: C3_NREL has the best in this study. Using discrete data obtained at every sampling time for
and the worst performance in the aspect of yaw actuator usage and yaw prediction, this feature of the predicted wind direction was fully taken
error reduction, respectively; C1 and C2 outperform the other two into consideration for constructing the two novel methods. Three kinds
controllers in terms of yaw error reduction, but C1 has a big payback of of wind direction prediction data (10 s, 30 s and 60 s mean values which
yaw actuator usage; overall, C2 has the most favorable performance. are provided by a hybrid ARIMA-KF model) were used by the first novel
yaw control method as the tracking references. The second novel
4.2.2. Statistical data comparison and discussion method employed the MPC with a finite control set, which utilized the
To further evaluate the performance of the four controllers, the predicted wind direction with 10 s mean value to estimate future yaw
numerical results from Fig. 12(b–d) are collected to obtain statistical error and achieved multiple control objectives through explicitly in-
data, which are convenient for comparison and analysis. Accordingly, cluding them into the quality function. The two novel methods are
the four performance indexes discussed in Section 2.3 are calculated transformed to two novel controllers, whose performance is compared
and summarized in Table 5.Table 5 shows the results of MAE (θye ) and to the one of two industrial yaw controllers. Finally, the proposed two
RMSE (θye ) . Among the four controllers, C3_NREL has the biggest novel yaw controllers developed by the proposed methods demon-
MAE (θye ) and RMSE (θye ) , which are 9.69 and 12.88, respectively. strated the superiority and the applicability through some simulation
C4_MY has smaller results, where MAE (θye ) and RMSE (θye ) are de- tests using wind direction data obtained from a wind farm located in
creased by 17.86% (being 7.96) and by 17.93% (being 10.57), re- Guangdong Province of China. That is, comparison results showed that
spectively. C1 and C2 have closer results, whereas C2 has the smallest the two novel controllers extracted 1% more wind power through an
values, where MAE (θye ) and RMSE (θye ) are decreased by 35.71% accurate wind direction tracking in comparison with the two industrial
(being 6.23) and by 36.49% (being 8.18), respectively. These numerical controllers. Also, C1 increased wind power extraction at the high ex-
results are in good agreement with time series of yaw error shown in pense of yaw actuator usage, while C2 showed its capability of max-
Fig. 12(b), and they reveal that the developed two novel controllers imizing power extraction and reducing yaw actuator usage at the same
noticeably outweigh the other two controllers in reducing yaw error. time, which owed to the quality function, constructed by adding pe-
Table 5 also shows the total yaw action time (tyaw ) and yaw action nalties on the two interesting control objectives.
count (Cyaw ) of the four controllers during the day. Among the four Although encouraging results have been obtained by the proposed
controllers, C3_NREL has the least yaw action time and action count, novel solution, there are some uncovered issues to be addressed in fu-
which are 85.4 min and 168 times, respectively. C4_MY has a con- ture studies, such as incorporating advanced machine-learning based
siderable yaw action time and action count, which are 359.4 min and prediction algorithms into the control solution, investigating the im-
876 times, respectively. It means that the yaw actuators keep working pacts of the prediction uncertainty on the control method, studying
for one fourth of the time on that day, and are activated per two adaptive algorithms for parameter adjustment corresponding to dif-
minutes. The lengthy working time and frequent start may bring high ferent wind conditions.
loads, thus shorten the lifetime of the yaw actuators. However, C1 has a
longer yaw time and frequenter action, which are increased by 45.6%
(being 523.3 s) and 57.2% (being 1377 times), respectively. The pay- Acknowledgments
back of reducing yaw error for C1 is enormous but reasonable, because
there is no time threshold for the C1. By comparison, C2 shows a sa- This work is supported by the National Natural Science Foundation
tisfactory behavior: the yaw time is reduced by 8.6%, being 328.9 s, of China under Grant 51777217, the Project of Innovation-driven Plan
whereas the yaw action count is decreased by 54.8%, going down to in Central South University, the Australian Research Council under
622 times. Grant DE160100675 and the Project funded by China Postdoctoral
Finally, Table 5 shows the result of the reduction factor (ξ ) of dif- Science Foundation (2017M622605). This work is also financially
ferent controller, the calculation of which is given in Appendix A. It can supported by the Basic Science Research Program through the National
be seen that C3_NREL has the most power reduction caused by the yaw Research Foundation of Korea (NRF) funded by the Ministry of
error, which is 4.53%, and C4_MY comes the second with the value of Education, NRF-2016R1A6A1A03013567.

Appendix A

The reduction factor ξ is calculated using Eqs. (6) and (7). To do this, θyej and their distribution probability ( f j ) are collected and calculated based
on the results in Fig. 12(b). The results of θyej and f j are shown in the form of histogram in Fig. 13(a) and (b), respectively. Based on these distribution
data, ξ of the four controllers can be calculated and obtained.

597
 D. Song et al. Energy Conversion and Management 157 (2018) 587–599

5 Fig. 13. Distribution probability of yaw error

under (a) C4_MY and C3_NREL; (b) C1 and C2.

C3_NREL
Distribution probability [%]

4
C4_MY

0
-30 -20 -10 0 10 20 30
Yaw error [degree]
(a)
6

C1
Distribution probability [%]

5
C2
4

0
-30 -20 -10 0 10 20 30
Yaw error [degree]
(b)

References [16] Kragh KA, Fleming PA. Rotor speed dependent yaw control of wind turbines based
on empirical data. AIAA aerospace sciences meeting including the new horizons
forum and aerospace exposition. 2012.
[1] Chehouri A, Younes R, Ilinca A, Perron Jean. Review of performance optimization [17] Bu FF, Huang WX, Hu YW, Shi K, Wang QS. Study and implementation of a control
techniques applied to wind turbines. Appl Energy 2015;142(4):361–88. algorithm for wind turbine yaw control system. Proceedings of the 2009 world non-
[2] Song DR, Yang J, Cai ZL, Dong M, Su M, Wang YH. Wind estimation with a non- grid-connected wind power and energy conference, Nanjing, China, 24–26
standard extended Kalman filter and its application on maximum power extraction September 2009. 2009. p. 1–5.
for variable speed wind turbines. Appl Energy 2017;190:670–85. [18] China Mingyang wind power. MY 1.5MW safety and control scheme; 2009.
[3] Bontempo R, Manna M. The axial momentum theory as applied to wind turbines: [19] Scholbrock AK, Fleming PA, Wright A, Slinger C, Medley J, Harris M. Field test
some exact solutions of the flow through a rotor with radially variable load. Energy results from lidar measured yaw control for improved power capture with the NREL
Convers Manage 2017;143:33–48. controls advanced research turbine. In: 33rd Wind energy symposium (AIAA).
[4] Burton T, Sharpe D, Jenkins N, Bossanyi E. Wind Energy Handbook. Wiley; 2011. Proceedings of SPIE - the international society for optical engineering, vol. 5685,
[5] Medici D. Experimental studies of wind turbine wakes: power optimisation and issue no. 6; 2014. p. 1080–4.
meandering. Ph.D. thesis, KTH, Mechanics. [20] Fleming PA, Scholbrock AK, Jehu A, Davoust S, Osler E, Wright AD, et al. Field-test
[6] Schepers JG. IEA annex XX: dynamic inflow effects at fast pitching steps on a wind results using a nacelle-mounted lidar for improving wind turbine power capture by
turbine placed in the NASA-Ames wind tunnel. ECN, report; 2007 reducing yaw misalignment. J Phys Conf Ser 2014;524(1):012002.
[7] Boorsma K. Power and loads for yawed flow conditions. ECN, report; 2012 [21] Kragh KA, Hansen MH. Potential of power gain with improved yaw alignment.
[8] Kragh KA, Hansen MH. Load alleviation of wind turbines by yaw misalignment. Wind Energy 2015;18(6):979–89.
Wind Energy 2014;17(7):971–82. [22] Spencer M, Stol K, Cater J. Predictive yaw control of a 5 MW wind turbine model.
[9] Pedersen TF, Paulsen US, Pedersen SM, Enevoldsen P. Operational experience and AIAA aerospace sciences meeting including the new horizons forum and aerospace
analysis of a spinner anemometer on a MW size wind turbine. Conference pro- exposition. 2013.
ceedings (online) European Wind Energy Association. 2008. [23] Hure N, Turnar R, Vasak M, Vasak M, Bencic G. Optimal wind turbine yaw control
[10] Ribrant J, Bertling LM. Survey of failures in wind power systems with focus on supported with very short-term wind predictions. IEEE international conference on
Swedish wind power plants during 1997–2005. IEEE Trans Energy Convers industrial technology. 2015. p. 385–91.
2007;22(1):167–73. [24] Tsioumas E, Karakasis N, Jabbour N, Mademlis C. Indirect estimation of the Yaw-
[11] Pérez JMP, Márquez FPG, Tobias A, Papaelias M. Wind turbine reliability analysis. Angle misalignment in a horizontal axis wind turbine. IEEE 11th international
Renew Sustain Energy Rev 2013;23(23):463–72. symposium on SDEMPED. 2017. p. 45–51.
[12] Farret FA, Pfitscher LL, Bernardon DP. Sensorless active yaw control for wind tur- [25] Azimi R, Ghofrani M, Ghayekhloo M. A hybrid wind power forecasting model based
bines. Proceedings of the 27th annual conference of the IEEE Industrial Electronics on data mining and wavelets analysis. Energy Convers Manage 2016;127:208–25.
Society, Denver, CO, USA, 29s November–2 December 2001, vol. 2. 2001. p. [26] Yesilbudak M, Sagiroglu S, Colak I. A novel implementation of kNN classifier based
1370–5. on multi-tupled meteorological input data for wind power prediction. Energy
[13] Wu X, Liu Y, Teng W. Modified hill climbing method for active yaw control in wind Convers Manage 2017;135:434–44.
turbine. IEEE control conference. 2012. p. 6677–80. [27] Mohammadi K, Alavi O, Mcgowan JG. Use of Birnbaum-Saunders distribution for
[14] Karakasis N, Mesemanolis A, Nalmpantis T, Mademlis C. Active yaw control in a estimating wind speed and wind power probability distributions: a review. Energy
horizontal axis wind system without requiring wind direction measurement. IET Convers Manage 2017;143:109–22.
Renew Power Gener 2017;10(9):1441–9. [28] Zameer A, Arshad J, Khan A, Raja MAZ. Intelligent and robust prediction of short
[15] Chen FQ, Yang JM. Fuzzy PID controller used in yaw system of Wind Turbine. IEEE term wind power using genetic programming based ensemble of neural networks.
international conference on power electronics systems and applications. 2009. Energy Convers Manage 2017:361–72.
p. 1–4. [29] Zhang C, Zhou JZ, Li CS, Fu WL, Peng T. A compound structure of ELM based on

598
D. Song et al. Energy Conversion and Management 157 (2018) 587–599

feature selection and parameter optimization using hybrid backtracking search al- [36] Mostafavi ES, Mostafavi SI, Jaafari A, Jaafari A, Hosseinpour F. A novel machine
gorithm for wind speed forecasting. Energy Convers Manage 2017;143:360–76. learning approach for estimation of electricity demand: an empirical evidence from
[30] Noorollahi Y, Jokar MA, Kalhor A. Using artificial neural networks for temporal and Thailand. Energy Convers Manage 2013;74(10):548–55.
spatial wind speed forecasting in Iran. Energy Convers Manage 2016;115:17–25. [37] Zhang YC, Liu KP, Qin L, An XL. Deterministic and probabilistic interval prediction
[31] Ouyang T, Kusiak A, He Y. Predictive model of yaw error in a wind turbine. Energy for short-term wind power generation based on variational mode decomposition
2017;123:119–30. and machine learning methods. Energy Convers Manage 2016;112:208–19.
[32] Song DR, Yang J, Liu Y, Su M, Liu AF, Joo YH. Wind direction prediction for yaw [38] Gosk A. Model predictive control of a wind turbine Master thesis Lyngby
control of wind turbines. Int J Control Autom Syst 2017;15(4):1720–8. (Denmark): Technical University of Denmark; 2011
[33] KRIWAN industie-Elektronik GmbH. INT30 wind direction sensors product manual; [39] Rodriguez J, Kazmierkowski MP, Espinoza JR, Zanchetta PH, Abu-Rub, Young A,
2016. < http://www.kriwan.com/en/search/?q=N234_N291_INT30_71000317/ et al. State of the art of finite control set model predictive control in power elec-
> [accessed 16.09.17]. tronics. IEEE Trans Ind Inf 2013;9(2):1003–16.
[34] Bivona S, Bonanno G, Burlon R, Gurrera D, Leone C. Stochastic models for wind [40] Song DR, Yang J, Dong M, Joo YH. Model predictive control with finite control set
speed forecasting. Energy Convers Manage 2011;52:1157–65. for variable-speed wind turbines. Energy 2017;126:564–72.
[35] Su ZY, Wang JZ, Lu HY, Zhao G. A new hybrid model optimized by an intelligent [41] Cortés P, Kouro S, Rocca BL, Vargas R, Rodríguez J. Guidelines for weighting factors
optimization algorithm for wind speed forecasting. Energy Convers Manage design in model predictive control of power converters and drives. IEEE interna-
2014;85(9):443–52. tional conference on industrial technology. 2009. p. 1–7.

599