2019 IEEE 58th Conference on Decision and Control (CDC)

Palais des Congrès et des Expositions Nice Acropolis

Nice, France, December 11-13, 2019

An adaptive passivity-based controller for a wind energy

conversion system?
Rafael Cisneros1 Rui Gao2 Romeo Ortega3 and Iqbal Husain2

Abstract— A controller, based on passivity, for a wind en- been considered in works like [2] where, based on the
ergy conversion system connected to a dc bus is proposed. frequency-domain for a linearized model of the system, a
The system consists of a wind turbine, a permanent magnet control design is proposed. Backstepping methods have
synchronous generator, a rectifier and a load. Guarantee-
ing stability and endowed with adaptive properties, the been used in [3], [4], [5]. On the other hand, controller
controller regulates the wind turbine angular velocity to a designs considering the complete dynamics have also
desired value—in particular, the set-point is selected such been reported. See, for example, [6] for a sliding mode
that the maximum power from the wind is extracted—and control of system based on a fully actuated—hence,
maximizes the generator efficiency. The fast response of the easier to control—doubly-fed induction generator. Also,
closed-loop system makes possible to operate under fast-
changing wind speed conditions. To assess the controller an Akagi’s PI-like controller is presented in [7]. Even
performance, realistic simulation results are included. though in the reported works their corresponding control
tasks are satisfied, most of them lack of a stability proof
I. I NTRODUCTION of the resulting overall closed-loop system.
The Future Renewable Electric Energy Delivery and The contribution of this paper is in the line of [8].
Management (FREEDM) system is the term to designate Passivity-based control (PBC) is the generic name of the
the future power distribution systems [1]. These systems employed control design methodology which achieves
are conceived to facilitate distributed renewable energy stabilization by exploiting the passivity properties of the
generation, integration and management. The advantages system. Contributions regarding PBC in WECS has been
of these promising systems include power flow control reported in [9], [10]. In these works, a standard Passivity
and an efficient plug-and-play interface for power dis- Based Control (sPBC) is derived for a battery-connected
tributed renewable energy resources. and a grid-connected WECS, respectively. Even though
Towards this renewable energy systems integration, in in these works the stability margins exhibited by the
this work we are interested in the control design of wind proposed controllers are reasonable, a limited response
energy conversion systems (WECS). Providing a theo- in the speed turbine regulation was observed.
retical framework for the controller design allows power In this contribution, we consider a system consisting
engineer practitioners to apply the control law with con- of a wind turbine, a permanent magnet synchronous
fidence and considerably simplifies the commissioning generator, a rectifier and a load which is connected to the
stage—hence, is a topic of paramount importance. Since distribution grid dc bus. To control the system, the sPBC
the WECSs are composed of a wind turbine and an technique is revisited and an adaptive version of it is
electrical generator, a common practice in the control proposed. Also, the resulting controller notably improves
realization is to dissociate these elements invoking dif- performance compared to aforementioned works.
ferences between their time scale. Thus, the electrical The remainder of this article is organized as follows.
dynamics of the generator is neglected assuming that Section II presents the system configuration and mod-
it is “fast enough”. This simplifying assumption has eling. The main result is given in Section III, where
the proposed controller is introduced for a more general
? This work has been supported by the ERC Program of the National class of systems and then applied to the wind energy
Science Foundation under Award Number EEC-0812121. system here considered. Section IV is devoted to the
1 R. Cisneros is with Instituto Tecnológico Autónomo de México,
simulation results. In Section V, concluding remarks
Mexico City, 01080 CDMX, Mexico. E-mail: [email protected].
2 R. Gao and I. Husain is with FREEDM Systems Center, North finalize this work.
Carolina State University, Raleigh, NC 27606, USA. E-mails: ihu- Notation: Unless indicated otherwise, all vectors in
sain2[rgao]@ncsu.edu. the paper are column vectors. For i = 1, · · · , 4, ei ∈ R4
3 R. Ortega is with Laboratoire des Signaux et Systemès,
CNRS-CentraleSupélec, 91192, Gif-sur-Yvette, France. E-mail: or- corresponds to the standard basis vector with the ith-
[email protected]. element equal to one. For scalar entries xk , col(xk )

978-1-7281-1398-2/19/$31.00 ©2019 IEEE 4852

Authorized licensed use limited to: East Carolina University. Downloaded on July 29,2020 at 13:22:49 UTC from IEEE Xplore. Restrictions apply.
 for the shaft’s angular speed, i.e.,
vw λ ?
rL rS ωm? := .
C r
vw 2) Turbine model: The dynamic equation of a one-
VS
mass turbine is obtained from Newton’s equation of
PMSG
motion
J ω̇m = Tm − Te (2)
Rectifier Voltage where J is the rotor inertia. Also, for ωm > 0, Tm is
Source
the mechanical torque by the wind tothe turbine shaft
Fig. 1: System under consideration. Pw 1

3 1 rωm
Tm = = ρAvw Cp , (3)
ωm 2 ωm vw
and Te , to be defined, is the electrical torque provided
denotes a column vector and diag(xk ) a diagonal matrix. by the generator.
When clear from the context the arguments in the
functions will be omitted. B. Permanent magnet synchronous generator and power
converter
II. S YSTEM MODELING AND CONTROL OBJECTIVES The dynamic equations of the generator in dq-
Shown in Fig. 1, the WECS is composed of a turbine, coordinates are [12]
a surface-mounted PMSG and rectifier connected to a Li̇d = −rid + Liq ωe − vd
(4)
load, a capacitor and a voltage source. This source sub- Li̇q = −riq − Lid ωe + φωe − vq
stitutes the dc voltage to which the system is connected. where id , iq , vd , vq are the q and d components of
the current and voltage. Also, R and L are the stator
A. Wind turbine resistance and inductance, φ is the permanent magnetic
The mechanical power captured by the turbine is flux and ωe is the electrical frequency. The electrical
given by the power function frequency satisfies the relation
p
1 3 ωe = ωm . (5)
Pw = ρAr Cp (λ)vw , 2
2 where p is the number of pole pairs. The electrical torque
where vw is the wind speed, which is assumed to be
Te is given by
known and constant—in fact, this later is a standard con- 3p
sideration when designing the control for these systems. Te (iq ) = φiq . (6)
22
Also, ρ is the air density, Ar is the area swept by the The input voltages of the generator are
blades and Cp is the turbine’s coefficient power, which vd = vc u1 , vq = vc u2 (7)
is function of the tip-speed ratio λ defined as
rb ωm where u1 and u2 are dq-coordinates of the averaged
λ := , (1) control signals. Finally, from Kirchhoff’s current law
vw
where ωm is the shaft’s rotational speed, and rb is the Vs
C v̇c = −re vc + + id u1 + iq u2 (8)
blade’s radius. rs
The turbine’s power coefficient also depends on the where C is the capacitance value, re := rrLL+r rs , rL is a
s

blade pitch angle, commonly symbolized as β. However, resistive load and rs and Vs are, respectively, the supply
in this paper it is assumed that the turbine operates in the internal resistance and its dc voltage. To prevent damage
regime of maximum aerodynamic efficiency, also known to the devices, the capacitor voltage in the prototype
as Region 2 [11]. In this regime, β is constant—typically is constrained to operate at positive values. Thus, it is
zero. For this reason, this control input has been omitted. assumed that vc > 0.
C. The overall system
1) Maximum power extraction: A typical shape of the
function Cp is depicted in Fig. 2. The system is required Substituting (3) and (6) in (2), also (7) and (5) in (4)
to operate at and gathering the resulting equations together with (8)
yields the overall system
λ? := arg max Cp (λ). p
The turbine is required to operate at this value to Li˙d = −rid + Liq ωm − u1 vc (9a)
2
maximize the power extracted from the wind. Using (1), p p
Li˙q = −riq − Lid ωm + φωm − u2 vc (9b)
it is possible to obtain the corresponding optimal value 2 2

4853

Authorized licensed use limited to: East Carolina University. Downloaded on July 29,2020 at 13:22:49 UTC from IEEE Xplore. Restrictions apply.
 C1. Minimize the copper loss and maximize the effi-
ciency of the generator, making the system operate
at x1 = 0;
C2. Operate at the maximum power extraction point
x3 = ωm? .
In the next Lemma, the equilibrium points of (10) are
stated such that C1 and C2 are met.
Lemma 1: The assignable equilibrium points x? for
the system (10) consistent with C1 and C2 are defined
by the set
4
Fig. 2: Function Cp (λ). E := x | x1 = 0, x2 = Tm? , x3 = ωm? ,
3pφ
s ) (12)
  Vs 1 Vs2
1 3 1 rωm 3p x4 = ± + 4re h
J ω̇m = ρAvm Cp − φiq (9c) 2re rs 2re rs2
2 ωm vw 22
Vs where
C v̇c = −re vc + + u1 id + u2 iq . (9d) 16r 2 2
rs Tm? := Tm (ωm? ), h := − 2 2 Tm? + ωm? Tm? .
9p φ 3
Proof: At the equilibrium, system (10) satisfies
III. SPBC DESIGN
0 = −[C(x3 , u) + R]x + Φ(x3 ) + θ. (13)
By multiplying both sides of (9c) by 23 and J1 = 32 J1 , It can be expressed as
(9) admits the Euler-Lagrange representation
0 = [J (x3 ) − R]x + G(x)u + Φ(x3 ) + θ. (14)
where
Σp : Dẋ + [C(x3 , u) + R] x = Φ(x3 ) + θ, (10) 
0 p

2 Lx3 0 0
where x = col(id , iq , ωm , vc ), − p Lx3 0 p
2 φ 0 ,

2 Vs J (x3 ) =  2
p
Φ(x3 ) := Tm (x3 )e3 , θ := e4 ,
 0 −2φ 0 0
3 rs 0 0 0 0
D :=diag(L, L, J1 , C) > 0, 
−x4 0

R :=diag(r, r, 0, re ) ≥ 0,  0 −x4 
G(x) =   .
− p2 Lx3 0 0 
 
0 0 u1
 p Lx3 0 − p2 φ u2  x1 x2
C(x3 , u) :=  2
p

A full-rank left annihilator ofG(x)  is the matrix
 0
2φ 0 0
−u1 −u2 0 0 ⊥ x>
G (x) = > .
are the external known forces and external unknown e3
forces, generalized inertia, damping and interconnection Multiplying (13) by G⊥ (x) yields
matrices, respectively. The dc link voltage Vs is assumed 0 =G⊥ (x)[(J (x3 ) − R)x + Φ(x3 ) + θ]
to be constant and Vrss to be unknown. The system energy
−x Rx + 32 Tm x3 + Vrss x4
 > 
is given by = 2 p .
1 3 Tm − 2 φx2
H(x) = x> Dx Setting x1 = 0 and x3 = ωm? , the last equation is
2
whose time-derivative is equivalent to
2 Vs 
−rx22 + 23 Tm? ωm? − re x24 + Vrss x4

Ḣ = −r(x21 + x22 ) − re x24 + Tm x3 + x4 (11) 0= . (15)
} |3 {z } rs 2 p
3 Tm? − 2 φx2 .
| {z
Electrical Dissipation
| {z }
Mechanical Electrical From the second equation it follows that
Power Power 4
Note that C(x3 , u) does not contribute to the power x2 = Tm? .
3pφ
balance. This is captured by the fact that C(x3 , u) = Finally, substituting the latter expression into the first
−C > (x3 , u). See [8] for an explanation of EL systems. equation of (15) yields the quadratic equation
Vs
A. Control objectives and assignable equilibrium points re x24 − x4 − h = 0.
rs
The control objectives are: Its roots are those in E. 

4854

Authorized licensed use limited to: East Carolina University. Downloaded on July 29,2020 at 13:22:49 UTC from IEEE Xplore. Restrictions apply.
 A standing assumption along the text is that the Thus, from Lemma 2, trajectories of Σp converge to
equilibrium points exist. Particularly, we assume that those of Σd . In the next step, we set xd1 = 0 and xd3 =
Assumption 1: ωm? in (17) with (20). This yields
V2 p
h > − s 2. (16) 0 = Lxd2 x3 − xd4 u1 + κP 1 x1 + κI 1 z1 (24a)
4re rs 2
p
which assures real values of x4? 1 . Lẋd2 = − rxd2 + φωm? − xd4 u2 + κp2 x̃2 + κI 2 z2
2
(24b)
B. Standard passivity-based controller 2 p
As explained in [8], the sPBC is a dynamic controller. 0 = Tm − φxd2 + κP 3 x̃3 + κI 3 z3 (24c)
3 2
Then, the first step of design consists in defining its C ẋd4 = − re xd4 + xd2 u2 + κP 4 x̃4 + κI 4 z4 (24d)
dynamics which are the EL system ż =x̃. (24e)
Σd : Dẋd + [C(x3 , u) + R]xd = Φ(x3 ) − vP I , (17) The last step consists in obtaining the explicit controller
Subtracting (17) to (10) yields the error system expression and dynamics. They come after (24) and are
Dx̃˙ + [C(x3 , u) + R]x̃ = ṽP I . (18) introduced in the following proposition.
where x̃ := x − xd and ṽP I := vP I + θ. Considering the Proposition 1. Consider the system (10) in closed-loop
storage function with u = β(x,  xd , z)pwhere 
1 1 Lx2d x3 + κP 1 x1 + κI 1 z1
S(x̃) = x̃> Dx̃, β= 2
p
2 xd4 −rxd2 + 2 φωm? − Lϕ + κP 2 x̃2 + κI 2 z2
it follows that (25a)
Ṡ = −x̃> Rx̃ + x̃> ṽP I , (19) re 1 xd2  p
ẋd4 = − xd4 + − rxd2 + φωm? − Lϕ
i.e., system (18) defines a passive mapping ṽP I 7→ x̃. C C xd4 2
Define vP I via the PI controller  κI 4 κP 4
+ κP 2 x̃2 + κI 2 z2 + z4 + x̃4 (25b)
vP I = − KP x̃ − KI z, ż = x̃. (20) C C
ż =x̃, (25c)
Here, Ki , Kp > 0 are diagonal matrices. Notice that,
defining z̃ := z − Ki−1 θ, the PI controller (20) can be where, for k = 1, · · · , 4, gains κP k , κI k > 0, and
written in the incremental form 4
xd2 = Tm + κP 3 x̃3 + κI 3 z3 (26)
3pφ
ṽP I = − KP x̃ − KI z̃, z̃˙ = x̃ (21)   
1 4 ∂Tm 3p
Substituting (21) into (18) produces ϕ= + κP 3 Tm − φx2 + κI 3 x̃3 .

D 0
 
J (x3 ) + R + KP KI
 J 3pφ ∂x3 22
ΣΨ : Ψ̇ + Ψ=0 (27)
0 KI −KI 0
(22) Then,
where Ψ := col(x̃, z̃). P1. For all initial conditions
 (x(0),
 xd4 (0), z(0)),
Lemma 2: The trajectories of the error system (22) x1
lim = 0.
are bounded and t→∞ x3 − ωm?

lim Ψ(t) = 0 (23) P2. Defining v̄c? as the greatestsroot of (12), namely,
t→∞
Proof. Consider the Lyapunovfunction candidate Vs 1 Vs2
v̄c? := + + 4re h,
1 D 0 2re rs 2re rs2
W(x̃, z̃) = Ψ> Ψ.
2 0 KI 4
(x, xd4 , z) = (0, 3pφ Tm? , ωm? , v̄c? , v̄c? , KI−1 θ) ∈ R9 is
Defining Ra := KP + R, the time-derivative of W is an asymptotically stable eq. point.
clearly given by Proof. From (24c), we get
Ẇ = −x̃> Ra x̃ ≤ 0, 4
xd2 = Tm + κP 3 x̃3 + κI 3 z3 . (28)
from which we conclude that the trajectories of (22) 3pφ
are bounded and the zero equilibrium of (22) is stable. Thus,  
Invoking LaSalle-Yoshizawa theorem [13], we also 4 ∂Tm
ẋd2 = ẋ3 + κp3 x̃˙ 3 + κI 3 ż3
conclude that x̃ → 0. Therefore, with x̃ = 0 in (22), 3pφ ∂x3
 
we conclude that z̃ → 0.  4 ∂Tm
= + κP 3 ẋ3 + κI 3 x̃3
3pφ ∂x3
1 Note that we have excluded the case when equality holds. The
reason is explained in the proof of Proposition 1.

4855

Authorized licensed use limited to: East Carolina University. Downloaded on July 29,2020 at 13:22:49 UTC from IEEE Xplore. Restrictions apply.
 TABLE I: WECS Parameters nominal values
Item Nominal Value
ΣΨ ν
Σd x1d = 0 Turbine
x3d = ωm? Inertia J = 7.856 kg· m2
t, Ψ u=β x4d Blades radius rb = 1.84 m
PMSG
Eq. 22 Eqs. 24a-24d Nominal Power Sn = 5 kVA
Poles P = 28
Fig. 3: Cascade interconnection of the closed loop. Synchronous resistance r = 0.3676 Ω
Synchronous inductance L = 3.55 mH
Flux φ = 0.2867 Wb
   Friction coefficient f = 3.035 × 10−4 N·m·s
1 4 ∂Tm 3p Rectifier & Electrical Parameters
= + κp3 Tm − φx2 + κI 3 x̃3 Capacitance C = 3.3 mF
J 3pφ ∂x3 22 System Load rL = 60 Ω
Power Supply Voltage Vs =400 V
=:ϕ Power Supply Resistance rs = 0.1 Ω
The state feedback β is obtained after solving (24a)-
(24b) for u. Then, substituting u2 into (24d) gives
(25b). The integral actions (25c) are those in (24e). v̄c? is asymptotically stable in the three cases. Avoiding
This completes the controller equations. P1 immediately the equality in the discriminant condition of Remark 1
follows from Lemma 2. To prove P2, first note that the excludes the system from having stable trajectories on
closed-loop system admits the cascade structure of Fig. the right side and unstable ones on the right—Fig. 4(b).
3. Its equations have the form From the latter and since Ψ → 0 from Lemma 2,
Ψ̇ =fψ (t, Ψ) (29a) P2 follows after invoking for (29), standard stability
ẋd4 =fd (xd4 , ν(Ψ, t)) (29b) arguments of cascaded systems [14]. 
The system (29b) corresponds to the left-side block in
Fig. 3. On the other hand, (29b)—the right-side block IV. SIMULATION RESULTS
in the figure—corresponds to (25b) expressed in terms The considered system is shown in Fig. 1. The power
of Ψ. That is, coefficient is given by [15]
re h Vs ν rω
fd (xd4 , ν) = − xd4 + + + . Cp ( ) = c1 [c2 η − c5 ] exp (−c6 η) ,
C Cxd4 Crs Cxd4 vw
with where c1 = 0.5, c2 = 116, c5 = 5, c6 = 21 and
p vw
ν =xd2 [−rxd2 + φωm? − Lϕ + κP 2 x̃2 + κI 2 z̃2 ] η= − 0.035.
2 rωm
2 16r 2 The curve Cp vs. λ is shown in Fig. 2. The maximum
+ κI 4 z̃4 + κP 4 x̃4 − Tm? ωm? + 2 Tm? .
3 9p φ point of the function corresponds to λ? = 7.954. Also,
(30) the system parameters are shown in Table I. A real
Likewise,  expressing xd2 and  ϕ in terms of Ψ yields wind speed profile was employed for the simulations.
1 4 ∂Tm As explained, the control objective is to maintain id = 0
ϕ= + κp3 [Tm
J 3pφ ∂x3 and ωm? = vwrλ? . In Fig. 5, the simulation results for
ωm and id are shown. Note that the closed loop system

3p
− φ(x̃2 + xd2 ) + κI 3 x̃3 (31) exhibits a good performance while tracking ωm? . The
22
and xd2 = 3pφ 4
Tm + κP 3 x̃3 + κI 3 z̃3 . chosen controller gains in the simulation are κP 1 =
κP 3 = 90, κI 1 = κI 3 = 40, κP 4 = 45 κI 4 = 20 and
Note that xd2 and ϕ are bounded since Tm , ∂T m
x3 κP 2 = κI 2 = 0.1. It is worth mentioning that the closed
and Ψ are bounded—see Section II-A and Lemma 2.
loop is robust under parameter variations. Unfortunately,
Therefore, ν is bounded and decays to zero as Ψ → 0.
due to lack of space, these plots and the simulation
On the other hand, setting Ψ = 0,
results for iq , vc and u are not presented.
re h Vs
ẋd4 =fd (xd4 , 0) = − xd4 + + . (32)
C Cxd4 Crs V. C ONCLUSIONS
The equilibrium points of (32) are the roots of
Vs A sPBC control approach for a wind energy con-
re x2d4 − xd4 − h = 0,
rs version system with guaranteed stability is reported.
which are those presented in (12) for x4 . Concerning Respect to previous works [16], [9], [17] this approach
the stability, three different scenarios can be presented: is endowed with adaptive properties and the closed-loop
h = 0, h > 0 and h < 0. Shown in Fig. 4, xd4 = system presents a faster response time.

4856

Authorized licensed use limited to: East Carolina University. Downloaded on July 29,2020 at 13:22:49 UTC from IEEE Xplore. Restrictions apply.
 ẋd4 ẋd4 ẋd4

v c? v̄c? v c? =v̄c? = rVrs

v c? v̄c? v c? =v̄c? xd4
e s
xd4
xd4
Vs2
h=−
2r
4rs e

(a) Case h > 0. (b) Case h < 0. (c) Case h = 0.

Fig. 4: Plot of (32) for the different scenarios. The zero-crossing points are the equilibrium points. Notice that,
V2
excluding the case when h = − 4R2SRe , v̄c? is always a stable equilibrium point.
S

11 6 45

4
10
2 40

9 0

-2 35
8
-4

7 -6 30
(a) vw (b) id (blue) and id? = 0 (red). (c) ωm (blue) and ωm? (red).

Fig. 5: Simulation Results (time scale : 16.7 sec/div).

R EFERENCES control authority,” IEEE Journal of Emerging and Selected

Topics in Power Electronics, vol. 1, no. 4, pp. 247–259, 2013.
[1] A. Q. Huang, M. L. Crow, G. T. Heydt, J. P. Zheng, and S. J. [10] F. Mancilla-David and R. Ortega, “Adaptive passivity-based
Dale, “The future renewable electric energy delivery and man- control for maximum power extraction of stand-alone windmill
agement (FREEDM) system: The energy internet,” Proceedings systems,” Control Engineering Practice, vol. 20, no. 2, pp. 173–
of The IEEE, vol. 99, no. 1, pp. 133–148, 2011. 181, 2012.
[2] W. E. Leithead and B. Connor, “Control of variable speed wind [11] L. Y. Pao and K. E. Johnson, “A tutorial on the dynamics and
turbines: Design task,” European Journal of Control, vol. 60, control of wind turbines and wind farms,” in American Control
no. 3, pp. 1122–1132, 2013. Conference, 2009, pp. 2076–2089.
[3] R. Galeazzi, K. Borup, H. Niemann, N. Poulsen, and [12] P. C. Krause, S. D. Sudhoff, S. Pekarek, and O. Wasynczuk,
F. Caponetti, “Adaptive backstepping control of lightweight Analysis of Electric Machinery and Drive Systems. IEEE Press
tower wind turbine,” in American Control Conference (ACC), Wiley, 2013.
2015, 2015, pp. 3058–3065. [13] M. Krstic, I. Kanellakopoulos, and P. Kokotovic, Nonlinear and
[4] R. Galeazzi, M. Gryning, and M. Blanke, “Observer backstep- adaptive control design. Wiley-Blackwell, 1995.
ping control for variable speed wind turbine,” in American [14] E. D. Sontag, “Further facts about input to state stabilization,”
Control Conference, 2013, pp. 1036–1043. IEEE Transactions on Automatic Control, vol. 35, no. 4, pp.
473–476, 1990.
[5] Z. Lu and W. Lin, “Asymptotic tracking control of variable-speed
[15] S. Heier, Grid Integration of Wind Energy. Wiley, 2014.
wind turbines,” IFAC World Congress, vol. 18, no. 1, pp. 8457–
[16] R. Cisneros, R. Gao, R. Ortega, and I. Husain, “PI passivity-
8462, 2011.
based control for maximum power extraction of a wind energy
[6] B. Boukezzar and M. M’Saad, “Robust sliding mode control
system with guaranteed stability properties,” International Jour-
of a DFIG variable speed wind turbine for power production
nal of Emerging Electric Power Systems, vol. 5, pp. 567–57,
optimization,” in Mediterranean Conference on Control and
2016.
Automation, 2008, pp. 795–800.
[17] R. Gao, I. Husain, R. Cisneros, and R. Ortega, “Passivity-based
[7] M. Chinchilla, S. Arnaltes, and J. C. Burgos, “Control of and standard PI controls application to wind energy conversion
permanent-magnet generators applied to variable-speed wind- system,” in IEEE Energy Conversion Congress and Exposition,
energy systems connected to the grid,” IEEE Transactions on 2016, pp. 1–5.
Energy Conversion, vol. 21, no. 1, pp. 130–135, 2006.
[8] R. Ortega, J. A. Loria, P. J. Nicklasson, and H. Sira-Ramı́rez,
Passivity-based Control of Euler-Lagrange Systems. Springer,
1998.
[9] R. Cisneros, F. Mancilla-David, and R. Ortega, “Passivity-based
control of a grid-connected small-scale windmill with limited

4857

Authorized licensed use limited to: East Carolina University. Downloaded on July 29,2020 at 13:22:49 UTC from IEEE Xplore. Restrictions apply.