Fault-Tolerant Control of Wind Turbine System Using Linear Parameter-Varying Model

Li, Shanzhi; Aitouche, Abdel; Christov, Nicolai

doi:https://doi.org/10.1155/2022/1290639

Mathematical Problems in Engineering

On this page

Abstract Introduction Preliminaries Conclusions Data Availability Conflicts of Interest Acknowledgments References Copyright Related Articles

Research Article | Open Access

Volume 2022 | Article ID 1290639 | https://doi.org/10.1155/2022/1290639

Fault-Tolerant Control of Wind Turbine System Using Linear Parameter-Varying Model

Shanzhi Li,¹Abdel Aitouche,²and Nicolai Christov³

Academic Editor: Jian Zhang

Received23 Feb 2022

Accepted06 Jun 2022

Published14 Jul 2022

Abstract

This paper proposes a fault-tolerant control scheme for WTS with actuator faults and disturbance, based on a linear parameter-varying (LPV) model. First, a LPV model of WTS subject to disturbances and actuator fault is obtained. Using this model, unknown input observer (UIO) is developed to estimate the WTS fault and state variables. Then, an integral tracking controller for LPV systems is developed. Based on the Lyapunov and theory, the convergence of UIO and the stability of the closed-loop system are analyzed and ensured. The proposed fault-tolerant controller is applied to a wind turbine system with pitch and torque actuator faults, and its performances are compared with those of controller without integral action.

1. Introduction

Wind power is one of the main renewable energy sources and attracts researchers’ and governments’ attention around the world. The wind turbine systems (WTSs) are being used more and more widely. WTSs are complex nonlinear systems such that it is difficult to model and build a suitable controller [1].

Recently, linear parameter-varying (LPV) models have been used in the control design of different nonlinear systems due to the advantages of representation with highly nonlinear and uncertain models. The most widely used control frameworks for LPV models are the gain scheduled control [2] and control [3]. Such control approaches for WTS have been developed in [4, 5]. Also, using polytopic LPV models, robust gain scheduling control [6], switching control [7], and other linear control methods [8] have been proposed to regulate the WTS power or pitch angle. Because of the similarity of polytopic LPV and Takagi–Sugeno (TS) fuzzy models, the TS fuzzy control approach has also been applied for WTS [9]. Taking into account the advantages of the LPV models in nonlinear system control design, in this paper, we will use the LPV model to describe the dynamics of WTS.

The wind turbine system consists of a wind turbine, gearbox, generator, sensors, actuators, etc. It is inevitably subject to various sensors and actuators faults, such as rotor speed and pitch angle sensor faults and torque actuator and pitch actuator faults [10]. In order to improve WTS’s reliability and safety, fault-tolerant control (FTC) schemes have been proposed. There are passive and active fault-tolerant controllers. The passive fault-tolerant controllers are fixed in the operation period and need to be robust against a class of presumed faults [11, 12]. In turn, the active fault-tolerant controllers use algorithms to estimate and compensate actively the effect of faults or system component failures [13–15]. A number of fault-tolerant control (FTC) methods have been used for WTS: model-based methods [16, 17], data-driven methods [18, 19], etc. However, there exist difficulties in the application of these methods to WTS. In reference [20], the authors introduced a FTC approach based on a discrete LPV model using an interval observer. In [3], a full state robust controller was designed for WTS. In [20], the complexity of the estimation algorithm makes it hard to compute results online, while in [3], it is difficult to find a candidate Lyapunov function allowing to give theoretical guarantees.

To overcome these limitations, a new fault-tolerant control strategy for WTS is proposed in this paper. An unknown input observer (UIO) based on the polytopic LPV model is designed to estimate the state variables. In order to minimize the effect of the disturbances, both the observer and controller design are based on control theory. To reduce tracking errors, an augmented system is built by adding an integrator. In addition, a residual-signal generation system is built for fault diagnosis detection. The proposed controller is applied to a model of WTS with actuator faults, and illustrative experiments are carried out. This paper’s primary contributions can be summarized as follows. (1) An unknown input observer based on the LPV model for estimating actuator fault is developed and sufficient conditions for the UIO are also given. (2) Different from the previous work in [21], a fault-tolerant control subject to actuator fault with integrator is designed and performs well.

The paper is organized as follows. In the next section, a LPV model of WTS is presented. Section 3 deals with the design and convergence analysis of UIO. In Section 4, an integral controller for LPV systems is designed and the stability of the closed-loop system is analyzed. The results of application of the proposed controller for 5 MW variable-pitch wind turbine benchmark are shown in Section 5. Conclusions are given in Section 6.

2. Problem Formulation and Preliminaries

The WTS is composed of aerodynamic subsystem, drive-train subsystem, pitch subsystem, etc. In order to obtain an appropriate dynamical model, some components such as tower model are neglected. The output of the aerodynamic torque is considered as a nonlinear function of the wind speed , rotor speed , and pitch angle . The conversion efficiency from wind energy to mechanical power is described by the power coefficient . The aerodynamic torque is [22]where is the wind speed, is the pitch angle, is the rotor-plane radius, is the tip speed ratio, and is the air density. The aerodynamic torque can be linearized aswhere and arewith .

Generally, the rotor turbine and the generator are connected by a gearbox and induction generator system. According to reference [22], the drive train model can be treated as a two-mass system:where is the gearbox ratio, are the stiffness damping coefficients, respectively, is the generator speed, and are the rotor and generator inertia, respectively, and is the torsion angle of the drive train.

Usually, the electrical subsystem consists of generator and converter and is much faster than the mechanical subsystem. Hence, the electrical subsystem model can be simplified to the first-order equationwhere is the reference generator torque signal and is the time constant of the electrical subsystem.

The pitch subsystem can be modeled as a first-order differential equation:where is the time constant of the pitch subsystem and is the reference of the pitch angle.

According to reference [22], the parameters , , can be evaluated over the entire range of wind speeds. Thus, they can be seen as functions that are only related to the wind speed, i.e., , , . Defining the state vector , input vector , and output vector , a LPV model of the WTS can be given aswhere

For a general system, the LPV model can be formulated aswhere , , and are the state vector, input vector, and output vector and and are disturbance vector and fault vector. is a constant matrix, and the matrices depend on the time-varying parameter bounded as

and are the symbolic variable. The matrices , , , and can be written as polytopic matrices:where are weights of the LPV subsystems. Suppose that

The polytopic representation of system (9) is

Assume that(A1)The pair is observable, .(A2)The matrix is of full row rank and is of full column rank for .(A3)The disturbance and fault are functions with bounded first time derivative.

3. Unknown Input Observer Design for LPV Systems

3.1. Unknown Input Observer for LPV System

For system (13), an UIO can be designed aswhere are matrices with appropriate dimension and the matrices , will be chosen to guarantee the convergence of the fault estimation. The residual and the fault estimation error are defined as and .

Defining the state estimation error and the matrix , where is a unit matrix with appropriate dimension, we obtain that

If the following conditions are satisfied:then the derivative of the state estimation error is

If the fault estimation error tends to zero and the matrix is a Hurwitz matrix, the state estimation error will converge to zero.

3.2. Stability and Convergence Analysis for UIO

Lemma 1 (see [23]). Given a scalar and a symmetric positive definite matrix , the following equality holds:

Theorem 1. If there exist symmetric positive definite matrices , such that the LMIs, are satisfied,and the adaptation law for estimation of the fault is chosen asthen the state estimation error and the fault estimation error converge to a small set.

Proof. Choose the candidate Lyapunov functionThe time derivative of isAccording to equation (21), we haveFrom (23) and (24), one obtainsSince the derivative of the fault is assumed to be bounded, we have , and . According to Lemma 1, one obtainswhere is the maximum eigenvalue of the matrix .
Note that if the fault signal is constant, i.e., , adaptive law (21) becomes an adaptive law for estimation of constant fault. In this case, a proportional integral observer is developed in [24] to estimate the fault by assuming that the fault is a step function. Furthermore, considering time-varying faults, an adaptive polytopic observer for singular delayed LPV systems is developed in [25].
Substituting equation (20) into (25) and using (26), the time derivative of becomesSetting the vector , can be written aswhere .
If there exists matrix , using Theorem 1 and equation (26), one obtainswhere . Then, , for . According to the Lyapunov stability theory [26], both the state estimation error and the fault estimation error converge to a small set.
From foregoing condition (16), one hasThe development of equation (30) givesDenoting , , and , we havewhere is a pseudo-inverse of . The matrices and can be obtained from the block matrix . According to (16), we have

Corollary 1. If there exist a symmetric positive definite matrix and matrices , , such thatwhere , , , , then UIO (14) exists and the estimation errors converge to zero.

Proof. Replacing by in LMI (34), Corollary 1 can be easily proved according to Theorem 1.
By solving LMI (34), the matrices and can be obtained. Then, the remaining matrices are calculated in turn.

4. Integral Controller Design for LPV Systems

In this section, in order to track the reference , an integrator is added to the standard state feedback controller. Then, the system input is

Define auxiliary variable such thatwhere .

The closed-loop system can be written aswhere , , , , , , and The parameter can be tuned to satisfy the controllability condition for calculating the parameters of the controller.

For performance criterion [27], the condition for convergence of tracking error to zero is

Theorem 2. For system (37), if condition (38) is satisfied and there exist positive definite matrix and matrix such that for the following LMIs holds:where , then the closed-loop system is asymptotically stable with performance.

Proof. Consider the Lyapunov function and the functionIntegrating function , one obtainsSince , in order to ensure , the condition should be satisfied. Note that equation (38) is a sufficient condition for ensuring the stability of closed-loop system.
The function can be rewritten asPre and post-multiplying , by , one obtainswhere and .
The matrices and can be computed by solving LMI (43). Then, the matrices and can be obtained by partitioning the matrix .

5. Fault-Tolerant Control for WTS

In this section, fault-tolerant control is proposed for WTS in which actuator fault occurs. The scheme of the proposed control is given in Figure 1.

UIO is first designed using the method proposed in Section 3. If there does not exist a fault, the norm of the residual maintains under a threshold if the matrix in equation (17) is a Hurwitz matrix. The WTS fault estimate is

If a fault occurs, will be over the threshold. Thus, comparing the norm of the residual with the threshold, the fault can be detected. The threshold can be obtained by experiment. Note that when the system is in transient process (the reference is rapidly changed) or in initial stage, the residual should be neglected because of the initial errors of UIO start-up.

If there is no fault, the norm of the residual will be less than certain threshold and the system will operate in normal mode. The fault estimation subsystem will do not work and the fault estimate is set to zero. When the real time residual is over the threshold, a fault is detected and the system goes to the fault mode.

From equation (44), it can be seen that the fault estimate is zero in normal mode. The compensation of the actuator fault will be also zero in this mode. In fault mode, the fault is estimated and compensated. In these two modes, the control law iswhere is a pseudo-inverse of .

The main steps of the proposed FTC design method can be summarized as follows.

Step 1. Compute the matrix from equation (30) and obtain the matrices .

Step 2. Solve LMI (34) and obtain the matrices and and then calculate and .

Step 3. Calculate the matrices from equation (16).

Step 4. Calculate the matrices , , from equation (33). Thus, all UIO parameters are obtained.

Step 5. Solve LMI (43) and obtain the matrices and then calculate the matrices , from .

6. Simulation Results for FTC of WTS

This section presents the results obtained by applying the proposed FTC to a LPV model of the variable-pitch WTS described in [28].

Note that the time-varying parameter in the WTS model of type (9) is the wind speed . Assuming that , the weights of LPV model can be calculated as

In variable-pitch WTS, the pitch subsystem will start working above the rated wind speed. If the wind speed exceeds the cut-out wind speed, the pitch subsystem will be forced to stop working in order to protect itself. In this paper, the interval of wind speed is set as .

Using the WTS parameters values given in [28], one obtains a WTS polytopic model of type (13) with

The corresponding UIO parameters can be calculated by using YALMIP toolbox. One obtains

Using Theorem 1 and solving LMI (43), the following controller gain matrix is obtained:

The wind speed used in the simulations is shown in Figure 2. The initial states are set as and . According to reference [29], the actuator faults can be caused by both the pitch and converter systems and result in either an offset or changed dynamics of pitch angle and torque. In this paper, the following pitch actuator fault signal and torque actuator fault signal are considered:

The obtained simulation results are shown in Figures 3–7. The simulated and estimated signals of pitch and torque actuator faults are displayed in Figure 3. The simulated and estimated values of the system states are shown in Figure 4. The corresponding estimation errors are illustrated in Figure 5 where the actuator fault effect can be seen.

Figures 6 and 7 display the WTS pitch angle and generator speed values obtained applying the proposed FTC method (denoted as “CF”), control method without fault compensation (denoted as “NF”), and state feedback control without integrator (denoted as “SF”). The “NF” controller is obtained by removing the fault compensation in the “CF” controller. The “SF” controller is designed using the method presented in [21]. For the “SF” controller, the fault and state observers are the same as for the “CF” controller.

These figures show that the proposed fault-tolerant controller can ensure better performances in fault and fault-free situations than controller without fault compensation and state feedback controller without integrator.

7. Conclusions

In this paper, a FTC scheme with actuator fault diagnosis based on UIO for WTS is proposed. Considering WTS disturbance and pitch actuator and torque actuator faults, an UIO is first designed to estimate the system fault and state variables. Then, an integral controller for WTS is designed and the stability of the closed-loop system is analyzed. The proposed FTC approach is tested for a variable-pitch wind turbine benchmark. The obtained simulation results demonstrate the high efficiency of the proposed FTC with actuator fault diagnosis.

Data Availability

The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Acknowledgments

This study was supported in part by the Ningbo Natural Science Foundation (no. 2019A610128).

References

M. Hung Do and D. Soffker, “State-of-the-art in integrated prognostics and health management control for utility-scale wind turbines,” Renewable and Sustainable Energy Reviews, vol. 145, Article ID 111102, 2021.
View at: Publisher Site | Google Scholar
C. E. de Souza and J. Osowsky, “Gain-scheduled control of two-dimensional discrete-time linear parameter-varying systems in the roesser model,” Automatica, vol. 49, no. 1, pp. 101–110, 2013.
View at: Publisher Site | Google Scholar
H. S. Ali, M. Darouach, M. Zasadzinski, and M. Alma, “An H∞ LPV control for a class of LPV systems using a descriptor approach: application to a wind turbine mode,” IFAC-PapersOnLine, vol. 48, no. 26, pp. 213–217, 2015.
View at: Publisher Site | Google Scholar
F. D. Bianchi, R. J. Mantz, and C. F. Christiansen, “Gain scheduling control of variable-speed wind energy conversion systems using quasi-lpv models,” Control Engineering Practice, vol. 13, no. 2, pp. 247–255, 2005.
View at: Publisher Site | Google Scholar
E. B. Muhando, T. Senjyu, A. Uehara, and T. Funabashi, “Gain-scheduled ${\cal H}_{\infty}$ control for WECS via LMI techniques and parametrically dependent feedback Part II: controller design and implementation,” IEEE Transactions on Industrial Electronics, vol. 58, no. 1, pp. 57–65, 2011.
View at: Publisher Site | Google Scholar
F. Lescher, J. Y. Zhao, and P. Borne, “Robust gain scheduling controller for pitch regulated variable speed wind turbine,” Studies in Informatics and Control, vol. 14, no. 4, p. 299, 2005.
View at: Google Scholar
L. Fabien, J. Zhao, and B. Pierre, “Switching lpv controllers for a variable speed pitch regulated wind turbine,” in Proceedings of the IEEE Multiconference on Computational Engineering in Systems Applications (IMACS’06), pp. 1334–1340, IEEE, Beijing, China, October 2006.
View at: Google Scholar
S. Bououden, M. Chadli, and H. R. Karimi, “Robust predictive control of a variable speed wind turbine using the lmi formalism,” in Proceedings of the 2014 European Control Conference (ECC), pp. 820–825, IEEE, Strasbourg, France, 2014.
View at: Publisher Site | Google Scholar
E. Kamal, A. Aitouche, R. Ghorbani, and M. Bayart, “Robust nonlinear control of wind energy conversion systems,” International Journal of Electrical Power & Energy Systems, vol. 44, no. 1, pp. 202–209, 2013.
View at: Publisher Site | Google Scholar
H. Habibi, I. Howard, and S. Simani, “Reliability improvement of wind turbine power generation using model-based fault detection and fault tolerant control: a review,” Renewable Energy, vol. 135, pp. 877–896, 2019.
View at: Publisher Site | Google Scholar
R. Wang, H. Jing, H. R. Karimi, and N. Chen, “Robust fault-tolerant H∞ control of active suspension systems with finite-frequency constraint,” Mechanical Systems and Signal Processing, vol. 62-63, pp. 341–355, 2015.
View at: Publisher Site | Google Scholar
Y. Zhang and J. Jiang, “Bibliographical review on reconfigurable fault-tolerant control systems,” Annual Reviews in Control, vol. 32, no. 2, pp. 229–252, 2008.
View at: Publisher Site | Google Scholar
Y. Vidal, L. Acho, and F. Pozo, “Robust fault detection in hysteretic base-isolation systems,” Mechanical Systems and Signal Processing, vol. 29, pp. 447–456, 2012.
View at: Publisher Site | Google Scholar
E. Kamal, A. Aitouche, R. Ghorbani, and M. Bayart, “Robust fuzzy fault-tolerant control of wind energy conversion systems subject to sensor faults,” IEEE Transactions on Sustainable Energy, vol. 3, no. 2, pp. 231–241, 2012.
View at: Publisher Site | Google Scholar
B. Li, Q. Hu, G. Ma, and Y. Yang, “Fault-tolerant attitude stabilization incorporating closed-loop control allocation under actuator failure,” IEEE Transactions on Aerospace and Electronic Systems, vol. 55, no. 4, pp. 1989–2000, 2019.
View at: Publisher Site | Google Scholar
P. Kühne, F. Pöschke, and H. Schulte, “Fault estimation and fault-tolerant control of the fast nrel 5-mw reference wind turbine using a proportional multi-integral observer,” International Journal of Adaptive Control and Signal Processing, vol. 32, no. 4, pp. 568–585, 2018.
View at: Publisher Site | Google Scholar
A. Azizi, H. Nourisola, and S. Shoja-Majidabad, “Fault tolerant control of wind turbines with an adaptive output feedback sliding mode controller,” Renewable Energy, vol. 135, pp. 55–65, 2019.
View at: Publisher Site | Google Scholar
V. Pashazadeh, F. R. Salmasi, and B. N. Araabi, “Data driven sensor and actuator fault detection and isolation in wind turbine using classifier fusion,” Renewable Energy, vol. 116, pp. 99–106, 2018.
View at: Publisher Site | Google Scholar
X. J. Li and N. Wang, “Data-driven fault estimation and control for unknown discrete-time systems via multiobjective optimization method,” IEEE Transactions on Cybernetics, vol. 52, no. 5, pp. 3289–3301, 2022.
View at: Publisher Site | Google Scholar
P. Lluís Negre, V. Puig, and I. Pineda, “Interval lpv identification and fault diagnosis of a real wind turbine,” IFAC Proceedings Volumes, vol. 45, no. 16, pp. 1689–1694, 2012.
View at: Publisher Site | Google Scholar
T. Bakka, H. Karimi, and S. Christiansen, “Linear parameter varying modelling and control of an offshore wind turbine with constrained information,” IET Control Theory & Applications, vol. 8, no. 1, pp. 22–29, 2014.
View at: Publisher Site | Google Scholar
F. D. Bianchi, H. De Battista, and R. J Mantz, Wind Turbine Control Systems: Principles, Modelling and Gain Scheduling Design, Springer, Verlag London, p. 205, 2007.
K Zhang, B. Jiang, and V. Cocquempot, “Adaptive observer-based fast fault estimation,” International Journal of Control, Automation and Systems, vol. 6, no. 3, p. 320, 2008.
View at: Google Scholar
H. Hamdi, M. Rodrigues, C. Mechmeche, D. Theilliol, and N. B. Braiek, “Fault detection and isolation in linear parameter-varying descriptor systems via proportional integral observer,” International Journal of Adaptive Control and Signal Processing, vol. 26, no. 3, pp. 224–240, 2012.
View at: Publisher Site | Google Scholar
M. Rodrigues, H. Hamdi, N. Benhadj Braiek, and D. Theilliol, “Observer-based fault tolerant control design for a class of lpv descriptor systems,” Journal of the Franklin Institute, vol. 351, no. 6, pp. 3104–3125, 2014.
View at: Publisher Site | Google Scholar
A. H. Hassanabadi, M. Shafiee, and V. Puig, “Uio design for singular delayed lpv systems with application to actuator fault detection and isolation,” International Journal of Systems Science, vol. 47, no. 1, pp. 107–121, 2016.
View at: Publisher Site | Google Scholar
A. Zemouche and M. Boutayeb, “Sobolev norms-based state estimation and input recovery for a class of nonlinear systems. design and experimental results,” IEEE Transactions on Signal Processing, vol. 57, no. 3, pp. 1021–1029, 2009.
View at: Publisher Site | Google Scholar
F. A. Shirazi, K. M. Grigoriadis, and D. Viassolo, “Wind turbine integrated structural and lpv control design for improved closed-loop performance,” International Journal of Control, vol. 85, no. 8, pp. 1178–1196, 2012.
View at: Publisher Site | Google Scholar
F. P. Odgaard, J Stoustrup, and K Michel, “Fault-tolerant control of wind turbines a benchmark model,” Control Systems Technology IEEE Transactions on IFAC Proceedings Volumes, vol. 42, no. 8, pp. 155–160, 2013.
View at: Google Scholar

Copyright

Copyright © 2022 Shanzhi Li et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

PDF Download Citation

Download other formats

Order printed copies