Predictive control method based on a single linear model for reducing structural loads in onshore, offshore and floating wind turbines

Abstract

The present invention relates to a method for controlling the angles of attack of wind turbine blades, based on a linear model of the wind turbine using a parameter indicative of the acceleration of the wind turbine tower, the linear model being constant across the entire wind turbine operating range. The invention further relates to processing means configured to execute the model-based predictive control. The invention further relates to a method for controlling the position of a moving mass coupled to the wind turbine, based on a linear model of the wind turbine using a parameter indicative of the acceleration of the wind turbine tower and a parameter indicative of the position of the moving mass, the linear model being constant across the entire wind turbine operating range.

Description

[0001] DESCRIPTION

[0002] A predictive control method based on a single linear model to reduce structural loads on land-based, offshore, and floating wind turbines.

[0003] TECHNICAL SECTOR

[0004] The present invention relates to a method for controlling angles of attack of blades of a wind turbine, to a method for controlling a position of a moving mass coupled to a wind turbine, and to processing means configured to execute said controls.

[0005] BACKGROUND

[0006] Systems for reducing the structural loads to which wind turbines are subjected during operation are already known in the state of the art, especially when operating at wind speeds above the nominal speed. However, a system that improves the reduction of oscillations (e.g., vibrations) that occur in a rotor, tower, and / or the floating structure of a wind turbine is desirable. Traditional methods for reducing structural loads based on SISO (i.e., single-input, single-output) control are highly sensitive to modeling errors, which requires retuning to maintain adequate control when changes occur in turbine dynamics (e.g., due to ice formation on the blades or on the floating structure of a floating wind turbine).Furthermore, these methods do not compensate for the delay of the actuators used, nor do they adequately manage the operating constraints of the installation, which significantly limits the performance that can be achieved.

[0007] The use of MIMO (multiple input, multiple output) methods is also known, for example, those based on model-based predictive control techniques. However, these methods have disadvantages that make them difficult to apply to reducing the structural load of a wind turbine.

[0008] DESCRIPTION OF THE INVENTION

[0009] In order to overcome the drawbacks of the prior art, a first aspect of the present invention relates to a method for controlling angles of attack of blades of a wind turbine, the method comprising: providing a model of the wind turbine; the model being linear and invariant over the entire operating range of the wind turbine; and applying model-based predictive control (MPC), using the linear model to determine, with processing means and based on a parameter indicative of acceleration of a tower of the wind turbine, an angles of attack control signal for the blades of the wind turbine; the parameter indicative of tower acceleration being an input to the model-based predictive control.

[0010] The first aspect of the invention enables compact control of the wind turbine, requires limited computational capacity, and contributes to, and may even guarantee, the convergence of the MPC control optimization algorithm. This contrasts with other approaches based on the MPC methodology, which, due to how they manage the high nonlinearities present in both the aeroelastic behavior of the rotor and, where applicable, the hydrodynamic behavior of a floating structure of a floating wind turbine, cannot guarantee convergence and require high computational capacity for their practical implementation. Robustness in turbine control can be achieved through the use of recessive prediction horizons and a state estimator.This control methodology eliminates the need for controller retuning, compensates for delays caused by actuators, optimally manages installation constraints, and guarantees the stability of the control algorithm in advance.

[0011] The linear and invariant model across the entire operating range (performance) of the wind turbine used in the control is a model that does not vary with changing turbine operating conditions (i.e., it is independent of the turbine operating conditions), and therefore, the linear model does not vary over time.

[0012] Furthermore, the proposed control is compatible with controls already implemented in existing wind turbines and can also be complemented. The invention is easily applicable to existing turbines because the proposed control requires relatively little computing power, generally assumed by the control platforms currently installed in most wind turbines, and merely requires a software update (i.e., no modification of the turbine's mechanical components). The present invention has been shown to reduce the fore-aft vibrations (F-A-T) of a wind turbine tower by nearly 70%, whether on land, at sea, or floating. Similar reductions can be achieved in the pitching movements of the floating structure of a floating wind turbine.On the other hand, it should be noted that, compared to the first aspect of the invention, the model-based predictive controls known in this context require relatively high computing power, which is not available by default in most wind turbines installed to date. Furthermore, they do not guarantee control stability, partly due to the use of a nonlinear internal model or different linearized models for different turbine operating points.

[0013] The angles of attack of the blades can be equal. Therefore, the blade angle of attack control signal keeps the blade angles equal (i.e., it is a control signal with the same angle for all blades). This relationship between angles of attack is common in wind turbines.

[0014] The angles of attack of the blades can be different from each other. Therefore, the blade angle of attack control signal allows the angle of attack of each blade to be different from the angles of attack of the other blades. Allowing different blade angles of attack can be useful to reduce the stress load on one of the blades when it is subjected to a high load. This blade control is not as common as the previous one for wind turbines, but it can be useful (especially in wind turbines with particularly large blades).

[0015] In some embodiments, the method comprises adjusting the angles of attack of the blades based on the angles of attack control signal of the blades.

[0016] In some embodiments, the model-based predictive control is a predictive control based on a single linear model (or as it is known in English “Linear Unique Model Predictive Control” (LUMPC)).

[0017] In some embodiments, the method comprises providing a processing block (the processing block being outside the model used for MPC prediction) of a non-linear model (e.g., of a non-linear function), the non-linear model relating an aerodynamic thrust of the turbine (e.g., an incremental aerodynamic thrust on the rotor of the wind turbine) to an adjustment of the angles of attack of the blades (e.g., to an incremental adjustment of the angles of attack of the blades);and the step of applying a model-based predictive control to determine, with processing means and based on a parameter indicative of acceleration of a wind turbine tower, a control signal for angles of attack of the wind turbine blades comprises: applying the model-based predictive control using the linear model to determine, with processing means and based on the parameter indicative of acceleration of the wind turbine tower, a signal indicative of aerodynamic thrust control of the wind turbine; the signal indicative of aerodynamic thrust control of the wind turbine being an output of the model-based predictive control; executing, with processing means, the processing block using the signal indicative of aerodynamic thrust control, determining, as a result of said processing, a second signal (e.g., an additive signal) for angles of attack control of the wind turbine blades;and wherein the blade angle of attack control signal is the second blade angle of attack control signal or the method comprises determining, with processing means, the blade angle of attack control signal based on the second blade angle of attack control signal;

[0018] To ensure that the internal model of the LLIMPC controller is linear and unique across the entire operating range, some processing blocks external to the LLIMPC controller itself may be required. These external blocks allow variables that maintain a highly nonlinear physical relationship to be linked. Thanks to this, once the invariant, linear internal model used by the LLIMPC controller to predict the dynamic evolution of the degrees of freedom of interest in the wind farm has been defined, the implementation of these external blocks can be developed using the measured variables available for the control system.

[0019] In this way, one or more nonlinearities that have a relatively significant impact on turbine control are removed from the linear model. This nonlinearity is thus taken into account in turbine control, while simultaneously leveraging the advantages of predictive control, which uses an internal, linear, and invariant model to optimize the dynamic evolution of the system over the recessive time horizon considered. This combination has been observed to achieve particularly stable and robust control, allowing for a significant reduction in the structural load on the wind turbine during operation.

[0020] In some embodiments, the parameter indicative of acceleration of a wind turbine tower is a parameter indicative of acceleration of an upper portion (e.g., an upper end) of the tower. The upper portion of the tower is closer to the rotor than the lower portion, and therefore, comparatively, an acceleration of the upper portion allows a more accurate understanding of the acceleration (and other position-related parameters) of the wind turbine rotor, which is the element that generally suffers the greatest disturbance due to the aerodynamic thrust of the incident wind, which is one of the possible sources of structural loading during the operation of the installation. In some embodiments, the turbine is a floating offshore wind turbine (FOWT), the floating wind turbine comprising a floating structure, and the floating structure supporting the wind turbine tower.Furthermore, an input to MPC is a parameter indicative of the floating structure's acceleration; and the blade angle of attack control signal is determined based on this parameter indicative of the floating structure's acceleration. Because oscillations of the floating structure (e.g., due to waves in the water where the floating structure is floating) have a relatively high impact on turbine stability, considering the floating structure's acceleration by MPC allows for improved control and further reduction of the structural load on the tower and / or floating platform.

[0021] In some embodiments, the parameter indicative of the acceleration of the floating structure is a parameter indicative of the pitch acceleration of the floating structure. The use of pitch acceleration has been found to be particularly advantageous for minimizing the energy of oscillations of the floating structure.

[0022] In some embodiments, the wind turbine is mechanically coupled to a movable mass to adjust a spatial weight distribution of the wind installation.

[0023] In some embodiments, a parameter indicative of the position of the movable mass is an input to the model-based predictive control; and the angle-of-attack control signal (e.g., an incremental angle-of-attack control signal) for the blades is determined based on the parameter indicative of the position of the movable mass.

[0024] In some wind turbines, a movable mass (e.g., multiple movable masses) is used to improve the spatial distribution of the turbine's weight while the turbine is in operation, for example, by stabilizing the turbine. The movement of this movable mass reduces the structural load on the turbine under harsh operating environmental conditions (e.g., intense wind and / or wave(s)). In this way, the movable mass reduces variations in the control signals for the blade angles of attack and / or the generator torque, thereby minimizing unwanted variations in the power generated by the generator and / or fatigue of the actuators involved in adjusting these angles of attack and generator torque.

[0025] In some embodiments, the motion of the moving mass is adjustable by a controller and an actuator (i.e., they constitute a device known as an "Active Mass Damper (AMD)"). In some embodiments, the motion of the moving mass is self-regulated during operation of the wind turbine, without the need for a controller (i.e., they constitute a device known as a "Tuned Mass Damper (TMD)"), for example, by making use of passive mechanical elements such as springs and some type of passive dampers, which cause said TMDs to behave as harmonic oscillators. In some embodiments, instead of using springs, the mass moves on mechanical guides provided with a certain curvature.For example, mechanical guides can be shaped like an inverted arc of a circle, that is, an arc of a circle with the concave part of the arc facing upward. By appropriately selecting the physical characteristics of the mass and the aforementioned mechanical elements, the natural frequency of the TMD motion can be set so that it matches ("tunes") with a frequency of interest to the system to which said TMDs are coupled. At this point, we recall that the natural frequency of a dynamic system, described with a single degree of freedom, is the frequency of the applied excitation force at which said system exhibits the greatest amplitude oscillatory response.

[0026] In some embodiments, moving mass systems are used that constitute a hybrid of the "active mass damper" and "tuned mass damper" types, known as "hybrid mass dampers" (HMD). Finally, hybrid mass dampers can be used that allow the natural frequency of their movement to be adjusted, in passive mode, during operation, known as "adjustable hybrid mass dampers" (AHMD). Since the aforementioned adjustable hybrid mass dampers (AHMD) can adapt to changing circumstances of the wind turbine installation (waves, changes in rotor vibration dynamics, etc.) and can also be used in both passive and active modes, they are the most flexible to use, so they are the ones we will consider by default from now on.Obviously, the first aspect of the invention can be applied to wind installations that do not have AHMD stabilization systems, or that are only passive (TMD) or only active (AMD) systems, adjustable or not.

[0027] In some embodiments, a parameter indicative of the position of the mass (e.g., of an AHMD installed on the wind turbine) is an input to the model-based predictive control; and the blade angle of attack control signal is determined based on the parameter indicative of the position of the moving mass.

[0028] The input associated with the parameter indicating the actual position of the moving mass allows for improved control implemented with model-based predictive control, since the spatial weight distribution of the turbine at any given time is known in greater detail.

[0029] In some embodiments, the method comprises applying model-based predictive control using a single linear model to determine, with processing means and based on the parameter indicative of the position of the movable mass, a control signal for the movement of the movable mass.

[0030] In this way, model-based predictive control, which uses the same linear model as for adjusting the blade angles of attack, can be used to adjust the position of the moving mass.

[0031] In some embodiments, the movable mass is installed inside the turbine nacelle. In some embodiments, the movable mass is installed on the floating structure (e.g., inside the floating structure) of the turbine that supports the turbine tower. In some embodiments, movable masses are installed on the nacelle and the floating structure. The arrangement of the mass in the nacelle of the wind turbine makes it possible to adjust the spatial weight distribution of the nacelle, thereby improving, above all, the stability of the tower. The arrangement of the mass in the floating structure makes it possible to adjust the spatial weight distribution of the floating structure, thereby improving, above all, the stability of the floating structure.By designing the LUMPC controller that generates the additive control signal for the angles of attack of the blades and the positioning of the moving masses, the stability of the installation as a whole can be considered, since the dynamics of its different mechanical parts (rotor, nacelle, tower, platform) are closely coupled.

[0032] In some embodiments, controlling the movable masses coupled to the system makes it possible to substantially reduce the need for the control signal additive to the rotor blade angles of attack. In some embodiments, controlling the movable masses coupled to the system makes it possible to completely eliminate the need for using the control signal additive to the rotor blade angles of attack.

[0033] In some embodiments, the moving mass motion control signal is an output of the model-based predictive control. It has been observed that no further processing of the model-based predictive control output is required to obtain a moving mass motion control signal that allows for reducing the structural load on the wind turbine during operation.

[0034] In some embodiments, an objective of the model-based predictive control is to minimize a structural load on the tower. For example, to minimize oscillations (e.g., vibrations) about an imaginary axis perpendicular to a blade rotation axis and a height direction of the tower. These oscillations, in the context of wind turbines, are referred to as back-and-forth oscillations. Minimizing oscillations can be achieved by setting, as a control objective, a minimization of tower deflection and / or its time derivative. Alternatively, preferably with a lower priority than minimizing tower deflection and / or its time derivative, a minimization of floating structure pitch and / or its time derivative can be added to the control objective.The lowest priority can be achieved by assigning a lower weight to the floating structure pitch minimization objective than to the tower deflection minimization objective.

[0035] The control objective can be implemented as a cost function to be optimized by processing means; the cost function comprises a value predicted, using the linear model, by predictive control.

[0036] In some embodiments, an objective of model-based predictive control is to minimize a structural load on the floating structure. For example, to minimize the pitch oscillations (e.g., vibrations) of the floating structure and / or their time derivative. Minimizing the pitch of the floating structure can be achieved by setting, as a control objective, minimizing the energy of such oscillations.

[0037] In some embodiments, the angle of attack control signal is determined based on an output of the model-based predictive control and an output of a blade angle of attack control other than the model-based predictive control. Thus, the model-based predictive controller proposed in the invention is optional and complements a base blade angle of attack controller, allowing for improved control of the turbine by reducing the structural load, while maintaining the base controller, which may be pre-installed on the turbine. In fact, one of the recommended implementations of the invention is posed as a simple update of the wind farm's control software.

[0038] In some embodiments, the objective of the model-based predictive control is to minimize oscillations about an imaginary axis perpendicular to a blade rotation axis and a tower height direction.

[0039] In some embodiments, the blade angle of attack control signal is determined based on: the output of the blade angle of attack control other than the model-based predictive control, and the second blade angle of attack control signal, which may be used additively to the former output. A second aspect of the present invention relates to processing means configured to execute model-based predictive control using a linear model to determine, based on a parameter indicative of acceleration of a wind turbine tower, a blade angle of attack control signal for a wind turbine; the parameter indicative of tower acceleration being an input to the model-based predictive control; the linear model being invariant over the entire operating range of the wind turbine.

[0040] The processing means of the second aspect of the invention are configured to allow the model-based predictive control of the first aspect of the invention to be executed.

[0041] The second aspect of the invention allows for similar advantages to those of the first aspect of the invention.

[0042] A third aspect of the present invention relates to a method for controlling a position of a movable mass coupled to a wind turbine, the mass being for adjusting a spatial weight distribution of the turbine, the method comprising: providing a linear model of the wind turbine; the linear model being invariant over all operating range of the wind turbine; and applying a model-based predictive control using the linear model to determine, with processing means and based on a parameter indicative of acceleration of the wind turbine tower and a parameter indicative of the position of the movable mass, a control signal of the movement of the movable mass; the parameter indicative of acceleration of the tower being an input of the model-based predictive control; and the parameter indicative of the position of the movable mass being an input of the model-based predictive control.

[0043] In some embodiments, the turbine is a floating offshore wind turbine, the turbine comprising a floating structure, and the floating structure supporting the turbine tower; an input to the model-based predictive control is a parameter indicative of acceleration of the floating structure; and the determination of the control signal for the motion of the moving mass is performed based on the parameter indicative of acceleration of the floating structure.

[0044] As indicated below, the movement of the mass can be controlled in several ways. This control minimizes wind turbine oscillations.

[0045] In some embodiments, the movable mass is coupled to the turbine by a guide that allows the movable mass to move relative to the turbine; the movable mass is coupled to an electric machine by a mechanical transmission such that adjusting the electromagnetic torque of the electric machine allows adjusting the displacement of the movable mass along the guide; the motion control signal comprises at least one electromagnetic torque adjustment signal of the electric machine.

[0046] The guide may be coupled directly to the nacelle, to the floating structure, or there may be one guide coupled to the nacelle for a first mass and another guide coupled to the floating structure for a second mass. The mass may be inside the nacelle. The mass may be inside the floating structure.

[0047] For example, the method may comprise adjusting a displacement of the mass, the mass being mechanically coupled to an electric machine by a mechanical transmission, the mechanical transmission being such that an adjustment of the electromagnetic torque of the electric machine allows adjusting the displacement of the mass in oscillations of the turbine, the displacement being a displacement with respect to the turbine; the mass being coupled to the turbine with a mechanical coupling (i.e., the guide), the mechanical coupling of the mass with the wind turbine allowing the displacement of the mass caused by a gravitational force and by a force applied by the electric machine to the mass in the oscillations through the mechanical transmission, by adjusting the electromagnetic torque of the electric machine; the oscillations causing variations of an angle of inclination of the oscillating structure with respect to the direction of the gravitational force;the displacement being adjusted by adjusting the force applied by the electric machine to the mass; the method comprising: i) providing a control with a model of the dynamics of the oscillating structure, the mass, the mechanical transmission, the mechanical coupling and the electric machine;

[0048] (i) establishing as a target of the model-based predictive control a certain inclination of the turbine (e.g. inclination of the turbine tower or of the floating structure of the turbine), or the derivative thereof with respect to time, (iii) obtaining: (i) a parameter indicative of a position of the mass, (ii) a parameter indicative of an angle of inclination of the turbine, where the angle relative to a direction of the gravitational force is;and provide! a parameter indicative of a position of the mass and a parameter indicative of an angle of inclination to the control as inputs of the control, iv) execute the control, using the model, the control objective, the parameter indicative of a position of the mass and the parameter indicative of an angle of inclination of the turbine, to obtain an indicative parameter of adjustment of the force applied, by the electric machine, to the mass, through the adjustment of the electromagnetic torque, v) adjust, based on the indicative parameter of adjustment of the force through the adjustment of the electromagnetic torque, the force applied, by the electric machine, to the mass, and vi) repeat iii) av).;

[0049] In some embodiments, the moving mass carries the aforementioned electric machine, as well as the mechanism for multiplying the angular velocity of the machine's shaft with respect to the axis of a pinion, which is in turn engaged in a mechanical rack integral with the turbine. This mechanical transmission device, of the type known as rack-and-pinion, allows the mass to be propelled, thanks to the electromagnetic torque applied by the electric machine to its shaft, to control its oscillatory movement on rails (guides) shaped like an inverted circular arc—with the concave part facing upwards—and of a certain curvature. Using these curved rails allows the moving mass to be used also in passive mode (the electric machine does not apply any torque), without the need for a spring-based suspension mechanism, since such mechanisms are usually very inconvenient in practice.The hybrid nature (possible active-passive use) of the device can, in some embodiments, be given an adjustable nature by incorporating a continuously variable torque gearbox (CVT) into the shaft connecting the electric machine to the transmission pinion. This gearbox can provide any value for the transmission ratio within a predefined range, or alternatively, by using guides whose curvature can be changed at will, also continuously, within a range. These adjustments make it possible to select, within a range, the undamped natural frequency of oscillation of the moving mass, which can improve its performance when operating in passive mode.At this point, we recall that the undamped natural frequency of a dynamic system, described with a single degree of freedom, is the frequency of the applied excitation force at which the system exhibits the largest amplitude oscillatory response, if we neglect the damping of the system. This adjustment can also improve very important operational aspects when operating in active mode, making it very convenient, although not absolutely necessary in that mode. It should be noted that a multiplier could also be used to set a finite number of different torque transmission ratios within a range; however, in that case, the choice of the undamped natural frequency for the moving mass could only be implemented approximately.

[0050] In some embodiments, the movable mass is coupled to a portion of the turbine by a guide allowing the movable mass to move relative to the turbine; the movable mass being coupled to an electrical machine by a mechanical transmission provided with a multiplier with a variable transmission ratio, such that adjusting the electromagnetic torque of the electrical machine makes it possible to control the movement of the movable mass along the guide; the method comprising at least one of a) and b): a) estimating a frequency of a disturbance causing oscillation of the portion of the wind turbine to which the movable mass is coupled; and adjusting the transmission ratio of the multiplier based on the estimate of the frequency of the disturbance such that the undamped natural frequency of the movable mass is close to (e.g., substantially equal to) the frequency of the disturbance impacting the portion of the turbine to which the movable mass is coupled;and b) adjusting the gear ratio of the multiplier so that the undamped natural frequency of the moving mass is close to (for example, substantially equal to) the undamped natural frequency of the part of the turbine to which the moving mass is coupled;

[0051] For example, the gear ratio set in b) may be such that: w. a> m = - - — a> sm s +m where m s is the mass of the part of the wind turbine where the moving mass is attached, m is the mass of the moving mass, or m is the undamped natural frequency of the moving mass I s It is the undamped natural frequency of the part of the wind turbine where the moving mass is coupled.

[0052] The part of the wind turbine to which the moving mass is attached is, for example, the nacelle and tower. Another example of a part to which the mass is attached, in the case of a floating wind turbine, is the floating structure of the wind turbine.

[0053] In some embodiments, the movable mass is coupled to a portion of the turbine by a variable curvature guide (for example, the mechanical guide may have the shape of an inverted arc of a circle, i.e., an arc of a circle with the concave portion of the arc facing upwards) that allows the movable mass to move relative to the turbine; the movable mass being coupled to an electrical machine by means of a mechanical transmission such that adjusting the electromagnetic torque of the electrical machine allows the displacement of the movable mass along the guide to be controlled; the method comprising at least one of a) and b): a) estimating a frequency of a disturbance that causes oscillation of the portion of the wind turbine to which the movable mass is coupled;and adjusting the guide curvature based on the estimate of the disturbance frequency so that the undamped natural frequency of the moving mass is close to (e.g. substantially equal to) the frequency of the disturbance incident on the part of the turbine to which the moving mass is coupled; and b) adjusting the guide curvature so that the undamped natural frequency of the moving mass is close to (e.g. substantially equal to) the undamped natural frequency of the part of the turbine to which the moving mass is coupled. For example, the gear ratio adjusted in b) may be such that: w. a> m = - - — a> sm; s +m where m s is the mass of the part of the wind turbine where the moving mass is coupled, m is the mass of the moving mass, com is the undamped natural frequency of the moving mass and ms is the undamped natural frequency of the part of the wind turbine where the moving mass is coupled.

[0054] The part of the wind turbine to which the moving mass is attached is, for example, the nacelle and tower. Another example of a part to which the mass is attached, in the case of a floating wind turbine, is the floating structure of the wind turbine.

[0055] It is well known that the frequency of the incident disturbance is close to the oscillation frequency of the turbine part affected (for example, the oscillation frequency of the turbine tower and / or the oscillation frequency of the floating structure). Logically, in cases where the aim is to minimize tower oscillations, the mass is directly coupled to the tower (for example, to the nacelle) and the frequency of the disturbance affecting the tower (wind) is estimated. Logically, in cases where the aim is to minimize oscillations of the floating structure, the mass is directly coupled to the floating structure and the frequency of the disturbance affecting the floating structure (especially waves) is estimated.The adjustments in a), both in the case of curvature adjustment and gear ratio adjustment, use the estimated frequency of the disturbance affecting the different parts of the system at any given time. This estimate is based on the dynamic model of the system and measurements taken on the turbine (e.g., tower acceleration and pitching of the floating structure). However, the estimation of unmeasured input disturbances is usually performed by the MPC controller during normal operation, so that this can be used to obtain the frequency used in the adjustment in a), for example, if the device is operating in its active mode. Obviously, the estimation could also be implemented using processing external to that of the MPC controller.On the other hand, adjustments (b) of the method can be used when such an estimate of the disturbance frequency is unavailable, for example, during operation of the AHMD device in passive mode, with the MPC controller and its state estimator disabled. In this case, the frequency response of the entire system, i.e., the turbine and the coupled moving mass device, under a range of disturbance frequencies that sufficiently covers the operating conditions that the installation may encounter at its location, is used in the AHMD adjustment.

[0056] The adjustments a) are based on the theory of antiresonance which indicates that adjusting the natural frequency of the moving mass so that it is close to the frequency of the disturbance at any given time, allows the turbine oscillations caused by the disturbance to be minimized.

[0057] Adjustments (b) can be performed in numerous ways (see e.g. Michele Zilletti, Stephen J. Elliott, Emiliano Rustighi, “Optimisation of dynamic vibration absorbers to minimise kinetic energy and maximise internal power dissipation” Journal of Sound and Vibration, Volume 331, Issue 18, 2012, pp. 4093-4100 for a summary of different methods that exist to tune the natural frequency and damping of a dynamic vibration absorber). However, most of the proposed methods aim for a more or less “flat” frequency response over the whole range of disturbance frequencies of interest. Furthermore, all methods propose to achieve this by setting the natural frequency of the absorber device to be closer to the natural frequency of the oscillating structure, the smaller the moving mass is relative to the mass of the structure.Specifically, for the case in which the moving mass is at most 10% of the mass of the oscillating structure, the recommended undamped natural frequency for said mass is close to the undamped natural frequency of the structure. In some embodiments, the moving mass used is very small relative to the mass of the oscillating structure. Specifically, the moving mass devices typically installed in the nacelle or floating structure of a wind turbine rarely exceed 5% of the mass of the structure to which they are attached and are typically around 2%.Therefore, in these cases, the undamped natural frequency of the moving mass can be adjusted to be close to (e.g., coincident with) the undamped natural frequency of the turbine part to which it is coupled, i.e., the natural oscillation frequency of the tower or the floating structure depending on whether the moving mass is, respectively, coupled to the tower (e.g., in the nacelle) or to the floating structure. Once this frequency has been determined, the CVT transmission ratio and / or the guide curvature can be adjusted to implement this undamped natural frequency in the moving mass.

[0058] The third aspect of the invention allows for similar advantages to the first aspect of the invention, as it is also based on using model-based predictive control that uses the linear invariant model to improve control of the wind turbine (e.g., minimizing fore-aft oscillations of a floating structure of a floating wind turbine and / or of the tower of a wind turbine).

[0059] The various aspects and embodiments of the invention defined above can be combined with each other, provided they are mutually compatible. For example, two or more of the following can be combined: blade angle of attack control, electromagnetic torque control of the electric machine, torque transmission ratio control, and guide curvature radius control. The third aspect of the invention allows reducing mechanical fatigue of the blade actuators, as it allows blade angle of attack control to require smaller (smoother) changes in the blade angle of attack.

[0060] Additional advantages and features of the invention will become apparent from the following detailed description and will be particularly pointed out in the appended claims.

[0061] BRIEF DESCRIPTION OF THE DRAWINGS

[0062] To complement the description and in order to help better understand the characteristics of the invention, in accordance with some examples of practical implementation of the invention, a set of figures is attached as an integral part of the description, in which, for illustrative and non-limiting purposes, the following has been represented:

[0063] Figure 1 shows a portion of a wind turbine according to some embodiments of the invention. Figure 2 shows a control diagram of a wind turbine according to some embodiments of the invention.

[0064] Figure 3 shows the internal structure of a linear model according to some embodiments of the invention.

[0065] Figure 4 shows a linear model dedicated to the assembly comprising rotor, tower, floating structure and the two AHMDs installed in the wind turbine, according to some embodiments of the invention.

[0066] Figure 5 shows a linear model dedicated to the assembly comprising rotor, tower and an AHMD installed in the nacelle of the wind turbine, for the “onshore” and “monopile” cases.

[0067] Figure 6 shows a linear model dedicated to the assembly comprising rotor, tower, floating structure and the two AHMDs installed in the wind turbine, according to some embodiments of the invention for the “Barge” type FOWT case.

[0068] Figure 7 shows a linear model dedicated to the assembly comprising rotor, tower, floating structure and the two AHMDs installed in the wind turbine, according to some embodiments of the invention for the FOWT type “Spar-buo^' case.

[0069] Figures 8a and 8b show the speed of the tower tip in meters / second of the application example of the method of the invention for the case of gust and turbulent wind, respectively.

[0070] Figures 9a and 9b show the deflection of the tower tip in meters of the application example of the method of the invention for the case of gust and turbulent wind, respectively.

[0071] Figures 10a and 10b show the fore-aft bending moment of the tower in N m of the application example of the method of the invention for the case of gust and turbulent wind, respectively.

[0072] Figures 11a and 11b show the aerodynamic thrust (solid line) and its on-line estimate (dashed line) in Newtons of the application example of the method of the invention for the case of gust and turbulent wind, respectively.

[0073] Figures 12a and 12b show the angle of attack -pitch- of the rotor blades in degrees of the application example of the method of the invention for the case of gust and turbulent wind, respectively.

[0074] Figures 13a and 13b show the derivative of the angle of attack -pitch- of the rotor blades in degrees / second of the application example of the method of the invention for the case of gust and turbulent wind, respectively.

[0075] Figures 14a and 14b show the generator speed in rad / s of the application example of the method of the invention for the case of gust and turbulent wind, respectively.

[0076] Figures 15a and 15b show the generated power (c£> ge n*Tgen) in Watts of the application example of the method of the invention for the case of gust and turbulent wind, respectively.

[0077] DESCRIPTION OF A PREFERRED EMBODIMENT OF THE INVENTION

[0078] In describing the possible preferred embodiments of the invention, it is necessary to provide numerous details to facilitate a better understanding of the invention. However, it will be apparent to those skilled in the art that the invention can be implemented without these specific details. Furthermore, well-known features have not been described in detail to avoid unnecessarily complicating the description.

[0079] Figure 1 shows a wind turbine 10 according to some embodiments of the invention. For simplicity, the lower anchoring of the turbine tower 11 has not been shown. The turbine 10 comprises a tower 11, a nacelle 12 coupled, preferably rotatably, to the upper end of the tower 11, a rotor comprising a main shaft 13 coupled to the nacelle 12, blades 14 coupled to the main shaft 13, and a control system according to embodiments of the invention. The nacelle 12 and the blades 14 are rotated and oriented in the wind direction by a yaw angle adjustment system. In some embodiments, the nacelle 12 houses components (not shown) of the turbine 10 for generating electrical power from kinetic energy of the wind, including a generator, a gearbox, a transmission, and a brake assembly. In the operation of the turbine, wind energy causes a rotational movement of the blades.The rotational motion of the blades causes the generator to rotate, generating electrical power. The electrical power generated by the turbine 10 may be transmitted, for example, to the electrical grid. In some wind turbines, the gearbox may not be used, and at least one converter comprises components that may not be in the nacelle 12. The components of the wind turbine 10 illustrated are in their operating positions; specifically, the rotor is installed in the nacelle 12, and each of the blades 14 is installed on the rotor and the main shaft 13.

[0080] In some embodiments, the wind turbine is variable-speed, meaning the rotor rotation speed can be adjusted to the wind speed. The mechanical load supported by variable-speed turbines during operation is typically greater than the mechanical load supported by non-variable-speed turbines.

[0081] Figure 2 schematically shows a control of a floating wind turbine 20 in water (e.g., the wind turbine illustrated in Figure 1). The floating wind turbine comprises a floating structure. The floating structure supports the wind turbine tower. The tower and the floating structure are subjected to structural loads resulting from disturbances caused by, e.g., wind and waves.

[0082] The floating wind turbine 20 includes two movable masses with adjustable positions to configure a certain spatial weight distribution of the wind turbine. The movable masses are coupled, respectively, inside the floating structure (hereinafter called mass Mi, and also, MAHMDI) and inside the nacelle (hereinafter called mass M2, and also, MAHMD2), so as to allow adjusting the spatial weight distribution of the floating wind turbine. The adjustment of the positions xi and X2 (also called XAHMDI and XAHMD2, respectively) of the respective movable masses Mi and M2 allows acting on the oscillations of the wind turbine in order, for example, to stabilize the wind turbine, more specifically, to reduce fore-aft vibrations (i.e., forward and backward) of the tower and / or pitch oscillations of the floating structure, in the FOWT case.The aforementioned moving masses are considered to constitute hybrid mass dampers that allow the frequency of their motion to be adjusted, in a passive mode, during operation. These are known as adjustable hybrid mass dampers (AHMDs), introduced in the previous section. The AHMD devices considered in this example use electric machines connected to the moving masses, with the moving masses connected to the wind turbine structure via a rack-and-pinion mechanical transmission as positioning actuators. Furthermore, they include a ratio multiplier N in the transmission between the electric machine and the respective moving mass. gea ri, N gea r2 of the continuously variable type (CVT, from the English “Continuosly Variable Transmission"), so that the adjustment of the ratios N gear i, N gear2 allows, optionally, the undamped natural frequency of AHMD devices to be adjusted according to the operating circumstances, seeking to reduce the control action required for positioning the moving masses when the AHMD device operates in active mode, or to improve its performance when it operates in passive mode. The moving masses move on mechanical guides (rails, for example) that can be in the form of an inverted circular arc, with a certain curvature, which enables their use in passive mode, without the need for springs, etc.

[0083] Several sensors are connected to the wind turbine 20 to measure, for example: a parameter indicative of the generator rotation speed w gen (from which the angular velocity of the wind turbine rotor w can also be easily obtained) ro t) a parameter indicative of generator torque T gen, a parameter indicative of acceleration of an upper portion of the tower 0 t , a parameter indicative of the pitch acceleration of the floating structure 0 P , a parameter indicative of the angles of attack of the blades and parameters indicative of the position of the moving masses xi and X2.

[0084] In some embodiments, the parameter indicative of the generator rotation speed w gen is a parameter indicative of the generator's instantaneous rotational speed. For example, the generator rotational speed indicative parameter Wgen is a measure of the generator's instantaneous rotational speed.

[0085] In some embodiments, the generator torque indicative parameter T gen is a parameter indicative of the generator's instantaneous torque. For example, the generator torque indicative parameter T gen It is a measure of instantaneous torque of the generator.

[0086] In some embodiments, the parameter indicative of acceleration of an upper portion of the tower 0 t It is a parameter indicative of instantaneous acceleration of an upper portion of the tower 0 t . For example, the parameter indicating the acceleration of an upper portion of tower 0 t It is a measure of instantaneous acceleration of an upper portion of the tower.

[0087] In some embodiments, the parameter indicative of pitch acceleration of the floating structure 0 P It is a parameter indicative of the instantaneous angular acceleration of the pitching of the floating structure 0 P For example, the parameter indicating the pitch acceleration of the floating structure 0 P It is a measure of the instantaneous angular acceleration of the floating structure's pitch.

[0088] In some embodiments, the parameter indicative of blade angles of attack P is a measure of blade angles of attack.

[0089] In some embodiments, the parameters indicative of the positions of the moving masses xi and X2 are position measurements of the moving masses.

[0090] Figure 2 schematically shows a generator controller 23, a blade angle of attack controller 24, a controller configured to perform single linear model predictive control (LlIMPC) 25, and a state estimator 26. The state estimator is, for example, a Kalman filter. In some embodiments, such as this one, the state estimator is integrated with the controller 25.

[0091] To implement control, the turbine comprises processing means and, preferably, a memory. The processing means may be, for example, a CPU.

[0092] Generator controller 23 is configured to perform generator torque control. The generator torque parameter T gen , the parameter indicating the generator rotation speed w gen and a parameter indicating generator electrical power P* are inputs to the generator torque control. An output of the generator torque control is a generator torque control signal T* ge n. The generator torque control determines the generator torque control signal T* ge n based on the indicative parameter of the electric generator torque T gen , to the parameter indicating the generator rotation speed w gen already an indicative parameter of the generator's electrical power P*. An actuator receives the torque control signal from the generator T* ge ny adjusts the generator torque based on the generator torque control signal T* gen received.

[0093] The blade angle of attack controller 24 is configured to perform blade angle of attack control. The generator rotation speed parameter w gen is an input to the blade angle of attack control. An output of the blade angle of attack control is a first blade angle of attack control signal p*. The blade angle of attack control determines the first blade angle of attack control signal p* based on the generator rotational speed parameter w gen .

[0094] Generator torque control and blade angle of attack control can be controls already known in the state of the art. These controls typically aim to maximize the electrical power generated by the generator and, simultaneously with this maximization, set a rotation speed for the wind generator of, at most, the nominal rotation speed described for the wind turbine.

[0095] The linear single model predictive control (LUMPC), executed by the controller 25, has as inputs, for example, the parameter indicating the acceleration of an upper portion of the tower 0 t , the indicative parameter of pitch acceleration of the floating structure 6 P , the indicative position parameters of the moving masses xi and X2 and the adjustable ratios N ge an, N gear2 used by the AHMDs at any given time. The outputs of the linear single model predictive control (LUMPC) are: an incremental aerodynamic thrust control signal from the wind turbine (more specifically, an incremental aerodynamic thrust control signal from blades AF*-r) and the movable mass position control signals T*MI and T*M2. Actuators receive the movable mass position control signals and adjust the motion of the movable masses based on the received movable mass position control signals T*MI and T*M2. For example, the actuators can adjust the forces applied to movable masses FAI and FA2, as we will see later.Model-based predictive control introduces the control inputs and the estimation of the disturbances acting on the system (variables) into the linear model that describes the degrees of freedom of interest of the wind installation (invariant) to predict its future behavior and thus be able to optimize the available control signals in the sense of, among other possible objectives, minimizing the structural load.

[0096] In this way, measurements of the tower's acceleration and the floating platform's pitch are used to detect alterations in the aerodynamic thrust generated by the wind on the rotor and in the momentum created by the wave force on the platform's hull. Based on the dynamic model of the structure and its actuators, the behavior can then be predicted over a given time horizon, and the optimal way to manipulate the aforementioned control variables can be calculated to reduce structural vibrations. The model-based predictive control (MPC) methodology is used, which provides high performance and robustness, but uses a single internal linear model across the entire range of use (LlIMPC).

[0097] In some embodiments, the dynamics of the incremental blade thrust control signal AF*y are faster than the generator rotational speed dynamics. This difference in the frequency ranges of both physical variables allows the incremental blade thrust control signal AF*T to be transformed into an incremental blade angle of attack control signal Ap, which can be used as an additive term to ap* and corrective to the aerodynamic thrust to which the turbine rotor is subjected. Therefore, this control signal allows for some control of the aerodynamic thrust to minimize unwanted turbine movements such as, for example, fore-aft oscillations of the tower and / or pitching of the floating structure.

[0098] The state estimator 26 is configured to estimate an aerodynamic thrust of the turbine (preferably, an aerodynamic thrust on the rotor F ) based on the parameter indicative of acceleration of an upper portion of the tower 0 t , the indicative parameter of pitch acceleration of the floating structure 6 P , the indicative position parameters of the moving masses xi and X2 and the adjustable ratios N gea ri, N gea r2 that the AHMDs use at any given time.

[0099] As described in the previous section, in order to maintain the internal model used by the LLIMPC controller linear and invariant for the entire operating range of the turbine, some processing blocks are used which, external to the controller, allow the use of some well-known and very important relationships in this application, but of a non-linear nature. A first external processing block 27 is configured to estimate a relative wind speed with respect to the rotor Vp, t (i.e., the subtraction of the wind speed minus the speed of the top of the tower) based on the estimate of aerodynamic thrust on the rotor F , to the parameter indicative of the generator rotation speed w gen and the parameter indicating the angle of attack of the blades p. To do this, block 27 can estimate the relative wind speed with respect to the rotor Vp, t using a non-linear relationship between Vp, tand the variables F , w gen and P, which is well known, for example: where p is the air density, R is the rotor radius, 2 is the blade tip speed, ? is the angle of attack of the blades, ro t is the angular velocity of the rotor, obtained directly from w gen , and Or is the effective aerodynamic thrust coefficient, which depends, in a highly non-linear manner, on 2 and p. This coefficient C is represented as a surface with the axes of its base formed by the values ​​of 2 and p, and is usually computed as a look-up table created with data extracted from experimental measurements carried out on the wind turbine itself during its operation.

[0100] This expression (1.1), highly non-linear due to the coefficient C , allows us, through numerical procedures implemented by block 27 of Figure 2, to obtain an estimate of the relative wind speed at the rotor, from the estimate of the aerodynamic thrust made at each moment by block 26, as well as the speed of the generator / rotor and the angle of attack measured at that instant. For this calculation, one possibility includes numerically generating possible values ​​-candidates- of Vrei within its range and taking as an estimate, manipulating (1.1), the following:

[0101] Where, as we see, the cogen is used (or ro t) and ?which are measured at that instant, as well as F the estimate of the aerodynamic thrust T which leaves block 26 at that time.

[0102] A second external processing block 28 is configured to determine a second AP*T blade angle of attack control signal based on the estimate of relative wind speed with respect to the rotor. to the parameter indicating the generator rotation speed w gen and the incremental aerodynamic thrust control signal of the blades AF*T. To do this, the second external block 28 can determine the second control signal of the angles of attack of the blades Ap using the non-linear relationship (1.1) between AP*T and the variables V re w genand AF*T. Once the relative wind speed has been estimated by block 27, we can convert the increase in thrust, calculated by the LUMPC (AFT*), into the increase in the angle of attack A* that will effectively cause said change in the thrust suffered by the rotor. To do so, by means of block 28 of Figure 2, one possibility involves obtaining from expression (1.1) the increase in the effective aerodynamic thrust coefficient caused by AFT* for the relative wind speed at that instant:

[0103] Then, the intersection of the surface of the effective aerodynamic thrust coefficient Crf , ?) with a plane passing through the current value of the tip speed act, obtained from the cogen measurement ( or roi) and the estimate of the relative wind speed at the rotor at that instant. This gives a function of the coefficient C? which is now only a function of the angle of attack p. From this function and the CT calculated in (1.3), the increment A*, corresponding to the value of AFT*, calculated by the LUMPC at that instant, can be easily obtained.

[0104] Thus, as illustrated in Figure 2, the blade aerodynamic thrust control signal AF*T is converted, in a processing block 28, external to the LUMPC model-based predictive control (i.e., in a block that is not part of the linear model used by the model-based predictive control), into an additive blade angle of attack correction term that is added to the first blade angle of attack control signal p*, as can be seen in Figure 2.

[0105] The angle of attack control signal *T is calculated by combining the first blade angle of attack control signal * with the second blade angle of attack control signal AP*T, for example, by summing them. An actuator receives the angle of attack control signal p and adjusts the blade angles of attack based on the received angle of attack control signal P*T.

[0106] The model-based predictive control runs in parallel with the blade angle of attack control of the controller 24 and, preferably, the model-based predictive control does not alter the generator performance relative to the generator rotation speed thanks to the different frequency ranges of both physical variables.

[0107] Model-based predictive control does not control the generator speed. Relinquishing control of the generator speed by the model-based predictive controller—leaving it to the generator controller 23—brings implementation advantages, given that this control loop is subject to very high safety certification requirements, and therefore many turbine manufacturers are reluctant to make major changes to the generator controller 23. The present invention makes it possible to substantially reduce the structural load without compromising the performance of the baseline loop (i.e., the standard control loop) responsible for generator speed control.

[0108] External block 29 that calculates the N ratios gear i, N gear2 for the multipliers of the two AHMDs and which is optional, works differently depending on whether or not the estimates of the external disturbances are available (in this case, the aerodynamic thrust on the rotor and the wave moment on the floating structure). If these estimates are available - case of setting a) in the previous section -, which is usual if the control is active, the block proposes values ​​for N gear i, N gea r2 that adjust the undamped natural frequency of the AHMDs to the apparent frequency of said disturbances. Then, on the one hand, for the adjustment of N gear2 -the ratio of the AHMD2, installed in the nacelle (gondola) of the turbine-, an experimental characterization is needed that relates the relative wind speed and the estimate of the aerodynamic thrust affecting the rotor at any given time with the frequency with which the rotor oscillates collectively in a direction perpendicular to the plane of rotation (flapwise), which supposes the effective excitation of the tower structure. Furthermore, the estimate of the wave moment -in particular its apparent frequency- is used to tune the AHMD1 (adjust its undamped natural frequency to said apparent frequency) of the floating platform to the disturbance due to the waves affecting it at any given time.On the other hand, if the estimation of the disturbances at each moment is not available - case of adjustment b) in the previous section, for example, for the operation of the AHMD in passive mode, without the LlIMPC controller running -, block 29 proposes values ​​for the N ratios. gear i, N gear 2, so that the undamped natural frequencies of said AHMDs are set to values ​​close to (e.g., coincident with) the undamped natural frequencies of the floating structure and the turbine tower, respectively, taking advantage of the fact that the moving masses of the AHMDs are a small fraction of the masses of said structures, to which they are coupled. Finally, block 29 requires the use of the relationship between the ratios N gear i, N gear2, to be imposed on the CVT transmissions of AHMDs, and the undamped natural frequency thus installed in them. This relationship can be deduced from certain construction aspects of the mechanical transmission of AHMDs.

[0109] The undamped natural frequency of oscillation (due to the inverted circular arc profile guide) of the moving mass can be related to the gear ratio of the multiplier as follows: where: m is a total mass value (e.g., in kg) of the moving mass; g is the acceleration of gravity;

[0110] Ngear is the transmission ratio between the rotor of the electric machine and the moving mass;

[0111] R is a radius of the rack-and-pinion transmission gear wheel (pinion).

[0112] R2 is the radius of curvature of the inverted arc of a circle guide; Jgen_Hss is the moment of inertia of the electric machine rotor, calculated relative to the rotation axis where the electric machine is installed (high-speed axis of the CVT transmission);

[0113] T2 is the oscillation period (in seconds) desired for the moving mass (inverse of the undamped natural frequency -in hertz- desired for the AHMD moving mass).

[0114] The undamped natural frequency of oscillation (due to the inverted circular arc profile guide) of the moving mass can also be adjusted by altering the curvature R2 of the guide that couples the mass directly to the wind turbine. For example, the undamped natural frequency of the moving mass can be related to the curvature R2 as follows: where:

[0115] T2 is the natural period in seconds (inverse of the undamped natural frequency in hertz) that is to be imposed on the moving mass; and

[0116] Single linear model predictive control (LlIMPC) uses a single linear model to determine the dynamic evolution of the degrees of freedom of interest in the wind turbine installation in order to reduce the structural load. For example, this linear model may be the linear model 30 shown in Figure 3. The linear model 30 may include the actuator model 31 that adjusts the angle of attack of the blades, the models (32 and 33) of the actuators that adjust the motion of the moving masses, and the model 34 of the assembly comprising the rotor, tower, floating structure, and the two AHMDs installed on the wind turbine. These models are described below.

[0117] The inclusion of the model of the actuator that adjusts the angle of attack of the blades 31 is based on the hypothesis that the dynamics of aerodynamic thrust control are the same as those of the actuator that adjusts the angles of attack of the blades, thereby causing the required changes in aerodynamic thrust. For example, if the actuator that adjusts the angles of attack of the blades is hydraulic, the model of said actuator 31 can generate a second-order response, typical in this type of actuators, given by the expression: where, in this example, values ​​of the damping ratio ¿ = 0.7 and natural frequency WR = 2n rad / s are used, typical of this type of actuators.

[0118] The dynamics and mechanical transmission of the actuators that adjust the movement of the moving masses are taken into account in the actuator models (32 and 33) that adjust the movement of the moving masses. In the case that the turbine incorporates AHMD systems, there is the option of using, as manipulated variables of the LlIMPC control system, alternative or additional to the already mentioned AFT*, the torque setpoints (TMI* and TM2*) of the drives - normally electric machines coupled to the moving masses by means of a mechanical transmission - that move the masses of the aforementioned AHMDs. Obviously, the dynamics introduced by said drives, as well as the corresponding mechanical transmissions, are also taken into account in order to dynamically relate the aforementioned TMI* and TM2* with the forces FAI and FA2, which are used to move the moving masses Mi and M2 of the respective AHMDs, and which appear in model 34, described below.For example, a rack-and-pinion mechanical transmission is used to connect moving masses to the shaft of electrical machines. A mechanical speed multiplier is considered, inserted between the transmission pinion and the electrical machine. The system dynamics are described as follows:

[0119] 1 1

[0120] T Mi = - r,, T Mi + - r,, T Mi (1.5)

[0121] 1 lag 1 lag where a first order response has been considered -see (1.5)-, with a time constant Tiag, for the electromagnetic torque TM¡ that each electric machine applies at each instant (we use a Ti ag ,=0, 1 s., in this example) On the other hand, N gear¡ is the multiplier ratio (which takes continuous values ​​within a range, in the case of CVT type multipliers), RD¡ is the transmission pinion radius, D¡ are the damping coefficients and J¡ the moments of inertia, respectively, of the electric machines, measured from the high-speed shaft.

[0122] Below is a possible description of block 34, which is part of the LlIMPC internal model -see Figure 3- and is dedicated to modeling the assembly comprising the rotor, tower, floating structure and the two AHMDs installed on the wind turbine. In this case, in order to predict the dynamics that take place in the wind turbine structure due to wind thrust and pitching due to waves (in the case of FOWT floating turbines), the degrees of freedom (DOFs) of interest are, preferably -see Figure 4-: the rotor (specifically, the collective bending 40 of the blades with respect to the tower in the flapwise direction -perpendicular to the plane of rotation-), the tower (bending with respect to the floating platform in the fore-aft direction -fore-aft-) and, where appropriate,the floating structure (pitch around an imaginary axis perpendicular to the X and Z axes in the direction of the waves) and the location (Xi and X2) of the two mobile masses MAHMDI and MAHMD2 (also called Mi and M2 respectively), corresponding to the two stabilization elements AHMD (in the fore-aft direction of the tower -nacelle- and the pitch of the floating structure -platform-, respectively, that is, in the direction of the X axis of figure 4) incorporated. The movement of the structure in said DOFs are those that, normally, appear coupled in practice and affect to a greater extent the load that it suffers and that are intended to be reduced to the minimum possible.,

[0123] Therefore, for block 34, a physical model with five degrees of freedom (5 DOFs) is proposed, consisting of five masses, joined by flexible elements and dampers -see Figure 4-. The masses represent, respectively, the flexible portions of the collective rotor blades, the flexible part of the tower, the floating structure and, finally, the mobile masses of the AHMDs, placed in the nacelle and the floating platform, respectively. Specifically, the flexible part of the collective rotor is modeled as a first translational oscillator of mass ms (also called MB) connected to a first rotational oscillator 41 , of mass m? (also called MT) and moment of inertia IT, which represents the tower and which, in turn, rests -by means of a hinge 42- on a second rotational oscillator 43. This second rotational oscillator 43, of mass mp (also called MP) and moment of inertia l P, is dedicated to modeling the pitching motion of the floating platform and is supported by kinematic binding conditions, imposed by the state of the water in which the floating structure floats.The AHMD devices are modeled as a second and a third translational oscillator, of masses Mi and M2, coupled -inside, in fact- to the nacelle (rotational oscillator 41) and to the floating structure (rotational oscillator 43), respectively, by means of springs (in this example, as we have already seen, the springs are replaced by the use of guides in the shape of an inverted arc of a circle with a certain curvature), dampers (in this example, replaced by the damping introduced by the coupled electrical machines) and actuators (in this example, constituted from the electromagnetic torques applied by the stator of the coupled electrical machines and which are converted into force on the masses thanks to the continuously variable ratio CVT multiplier and the rack-and-pinion mechanical transmission).The mentioned actuators can apply forces FAI and FA2 on the moving masses Mi and M2, respectively, to allow active control of their positions on the X axis.

[0124] As illustrated in Figure 4, the flexible elements that connect the five masses introduce the stiffness (k) and damping (d) of, respectively, the blades (kb, db), the tower (k t , d t ), the floating platform (k p , d p) -in this case, due to the righting arm and hydrodynamic damping of the hull, as well as the behaviour of the mooring lines, the floating platform - and the attachment systems of the moving masses of the AHMDs (ki , k2, di , d2) (also called respectively kAHMDi , kAHMD2, dAHMDi , dAHMD2) -in this example, to be obtained from the curvature of the guides, the CVT torque transmission ratio and the construction characteristics of the electrical machines and the rack-and-pinion mechanical transmission

[0125] Regarding external disturbances, the rotor is subjected to the aerodynamic thrust of the wind FT, and the floating platform, in turn, is subjected to the hydrodynamic moment Mw. The aerodynamic thrust of the wind can be estimated based on the relative wind speed at the rotor, the angle of attack of the blades, and the rotor rotational speed. Specifically, the aerodynamic thrust depends nonlinearly on the relative wind speed, the angle of attack of the blades, and the rotor rotational speed. The hydrodynamic moment can be estimated based on the wave force of the water in which the floating structure floats and an acceleration of the turbine, which is represented as a mass added to the corresponding moment of inertia. Specifically, the hydrodynamic moment depends nonlinearly on the wave force of the water in which the floating structure floats and on the acceleration of the turbine.

[0126] Therefore, as can be seen in Figures 3 and 4, the inputs to block 34 of the internal model, dedicated to modelling the assembly comprising the rotor, tower, floating structure and the two AHMDs installed on the wind turbine, are, on the one hand, the external disturbances (the aerodynamic thrust FT applied at one end and the hydrodynamic moment Mw applied at the other) and, on the other hand, the signals used for the control action (the incremental aerodynamic thrust of the blades AFT and the forces FAI and FA2, used for the control of the AHMD devices) calculated by the LlIMPC controller at each instant in order to optimise, in some sense to be chosen by the designer, the dynamic behaviour of the complete structure.

[0127] The approach, outlined in Figure 4, of this representation of the mechanical assembly comprising the rotor, tower, floating structure and the two AHMDs installed on the wind turbine, can be used flexibly to cover various types of wind installations. For example, if the DOFs (degrees of freedom) due to the floating platform and the AHMD installed on it are eliminated, it would be the case of an offshore turbine anchored to the seabed (for example, monopile type wind turbines) and subject to disturbance due to sea force. Likewise, if this disturbance, due to wave momentum, is eliminated, the appropriate model for an onshore turbine would be obtained. Thus, in the case illustrated in Figure 5, the degrees of freedom are reduced to three, since there is no floating structure, nor any movable mass with adjustable position installed inside.This model can be used for land-based wind turbines (making Mw=0) or those anchored to the seabed (My 0).

[0128] On the other hand, taking advantage of the five DOFs, the case of FOWT turbines that use floating structures would be covered, both barge type (i.e., "barge" type as it is known in English) -see figure 6- which comprises the floating structure (i.e., the barge) and a mobile mass arranged on the floating structure; and buoy type (i.e., "spar-buoy" type as it is known in English) -see figure 7- which comprises the floating structure (i.e., the cylindrical buoy) and a mobile mass arranged on the floating structure. As shown in figures 6 and 7, the partial theoretical model of a barge-type floating wind turbine is identical to the partial theoretical model of a buoy-type floating turbine. It is only necessary to change the numerical value of the model parameters.These models differ primarily in the characteristics (e.g., dimensions and mass) of the second rotating oscillator (i.e., the floating structure) and in the stiffness and damping of the connection between the floating structure and the tower. In the buoy type, the tower and buoy are part of a single structure, and thus, the stiffness and damping of the connection between the two rotating oscillators actually describe the flexibility characteristics of that structure.

[0129] From the physical representation that includes the five DOFs, we can obtain the simplified mathematical model of the assembly comprising the rotor, tower, floating structure and the two AHMDs installed in the wind turbine -block 34 in Figure 3-, considering the measurements from an absolute reference system and assuming that the inclination angles of both rotational oscillators 6 and OP, measured with respect to the vertical, are small (sin 0¡=0, cos 0i=1), we have: where expression (1.7) describes the motion of the AHMD installed on the nacelle, expression (1.8) refers to the flapwise oscillation of the collective rotor, (1.9) to the fore-aft motion of the tower, (1.10) to the pitching of the floating platform and (1.11) to the dynamics of the AHMD installed on said platform.In addition to the parameters already described or introduced in Figures 4, 5, 6 and 7, Ri is the height at which the AHMD1 device is installed above the hinge of the rotational oscillator of the floating platform, R2 is the height at which the AHMD2 device is installed above the hinge 42 of the rotational oscillator of the tower, RB is the height of the mass MB above the hinge 42 of the rotational oscillator of the tower, RT is the height of the center of gravity of the tower above the hinge 42 of the rotational oscillator of the tower, RP is the height of the center of gravity of the floating platform with respect to the hinge of the rotational oscillator that describes the movement of the platform and g is the gravity.The parameters involved in the model in block 34 (M¡, l¡ R¡, k¡, di) are numerical values ​​that define the dynamics of interest of the assembly comprising the rotor, tower, floating structure and the two AHMDs installed on the wind turbine and, in this example, are considered known - they are extracted from the knowledge we have on the observed behavior - and constant for any operating point in which the installation is located. In reality, some of these parameters, such as kb and db in expression (1.8) (related to the natural frequency and the damping of the first flapwise vibration mode of the rotor), vary smoothly with the incident wind speed - see, for example, “Aeroelastic Instabilities of Large Offshore and Onshore Wind Turbines”, Gunjit Bir and Jason Jonkman, Journal of Physics: Conference Series, Vol. 75, n. 1 , pp.12069, 2007-, However, as we will see below, the robustness of the LlIMPC controller to the modeling errors caused by holding these parameters constant means that the performance achieved is hardly affected by this. However, it is advisable to take into account the estimate of the incident wind when adjusting the undamped natural frequency of the AHMD2 device (via N. gea r2) installed in the gondola -see figure 2 and commentary regarding the internal operation of block 29 of said figure-.

[0130] Thus, once these dynamics, included in block 34, have been incorporated into what was described above regarding blocks 31, 32 and 33 of figure 3, the internal model of the LLIMPC for calculating the prediction during the chosen time horizon is completely defined -see figure 3-, using the linear expressions (1,4)-(1.11).

[0131] As illustrated in Figure 3 and can be seen from the linear expressions (1 ,4)-(1 .11), the inputs to the linear internal model 30 are the perturbations (the aerodynamic wind thrust FT and the hydrodynamic moment M w ) suffered by the wind turbine in its operation and the control signals generated by the LLIMPC model-based predictive control, that is, AF*T, T*MI and T*M2. The outputs of the linear model 30 in Figure 3 that are declared as measured variables are, in this example: the acceleration of the upper portion of the tower, the positions of the moving masses of the AHMD devices installed in the nacelle and floating platform and the angular acceleration of the pitch of the floating structure.

[0132] As can be seen from the above, the internal model 30, depicted in Figure 3, used by the LUMPC controller in this application example, is indeed linear and invariant across the entire operating range of the wind turbine. That is, the linear internal model does not change with changes to any one or more of the turbine operating conditions: generator rotational speed, generator generated power, generator torque, wind speed, wind direction, wave conditions (e.g., wave direction and / or wave power), tower forward and aft acceleration, floating structure pitch acceleration, and blade angles of attack.

[0133] In some embodiments, devices may be used that allow previews of the input disturbances F to be obtained. T and M wFor example, a LIDAR can be used to measure the incident wind at a certain distance from the rotor and calculate the future aerodynamic thrust of the turbine F T For example, a buoy (or wave radar) can be used with one or more sensors to preview the hydrodynamic moment M w . In some embodiments, it can be predicted, at each instant and based on estimates of the hydrodynamic moment M wcaused by the incident waves, made up to that moment by the state observer, the future disturbance -preview- due to the incident waves that the floating platform will suffer during the prediction horizon used by the proposed controller. In some embodiments, said prediction can be made following any of the methods described in: F. Fusco and J.V. Ringwood, "Short-Term Wave Forecasting for Real-Time Control of Wave Energy Converters," in IEEE Transactions on Sustainable Energy, vol. 1, no. 2, pp. 99-106, July 2010. In these cases, the predicted preview of the effect of the wave disturbance can be used to assist in the control of the inclination of the structure in the face of future disturbances, for example, by allowing compensation for the delays presented by the mass movement actuators.

[0134] This description presents a physical model to facilitate understanding of the method. However, the linear model can be obtained in different ways, giving rise to various models. Some of these models are defined based on parameters that are usually related to the physical description of the installation (distances, moments of inertia, masses, etc.). If the model used is parametric, these parameters can be obtained from the internal model in various ways: from the construction description, through identification algorithms based on the use of experimental data measured on the turbine itself, etc.When using a parametric model, it is highly recommended to adjust its parameters using information obtained directly from the turbine in question ("customized" information) and to verify, exhaustively and also using experimental data obtained from that turbine, that the model predicts the behavior of the installation in question with sufficient fidelity.

[0135] In some embodiments (not illustrated), instead of using theoretical physical models, non-parametric models are used. Non-parametric models can be obtained by an appropriate choice of the number of degrees of freedom of the model and from experimental data sufficiently descriptive of the dynamic behavior to be modeled. Therefore, the differentiating element of this method is not, strictly speaking, the type of modeling, but the use of a single linear model for the entire operating range. However, a physical model (i.e., a model based on the physical description) is, in general, more advisable since it can facilitate the adjustment of the predictive control based on a single linear model LlIMPC, especially with regard to choosing and giving weights to the control objectives and defining the operating constraints to which the system is subject.Regarding the configuration of the model-based predictive controller, appropriate values ​​are chosen for the control horizon and the prediction horizon, so as to balance the action of “feedback” and “feedforward” (i.e., pre-feedback), keeping the necessary computational load as low as possible, while covering the temporal evolution of the main degrees of freedom taken into account. This choice takes into account the transient duration of the main dynamics of interest in the turbine and its actuators, as well as the degree of reliability of the linear model and the quality of the measurements obtained with the sensors.

[0136] Model-based predictive control is also adjusted, particularly with regard to the aggressiveness of the control action and the speed of the state estimator. This adjustment should be carried out with an effort to ensure that the estimation is as rapid as possible, within stability, and that the control is sufficiently aggressive, without causing excessive activity for the actuators (of the blade angle of attack and the moving mass), nor significantly impairing, and preferably aiding, the performance of the loops that control the generator speed in parallel.

[0137] To test the implementation, a study is conducted on the computational load associated with the resulting algorithm, considering the possible limitation in the number of iterations allowed for optimization. The need to impose restrictions on certain system parameters and the impact this may have on the practical implementation is also analyzed. Clearly, the implementation platform must be able to comfortably handle the calculated load, and at that point, the implementation of the present invention (for example, as an anti-vibration system) can be considered an update of the software related to the control of the angle of attack of the blades and the moving masses of the installed AHMD devices.

[0138] Logically, in those embodiments in which the wind turbine does not comprise any floating structure, the parameter indicating the pitch acceleration of the floating structure 6 is dispensed with. PSome examples of these systems are land-based wind turbines and wind turbines with towers attached directly to the ground and submerged in water.

[0139] Logically, in those embodiments in which the wind turbine does not comprise any of the movable masses with adjustable position to adjust the weight distribution of the wind turbine, the parameter indicative of the position of the absent movable mass is dispensed with, as well as its position control signal and the corresponding movement adjustment actuator. In the present disclosure, the controllers and external blocks (for example, block 27, block 28, block 29 and the block that combines the first blade angle of attack control signal p* with the second blade angle of attack control signal AP*T that have already been described and appear in Figure 2), which make up the control system of the wind turbine, may use the same processing means (for example, a single microprocessor) or use different processing means (for example, different controllers may use different microprocessors).

[0140] In this disclosure, the term “instantaneous” when used in conjunction with the term “Z” (such as “instantaneous Z”) refers to “Z” at a specific instant in time. For example, the term “instantaneous speed” refers to a speed at a specific instant in time, and the term “instantaneous torque” refers to torque at a specific instant in time.

[0141] Next, to conclude the description of a possible preferred embodiment, a possible application case is presented, including specific details to facilitate a detailed understanding of the method. First, each step of the method is described for the broadest application case (a floating wind turbine with two AHMDs installed). Now, the application to a simple case, a land-based wind turbine without AHMDs installed, is explicitly considered. The intention in choosing the land-based case is to illustrate the operation of the invention. It is believed that by eliminating several degrees of freedom, the operation of the control action is more clearly associated with the load reduction achieved. In particular, the method is applied to a benchmark turbine for developments proposed by the US National Renewable Energy Laboratory (NREL).This onshore turbine has a nominal power of 5 MW and has become a well-known and widely used case study, both in academia and industry.

[0142] For the selection of the internal prediction model, since this is an onshore wind turbine without coupled AHMD devices, only the degree of freedom corresponding to the rotational oscillator that describes the fore-aft motion of the tower is considered. Therefore, only the control signal based on the incremental aerodynamic thrust AF*T is used, which is obtained by adding an additive signal A* to the angle of attack of the rotor blades, a simplified internal model referred to a single degree of freedom and considering only the external disturbance due to the aerodynamic thrust on the rotor.

[0143] In the specific case of the land turbine that we are presenting as an application example, we only need the two external blocks 27 and 28 of Figure 2, which we have implemented programmatically in the manner described above.

[0144] In the specific case of the onshore turbine in the example, the physical data from the NREL 5 MW model are used to obtain the internal model parameters - see, for example, “Definition of a 5-MW reference wind turbine for offshore system development,” J. Jonkman, S. Butterfield, W. Musial, and G. Scott, NREL, Tech. Rep. TP-500-38060, 2009. -, namely, the tower mass (347,462 kg), the nacelle mass (240,000 kg), the hub mass (56,780 kg), the blade mass (3*17,740 kg), the tower structural damping ratio (0.08), the undamped natural frequency of the tower’s first fore-aft pitch mode (-=0.321 Hz), and the tower height ( / 7r=87.6 m). Thus, taking into account that, according to the study “Windkraftanlagen: Grundlagen, Entwurf, Planung und Betrieb" by R. Gasch, J. Twele, and P. Bade, Wiesbaden, Germany: Teubner, 2005, only a quarter of the tower's mass contributes to the oscillatory movement, we obtain: kT = 1,364.10 10 d T = l,0819- 10 9

[0145] 7 r = 3.3524- 10 9

[0146] On the other hand, the damping ratio ¿ = 0.7 and the natural frequency op = 2n rad / s are used as parameters corresponding to the hydraulic actuator of the angle of attack of the blades and the dynamic model (1.4).

[0147] In the specific application example, taking into account the dynamics of the NREL 5 MW wind turbine tower (-=0.321 Hz), it is estimated that one second is sufficient as a prediction horizon. Furthermore, since no preview is used and in order to lighten the computational load, a control horizon of 0.25 seconds is set. Considering that the sampling frequency of the control system is the usual one in these facilities (80 Hz), prediction and control horizons of 80 and 20 samples, respectively, are obtained.

[0148] In the specific case of the onshore turbine in the example, to test the potential capabilities of the method, a relatively strong emphasis is placed on achieving the control objective, which, in this case, consists of nullifying the tower's fore-aft velocity. However, to limit the activity of the hydraulic pitch actuators, a significant penalty has also been imposed on the angle of attack derivative.

[0149] In the specific case of the onshore turbine in the example, to test the system of the invention, the installation is subjected to two operational load situations. First, to study the extreme loads, a wind gust according to the IEC standard in V is applied to the wind turbine. out=25 m / s -see results in Figures 8a, 9a, 10a, 11a, 12a, 13a, 14a, 15a-, Then, to analyze the performance of the proposed system under fatigue loads -see results in Figures 8b, 9b, 10b, 11 b, 12b, 13b, 14b, 15b-, a turbulent wind field of type A according to IEC-61400-1 of average speed 25 m / s and 15 % turbulence, generated with Turbsim, is applied to the wind turbine -see, for example, “TurbSim user's guide,” B.J. Jonkman. NREL, Tech. Rep. TP-500-46198, 2009.-.

[0150] The tower tip velocity - Figures 8a and 8b -, the tower tip deflection - Figures 9a and 9b -, the tower fore-aft bending moment - Figures 10a and 10b -, the aerodynamic thrust and its on-line estimation - Figures 11a and 11 b -, the blade angle of attack - Figures 12a and 12b -, the time derivative of the angle of attack - Figures 13a and 13b -, the generator speed - Figures 14a and 14b - and the power generated by the installation - Figures 15a and 15b - are extracted by means of numerical simulations.

[0151] First, we examine the extent to which the control system is successful in reducing the structural load on the tower. Next, we analyze the potential impact of the invention's operation on the installation's performance. Finally, we draw conclusions from the above.

[0152] A. Performance of the invention:

[0153] It is observed in Figures 8a and 8b that the main control objective (eliminating the tower oscillation speed) is largely met, both under extreme load and under fatigue load and, therefore, it can be said that the LlIMPC control system works adequately.

[0154] The tower's bending fluctuations (Figures 9a and 9b) are also substantially reduced, although, as expected, they reflect the low-frequency variations in turbulent winds and the aerodynamic thrust they cause. Figures 10a and 10b show the reductions, of around 80%, caused by the tower's fore-aft bending moment MbT, which accumulates the effects of bending and its derivative with respect to time. This moment is usually used to quantify the structural load on the wind turbine tower at each instant of operation:

[0155] Figures 11a and 11b show that the controller's estimate of aerodynamic thrust accurately follows the estimated physical variable with little delay. This largely explains the success of the control action and confirms the accuracy of the internal model used.

[0156] B. Impact of the control action: The angle of attack -pitch- of the blades can be seen in Figures 12a and 12b. The effect of the additive component for the pitch setpoint generated by the control action of the invention can be seen in Figure 12a, where it advances the reaction of the control system to the arrival of the disturbance in the aerodynamic thrust due to the gust of wind incident on the rotor, thus largely preventing the tower from bending. Its effect can also be seen in Figure 12b, where it introduces higher frequency components from t=50 seconds, the instant at which the proposed LlIMPC control system is activated. The expected counterpart is greater activity and, therefore, fatigue in the hydraulic actuators of the angle of attack of the blades.However, we see that, in the case of gusts (extreme load), this does not occur (see Figure 13a). Thanks to the aforementioned anticipation of the control action, both the pitch angle and its derivative reach lower numerical values ​​than when the invention is inactive. On the other hand, when dealing with turbulent wind, a notable increase in the angle of attack derivative with respect to time is observed. In this case, although the pitch derivative never reaches -see Figure 13b- the limits usually imposed on hydraulic actuators used in this type of installation (typical Max pitch rate = 8 rad / s). The possibility of limiting the aggressiveness of the control action should be considered, or incorporating an additional device into the design (for example, by installing an AHMD in the nacelle) that allows reducing the activity of the pitch actuators.On the other hand, Figures 14a and 14b show that the operation of the invention does not impair, in fact, it even improves, the low-frequency performance of the baseline control system (the control system that pre-existed the application of this invention) for the speed of the cogen generator. However, for the turbulent wind case, new higher-frequency components are introduced that reflect the additive control action we used to reduce the load on the tower. These new components have no practical impact other than their transfer to the mechanical power (c£>). ge n*Tgen) generated, as can be seen in Figure 15b. These components are filtered in the DC bus that connects the generator power electronics with the power electronics of the grid access module, thus being eliminated from the generated electrical power.

[0157] The proposed controller has been implemented in C code and executed in a real-time kernel on a computing platform equipped with a 4-core Intel Core i5-2400 processor (3.10 GHz), with characteristics very similar to the devices found in current mid-range control platforms. In no case have more than 10 iterations been necessary to solve the quadratic LLIMPC optimization problem (as a safety measure, iterations have been limited to 20, maximum). In the worst case, for 10 iterations, the total execution time has been 7.25.10'. 5 seconds, which is very far from the sampling period used (1,25.10 -2seconds). In this way, it is not even necessary to include the computational delay in the internal model and the applicability of the proposed method to any current control platform is guaranteed in the recommended format, namely as an update to the wind turbine control software.

[0158] From the above, it can be deduced, firstly, that high potential performance (under optimal circumstances) can be expected from the proposed invention, achieving reductions of more than 80% in the structural load suffered by the installation, both under extreme loads (gusts) and fatigue (turbulent wind). On the other hand, it can also be deduced that the main operational problem of the proposed system is that it requires greater activity of the blade angle of attack actuators under fatigue loading conditions. In particular, in the case of land-based wind turbines, such activity is limited (see Figures 12a, 12b and 13a, 13b) and can be considered acceptable.However, in installations with higher vibration levels (floating wind turbines), it is highly advisable to install additional dampers - similar to the AHMDs considered in the invention - if they are not already present, in order to reduce or eliminate the use of the aforementioned angle of attack actuators with the aim of reducing the structural load.

[0159] In view of this description and figures, the person skilled in the art will understand that the invention has been described according to some preferred embodiments thereof, but that multiple variations can be introduced into said preferred embodiments, without departing from the object of the invention as it has been claimed.

[0160] In this text, the term "comprises" and its derivatives (such as "understanding," etc.) should not be understood in an exclusive sense. That is, these terms should not be interpreted as excluding the possibility that what is described and defined may include more elements, stages, etc.

Claims

CLAIMS 1. A method for controlling angles of attack of blades of a wind turbine, the method comprising: providing a linear model of the wind turbine; the linear model being invariant over the entire operating range of the wind turbine; and applying model-based predictive control using the linear model to determine, with processing means and based on a parameter indicative of acceleration of a tower of the wind turbine, an angle of attack control signal for the blades of the wind turbine; the parameter indicative of tower acceleration being an input to the model-based predictive control.

2. A method according to claim 1, the method comprising providing a processing block of a non-linear model, the non-linear model relating a turbine aerodynamic thrust to a blade angle of attack adjustment; and wherein the step of applying a model-based predictive control using the linear model to determine, with processing means and based on a parameter indicative of acceleration of a wind turbine tower, a wind turbine blade angle of attack control signal comprises: applying the model-based predictive control using the linear model to determine, with processing means and based on the parameter indicative of acceleration of the wind turbine tower, a signal indicative of wind turbine aerodynamic thrust control; the signal indicative of wind turbine aerodynamic thrust control being an output of the model-based predictive control;and executing, with processing means, the processing block using the aerodynamic thrust control signal, determining as a result of the processing a second blade angle of attack control signal of the wind turbine; and wherein the blade angle of attack control signal is the second blade angle of attack control signal or the method comprises determining, with processing means, the blade angle of attack control signal based on the second blade angle of attack control signal.

3. Method according to any one of the preceding claims, wherein the turbine wind turbine is mechanically coupled to a moving mass to adjust the spatial weight distribution of the turbine.

4. A method according to any one of the preceding claims, wherein a parameter indicative of the position of the movable mass is an input to the model-based predictive control; and the blade angle of attack control signal is determined based on the parameter indicative of the position of the movable mass.

5. Method according to claim 3 or 4, the method comprising applying the model-based predictive control using the linear model to determine, with processing means and based on the parameter indicative of the position of the moving mass, a motion control signal of the moving mass.

6. The method of claim 5, wherein the motion control signal of the moving mass is an output of the model-based predictive control.

7. Method according to any one of the preceding claims, wherein an objective of the model-based predictive control is to minimize a structural load of the tower.

8. A method according to any one of the preceding claims, the turbine being a floating wind turbine, the floating wind turbine comprising a floating structure, and the floating structure supporting the wind turbine tower; wherein an input to the model-based predictive control is a parameter indicative of acceleration of the floating structure; and wherein the blade angle of attack control signal is determined based on the parameter indicative of acceleration of the floating structure.

9. The method of claim 8, wherein an objective of the model-based predictive control is to minimize a structural load of the tower and a structural load of the floating structure.

10. Method according to any one of the preceding claims, wherein an objective of the model-based predictive control is to minimize oscillations around an imaginary axis perpendicular to a rotation axis of the blades and to a height direction of the tower.

11. Method according to any one of the preceding claims, wherein the angle of attack control signal is determined based on an output of the model-based predictive control and an output of a blade angle of attack control different from the model-based predictive control.

12. A method according to claim 2 and claim 11, wherein the blade angle of attack control signal is determined based on: the output of the blade angle of attack control other than the model-based predictive control, and the second blade angle of attack control signal.

13. Processing means configured to execute model-based predictive control using a linear model to determine, based on a parameter indicative of the acceleration of a wind turbine tower, a control signal for angles of attack of the wind turbine blades; the parameter indicative of tower acceleration being an input to the model-based predictive control; the linear model being invariant throughout the operating range of the wind turbine.

14. A method for controlling a position of a movable mass coupled to a wind turbine, the mass being used to adjust a spatial weight distribution of the turbine, the method comprising: providing a linear model of the wind turbine; the linear model being invariant over the entire operating range of the wind turbine; and applying model-based predictive control using the linear model to determine, with processing means and based on a parameter indicative of acceleration of a tower of the wind turbine and a parameter indicative of the position of the movable mass, a motion control signal for the movable mass; the parameter indicative of acceleration of the tower being an input of the model-based predictive control; and the parameter indicative of the position of the movable mass being an input of the model-based predictive control.

15. The method of claim 14, wherein the turbine is a floating offshore wind turbine, the turbine comprising a floating structure, and the floating structure supporting the turbine tower; wherein an input of the model-based predictive control is a parameter indicative of acceleration of the floating structure; and wherein determining the control signal of the movement of the turbine tower is a parameter indicative of acceleration of the floating structure. mobile mass is carried out based on the parameter indicating the acceleration of the floating structure.

16. A method according to claim 14 or 15, wherein the movable mass is coupled to the turbine by a guide allowing movement of the movable mass relative to the turbine; the movable mass being coupled to an electrical machine by a mechanical transmission such that adjusting the electromagnetic torque of the electrical machine allows adjusting the movement of the movable mass along the guide; the motion control signal comprising an electromagnetic torque adjustment signal of the electrical machine.

17. A method according to claim 14 or 15, wherein the movable mass is coupled to a portion of the turbine by a guide allowing movement of the movable mass relative to the turbine; the movable mass being coupled to an electrical machine by a mechanical transmission provided with a multiplier with a variable transmission ratio, such that adjusting the electromagnetic torque of the electrical machine makes it possible to control the movement of the movable mass along the guide; the method comprising at least one of a) and b): a) estimating a frequency of a disturbance causing oscillation of the portion of the wind turbine to which the movable mass is coupled; and adjusting the transmission ratio of the multiplier based on the estimation of the frequency of the disturbance such that the undamped natural frequency of the movable mass is close to the frequency of the disturbance affecting the portion of the turbine to which the movable mass is coupled;and b) adjust the torque transmission ratio so that the undamped natural frequency of the moving mass is close to the undamped natural frequency of the part of the turbine where the moving mass is coupled; 18. Method according to claim 14 or 15, wherein the moving mass is coupled to a part of the turbine by means of a variable curvature guide that allows a displacement of the moving mass with respect to the turbine; the moving mass being coupled to an electric machine by means of a mechanical transmission such that an adjustment of the electromagnetic torque of the electric machine allows controlling the displacement of the moving mass along the guide; the method comprising at least one of a) and b): a) estimating a frequency of a disturbance that causes oscillation of the part of the wind turbine to which the moving mass is coupled; and - adjusting the curvature of the guide based on the estimate of the frequency of the disturbance so that the undamped natural frequency of the moving mass is close to the frequency of the disturbance incident on the part of the turbine to which the moving mass is coupled; and b) - adjusting the curvature of the guide so that the undamped natural frequency of the moving mass is close to the undamped natural frequency of the part of the turbine to which the moving mass is coupled.

Images