Lidar-Based Multivariable Feedforward Control of Wind Turbines

Abstract

Summarizing, the present invention relates to a control system for a wind turbine includinga feed forward controller having a rotor effective wind speed of the wind turbine as an input parameter and having a plurality of output parametersa feedback controller wherein an input parameter is based on a rotor speed or a generator speed of the wind turbine and having at least one output parameter and wherein o one output parameter of the feed forward controller is provided to the feedback controller as an input parameter and o another output parameter of the feed forward controller is used as a feed-forward control parameter for controlling the wind turbine and o one output parameter of the feedback controller is used as a feed-back control parameter for controlling the wind turbine, as well as a wind turbine and a control method.

Claims

1. A control system for a wind turbine, comprising: a feed forward controller having a rotor effective wind speed of the wind turbine as an input parameter and having a plurality of output parameters:, and a feedback controller wherein an input parameter is based on a rotor speed or a generator speed of the wind turbine and having at least one output parameter and wherein, one output parameter of the feed forward controller is provided to the feedback controller as an input parameter and another output parameter of the feed forward controller is used as a feed-forward control parameter for controlling the wind turbine and one output parameter of the feedback controller is used as a feed-back control parameter for controlling the wind turbine.

2. The control system according to claim 1, wherein the feed-forward control parameter of the wind turbine controls a pitch angle of rotor blades of the wind turbine.

3. The control system according to claim 1, wherein the feed-back control parameter of the wind turbine controls a torque of a power generator of the wind turbine.

4. The control system according to claim 1, wherein the output parameter of the feed forward controller that is provided to the feedback controller as an input parameter is an updated generator torque rate.

5. The control system according to claim 1, wherein the output parameter of the feed forward controller that is used as a feed-forward control parameter of the wind turbine is an updated pitch angle of the rotor blades of the wind turbine.

6. The control system according to claim 1, wherein the rotor effective wind speed is obtained using a Lidar device.

7. The control system according to claim 1, wherein the output parameters of the feed forward controller are calculated based on the input parameter and assuming that no dynamics for a rotor and a tower motion of the wind turbine are desired.

8. The control system according to claim 1, wherein the output parameters of the feed forward controller are calculated such that an impact of the rotor effective wind speed of rotor and tower motion is compensated.

9. The control system according to claim 1, wherein a change over time of the rotor speed is reduced and/or a change over time of fore-aft displacement of a tower of the wind turbine is reduced and/or an acceleration of the fore-aft displacement of the tower of the wind turbine is reduced.

10. The control system according to claim 1, wherein the control system is used to control the wind turbine in a transition from an operation of aerodynamic optimality to a full load operation.

11. The control system according to claim 1, wherein the control system is limited to control the wind turbine in a transition from an operation of aerodynamic optimality to a full load operation.

12. The control system according to claim 1, wherein the feedback controller has exactly one output parameter and wherein the feed forward controller has exactly two output parameters.

13. A wind turbine comprising the control system according to claim 1, a power generator, and a plurality of rotor blades.

14. A method for controlling a wind turbine, the method comprising: obtaining a rotor effective wind speed of the wind turbine; providing the rotor effective wind speed as an input parameter to a feed forward controller; providing, by the feed forward controller, one output parameter referred to as a feed forward control parameter for controlling the wind turbine; providing, by the feed forward controller, another output parameter referred to as a feed forward input parameter as an input parameter of a feedback controller; providing by the feedback controller one output parameter referred to as a feedback control parameter for controlling the wind turbine wherein the feedback control parameter is based on the feed forward input parameter and a rotor speed or a generator speed of a power generator of the wind turbine.

15. The method according to claim 14, wherein one or more steps of the method are repeatedly carried out.

Description

BRIEF DESCRIPTION OF THE FIGURES

[0056] FIG. 1 illustrates the degrees of freedom for the reduced non-linear model.

[0057] FIG. 2 illustrates a review of feedback controller design with a wind speed step, showing the desired behaviour in gray and the full simulation model in black.

[0058] FIG. 3 shows a control loop with feedback and multivariable feedforward.

[0059] FIG. 4 illustrates the necessary pitch angle and generator torque to maintain the wind turbine in its steady state for wind speed deviations from 12 m/s.

[0060] FIG. 5 demonstrates the reaction to a small gust at 12 m/s with perfect wind preview, for feedback controller only (gray) and with additional feedforward (black).

[0061] FIG. 6 demonstrates the reaction to a turbulent wind field. The top frame showing rotor-effective wind speed (gray) and its lidar estimate (black). The remaining frames showing feedback controller only (gray) and with additional feedforward (black) using simulated lidar measurements.

[0062] FIG. 7 demonstrates power spectral densities for the 1 hour simulation: Feedback controller only (gray) and with additional feedforward (black). Corner frequencies of used filters are shown in red.

[0063] FIG. 8 shows the power coefficient at 12 m/s and rated rotor speed. By changing the minimum pitch angle from 0 deg (gray) to ?1 deg (black), the power coefficient can be increased from 0.4668 to 0.4699 (0.66%).

DETAILED DESCRIPTION OF THE INVENTION

[0064] I. Modeling of the Wind Turbine

[0065] In this study, a full model of a 5 MW reference turbine is used for simulations. A reduced version of the same turbine is used for the controller design.

[0066] A. Full Simulation Model

[0067] Simulations are done with the aero-elastic simulation tool FAST. In FAST, an onshore wind turbine structure is modeled by a flexible multibody system, which experiences external forces from aerodynamics. The structural model represents dynamics of flexible parts such as the tower, blades, and drive train. The following 15 degrees of freedom (DOF) are enabled in the simulations: first and second flapwise modes and first edgewise mode of three blades, first and second side-to-side and fore-aft tower bending modes, rotor motion and drive train flexibility. Two different types of wind input files can be loaded to the aerodynamic subsystem. Coherent time series of wind characteristics such as wind speed, direction, and shears are used for the extreme load calculations in Section IV-A. The fatigue simulations in Section IV-B are done with a turbulent three-dimensional wind field over the rotor disk generated with TurbSim. In both cases, aerodynamic forces along the blades are calculated iteratively by applying the Blade Element Momentum theory and transferred to the structural model. The described simulation tools have proven to have reliable accuracy which justifies its application as a full simulation model in this work.

[0068] B. Reduced Controller Design

[0069] The aero-elastic model is not useful for controller design due to its complexity and the iterative calculation of the aerodynamics. Here, the SLOW (Simplified Low Order Wind turbine) model from the flatness-based approach is used with some minor changes. Similar to the full simulation model, SLOW consists of a reduced servo-elastic and aerodynamic module, see FIG. 1.

[0070] In the servo-elastic part, only the first tower fore-aft bending mode and the rotational motion are considered:

[00001]$\begin{array}{cc}J.Math..Math.\stackrel{.}{?}+\frac{M_G}{i_{\mathrm{GB}}}=M_a& \left(1.Math.a\right)\\ m.Math..Math.{\stackrel{\mathrm{.Math.}}{x}}_T+c.Math..Math.{\stackrel{.}{x}}_T+k?\left(x_T-x_{0.Math..Math.T}\right)=F_a& \left(1.Math.b\right)\end{array}$

[0071] Equation (1a) models the rotor dynamics, where ? is the rotor speed, M.sub.a is the aerodynamic torque and MG the generator torque. Moreover, i.sub.GB is the gearbox ratio and J is the overall sum of the moments of inertia of rotor and hub about the rotation axis. Equation (1 b) describes the tower fore-aft dynamics, where F.sub.a is the aerodynamic thrust, x.sub.?. the tower top fore-aft displacement, x.sub.0?. the static tower top fore-aft displacement, and m, c, and k are the tower equivalent modal mass, structural damping, and bending stiffness, respectively.

[0072] In the aerodynamic part, the aerodynamic torque and thrust acting on the rotor with the radius R are

[00002]$\begin{array}{cc}{M}_a=\frac{1}{2}.Math.?.Math..Math.?.Math..Math.R^3.Math.\frac{c_P?\left(?,0\right)}{?}.Math.v_{\mathrm{rel}}^2& \left(2.Math.a\right)\\ F_a=\frac{1}{2}.Math.?.Math..Math.?.Math..Math.R^2.Math.c_T?\left(?,0\right).Math.v_{\mathrm{rel}}^2& \left(2.Math.b\right)\end{array}$

[0073] where ? is the air density, ? the tip-speed ratio, defined as

[00003]$\begin{array}{cc}?=\frac{?.Math..Math.R}{v_{\mathrm{rel}}}& \left(3\right)\end{array}$

[0074] and c.sub.? and c.sub.? are the effective power and thrust coefficients, respectively. Two dimensional look-up tables are used to obtain these coefficients, which are precalculated from steady state simulations with the full simulation model. The relative wind speed vrel is used to model the aerodynamic damping and is defined as the superposition of the tower top speed {dot over (x)}.sub.? and the rotor effective wind speed v.sub.0:

V.sub.rel=(v.sub.0?{dot over (x)}.sub.?) (4)

[0075] II. Simulation of Lidar Measurements

[0076] For the lidar-assisted control of the collective pitch and generator torque, a preview of the rotor effective wind speed v.sub.0, is necessary. Current lidar technology provides the possibility to measure the speed of aerosols in front of the turbine by back-scattered light. Due to limitations in the lidar measurements, only the lidar estimate of the rotor effective wind speed v.sub.0L can be provided. Here, the same generic wind field applied to the aero-elastic simulation is scanned with a lidar simulator. The scan trajectory is optimized to provide the best coherence bandwidth for the measurements on the NREL 5 MW wind turbine based on known work.

[0077] Taylor's frozen turbulence hypothesis, which assumes that the turbulent wind field moves unaffected with the average wind speed, is used in the simulation of the measurements as well as for the wind speed estimation. Here, all measurements are condensed to the lidar estimate of the rotor effective wind speed V.sub.0L.

[0078] III. Controller Design

[0079] In this section, the feedback controller for the transition region is designed. Then, the advantages and disadvantages of the flatness-based approach are discussed. Eventually, the lidar-assisted multivariable feedforward controller is derived for the cases of perfect and realistic wind preview.

[0080] A. Feedback Controller

[0081] In this work, only the transition (usually referred to as region 2.5) between the operation of aerodynamic optimality (region 2) and the full load operation (region 3) is considered. The baseline feedback controller for the 5 MW reference wind turbine leaves region 2 at 10.3 m/s and Q=11.7 rpm and then adjusts the generator torque M.sub.G linearly with increasing rotational speed until reaching region 3 at 11.3 m/s and ?.sub.rated=12.1 rpm.

[0082] However, commercial wind turbines often use a PI torque controller. The advantages are that the turbine can be operated with aerodynamic optimality over a larger range and the closed loop behavior can be tuned. The transition to region 2 is usually done by adjusting the lower limit of the torque PI controller using an optimal state feedback of region 2. Usually, a torque or power error term needs to be included in the pitch PI controller in addition to the speed error to have a smooth transition to region 3 and to prevent the pitch from acting during low wind speeds.

[0083] For this work, a generator torque feedback controller (FB) is designed by using the closed-loop-shaping method from the collective pitch controller design. The rotor motion (1a) is linearized at 12 m/s and the proportional and integral gains are chosen, such that the closed loop from wind speed v.sub.0 to generator speed ?.sub.G=?/i.sub.GB has a damping of 0.7 and a natural frequency of 0.6 rad/s. The response of the full simulation model to a wind speed step from 12 m/s to 12.1 m/s is close to the desired behavior, see FIG. 2. The deviations are due to the generator torque filter and the dynamics neglected in the design approach.

[0084] The rated power is increased to 8 MW to have a sufficiently large region 2.5 (ranging now from 10.6 m/s to 13.6 m/s) to test the designed feedback and feedforward controller. Increasing the rated power of wind turbines of the same size while keeping the same rotor and rotor speed has been done by industry and thus seems to a realistic scenario. The Senvion 6.2M126 with rated power of 6.2 MW is based on the 5M with 5 MW. The rated power of the Enercon E-126 was increased from 6 MW to 7.6 MW. More details can be found on the company websites.

[0085] The pitch feedback controller is not further considered in this paper, since all simulations are performed only in region 2.5, where the pitch angle is limited to ?.sub.min=0 deg .

[0086] FIG. 3 shows the overall control loop.

[0087] B. Pros and Cons of the Flatness-Based Approach

[0088] A flatness-based feedforward controller has already been introduced. Based on the wind speed preview and considering system constraints, trajectories of the rotor speed and tower motions are continuously designed during operation and with an inverse wind turbine model translated into trajectories for the pitch angle and the generator torque. The trajectories are planned to minimize the tower movements during the transition between partial and full load operations. The approach has the following advantages: [0089] The feedforward controller is nonlinear and can be used in all regions without scheduling. [0090] Tower and rotor motion are directly reduced by a feedforward of the pitch angle and generator torque. [0091] It can be combined with a conventional feedback controller. [0092] All feedforward signals have zero-mean and can be set to zero, if problems with the wind preview are detected. [0093] Is computationally less expensive compared to NMPC.

[0094] However, there are also disadvantages compared to the collective pitch feedforward controller used in known literature: [0095] The trajectory planning for the rotor and tower motion is difficult to tune. [0096] Pitch angle and generator torque trajectories are not directly designed and might result in extreme inputs. [0097] The overall concept is quite complicated.

[0098] The feedforward controller presented in the next subsections lacks these disadvantages, but abandons the first advantage by linearizing and simplifying the flatness-based approach for the region 2.5. The other advantages can be maintained.

[0099] C Multivariable Extension based on Simplified Calculations

[0100] The Multivariable Extension based on Simplified Calculations (MESCAL) is derived in three main steps: [0101] 1) Calculation of control actions. [0102] 2) Linearization and control actions. [0103] 3) Combination with feedback.

[0104] In the first step, the inverse model of the flatness-based controller is used to calculate the desired generator torque and pitch angle to mitigate the effect of changes in the rotor effective wind speed v.sub.0 to the rotor and tower motion for a given operating point. In contrast to the flatness-based feedforward controller, no dynamics for the rotor and tower motion are designed ({dot over (?)}.sub.d={dot over (x)}.sub.?,d={umlaut over (x)}.sub.?,d=0). With the desired rotor speed ?.sub.d=?.sub.rated and using (3), the desired tip speed ratio ?.sub.d is

[00004]$\begin{array}{cc}{?}_d=\frac{{?}_d.Math.R}{v_0}& \left(5\right)\end{array}$

[0105] With the desired tower top displacement x.sub.?,d and using (1 b) and (2b), the desired thrust coefficient is

[00005]$\begin{array}{cc}{c}_{T,d}=\frac{2.Math.F_{a,d}}{?.Math..Math.?.Math..Math.R^2.Math.v_0^2}.Math..Math.\mathrm{with}.Math..Math.F_{a,d}=k?\left(x_{T,d}-x_{0.Math..Math.T}\right)& \left(6\right)\end{array}$

[0106] Using a inverse ?(?, c.sub.?) of the look-up table c.sub.?(?,0) , one obtains the desired pitch angle

?.sub.d=?(?.sub.d, c.sub.96 ,d) (7)

[0107] Finally, the desired generator torque MG,d can be obtained using (1a) and (2a):

[00006]$\begin{array}{cc}{M}_{G,d}=i_{\mathrm{GB}}.Math.\frac{1}{2}.Math.?.Math..Math.?.Math..Math.R^3.Math.\frac{c_P?\left({?}_d,{?}_d\right)}{{?}_d}.Math.v_0^2& \left(8\right)\end{array}$

[0108] If the generator torque and pitch angle of the SLOW model follow the desired values M.sub.G,d and ?.sub.d, the rotor and tower motions are unaffected by changing wind speed v.sub.0. To visualize the control action, M.sub.G,d and ?.sub.d are calculated for the operating point at v.sub.0?=12 m/s and for wind speeds with ?0.5 m/s and are plotted in FIG. 4.

[0109] In the second step, M.sub.G,d and ?.sub.d are approximated by linear functions in v.sub.0 with regression coefficients a.sub.G,b.sub.G,a.sub.?, and b.sub.?:

M.sub.G,d?a.sub.G+b.sub.G(v.sub.0?v .sub.op) (9a)

?.sub.d?a.sub.p+b.sub.p (v.sub.0?v.sub.op) (9b)

[0110] In the third step, the feedforward actions are combined with the feedback controller in region 2.5 as depicted in FIG. 3. A generator torque rate updated is added to the integral term of the torque feedback controller similar to the collective pitch rate update used in known literature:

??.sub.FF=b.sub.?(v.sub.0?v.sub.0) (9a)

[0111] Performing the same for the pitch angle would not have the desired effect, since the integrator of the PI pitch controller will have negative, values in region 2.5. Therefore, a feedforward updated to the minimal pitch angle ?.sub.min is used:

??.sub.FF=b .sub.p(v.sub.0?v.sub.0) (11)

[0112] where v.sub.0 is a low pass filtered value of v.sub.0 to account for changing wind speeds and to avoid excessive pitch action in region 2.5. Here, a first-order linear filter with a cutoff frequency of f=0.01 Hz is used. The filter allows slow movements of the tower and thus fulfills a similar role to the tower trajectory planning of the flatness-based controller. However, the next section will show, that the tuning of f.sub.on is more intuitive compared to the trajectory planning. A low pass filter is used instead of a high pass filter, because v.sub.0 can be calculated from the wind preview before shifting it in time and thus less phase delay is achieved. Additionally, in a future work v.sub.0 can be used to adjust the minimum pitch angle ?.sub.min as proposed in known literature and to schedule b.sub.p, if necessary.

[0113] The feedforward controller might be derived directly from a linearized model with a similar outcome. Here, the relationship to the flatness-based controller is pointed out.

[0114] D. Adjustment for Realistic Wind Preview

[0115] Using a lidar system, the rotor-effective wind speed v.sub.0 cannot be measured perfectly as discussed in Section II. While in known work the measurement coherence is directly included in the control design, a prefilter is here used in addition to the controller. Previous work showed, that the transfer function between V.sub.0L and v.sub.0 is the optimal prefilter for the lidar estimate to remove all uncorrelated frequencies. A first-order low pass filter with a cut-off frequency of f.sub.off=0.134 Hz is fitted to the transfer function.

[0116] IV. Simulation Results

[0117] In this section, the multivariable feedforward controller is evaluated by simulations first using perfect wind preview and then using simulated lidar measurements.

[0118] A. Simulations Using Perfect Wind Preview

[0119] In a first simulation study, the feedforward controller is tested assuming perfect wind preview to verify that the design objectives (less rotor and less tower motion) can be achieved for the full simulation model.

[0120] Therefore, the full aero-elastic model is disturbed by a coherent gust at 12 m/s similar to known work, but only with 1 m/s amplitude (minimum to maximum) to stay within region 2.5. The proposed feedforward controller can achieve almost perfect cancellation of the effect from v.sub.0 to and ? and x.sub.?, see FIG. 5. The overshoot of the rotor speed (deviation from ?.sub.rated=12.1 rpm) can be reduced by 95.9% and the maximum tower base fore-aft bending moment M.sub.y? by 10.9% compared to the feedback controller, see Table I.

TABLE-US-00001 TABLE I Maximum values of simulation with perfect wind preview FB FB + FF [00007]$\frac{\mathrm{FB}+\mathrm{FF}}{\mathrm{FB}}.Math.\left[\%\right]$ ?? [rpm] 0.203 0.008 4.1 M.sub.yT [rpm] 79.4 70.7 89.1

[0121] The proposed feedforward controller demonstrates a good robustness against model uncertainties. Although the controller is designed with a nonlinear model with only two DOFs (rotor and tower motion) and static aerodynamics, it is able to almost perfectly cancel out the effect from the rotoreffective wind to the rotor speed and tower displacement for a full aero-elastic model with 15 DOFs. Thus, the results are consistent with the control objectives.

[0122] Simulations Using Simulated Lidar Measurements

[0123] In a second simulation study, the robustness against wind measurement errors of the simulated lidar system is examined. For this investigation, a turbulent wind field with a mean wind speed of ?=12 m/s, a very low turbulence intensity (7%) and a length of over 1 h is generated using TurbSim. The low turbulence is chosen to stay in region 2.5, which helps to isolate and to better understand the benefit of the proposed feedforward controller.

[0124] FIG. 6 illustrates a representative 5 min period of the simulation. In the top part of the figure, the time shift and a good agreement between the rotor-effective wind speed from the wind field and its lidar estimate can be observed. Due to the limitations of the lidar measurements and the not exact preview, a perfect performance similar to the previous section cannot be expected. However, with this more realistic wind preview, the variations in the rotor speed ? and tower top displacement x.sub.? are still reduced significantly. The effect of the multivariable feedforward controller in the frequency domain is visible in the Power Spectral

[0125] Densities (PSDs) in FIG. 7. The multivariable feedforward controller can significantly reduce the influence of the wind disturbance to the rotor speed at low frequencies, mainly by the generator torque rate update. Since the adaptive filter has a cut-off frequency at f.sub.off=0.134 Hz, the improvement minimizes above this frequency and no reduction is achieved at the damped eigenfrequency of the tower (0.322 Hz) and the 3P (three-times-per-revolution) frequency (0.601 Hz). In addition, the spectrum of the generator torque is reduced at low frequencies. This effect is similar to the collective pitch feedforward controller, where less pitch action is necessary to reduce the rotor speed variation. The tower base foreaft bending moment is also significantly reduced for low frequencies up to f.sub.off . However, the reduction starts at f.sub.on=0.01 Hz, since by (11) and the used low pass filter, pitch actions below this frequency are hindered.

[0126] Finally, Table II summarizes the results of the 1 h simulation at 12 m/s. Over 55% reduction in the standard deviation of the rotor speed can be achieved. For the calculation of the Damage Equivalent Loads (DELs), a reference number of cycles 2?10.sup.6 is used. Further, a Wohler exponent of 4 is assumed for the fatigue load calculation of the tower base fore-aft bending moment M.sub.y?, and the low-speed shaft torque M.sub.LSS. For M.sub.oop1, the out-of-plane blade root bending moment of blade 1, a Wohler exponent of 10 is applied. Besides the load reduction on the tower base (15%), additional load reductions on shaft and blade root (6% and 5%, respectively) are achieved. Taking into account the low turbulence intensity, the load reduction is promising.

TABLE-US-00002 TABLE II Results for the 1 h simulation with turbulent wind FB FB + FF [00008]$\frac{\mathrm{FB}+\mathrm{FF}}{\mathrm{FB}}.Math.\left[\%\right]$ STD (?) [rpm] 0.0346 0.0154 44.6 DEL(M.sub.yT) [MNm] 24.0 20.2 84.3 DEL(M.sub.LSS) [MNm] 2.64 2.47 93.7 DEL(M.sub.oop1) [MNm] 5.76 5.50 95.4 STD ({dot over (?)}) [deg/s] 0 0.0702 ? EP [MWh] 5.663 5.655 99.9

[0127] The improvements come with some worsening. The increase in pitch activity (represented by the standard deviation of the pitch rate) in considered to be not relevant, because in full load operation, the pitch rate is more than ten times larger. However, the loss in energy production (EP) of 0.14% is not insignificant. However, using both the multivariable feedforward and the adjustments of the minimum pitch angle might result in load reduction and increase in energy production. FIG. 8 shows the power coefficient c.sub.p at v.sub.0=12 m/s and ?=?.sub.rated (resulting in ?=6.65). By changing the minimum pitch angle ?.sub.min from 0 deg to ?1 deg, the power coefficient can be increased by 0.66%. The optimal minimum pitch angle changes with the mean wind speed and the benefit is increasing closer to region 3. Thus, it can be expected that lidar measurements can be used to adjust ?.sub.min and to increase the energy production even above the aforementioned losses.

[0128] V. Conclusions and Outlook

[0129] This paper presents a multivariable feedforward controller for wind turbines using lidar. The feedforward controller is designed to assist conventional feedback controllers for generator torque and collective pitch angle in the transition between partial to full load operations. The design is based on a flatness-based approach presented in previous work, but is simplified by a linearization and adjusted to avoid large pitch actions caused by large mean wind speed changes. Additionally, a PI generator torque controller is designed for the 5 MW reference wind turbine. The transition region is extended by increasing the rated power to 8 MW in order to have a sufficient large range for testing the concept.

[0130] Simulations with a full aero-elastic model and coherent wind show that the combined feedback-feedforward controller follows the design objectives and is able to keep rotor speed and tower motion constant assuming perfect wind preview. Promising load reduction is achieved in simulations with turbulent wind and a lidar simulator. The energy production is also slightly decreased, but possibilities to avoid the loss or even improve the energy production are outlined.

[0131] The invention can also be applied for the following: [0132] Design of a full feedback controller including a tower and drive train damper for the 8 MW wind turbine. [0133] Design a strategy to smoothly enable and disable the multivariable feedforward controller when entering and leaving the region 2.5 and to combine it with the collective pitch feedforward controller. [0134] Include the adjustment of the minimum pitch angle based on the lidar measurements as proposed in the known literature. [0135] Test the proposed multivariable controller in a detailed load analysis with a higher turbulence level. [0136] Determine the overall effect on energy production and load reduction of the concept.