WO2009153614A2

WO2009153614A2 - Means and method of wind turbine control for maximum power acquisition

Info

Publication number: WO2009153614A2
Application number: PCT/IB2008/002396
Authority: WO
Inventors: Clark Wilson Kitchener
Original assignee: Clipper Windpower Technology, Inc.
Priority date: 2008-06-20
Filing date: 2008-09-09
Publication date: 2009-12-23
Also published as: WO2009153614A3

Abstract

A turbine control system for a variable speed electrical generator in a wind turbine mounted atop a support tower. The wind turbine converts wind energy into a driving torque applied to the generator. The control system includes tachometers providing rotation rate directly, angle encoders providing the rotation angular position of the shaft, or incremental angle encoders indicating when the shaft angle has changed by a discrete amount. The wind speed is determined using rotation rate and its derivative as determined by differentiating the rotor angular position measurements independently of torques and turbine models. A further innovation is the use of a rotation rate estimator to produce the rotation rate and its derivative based on the pulse count data produced by incremental rotor angle encoder common in turbines. The estimator is designed to provide an estimation bandwidth consistent with good control while also eliminating undesired frequencies. A further innovation is the use of the wind estimate to calculate the power-optimum pitch, torque, power and rotor rotation rate operating points. A further innovation is the use of the optimum operating points as set points in a control system providing the turbine with maximum power generation in turbulent winds.

Description

MEANS AND METHOD OF WIND TURBINE CONTROL FOR MAXIMUM

POWER ACQUISITION

Kitchener Clark Wilson

BACKGROUND OF THE INVENTION

Field of the Invention

The present invention relates to fluid-flow turbines, and more specifically, to a means and method to provide pitch and torque control that maximizes the acquired wind power.

Description of Related Art

Most existing wind turbines operate at constant speed. The rotor drives a generator, such as an induction generator, at a constant rotational speed. Constant speed operation is required for a direct link with a constant frequency electrical utility grid. Other more modern wind turbines operate at variable speed to produce DC power, which a converter changes to AC power synchronous with the attached power grid.

Constant speed turbines adapt to increasing wind speed by detecting an increase in rotor rotation rate and then increasing the power generated. This increases the load torque on the rotor and keeps it from speeding up. If this is insufficient, the pitch of the rotor blades is reduced to take less rotational energy from the wind and keep the rotor from speeding up. If the rotor speed is not synchronized with the grid, power is not generated.

Variable speed turbines generate power at all wind speeds, but have favored operating points at each wind speed that generate optimal power without over-stressing the components. Striving to follow the operating points as wind speed increases, these turbines also sense rotor speed and use generator torque and rotor blade pitch for control.

Control strategies based merely on detecting deviations in rotor speed from a set point and adjusting pitch or generator torque to correct are limited in their power optimizing effectiveness. The speed change may be due to significant wind speed change and the set point may no longer be power-optimal. A different set point must be determined. Direct knowledge of wind speed is central to better control of the wind turbine and a means to estimate wind speed and track it over time is needed. Current patents in this field include HoUey US 5,155,375 and Holley US 5,289,041 and Wilson US 7,317,260.

To accurately track wind speed, it is necessary to determine the average wind speed over the area swept by the rotor blades of the wind turbine. Anemometers installed near to the area swept by the blades cannot accurately measure the average wind speed because they measure wind speed at a single location, whereas wind speed may vary over the area swept by the blades. Further, the blades alter the wind pattern and remove energy from the wind, and a sensor behind the blade will not reflect the wind in front. Hollev US 5, 155,375 and Hollev US 5,289,041 utilize a wind estimator using the difference between known generator torque and the estimated aerodynamic torque to predict rotor rotation rate and wind speed using a mathematical turbine model. An observer is used to correct the estimates using measured rotor rotation rate. The wind speed estimate is used to schedule the rotation rate set point and the generator torque in and effort to have the rotation rate follow the wind speed.

Wilson Patent US 7,317,260 discloses a turbine control system including a wind speed observer that estimates and tracks wind speed and support tower movement using a measured position of the support tower along with rotor rotation rate and blade pitch angle. These parameters are combined to estimate wind speed and tower movement. The wind speed observer is used in a turbine control system to properly adjust its operating point, to tune the controller (Proportional, Integral, Derivative-PID, state space, etc.), and to damp tower oscillations.

The prior art does not suggest how the wind speed can be used for scheduling to maximize power acquisition. It is desirable to provide a means and method to estimate the mean wind speed across the rotor of a wind turbine where such method is insensitive to estimated model characteristics.

It is desirable to provide a robust means and method to estimate the rotor rotation rate and its derivative for use in said wind speed estimation and other control functions. It is desirable to provide a means and method to use the wind speed and rotation estimates to select turbine operating points that maximize the acquired power in a manner that easily adapts to changing turbine conditions. It is desirable to use the turbine power maximizing operating points in a control system to extract the maximum power from the ever-changing wind.

SUMMARY OF THE INVENTION The present invention relates to a fluid-flow power generating system in which a turbine is mounted on a support tower which is held stationary in the horizontal axis with reference to the fluid flow, such as wind or water. The turbine includes a rotor connected to a rotor hub. The rotor has a main blade section with an adjustable pitch angle. The main blade may have an extender blade with an adjusting device connected to the extender blade. A motor moves the extender blade between a retracted position relative to the main blade section to a more extended position to expose more or less of the rotor to the wind.

A turbine control system is provided to 1) estimate the rotor rotation rate and its derivative; 2) estimate the mean fluid speed across the rotor of the turbine; 3) determine the power-optimum turbine operating point; and 4) using the operating point, control the turbine using pitch and torque to maximize the power acquired from the fluid flow.

In the present invention, wind estimation does not use aerodynamic torque and generator torque within an observer to predict the rotor rotation rate through a turbine model and observer. Instead, the fluid speed is determined using rotation rate and its derivative as determined by differentiating the rotor angular position measurements independently of torques and turbine models. An advantage of this is that model errors do not propagate into rotation rate and hence do not propagate into turbine control.

A further innovation is the use of a rotation rate estimator to estimate the rotation rate and its derivative based on pulse count data produced from incremental rotor angle encoders common in turbines. Further, the estimator is designed to provide an estimation bandwidth consistent with good control while also eliminating undesired frequency components.

A further innovation is the use of the estimated wind speed to specify the power- optimal operating point (blade pitch, generator torque, rotor rotation rate and power generation) in response to changing turbine parameters and changing wind conditions.

A further innovation is the use of the wind tracking power-optimal pitch, torque and rotation rate in a control system.

BRIEF DESCRIPTION OF THE DRAWINGS The invention will be described in detail with reference to the drawings in which:

FIGURE 1 illustrates a wind power-generating device in which the invention is embodied; FIGURE 2 is a block diagram of a turbine onto which the invention is applied;

FIGURE 3 is a block diagram of the control system as it operates with the turbine;

FIGURE 4 is a block diagram of a pitch and torque command generator;

FIGURE 5 is a flow chart of a method by which the invention is practiced;

FIGURE 6 is a graph setting forth the filter frequency response of an embodiment of the rotation estimator;

FIGURE 7 is a block diagram of the simulated system;

FIGURE 8 is a graph setting forth the simulated response of the system in a 14 m/s turbulent wind;

FIGURE 9 is a graph setting forth the simulated response of the system in a 14 m/s turbulent wind where the pitch set point is not used; and

FIGURE 10 is a graph setting forth the simulated response of the system in a 14 m/s turbulent wind where the actual C_Q characteristic is 25% greater than used by the control system.

In these figures, similar numerals refer to similar elements in the drawings. It should be understood that the sizes of the different components in the figures may not be to scale, or in exact proportion, and are shown for visual clarity and for the purpose of explanation.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Refer to FIGURE 1, which illustrates a wind power-generating device. The wind power-generating device includes an electric generator housed in a turbine 100, which is mounted atop a tall tower structure 102 anchored 104 to the ground. The turbine 100 is maintained in the horizontal plane and into the path of prevailing wind current by a yaw control mechanism. The turbine has a rotor 106 with variable pitch blades 108 which rotate in response to wind. Each of the blades has a blade base section 110 and may have blade pitch angle control capability and/or a blade extension section 114 which is variable in length to provide a variable diameter rotor. The rotor diameter may be controlled to fully extend the rotor at low flow velocity and to retract the rotor as flow velocity increases such that the loads delivered by or exerted upon the rotor do not exceed set limits.

The power- generating device is held by the tower structure in the path of the wind current such that the turbine 100 is held in place horizontally in alignment with the wind current. The electric generator is driven by the turbine to produce electricity and is connected to power carrying cables inter-connecting the generator to other units and/or to a power grid.

Conventional rotors utilize blades of fixed length, joined at a rotating hub. These blades may be of variable pitch (selectively rotatable about their longitudinal axes) in order to alter the angle of attack relative to the incoming wind flow, principally for power shedding in high flow velocities. Alternately, these blades may be fixed pitch or stall-regulated, wherein blade lift and therefore power capture falls off dramatically as wind speeds exceed some nominal value. Both variable pitch and stall regulated rotor blades with fixed diameters are well known in the art.

The wind working on the blades results in aerodynamic torque as a function of the wind speed, rotor speed, and blade pitch. The aerodynamic torque, varying with the wind speed, is applied to the turbine rotor to turn a drive train connected to the generator. The drive train has a rotor shaft and gears which increase the rotational speed at the generator. A power converter converts the electrical power output from the generator to electrical power compatible with an electrical utility grid. Control of the turbine involves turning the rotor into the prevailing wind current and altering either or both of the blade pitch (to alter the aerodynamic torque) and the generator torque (to alter the load on the aerodynamic torque).

The present invention provides a method and a controller for operating a wind or water turbine in order to obtain greater efficiency in conversion of wind or water energy to electrical energy. The controller controls the fluid-flow turbine to compensate for varying fluid speeds with knowledge of the fluid speed.

Overview

FIGURE 2 and FIGURE 3 are together a block diagram which illustrates the control system 301 for a pitch angle-regulated wind turbine 200. Although wind is used here for illustrative purposes, any fluid is applicable. The present invention has application to turbines that are not pitch angle regulated (e.g., stall regulated), in which case the pitch angle is taken as a constant in the control system algorithm. The present invention would also have application to turbines that utilize extendible rotor blades such as described in US patent 6,726,439.

A wind current 202 exerts a force on the rotor blades 204 which output aerodynamic torque 206 to the shaft/gearbox 210. Sensors provide outputs 208 of the blade pitch angles and blade length. Blade pitch and length actuators respond to the pitch and blade length commands 201 generated by the control system 301.

The aerodynamic torque 206 is transformed by the shaft/gearbox 210 to a mechanical torque 212 operating on generator/converter 214. Sensors in the gearbox 210 provide an output 216, which is shaft/gearbox rotation. The sensors are instruments attached directly to the rotor shaft or to intermediate shafts within the gearbox. The instruments are one or more of tachometers providing rotation rate directly, angle encoders providing the rotation angular position of the shaft, or incremental angle encoders indicating the shaft rotation angle has changed by a discrete amount. The incremental encoder is the simplest and most reliable and is used in this specification for illustration.

The generators 214 spin in response to the torque 212 and generate electrical power which the converter 214 makes compatible with the power grid 222. Sensor outputs 218 reflect the load torque applied by the generators. The generators apply load torques according to the generator torque command signal 220 generated by the control system 301, as shown in Figure 3.

The control system 301 includes a rotation estimator 302, a wind speed estimator 306, a power optimum set point calculator 310, and a command generator 314.

The rotation estimator 302 accepts the shaft/gearbox 210 rotation information 216 and generates estimates 304 of the rotor rotation rate and of its acceleration. The wind estimator 306 accepts rotation estimator 302 rotation rate outputs 304, the generator/converter 214 torque output 218, and the rotor blades and actuators 204 outputs 208. The wind estimator 306 outputs estimated wind speed 308.

The power optimum set point calculator 310 accepts the estimated wind speed 308. It calculates the optimum turbine operating set points (total generated turbine power, load torque, pitch and rotation rate) 312.

The command generator 314 accepts the set point calculator 310 set points 312, the generator/converter 214 torque output 218, and the rotor blade and actuators 204 outputs 208. Using any number of control schemes (feed forward, PID, state space, etc.), the command generator 314 generates the pitch and length commands 201 and the generator load torque command 220 used by the turbine 200.

Rotor Rotation Rate and Acceleration Estimator 302

For various reasons, turbines generally have several encoders on various shafts connected through the gearbox. The rotor itself is the slowest shaft (Low Speed shaft or LS), and there is generally at least one faster shaft geared up from the LS and referred to as the High Speed shaft (HS). Simultaneous with rotation rate estimation, there is often a need to eliminate specific vibrations from the rotation estimates. These vibrations include gearbox vibrations, the three blade thrice per rotation tower shadow, and blade flexible modes. Often these vibrations have narrow frequency bands and notch filters are effective so long as they do not affect the rotation estimation. Regarding multiple filters affecting each other, the issue is primarily the compounding of phase delays one to another. The solution is to design a single filter that incorporates all the desired filters such that the design process automatically accounts for all filters simultaneously and compensates for any interactions. This is the approach taken here: the rotation filter is designed while including notch filters so they do not interfere. Nominal Trajectory Rotation Filter: From the Taylor series, the future value of a time dynamic phenomenon is known if the current value and all of its time derivatives are known. The approach taken here is to consider that the current value of the LS rotation rate in radians/s, Ω_LS, and two of its derivatives are significant with higher order derivatives lumped into a random white sequence ε. Thus

where the random sequence ε is modeled as zero mean white noise having standard deviation σ_ε representing the higher order derivatives. More or fewer derivatives can be included or excluded, but two are used here to provide good rotation rate and acceleration estimation as needed for the following wind speed estimator. In state space format the equations are

The well known discrete solution, with sampling period T, to this stochastic differential equation is

where the state noise vector, υ, has co variance

The incremental shaft encoder produces a pulse whenever the shaft on which it is mounted rotates a discrete angle. A measurement is produced by counting the number of pulses since the last T-cycle. Since this count is effectively pulses per T, it is a rough rotation rate and the HS incremental count sum, ΔC_HS, is related to the state vector as

where N is the number of encoder pulses per revolution, K is the gear ratio from LS to HS, and μ is the measurement noise having zero mean and standard deviation σ_c. The rotation estimator is a differentiator that acts by integration. It solves the problem of "what should be integrated to produce the given data". The value integrated must be the desired derivative. This integrating approach to differentiation avoids the problem of increased noise, which would occur if actual differentiation were performed. Notch Filter Harmonic Elimination: Harmonic vibrations, independent of the nominal trajectory, are considered in the same manner as described in copending Wilson International Application number PCT/IB2007/001875, "Wind Turbine Damping of Tower Resonant Motion and Symmetric Blade Motion Using Estimation Methods". In the International application, the harmonic elimination is included within the same filter that tracks the nominal trajectory, so that the filter knows both exist. In this manner their gain-phase interference is minimized by design.

The harmonic vibration, h (encoder pulse counts) is modeled as a sinusoid

where ω_h is its known natural frequency (radian/s). The δ_h is a white noise sequence reflecting that the harmonic model may not perfectly reflect the vibration and has a standard deviation σ_h pulse counts. In state space format,

Using linear system theory, the discrete solution is

where

The state noise vector χ has covariance

Combining the nominal and harmonic equations, y

where the co variance of the combined state noise vector s is

Given values of the parameters σ_ε, σ_c, and σ_h, a steady state Kalman filter having steady state gain K is designed having the classic form

where is the actual measurement. Re-written as

the rotation estimator is time invariant linear single input multiple output system. Parameter value selection involves:

• Setting σ_c: count measurement noise (encoder counts) arbitrarily set to 1.

• Setting σ_ε: nominal trajectory state noise (radian/s/s/s/s) selected to set the rotation rate and rotation acceleration bandwidths to desired value • Setting σ_h: harmonic oscillator state noise (counts) selected to set the bandwidth of the

rejection filter

An alternate embodiment of the harmonic oscillator includes a damping term as

where the damping coefficient ξ_h allows adjustment of the depth of the signal rejection. Alternate embodiments of the rotation estimator 302 include the use of multiple encoders within the filter, multiple harmonic oscillators for signal rejection, and the use of other than a steady state Kalman approaches to the filter design. All these alternate embodiments are considered to be within the scope of the present invention.

Wind Speed Estimator 306, Figure 3

Ignoring for simplicity the turbine efficiencies and internally consumed power, the classic equation representing the aerodynamic torque produced by a given wind is

where p = air density (kg/m³) R = rotor radius (m) A = rotor swept area (m²) C_Q = blade aerodynamic torque coefficient λ = tip speed ratio (unit less) β = blade pitch angle (degrees)

Vwind = wind speed (m/s)

and . Using a simplified model of the rotor dynamics, the

turbine torque balance equation requires that

where I = combined rotational inertia of the rotor, shaft, gearbox, and generators as seen in by the rotor LS shaft (N-m) and

rotation rate and acceleration of the low speed rotor shaft (from the rotation estima tor 304) the load torque applied by the generator (from the sensors 218)

Combining (1) and (2):

All the terms of (3) are known and the tip speed ratio λ can be estimated using a

Newton-Raphson algorithm to solve

The Newton-Raphson iterative solution is presented in Table 1 where the partial of C_Q with respect to λ is determined by finite differences. Initial value of λ_opt is λ _opt from the last time series and, since λ_opt does not change greatly, only one or two steps of this iteration are needed.

Table 1: Newton-Raphson algorithm for tip speed calculation

To reduce wind speed estimation scatter due to brief gusts, the resulting λ_opt estimates are low-pass filtered to produce a smoothed wind speed estimate, V^_t, using

where ω_wj_n(j is the filter bandwidth in radian/s. Although first consideration would be to limit the bandwidth of the rotation estimator 302 and thus that of the wind estimate; this can lead to poor wind estimates due to the resulting phase delay of the rotation acceleration.

Air density, efficiency, power loss, inertia and C_Q are parameter inputs to this algorithm and can be changed in real-time. The wind speed estimation so produced is the effective or equivalent wind seen by the rotor uniformly across its area.

Power Optimum Set Point Calculator 310

Using (1), the aerodynamic power available from a given wind is

Ignoring efficiencies and power loss, the power acquired by the turbine is then

Given the wind speed estimate from (4), the optimal values of Ω_LS and β can be determined by maximizing (5) with respect to Ωis and β subject to the inequality constraints

where Ω_max is the maximum high wind turbine operational speed, βmin is the minimum allowed pitch, and P_max is the maximum turbine output power. Numerical optimization is implemented using a two-step process involving gradient and Newton-Raphson solutions.

The first step is to optimize (5) globally with respect to pitch and rotation rate using a gradient algorithm. The algorithm steps are presented in Table 2 where the partials of C_Q with respect to pitch β and tip speed ratio λ are determined by finite differences. The initial values of β_opt = β_mi_n and Ω_opt = Ω_max /2, and k_β and k^ are step size parameters with values around 1x10^-4 and 1x10 ^-8 respectively.

If the resulting P_turbin_e ≤ Pmax then the solution is complete and a second step is not needed. Otherwise

and a second step is initiated to optimize (5) using the Newton-Raphson algorithm to set β as presented in Table 3 where the partial of C_Q with respect to pitch β is determined by finite differences. The optimal torque is then

Lastly, the set points are made equal to the optimal values:

Air density, efficiency, P_max, β_min, Ω_max and C_Q are parameter inputs to this algorithm and can be changed in real-time.

Command Generator 314, Figure 3 The command generator embodiments are many including Proportional-Integral-

Derivative (PID) control loops, state space multiple-input, and multiple-output (MIMO) optimal controllers. A relatively simple embodiment is shown in FIGURE 4. The generator torque set point 312 output of the set point calculator 310 is used as the generator torque command 220 to the turbine 200. The rotation rate set point 312 from the calculator 310 is used as the set point for a conventional pitch PID compensator 402. The rotation rate estimate 304 from the rotation estimator 302 is subtracted from the rate set point to form the PID error input signal 400. The PID produces a pitch modulation 404 which is added to the pitch set point output 312 of the calculator 310 with the result being the pitch command 201 to the turbine. This arrangement combines feed-forward with the PID correction: the pitch set point is responsible for major pitch changes while the rotation rate set point and rotation rate estimate are responsible for pitch corrections. The corrections account for temporal variations due to wind gusts, and to deviations between the actual and the modeled characteristics used by the wind estimator 306 and the set point calculator 310. The PID compensator 402 response to the error signal 400 is used to suggest the model characteristics are deviating and should be investigated. For example, if the PID is simply a Proportional-Integral compensator (PI) with an internal error integral, the DC value of the integral suggests a consistent correction factor while ideally the integral should be close to zero. In an alternate embodiment, significant parameter deviations are detected by comparing the deviations of pitch, torque and power with their respective set points.

Collective versus Independent Pitch Control

The embodiment described thus far is a collective pitch implementation where all three blades are given the same pitch command. Independent pitch control, where each blade is differently pitched by separate pitch commands, is a more sophisticated implementation that can better handle vertical and horizontal wind shear and other phenomena (see the above referenced Wilson et al. International Application number PCT/IB2007/001875). Generally, independent blade pitch control techniques take a collective pitch command and add small modulations, which may be different for each blade.

The invention embodied herein is adaptable to independent blade pitch control by adding modulations to the pitch command 201 for each blade individually.

Method of Operation

Refer to FIGURE 5, which is a flow chart of a method by which the invention is practiced. The flow starts at block 500. Block 502 initializes the rotation estimator 302. Thereafter the flow moves in discrete time steps every T seconds. Block 504 acquires the shaft encoder rotation data 216. Block 506 estimates the rotation rate, acceleration and jerk using the rotation estimator 302 with harmonic blocking using the data acquired in block 504 and previous rotation rate, acceleration and jerk (derivative of acceleration) provided by block 502 or the previous operation of block 506. The rotor blade 204 sensor data 208 and generator/inverter 214 data 218 are acquired in block 508. In block 510 the wind estimator 306 is operated using the data acquired in block 508 and the rotation rate and acceleration provided by block 506. The optimum turbine operating point is calculated in block 512 using the wind speed estimate of block 510. In block 514 the command generator 314 determines the pitch command 201 and generator torque command 220 using the acquired data of block 508, the calculated operating points of block 512 and the estimated rotation rate of block 506. In block 516 the commands are sent to the turbine 200 pitch and torque actuators and the flow continuously loops back to block 504.

A Simulation Example

FIGURE 6 illustrates the frequency response of two examples of the rotation rate estimator having the gearbox ratio of 20: 1. Both examples have

T = 0.05 s sampling period

K = 20 gearbox ratio

N = 8192 pulses per revolution cϋh = 2π(0.575) rad/s 3P harmonic frequency for 11.5 rpm turbine σ_ε = 0.01 rad/s/s/s/s nominal trajectory state noise to produce around a 2 Hz Ω_LS bandwidth (3db down from DC). The typical control bandwidth of large turbines is around 0.1 Hz, and the Ω_LS bandwidth must be higher; here it is 2Ox and should be sufficient (too low a bandwidth induces phase delays in the

and Ω

estimates that make for unstable wind speed estimates). The first example (thin solid line) uses σj, = 0 harmonic noise sigma and presents a very narrow notch at the 3P frequency. The second example (thick dotted) uses σh = 0.1 counts noise sigma to produce a useable ±0.1 Hz pass-band about the center frequency. The resulting estimator gain matrices are

r= 0.6318945395612 0.0223920904671 0.0001763590737 -0.0002781566363 -0.0000140611100

-1.3131651874112 0.9015126109442 0.0461699348701 -0.0009922852300 -0.0000501610603

-2.3425065730515 -0.1756879929789 0.9931676891619 -0.0017701007428 -0.0000894804513

64.5643224478368 0.0484232418359 0.1883126071395 1.0325162637161 0.0521947813292

-197.6979819317763 -14.8273486448832 -0.5766191139677 -0.7984732901470 0.9761768445327

K= 0.000282757488347

0.001008698131602

0.001799379107940

-0.049594607019447 0.151860243408667

FIGURE 7 illustrates the mathematical details of the simulated system. The IOMW turbine 200 consists of a simplified model of the tower motion, lagging pitch (1 Hz) and torque (0.5 Hz) actuators, rotor and generator motion, and shaft windup. The commanded pitch and torque are used as the sensed values. The PID compensator is implemented as a

Proportional-Integral (PI) compensator in series with a second order pitch 0.5 Hz low-pass filter. The wind speed estimator uses a 0.2 Hz low-pass (ω_wj_nd). The only input to the system is the wind speed V_wj_n(1 taken from a simulated 14 m/s turbulent stream.

FIGURE 8 shows the response of the system under full control of the power maximizing controller. In the gray charts, the thick dark lines are the optimum set points, and the superimposed white lines are the system responses. The wind speed is smoothly interpolating the actual wind speed; the pitch is modulating the set point; the rotation rate has modest gust driven deviations from the set point; and the power generated closely tracks the optimum power. FIGURE 9 shows the response of another embodiment where the only change is to remove the β_set pitch feed-forward term. Only the PI compensator output is used to drive the pitch command. Advantages of this approach are there is less pitch activity and less tower stimulation, and the disadvantage is that rotation rate (RPM) control suffers somewhat. FIGURE 10 illustrates the method sensitivity to model parameter errors. The system is under full control with set points for pitch, rotation rate and torque, but the actual C_Q curve has 25% larger values than modeled. As expected, the wind speed estimates are biased upward, but there are no other significant deviations. There is little sensitivity to model parameter errors.

The invention has been described with reference to a wind turbine mounted atop a tower. It will be understood by those skilled in the art that the principles of the invention apply to any fluid-flow system, such water powered turbines which are situated in the path of water current and wind and water powered turbines which are situated in the path of both wind and water currents.

Claims

C l a i m s

1. A turbine control method for a variable speed electrical generator in a fluid-flow turbine comprising steps of:

A. initializing a rotation estimator 302;

B. sensing shaft rotated position change 216;

C. estimating rotation rate 304 using sensed data acquired in step B and previous rotation rate in step A or a previous operation of step C; D. acquiring 208 rotor blade sensor data 204 and generator/inverter 214 data 218; and

E. estimating wind speed 306 using the data acquired in step D and the rotation rate provided by step C;

2. The method of claim 1 further comprising a step of:

F. calculating a turbine operating point 312 using the wind speed estimate of step E.

3. The method of claim 2 where the turbine operating point 312 is calculated maximizing 310 the acquired power.

4. The method of claim 1 or 2 further comprising steps of:

G. determining 314 a pitch command 201 and generator torque command 220 using the acquired data of step D, the calculated operating points of step F and the estimated rotation rate of step C; and H. sending the commands to the turbine 200 pitch and torque actuators.

5. The method of claim 1, 2 or 4 further comprising the step of: I. looping method flow back to step B.