METHOD FOR DETECTING IRREGULAR TURBINE OPERATION USING
DIRECT AND INDIRECT WIND SPEED MEASUREMENTS
FIELD
[0001] The subject matter described herein relates to methods for operating a
5 wind turbine and to wind turbines, and more particularly to methods for operating
a wind turbine including a wind characteristics sensor for measuring wind
characteristics and at least one wind turbine state sensor for measuring a state of
the wind turbine from which an estimation of the wind characteristics is obtained.
BACKGROUND
10 [0002] Wind turbines typically include a tower and a nacelle mounted on the
tower. A rotor is rotatably mounted to the nacelle and is coupled to an electric
generator by a shaft. A plurality of blades extend from the rotor. The blades are
oriented such that wind passing over the blades turns the rotor and rotates the
shaft, thereby driving the generator to generate electricity.
15 [0003] A wind turbine converts wind energy into mechanical energy, e.g. into
rotational kinetic energy, and the mechanical energy is typically further converted
to electrical energy by a wind turbine generator. A blade pitch angle, i.e. an angle
of attack of a blade of the rotor of the wind turbine with respect to the direction of
the wind flow, can be adjusted in order to control force and/or torque acting on the
20 blade. The rotational speed of the rotor of the wind turbine and the electrical
power generated by the wind turbine generator, driven by the rotor through the
shaft of the wind turbine, can therefore be controlled adjusting the pitch angle of
the blades of the wind turbine.
[0004] A blade pitch angle may be adjusted for each blade individually or
25 collectively for one or more blades of the wind turbine. As the wind speed
changes, the blade pitch angle of one or more blades of the wind turbines is
adjusted to keep rotor speed and torque within operating limits for maximizing
3
efficiency of the generation of electrical energy by the wind turbine generator,
whilst minimizing the risks of damages to the wind turbine due to e.g. sudden
wind gusts.
[0005] A wind turbine may reach a stall condition, i.e. a condition such that if the
5 angle of attack of one or more blades is increased the maximum power that the
wind turbine generates begins to decrease. For actual wind conditions, an angle of
attack of one or more blades for which a further increase of the angle of attack
produces a decrease in power is an angle of attack producing a stall condition. The
minimum angle of attack producing a stall condition is called critical angle of
10 attack for the actual wind conditions.
[0006] A wind turbine may be operated in a stall condition, but when the angle of
attack of the one or more blade is further increased, a significant stall condition or
deep stall condition may result. It is not desirable to operate a wind turbine in a
significant stall condition or deep stall condition.
15 [0007] Typical critical angles of attack are in the range of 15 to 20 deg. Generally
a wind turbine is said to be in a stall condition, i.e. stalling, if the angle of attack
exceeds the critical angle. To avoid any stall on parts of the blade, angles of attack
are typically required to be around 3 to 5 deg. below the critical angle of attack
during the operation of a wind turbine. Therefore, a significant stall condition or
20 deep stall condition may be any condition wherein the wind turbine is stalling
when the angle of attack exceeds the critical angle of attack.
[0008] In a significant stall condition, turbulence of wind flow may result in a
chaotic or irregular dynamic of the wind flowing at the wind turbine. An operation
in a significant stall condition may be part of a wind turbine operation, but a
25 significant stall condition is usually undesired due to the prevalence of negative
effects like e.g. chaotic or irregular wind flow and/or power decrease.
Furthermore, excessive wind speeds or wind gusts may damage the wind turbine
and an operation in a significant stall condition in the presence of intense winds
4
may pose a significant risk of damage to blades and/or other wind turbine
components.
[0009] A malfunctioning or a disturbed condition of the wind turbine may result
from different causes like e.g. an icing of the blades of the wind turbine, deposited
5 dirt on the blades of the wind turbines, an aging of wind turbine components or
from other external or internal factors affecting the functioning of the wind
turbine.
[0010] Thus, it would be beneficial to reliably detect and/or prevent a significant
stall condition of the wind turbine or a malfunctioning or a disturbed condition of
10 the wind turbine.
SUMMARY
[0011] According to one aspect, a method for operating a wind turbine is
provided, the wind turbine including a wind characteristics sensor for measuring a
wind characteristic and at least one wind turbine state sensor for measuring a state
15 of the wind turbine, the method including: determining or adjusting one or more
wind characteristics relationships; and, performing an operation phase, the
operation phase including: measuring the wind characteristics with the wind
characteristics sensor, thereby obtaining measured wind characteristics; measuring
the state of the wind turbine with the at least one wind turbine state sensor and
20 determining an estimated wind characteristic from the measured state of the wind
turbine and parameters of the wind turbine; comparing the estimated wind
characteristics to the expected wind characteristics determined from the measured
wind characteristics, wherein the expected wind characteristics is determined
based on the one or more wind characteristics relationships; and, operating or
25 shutting down the wind turbine based at least in part on the comparison result.
[0012] Accordingly, the present disclosure aims at accurately measuring wind
characteristics of the wind present at the wind turbine, such as wind speed and/or
wind direction and/or wind shear, the presence of turbulences in the wind flow,
5
etc. In order to do so, a wind characteristic is measured with a wind characteristics
sensor. In addition, a measurement of a state of the wind turbine, which may e.g.
include a speed of the rotor and/or a torque of the rotor of the wind turbine and/or
generated power of the wind turbine, is carried out with at least one wind turbine
5 state sensor.
[0013] According to a further aspect, a wind turbine is provided wind turbine
including at least one wind measurement sensor; and a wind turbine state sensor
to measure a state of the wind turbine for estimating wind characteristics at the
wind turbine location; a control system configured to control the wind turbine
10 based at least in part on inputs formed by measured wind characteristics measured
by the wind measurement sensor, and by measured wind turbine states measured
by the wind turbine state sensor.
[0014] Further aspects, details and advantages are apparent from the following
description, the accompanying drawings and the dependent claims.
15 BRIEF DESCRIPTION OF THE DRAWINGS
[0015]The present disclosure will be explained in view of the following
exemplary drawings.
[0016] FIG. 1 shows a wind turbine with a nacelle, a rotor, and rotor blades
according to embodiments of the present disclosure.
20 [0017] FIG. 1A shows details of a wind turbine, showing in particular a shaft of
the wind turbine and a wind turbine generator according to embodiments of the
present disclosure.
[0018] FIG 1B illustrates a method for operating a wind turbine according to
embodiments of the present disclosure.
25 [0019] FIG 2 illustrates determining or adjusting one or more wind
characteristics relationships according to methods of the present disclosure.
6
[0020] FIG 3 illustrates an operation phase of a method for operating a wind
turbine according to embodiments of the present disclosure.
[0021] FIG 4 illustrates details related to a method for operating a wind turbine
according to embodiments of the present disclosure.
5 DETAILED DESCRIPTION OF THE INVENTION
[0022] Reference will now be made in detail to the various embodiments, one or
more examples of which are exemplarily illustrated in the figures.
[0023] FIG. 1 shows a wind turbine 10, the wind turbine including a tower 12,
placed on a support system 14, a nacelle 16, with a rotor 18, connected to a
10 rotatable hub 20. One or more rotor blades 22 are configured to convert the
kinetic energy of the wind into rotational kinetic energy of the rotor 18. Each
blade has a blade root portion 24, a load transfer region 26 where the rotation is
transmitted to the rotatable hub 20. When a wind component flows in the direction
28, the rotor and the rotatable hub rotate around an axis of rotation 30. Along the
15 rotor blades 22, pitch axes 34 are shown in FIG. 1.
[0024] A control system 36, which may be located at the wind turbine like in
FIG. 1 or elsewhere, is configured to control a pitch angle of the rotor blades,
related to an angle of attack with respect to the wind direction, in order to control
e.g. the speed or a torque of the rotor blades of the wind turbine, wherein the
20 speed or torque are imparted to the rotor by the wind. The wind turbine further has
a yaw axis 38 for orienting the rotor blades with respect to different wind
directions around the tower 12. A processor 40 may be part of the control system
36.
[0025] As shown in FIG. 1A, the nacelle 16 of the wind turbine further includes a
25 wind turbine generator 42 for the generation of electric energy from the rotational
kinetic energy of the rotor, this rotational kinetic energy results from the kinetic
energy of the wind in function of the pitch angles of the rotor blades.
7
[0026] In the present disclosure, it is intended that wind characteristics may
include one or more wind speeds, one or more wind shears, one or more temporal
or spatial derivatives of wind speeds, one or more wind directions. For example, a
wind characteristic may be a scalar related to an amplitude of a wind speed, for
5 example a wind speed in the direction 28 shown in FIG. 1 at the wind turbine
location. A wind characteristic may e.g. also be a vector related to a wind speed at
the wind turbine location, or a set of scalars or a set of vectors related to one or
more wind speeds at or near the wind turbine location, where the wind speeds
may be wind speeds at a given position in space, or averaged spatial or temporal
10 wind speeds at or near the wind turbine location. For example, wind
characteristics may be described in terms of ordered tuples of real numbers related
to a wind speed at or near the wind turbine location.
[0027] In some embodiments, the wind characteristics may be a magnitude of a
wind speed, in particular of a scalar wind speed or of a vector describing a wind
15 speed. For example, the wind characteristics may be measured in m/s.
[0028] In FIG. 1A further details of a wind turbine 10 are illustrated and un
particular of a nacelle 16 of the wind turbine 10. In particular a rotor shaft 44
transmits the kinetic energy to the wind turbine generator for the generation of
electric energy from the kinetic energy of the wind. The rotor shaft presents a
20 longitudinal axis 45 that forms an axis of rotation of the rotor shaft. A gearbox 46
may be used in order to control a rotational speed and torque of a high speed shaft
48 driving the wind turbine generator. The wind turbine generator 42 is driven by
the rotational kinetic energy of the high speed shaft 48 driven by the rotor shaft 44
through the gearbox 46 for the generation of electric energy. The rotor shaft 44
25 therefore transmit a rotational movement to the high speed shaft 48 through the
gearbox 46, and the rotational speed of the rotor shaft 44 is typically lower than
the rotational speed of the high speed shaft 48. The rotor shaft 44 is coupled to the
blades of the rotor 18 of the wind turbine and when the wind imparts a rotational
movement to the rotor, the rotor shaft rotates accordingly.
8
[0029] FIG. 1A further shows a coupling 50 between the high speed shaft 48 and
the wind turbine generator 42, supports 52 and 54, a yaw drive mechanism 56 for
the rotation of the nacelle around a yaw axis 38 for orienting the rotor with respect
to a wind speed direction 28. A wind characteristics sensor 58 may measure a
5 wind characteristics at the wind turbine location, e.g. a wind speed flowing in the
direction 28. The wind characteristics sensor 58 of the wind turbine may be for
example an anemometer. Generally, and not limited to any other feature described
in relation to Fig. 1A, the anemometer of the wind turbine may be located on top
of the nacelle.
10 [0030] Bearings 60, 62 may support the shaft or other components of the wind
turbine, as shone in FIG. 1A. The wind turbine may further include a pitch
assembly 66, that may include a pitch drive assembly 68 for the control of a pitch
angle of one or more blades. The assembly may include sensors 70, pitch bearings
72, a pitch drive motor 74, a pitch drive gearbox 76, a pitch drive pinion 78, for
15 one or more rotor blades.
[0031] An overspeed control system 80 may be present. Cables 82 for
transmitting signals from or to a control system of the wind turbine are further
indicated in FIG 1A. Finally an actuator 84 may provide an actual pitch angle of
the wind turbine blades, blades connected to the cavity 86 presenting an inner
20 surface 88 ad an outer surface 90.
[0032] As used herein, the term “blade” is intended to be representative of any
device that provides a reactive force when in motion relative to a surrounding
fluid, like air forming the wind at the wind turbine location. As used herein, the
term “wind turbine” is intended to be representative of any device that generates
25 rotational energy from wind energy, and more specifically, converts kinetic
energy of wind into mechanical energy. A “wind turbine generator” typically
further converts the mechanical energy to electrical energy trough a wind turbine
generator.
9
[0033] Although every commercial wind turbine is typically equipped with an
anemometer on the nacelle, these anemometers are generally not used as inputs
for the turbine control as their readings are too unreliable. Instead, some modern
wind turbines use model-based estimation techniques to calculate the wind speed
5 based on the performance of the turbine itself. However, these estimators rely on
accurate model information to be stored in the controller or on assumptions about
the wind turbine operation or the conditions affecting the wind turbine. As a
consequence, the estimators cannot be used to detect abnormal turbine operations
such as icing or stalling, as in these instances/situations the model parameters are
10 no longer correct and the estimator does not report the correct wind speed
anymore. Furthermore, if such an abnormal operation remains undetected the
controller will control the turbine incorrectly, e.g. drive it into deep stall, which
can cause additional loss of power production or even damage the wind turbine or
some components of it.
15 [0034] FIG. 1B illustrates a method 100 for operating a wind turbine according to
embodiments of the present disclosure. The method 100 for operating a wind
turbine includes determining or adjusting 102 one or more wind characteristic
relationships, and performing 104 an operation phase.
[0035] As used herein, a sensor for measuring wind characteristics may in
20 particular be a wind turbine anemometer. Methods of the present disclosure allow
in particular to calibrate the wind turbine anemometer or a sensor for measuring
wind characteristics, and the measured wind speed of the wind turbine
anemometer or of the sensor for measuring wind characteristics becomes a more
reliable and usable quantity for the control and the monitoring of the wind turbine
25 system. Modeled wind is reliable if the wind turbine operates in normal
undisturbed conditions. If the wind turbine does operate offline or in a stall or
disturbed condition, the wind speed obtained based on models may be wrong and
thus the turbine may not be operated at its optimum operating parameters or may
be even exposed to damage.
10
[0036] The present disclosure provides a highly accurate redundant wind speed
measurement that will be used to detect e.g. blade icing, blade failures and other
turbine abnormalities or disturbed conditions detectable through wind speed
deviations. In some embodiments, also a significant or deep stall condition is
5 detectable. During non-operational times of the wind turbine and/or during a stall
or disturbed condition of the wind turbine, the wind speed measurement by the
wind characteristics sensor, e.g. the anemometer, typically is much more accurate.
Therefore, the wind speed measured by a wind characteristics sensor or
anemometer may be used during a stall or disturbed condition in place of the wind
10 speed obtained/estimated with the use of models for an accurate control or in
order to prevent damages, provided that e.g. a systematic error affecting the wind
speed measurement obtained by the wind characteristic sensor, e.g. an
anemometer, is handled properly. This allows furthermore for possible power
calculations, improved accuracy and performance of a return into operation after
15 e.g. a calm or a storm condition, etc. Methods of the present disclosure also enable
the possibility of power curve measurements based on e.g. nacelle anemometry.
[0037] When for example an anemometer is calibrated against a met mast, severe
drawbacks are present. E.g. due to the distance between the met mast and the
turbine the correlation between the wind characteristics at the two locations is not
20 or not always good. Furthermore, this type of calibration is only applicable to
specific wind turbines. For turbines without a met mast, such a calibration taken
e.g. from another turbine may not be applicable and/or a significant variability in
function of factors like e.g. the local terrain configuration may affect the quality or
reliability of the calibration. When using wind speed measurement equipment that
25 measures wind characteristics in front of the rotor, such as e.g. a LIDAR, a
calibration may be less necessary. However, devices like LIDARs tend to be
expensive.
[0038] As used herein, it is intended that the state of the wind turbine may e.g.
include a rotor speed and/or a generated electrical power by the wind turbine
30 generator and/or a torque of the rotor and/or the rotor shaft. It is intended that
11
parameters for the wind turbine operation may e.g. include pitch angles of blades
of the wind turbines or e.g. a torque of a generator of the wind turbine and/or a
configuration of a gearbox of the wind turbine. Parameters are assumed to be
known quantities.
5 [0039] Wind characteristics may include one or more wind speeds, one or more
wind directions, one or more wind accelerations at one or more locations at or
near the wind turbine location, and/or wind turbulence. It is intended that both
wind characteristics and wind turbine states and parameters may be one or more
scalar and/or one or more vectors describing one or more quantities.
10 [0040] Wind characteristics at the wind turbine location may be measured using
different types of sensors It may be possible to use a wind measurement mast
positioned at some distance from the wind turbine, e.g. upstream of the wind
turbine. The value of the wind characteristics measured at the measurement mast
may however be different from the values at the wind turbine location, e.g. the
15 surrounding terrain and/or objects may produce a significant difference of values
of wind characteristics measured at the mast with respect to wind characteristics at
the wind turbine location.
[0041] When measuring the wind characteristics at the wind turbine location with
a local wind characteristic sensor, such as an anemometer placed at the wind
20 turbine, the measurement is typically affected by an error, e.g. a systematic error,
due to the presence of the wind turbine and the wind turbine blades themselves.
Therefore, the value of wind characteristics measured by a local sensor at the
wind turbine location, e.g. by an anemometer located at the wind turbine, cannot
be used directly for determining the real wind characteristics at the wind turbine
25 location due to the effects of the presence of the wind turbine and the wind turbine
blades themselves.
[0042] Under nominal circumstances, it is beneficial to use the wind turbine itself
as a measurement instrument for determining wind characteristics at the wind
turbine location. Knowing actual relevant operational parameters of the wind
12
turbine, such as the pitch angles of the blades, under nominal circumstances, the
state of the wind turbine, e.g. a rotor speed and/or the power output, correlates
with the wind characteristics at the wind turbine location, e.g. with a local wind
speed. It is therefore possible to estimate the wind characteristics from a state of
5 the wind turbine, given actual known values of parameters of the wind turbine.
Therefore, under nominal circumstances, the wind turbine itself may replace
sensors for measuring wind characteristics at the wind turbine location. But, if a
stall condition occurs, e.g. a significant or deep stall condition, or a disturbed
condition, e.g. in the presence of ice or dirt on the blades, the wind turbine may
10 not be used anymore for estimating wind characteristics, given that a correlation
between the true actual wind characteristics and a state of the wind turbine for
actual values of parameters, such as pitch angles, becomes irregular or chaotic or
unreliable or affected by significant errors.
[0043] Therefore, it is beneficial to detect a stall or a disturbed condition without
15 relying on the wind characteristics estimated from a state of the wind turbine, but
also avoiding a situation where only a wind characteristic sensor, like e.g. an
anemometer, is used directly, given that the wind characteristics sensor is
typically affected by significant systematic or statistical errors.
[0044] Detecting a stall or a disturbed condition is beneficial for operating a wind
20 turbine e.g. for avoiding damages to the wind turbine and/or for improving the
delivery of output power.
[0045] Some type of sensors, like LIDARs, may be able to measure wind
characteristics in the proximity of a wind turbine, that can be used for determining
reliably wind characteristics at the wind turbine location with sufficient accuracy
25 and precision, however LIDARs may be expensive or unpractical at least under
some circumstances.
[0046] It is therefore beneficial to calibrate a wind characteristic sensor, e.g. a
local anemometer, at the wind turbine location in order to overcome the
13
systematic errors introduced by the presence of the wind turbine that typically
affect said wind characteristic sensor.
[0047] It is furthermore beneficial to compare wind characteristics estimated from
a measured state of the wind turbine, under consideration of operational
5 parameters, with the wind characteristics obtained by the calibrated wind
characteristics sensor. Under normal circumstances, the wind characteristics
estimated from the measured state of the wind turbine are more reliable and
accurate, but nevertheless the value of the wind characteristics obtained with the
calibrated wind characteristics sensor is close to the value of the estimated wind
10 characteristics. That is, both values are comparable, whereas without calibration
the wind characteristics sensor is affected by a significant error, but the estimated
wind characteristics value will be typically more precise and accurate under
normal circumstances.
[0048] Under a significant stall condition or a disturbed condition the wind
15 characteristics estimated from the measured state of the wind turbine may be
erroneous and may be significantly different from the values of the wind
characteristics obtained with the calibrated wind characteristics sensor. Therefore,
the values obtained by the calibrated wind characteristics sensor may be used for a
plausibility check of the wind characteristics estimated from the measured state of
20 the wind turbine.
[0049] Comparing wind characteristics obtained from a measured state of the
wind turbine with wind characteristics obtained with the calibrated wind
characteristics sensor is beneficial in particular in order to detect a stall and a
disturbed condition of the wind turbine. In particular, a comparison may be based
25 on a difference between the estimated wind characteristics based on a measured
state of the wind turbine and the wind characteristics obtained with the calibrated
sensor for measuring wind characteristics, e.g. a calibrated anemometer. Without
calibration, a significant error may affect the wind characteristics sensor, e.g. the
anemometer and therefore the measurement can be erroneous.
14
[0050] In FIG. 2, a calibration phase of a method for operating a wind turbine is
illustrated according to some embodiments of the present disclosure. In particular
FIG. 2 illustrates how a relationship of one or more wind characteristic
relationships is determined or adjusted. As used herein, the term “calibration
5 phase” may therefore refer to the determination or adjustment of one or more
wind characteristics relationships, i.e. a determination or adjustment 200 of one or
more wind characteristics relationship is performed in a calibration phase. A wind
characteristics relationship may be implemented by any data structure capable of
associating information on wind characteristics to other information on wind
10 characteristics. For example, a wind characteristic relationship may be a transfer
function that associates to a vector, e.g. to a vector describing measured wind
characteristics, another vector, e.g. a vector describing estimated expected wind
characteristics. A wind characteristic relationship may also be implemented as a
set of ordered pairs of vectors, wherein for each ordered pair the first component
15 is a vector related to e.g. measured wind characteristics and the second component
is a vector related to e.g. expected estimated wind characteristics. The expected
estimated wind characteristics may form an expected value for wind
characteristics estimated with the use of physical models of the wind turbine
based e.g. at least in part on a measured state of the wind turbine.
20 [0051] A calibration phase for determining or adjusting one or more wind
characteristic relationships may include measuring 202 wind characteristics with
the wind characteristics sensor of the wind turbine, e.g. with the wind
characteristics sensor 58 of FIG 1A, thereby obtaining wind characteristics data,
measuring 204 the state of the wind turbine with at least one wind turbine state
25 sensor and determining an estimated wind characteristics of the wind turbine from
the measured state of the wind turbine and parameters of the wind turbine. The
wind turbine state sensor, may e.g. in particular measure a rotational speed of the
rotor shaft 44 of the wind turbine.
[0052] Further parameters, like e.g. a pitch angle of one or more rotor blades, may
30 be considered for the determination of the estimated wind characteristics. It is
15
intended that the estimated wind characteristics are based in particular on a
physical model of the wind turbine. As shown in FIG. 2, a calibration phase may
further includes determining or adjusting 206 a relationship between the measured
wind characteristics of the wind turbine and the estimated wind characteristics of
5 the wind turbine. The relationship may in particular be based upon measured wind
characteristics and/or estimated wind characteristics and of a historical sequence
of said characteristics stored in a convenient data structure, e.g. in a list of ordered
pairs stored e.g. in a memory of the control system 36 and/or of the processor 40.
It is intended that the relationship in block 206 may be identified with e.g. a
10 transfer function.
[0053] Wind characteristics measured by the wind characteristics sensor 58 of the
wind turbine, e.g. by a local anemometer, are indicated with the symbol
𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑. With the symbol 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 a state of the wind turbine and with the
symbol 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 operational parameters of the wind turbine are indicated
15 respectively.
[0054] It is intended that 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑, 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒, 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 may be scalars or vectors.
In some alternative embodiments these quantities may alternatively refer to a
wind turbine of the same type or to quantities related to a simulation of the wind
turbine.
20 [0055] The state 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 of the wind turbine may include e.g. rotor speed, rotor
torque and/or e.g. a rotational speed of the rotor shaft 44 of the wind turbine
and/or a torque of the rotor shaft 44 and/or a power output of the wind turbine
generator 42. It is intended that an estimation of wind speed characteristics at the
wind turbine location is possible when the state of the wind turbine is measured,
25 possibly with the consideration of the operational parameters of the wind turbine
that are assumed to be known.
[0056] The operational parameters 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 may include e.g. pitch angles of the
rotor blades, a torque parameter of the wind turbine generator, the actual
configuration of a gearbox, etc.
16
[0057] Knowing the state 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 of the wind turbine and the operational
parameters 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 of the wind turbine it is possible to estimate wind
characteristics at the wind turbine location. The estimated wind characteristics in
function of the state and the parameters of the wind turbine are indicated with
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 = 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 5 , 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒)
[0058] It is intended that 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 and 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑, may be scalars or vectors and
that they can be compared to each other with the use of e.g. a suitable metric, such
as an Euclidean distance between scalars or vectors. The computation of
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒, 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 ) may be in particular based on model based
10 estimation techniques and in particular on the use of physical models of e.g. the
wind turbine and/or of wind turbine components.
[0059] In some alternative embodiments where 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒, 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 relate to a wind
turbine of the same type, 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 also relates to a wind turbine of the same
type. In some alternative embodiments where 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 , 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 relate to a
15 simulation of the wind turbine, 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 also relates to a simulation of the wind
turbine.
[0060] Assuming that the operational parameters of the wind turbine 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 are
known, for brevity it is stated that an estimation of wind characteristics at the
wind turbine location is obtained from a state 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 of the wind turbine, and
20 equivalently it may be written that 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 = 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒) assuming
implicitly the dependency on 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒, with 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 known.
[0061] When a significant stall or disturbed condition is not present, 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑
may be a good estimate of the actual wind characteristics at the wind turbine
location, whereas in a significant stall or disturbed condition 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 may
25 deviate significantly from the true value of the wind characteristics at the wind
turbine location.
17
[0062] On the other side, 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 may be affected by a significant error, and in
particular by a systematic error due to the presence of the wind turbine or the
wind turbine blades.
[0063] If it is determined that the wind turbine is operating in a regular condition,
5 i.e. not in significant stall condition and not in a disturbed condition, the sensor
for measuring 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 may be calibrated using the information obtained from
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑, in order to account for the systematic error that affects 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑.
[0064] In order to eliminate or at least mitigate the systematic error that affects
𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑, in a calibration phase values of 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 and values of 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 may
be measured repeatedly at different time instants 𝑡1, 10 𝑡2, … ,𝑡𝑛. In this case, it is
assumed that also values of 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 are known at the time instants 𝑡1,
𝑡2, …,𝑡𝑛.
Then, in some embodiments, a sequence S of ordered pairs is determined
𝑆 =
(

(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡1
), 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡1
), 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡1
)) ),
(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡2
), 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡2
), 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡2
)) ),
…
(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝑛
),𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡𝑛
), 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡𝑛
)) ) )

Where 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝑖
) indicates the value of 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 at time instant
𝑡𝑖
, 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡𝑖
) indicates the value of 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 at time instant 𝑡𝑖 and 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 (𝑡𝑖 15 )
indicates the value of 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 at time instants 𝑡𝑖
, for 𝑖 = 1, … , 𝑛.
[0065]Assuming that 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 is known, it is written for compactness
𝑆 =
(

(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡1
), 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡1
)) ),
(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡2
), 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡2
)) ),
…
(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝑛
), 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡𝑛
)) ) )

And for even more compactness it is written
18
𝑆 = (
(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡1
), 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡1
) ),
(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡2
), 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡2
) ),
…
(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝑛
), 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡𝑛) )
)
With 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡𝑖
) = 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡𝑖
)) =
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡𝑖
), 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 (𝑡𝑖
)) for 𝑖 = 1, … , 𝑛.
[0066] In some embodiments, the time instants 𝑡𝑖 may identify time intervals of
fixed or variable length and 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝑖 5 ) may be an average measured wind
speed over the time interval identified by 𝑡𝑖
. For example 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝑖
) may be
an average measured wind speed of wind speeds during an interval related to 𝑡𝑖
,
e.g. during an interval [𝑡𝑖 − ∆𝑡
,𝑡𝑖
] with ∆𝑡 a predetermined time delay. For
example 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝑖
) may be a moving average, like a simple moving average
or an exponential moving average at time instant 𝑡 10 𝑖 of instantaneous wind
speeds. It is intended that in these embodiments also 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡𝑖
) may be an
average estimated wind speed over the time interval identified by 𝑡𝑖
, e.g. over
the interval [𝑡𝑖 − ∆𝑡
,𝑡𝑖
] and/or that 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡𝑖
) may also be a moving
average, like e.g. a simple moving average or exponential moving average, in
15 particular a moving average with an identical or similar sample window as the
moving average identified by 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝑖
).
[0067] In some embodiments, the sequence of ordered pairs S can be used in
order to determine a relationship between measured and estimated wind
characteristics at the wind turbine location. The relationship may be e.g. a transfer
20 function and may be stored in a memory of e.g. a local controller or processor of
the wind turbine or elsewhere.
[0068] In some embodiments, a relationship between the measured wind
characteristics and the estimated wind characteristics may be obtained by other
means, e.g. using at least in part interpolation and/or regression analysis and/or
25 Monte Carlo methods, based on measurements of 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 and computations of
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 based at least in part on the state 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 of the wind turbine. In some
19
embodiments, interpolation and/or regression analysis and/or Monte Carlo
methods may be based on S.
[0069] In some alternative embodiments, the relationship between measured
wind characteristics and estimated wind characteristics may be obtained in a
5 similar way and in particular based on a sequence of ordered pairs
𝑆 = (
(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡1
), 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡1
) ),
(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡2
), 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡2
) ),
…
(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝑛
),𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡𝑛) )
) obtained as described, but wherein
the measured wind characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑
(𝑡𝑖
) and the measured state and
parameters 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡𝑖
), 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒(𝑡𝑖
) of the wind turbine are related to a wind
turbine of the same type of the wind turbine. Therefore in some embodiments
𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝑖
) and 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡𝑖 10 ) are related to a wind turbine of the same type of
the considered wind turbine, and the relation S is based on the wind turbine of the
same type. Therefore, in some embodiments of the present disclosure, relations
based on S are based on a wind turbine of the same type.
[0070] In yet other alternative embodiments, the sequence of ordered pairs S may
15 be obtained by a simulation of the wind turbine and therefore a relationship
between measured wind characteristics and estimated wind characteristics based
on S may be obtained by simulation.
[0071] It is intended that values in S are not based on a significant stall condition
or disturbed condition of the wind turbine, i.e. for all time instants or time
intervals 𝑡1, 20 𝑡2, …,𝑡𝑛 the wind turbine is not in a significant stall condition or
disturbed condition. In embodiments in which a wind turbine of the same type of
the wind turbine is considered for obtaining S, it is intended that the wind turbine
of the same type is not in a significant stall condition or disturbed condition for all
time instants or time intervals 𝑡1,
𝑡2, … ,𝑡𝑛. In embodiments where S is obtained by
25 simulation, a significant stall condition or disturbed condition of the wind turbine
is not simulated and for all simulated time instants or time intervals 𝑡1,
𝑡2, … ,𝑡𝑛 a
20
significant stall condition or disturbed condition of the wind turbine is not
simulated.
[0072] Increasing the number n of time instants 𝑡1,
𝑡2, … ,𝑡𝑛 in a calibration phase
under various wind characteristics, the number of ordered pairs in the sequence S
5 increases and for each possible output 𝜔 of the sensor for measuring wind
characteristics at the wind turbine location, e.g. for each possible output 𝜔 of the
anemometer, typically either some pairs in the sequence S have 𝜔 as the first
component, or have a first component that is close to 𝜔. In some embodiments,
interpolation or regression may be alternatively used for obtaining missing data.
[0073] Let 𝑇[𝜔] = {𝑡𝑖
|𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝑖 10 ) ≅ 𝜔} be the set of time instants for which
𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 is equal to or close to 𝜔, with 𝜔 a possible output of the sensor for
measuring wind characteristics, e.g. a local anemometer. The set 𝑇[𝜔] is an
ordered set and it can be written 𝑇[𝜔] = {𝑡𝜔,1,𝑡𝜔,2, … }.
[0074] The symbol ≅ indicates equality or quasi equality where two scalars or
15 vectors are considered to be equal or quasi equal if their distance according to a
suitable metric is below a fixed bound. This fixed bound may be determined based
upon the characteristics of the sensor for measuring wind characteristics, e.g. upon
a variance affecting the outputs of said sensor and/or upon the tolerance of
components or parts included in the wind turbine.
20 [0075] Let 𝑆[𝜔] be the subsequence of S containing exactly those pairs in S for
which the first component is equal to 𝜔 or close to 𝜔. For sufficiently large n, the
subsequence 𝑆[𝜔] is expected to be nonempty and will contain in an ordered way
all ordered pairs in the sequence S having as first component a value equal to or
close to 𝜔, i.e.
25 𝑆[𝜔]
= ((𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝜔,1), 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡𝜔,1) ), (𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝜔,2), 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡𝜔,2) ), … )
with 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝜔,1) ≅ 𝜔, 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑(𝑡𝜔,2) ≅ 𝜔, ….
21
[0076] An expected value 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔) is then associated to 𝑆[𝜔], as an
expected value of the sequence 𝑆𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑[𝜔] ≔ (𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡𝜔,1) ,
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡𝜔,2), … ) obtained from 𝑆[𝜔] replacing the ordered pairs in 𝑆[𝜔] by
their second components. The expected value 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔) may e.g. be an
5 arithmetic average, or a geometric average or a median of the sequence
(𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡𝜔,1) , 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑡𝜔,2), …). In some alternative embodiments
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔) may be obtained from S by interpolation or regression.
[0077] Therefore, the expected value 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔) forms an expected value for
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 when the output value of the wind characteristic sensor measuring a
10 wind characteristics at the wind turbine location is or is close to 𝜔, e.g. when an
anemometer outputs 𝜔 or a value close to 𝜔. The expected value 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔) is
an expected wind characteristic value for the estimated wind characteristics,
determined from the measured wind characteristics, and is determined based on
the sequence S. The notation 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 remarks that an expected value of the
15 estimated wind characteristics is indicated. Therefore, 𝐸 indicates the expectation.
In alternative embodiments where a wind turbine of the same type is used,
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔) is related to the wind turbine of the same type, i.e. 𝜔 refers to a
possible output of e.g. an anemometer or wind characteristics sensor of the wind
turbine of the same type. In alternative embodiments where a wind turbine is
20 simulated, 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔) is related to the simulated wind turbine, i.e. 𝜔 refers to a
possible output of e.g. a simulated anemometer or simulated wind characteristics
sensor of the simulated wind turbine.
[0078] S and/or 𝑆[𝜔] and/or 𝑆𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑[𝜔] and/or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔) may be stored
as functions of 𝜔 with the use of any suitable data structure and on any suitable
25 device and/or medium and/or with the use of any suitable system. In particular
vectors containing pairs or vectors containing pairs of scalars or pairs of vectors
or lists containing pairs or lists containing pairs of scalars or pairs of vectors or
hash tables, or any nested combination of said data structures may be used,
wherein said data structures may be stored on and manipulated by any suitable
22
memory or computer or medium, either remotely or locally at the wind turbine
location. Related data may be transmitted on e.g. a network or a transmission line,
on one or more cables and/or on one or more waveguides or with the use of a
wireless communication system. The data structures may be stored permanently
5 or only for a required time interval, e.g. data structures implementing instances of
S and/or 𝑆[𝜔] and/or 𝑆𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑[𝜔] may be deleted once an instance of
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔) is obtained e.g. in a calibration phase considering the wind turbine
or a wind turbine of the same type or a simulated wind turbine.
[0079] It is intended that 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 may form a transfer function from the wind
10 speed measured using the wind characteristics sensor, e.g. an anemometer
installed on the wind turbine, to the expected wind speed estimated from the
turbine behavior. In alternative embodiments where a wind turbine of the same
type is used, it is assumed that 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 for the wind turbine of the same type is
identical or close to the result that would be obtained for 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 on the actual
15 physical wind turbine. In alternative embodiments where a wind turbine is
simulated, it is assumed that 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 for the simulated wind turbine is identical
or close to the result that would be obtained for 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 on the actual physical
wind turbine.
[0080] The transfer function 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 may in particular be generated from S
20 considering situations where the wind turbine is known to be operating at or near
the optimum, e.g. during wind turbine validation and where a significant stall
condition and a disturbed condition are not present.
[0081]In some embodiments of the present disclosure, in a calibration phase, one
or more transfer functions 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,1,𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,2, … , 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝜈 may be
25 obtained, with 𝜈 ≥ 1, the one or more transfer functions forming a finite sequence
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 = (𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,1, 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,2, … ,𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝜈).
of transfer functions. Some transfer functions in the sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 may
be based on measurements related to the wind turbine, some other transfer
23
functions in 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 may be obtained based on measurements related to a
wind turbine of the same type of the wind turbine. Yet other transfer functions in
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 may be obtained based on simulations of the wind turbine, i.e.
measurements are replaced by simulations based e.g. on physical models of the
5 wind turbine. Also, different relationships in the sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 may be
related to different time periods, when e.g. calibration phases are repeated.
Different transfer functions, e.g. obtained with measurements and/or simulations,
e.g. related to the wind turbine or a wind turbine of the same type, may be
combined in order to form a single transfer function in the sequence
10 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄. The combination may e.g. be based on averaging, weighted
averaging, interpolation, etc. Furthermore, in embodiments where a sequence
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 is obtained, an overall transfer function 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 may be
obtained from the sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄, e.g. by averaging
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔)
15 = 𝐴𝑉𝐸𝑅𝐴𝐺𝐸((𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,1(𝜔), 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,2(𝜔), … ,𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝜈(𝜔)))
for any value 𝜔 in the domain of the transfer functions in the sequence, and
where AVERAGE may indicate any average, e.g. a weighted average wherein
e.g. more recently obtained transfer function receive a greater weight in the
averaging operation. AVERAGE may also indicate for example an arithmetic
20 mean or a geometric mean or a median. Missing data for some 𝜔 may, in some
embodiments, be obtained e.g. by interpolation or regression.
[0082] In some embodiments, 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔) identifies a transfer function, i.e. a
relationship between measured wind characteristics and the expected estimated
wind characteristics. A value 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 of estimated wind characteristics is in
25 relation with a measured value 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 of wind characteristics considering
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑, if and only if 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 = 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) holds. It is
intended that the relation 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 forms a transfer function. It is intended
therefore that the transfer function 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 identifies a relationship between the
24
measured wind characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 and the estimated wind characteristics
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑.
[0083] In some embodiments a transfer function 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 identifying a
relationship between the measured wind characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 and the
5 estimated wind characteristics 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 may be obtained by other means, e.g.
using at least in part interpolation and/or regression analysis and/or Monte Carlo
methods, based on measurements of 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 and computations of 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑
based at least in part on the state 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 of the wind turbine.
[0084]In some alternative embodiments 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 is alternatively obtained
10 considering a wind turbine of the same type or a simulated wind turbine.
[0085] A calibration phase for determining or adjusting 200 one or more wind
characteristic relationships, i.e. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 and/or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄, takes place when
it is known that a significant stall condition or a disturbed condition of the wind
turbine is not present, therefore during calibration 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 is close to the actual
15 value of the wind characteristics at the wind turbine. When a relationship is
determined or adjusted according to block 206, the relationship may be 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑
or a relationship in the sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄.
[0086] In some embodiments, in a calibration phase, a wind turbine with a wind
speed estimator as part of its controller software may be run in an environment
20 where it is known that the estimator performs as expected, i.e. where 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 is
close to an actual wind condition at the wind turbine location. For example, it is
ensured that the blades are clean, and e.g. an anemometer forming the wind
characteristics sensor of the wind turbine is functioning correctly. During said
calibration phase the data from e.g. the turbine anemometer, i.e. 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑, and
25 the data from the wind speed estimator, i.e. the 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 is collected and a
transfer function 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 between values of 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 and values of
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 is computed, e.g. as described above for some embodiments of the
present disclosure. The transfer function 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 allows the calculation of the
25
expected output from the estimator based on the wind speed measured by e.g. the
anemometer, i.e. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) is expected to be close to 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑.
[0087] FIG. 3 shows an operation phase 300 of a method for operating a wind
turbine according to some embodiments of the present disclosure, the operation
5 phase including measuring 302 wind characteristics with the wind characteristic
sensor, e.g. with the wind characteristics sensor 58, thereby obtaining wind
characteristics data; measuring 304 the state of the wind turbine with at least one
wind turbine state sensor and determining estimated wind characteristics from the
measured state of the wind turbine and parameters of the wind turbine.
10 [0088]The operation phase 300 further includes comparing 306 the estimated
wind characteristics to an expected wind characteristic determined from the
measured wind characteristics, wherein the expected wind characteristics is
determined based on the one or more wind characteristics relationships, i.e.
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 and/or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄, e.g. based on one or more relationships between
15 wind characteristics determined or adjusted 200 in one or more calibration phases
, as indicated e.g. by the block 206. The operation phase 300 further includes
operating or shutting down 308 the wind turbine based at least in part on the
comparison. A relationship determined or adjusted 206 as described in FIG. 2 may
form 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or an element/component in the sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 from
20 which the expected wind characteristics is determined in function of the measured
wind characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑, measured according to block 302.
[0089] During the operation phase, i.e. in particular when a calibration phase is
not performed, when the wind characteristics sensor for measuring a wind
characteristics outputs a value 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑, an expected value
25 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) determined e.g. based on the relationship 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or on
the sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 obtained in one or more calibration phase gives an
expected value of 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑.
26
[0090] The expected value 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) function of 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑
therefore mitigates the errors that directly affect 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 and, at least if a
significant stall condition and a disturbed condition of the wind turbine are not
present, the expected value 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) approximates 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 that
5 is as described close to the real wind characteristics at the wind turbine, whereas
𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 directly is typically affected by a significant error and differs
significantly from 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 and therefore from the true wind characteristics.
[0091] After a calibration, during an operation phase the measured wind
characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 are used for obtaining an expected wind characteristic
10 value determined from the measured wind characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑, the expected
value being e.g. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) based on 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑. Whenever e.g.
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) differs significantly from 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 something
unexpected may be occurring, and in particular a significant stall condition or a
disturbed condition of the wind turbine may occur that causes the two values to be
15 significantly different.
[0092] It is therefore beneficial to operate the wind turbine during an operation
phase based on the comparison of e.g. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) with 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑. In
particular the operation phase may follow one or more calibration phases. When
e.g. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) is close to 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 a significant stall condition or a
20 disturbed condition of the wind turbine may be absent and the wind turbine is
operated according to 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 , i.e. according to 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 =
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒) = 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒, 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒), because the wind
characteristic estimated from a state 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 of the wind turbine and/or from a
state 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 of the wind turbine together with parameters 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 of the wind
25 turbine is more accurate than the wind characteristic 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 measured by a
wind characteristics sensor and also than the expected wind characteristics,
e.g. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑).
[0093] In some embodiments, if the expected wind characteristics, e.g.
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), differs significantly from 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑, e.g. when a value of
27
a magnitude of a difference between e.g. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) and 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 is
above a predetermined threshold, e.g. a threshold between 0.5 m/s and 2 m/s, ,
then the wind turbine is likely in a significant stall condition or in a disturbed
condition and therefore the wind turbine may e.g. be operated according to
5 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), because the value 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 is in this case typically
unreliable and inaccurate whereas 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) being based on
𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 may possibly be more accurate. Alternatively and/or in dependence of
the magnitude of the difference, the wind turbine may be shut off completely in
order to prevent possible damage to the wind turbine. If 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑)
10 differs significantly from 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 the wind turbine may therefore be shut off or
halted or operated in a very conservative way to prevent damage, e.g. when a
value of a magnitude of a difference between 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) and
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 is above a predetermined threshold, e.g. a threshold between 0.5 m/s
and 2 m/s. In some embodiments said threshold may be any value greater than e.g.
15 0.5 m/s.
[0094] In some embodiments, during operation of the wind turbine after the
calibration phase has been completed, the turbine constantly calculates the
expected wind speed 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) in regular intervals, e.g. in real time
or near-real time, or e.g. hourly, daily, or weekly. If the expected estimated wind
speed 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 20 ) and the model-based estimated wind speed
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 differ by more than a certain threshold, e.g. a threshold between 0.5
m/s and 2 m/s, several actions might be taken, according to embodiments of the
present disclosure. In one embodiment, a turbine controller of the wind turbine
switches to use the expected estimated wind speed 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑)
25 obtained from 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑, e.g. from a local anemometer, instead of the modelbased estimate 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 as an input to the main controller. In some
embodiments, a message will be generated indicating that the turbine needs to be
inspected. In some embodiments, a turbine will switch to a safer mode of
operation that protects it from potential damage due to certain conditions such as
30 an increased pitch angle to avoid stall. In some embodiments, the pattern of the
28
mismatch between expected value 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) and actual value
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 is compared against pre-computed or pre-determined fault patterns
stored in a software or a memory related to the wind turbine or the wind turbine
controller and, in some embodiments, an action is taken based on the particular
5 fault pattern.
[0095] More generally; embodiments of the present disclosure relate to a method
for operating a wind turbine, the wind turbine including a wind characteristics
sensor for measuring a wind characteristic 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑and at least one wind turbine
state sensor for measuring a state 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒of the wind turbine, the method
10 comprising: determining or adjusting one or more wind characteristics
relationships, i.e. a relationship 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or a sequence of relationships
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄; and, performing an operation phase, the operation phase including:
measuring the wind characteristics with the wind characteristics sensor, thereby
obtaining measured wind characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑; measuring the state
15 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 of the wind turbine with the at least one wind turbine state sensor and
determining an estimated wind characteristics 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 from the measured state
of the wind turbine and parameters of the wind turbine 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 =
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒) = 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒, 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒),; comparing the estimated
wind characteristics 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 to an expected wind characteristics
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 20 ) determined from the measured wind characteristics
𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑, wherein the expected wind characteristics 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) is
determined based on the one or more wind characteristics relationships, i.e. on
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or on 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄; and, operating or shutting down the wind turbine
based at least in part on the comparison result.
[0096] For example, if the expected wind characteristics 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 25 )
is based on a sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 =
(𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,1, 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,2, … , 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝜈) of relationships obtained e.g.
considering the wind turbine and/or a wind turbine of the same type and/or
29
simulations, then the expected wind characteristics 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) may be
obtained by averaging
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑)
= 𝐴𝑉𝐸𝑅𝐴𝐺𝐸((𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,1(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑),𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,2(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), … ,𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝜈(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑)))
5 [0097] In some embodiments; interpolation or regression may be used.
[0098] In some embodiments, determining or adjusting one or more wind
characteristics relationships, i.e. determining or adjusting 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or one or
more relations 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑖
in a sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 and therefore determining
or adjusting the sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄, is performed when the wind turbine is
10 not in a significant stall condition and not in a disturbed condition and includes:
measuring the wind characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 of the wind turbine with the wind
characteristics sensor of the wind turbine, thereby obtaining measured wind
characteristics of the wind turbine; measuring the state 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 of the wind
turbine with the at least one wind turbine state sensor and determining an
15 estimated wind characteristics of the wind turbine from the measured state of the
wind turbine and parameters of the wind turbine 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 =
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒, 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 ), determining or adjusting a relationship 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑
or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑖
, with i indicating the i-th relationship currently determined or
adjusted, between the measured wind characteristics of the wind turbine and the
20 estimated wind characteristics of the wind turbine; and adjusting the one or more
wind characteristics relationships, i.e. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 to include the
relationship 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑖 between the measured wind characteristics
of the wind turbine and the estimated wind characteristics of the wind turbine.
[0099] In some embodiments, determining or adjusting one or more wind
25 characteristics relationships i.e. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 includes: operating a
wind turbine of the same type as the wind turbine when the wind turbine of the
same type is not in a significant stall condition and not in a disturbed condition,
the wind turbine of the same type including a wind characteristic sensor and at
30
least one wind turbine state sensor; during the operation of the wind turbine of
the same type, measuring wind characteristics of the wind turbine of the same
type with the wind characteristics sensor of the wind turbine of the same type,
thereby obtaining measured wind characteristics of the wind turbine of the same
5 type; and measuring the state of the wind turbine of the same type with the at least
one wind turbine state sensor of the wind turbine of the same type and
determining an estimated wind characteristics of the wind turbine of the same type
from the measured state of the wind turbine of the same type and parameters of
the wind turbine of the same type; determining or adjusting a relationship between
10 the measured wind characteristics of the wind turbine of the same type and the
estimated wind characteristics of the wind turbine of the same type; and adjusting
the one or more wind characteristics relationships i.e. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄
to include the relationship between the measured wind characteristics of the wind
turbine of the same type and the estimated wind characteristics of the wind turbine
15 of the same type.
[0100] In some embodiments, determining or adjusting one or more wind
characteristics relationships i.e. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 includes: simulating a
wind and a wind turbine operation for the wind turbine without a significant stall
condition and without a disturbed condition of the wind turbine, the simulation
20 being based at least in part on a model of the wind turbine; obtaining simulated
wind characteristics, simulated state and simulated parameters of the wind turbine,
determining simulated estimated wind characteristics from the simulated state of
the wind turbine and the simulated parameters of the wind turbine; determining or
adjusting a relationship between the simulated wind characteristics and the
25 simulated estimated wind characteristics; and, adjusting the one or more wind
characteristics relationships i.e. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 to include the
relationship between the simulated wind characteristics and the simulated
estimated wind characteristics.
31
[0101]In some embodiments, the one or more wind characteristics relationships,
i.e. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄, are further combined into a single combined
relationship, and the expected wind characteristics is based on the single
combined relationship. For example for a sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 of relationships
5 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 = (𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,1, 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,2, … , 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝜈) obtained e.g.
considering a wind turbine and/or a wind turbine of the same type and/or
simulations, a single combined relationship may associate to each value 𝜔 in the
domain of the relationships in the sequence, an average value
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔) =
10 𝐴𝑉𝐸𝑅𝐴𝐺𝐸((𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,1(𝜔),𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,2(𝜔), … ,𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝜈(𝜔))) In some
embodiments missing data for some 𝜔 may be obtained e.g. with interpolation or
regression.
[0102]In some embodiments, during normal turbine operation, i.e. during the
operational phase, the data originating from the wind characteristics sensor, e.g.
15 the anemometer, and the estimated wind speed is continuously evaluated and
compared using e.g. the determined transfer function in a statistical sense. If the
match between the expected estimated wind characteristics
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) and the estimated wind characteristics 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 cannot
be obtained with predetermined requirements, then the wind turbine is assumed to
20 be not operating as intended, e.g. due to icing, blade fouling, or stall, and a
message to a remote control center is generated so that appropriate steps to
remedy the problem can be taken.
[0103] FIG. 4 summarizes details related to a method for operating a wind
turbine according to some embodiments of the present disclosure. FIG. 4 shows
25 that a wind characteristics sensor 402, that may be e.g. the wind characteristics
sensor 58, provides measured wind characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑, as indicated by 406,
and at least one sensor measuring 404 the state of the wind turbine provides a
measure of the state 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒, of the wind turbine as indicated by 408, wherein the
state may e.g. include a rotational speed or a torque of a rotor and/or of a shaft of
32
the wind turbine, e.g. the rotor shaft 44 and/or include e.g. a power output of the
generator. Parameters 410 of the wind turbine are assumed to be known, e.g. pitch
angles of blades of the wind turbine are assumed to be known and/or the
configuration of the gearbox. The parameters 410 are indicated with 𝑝𝑡𝑢𝑟𝑏𝑖𝑛𝑒 , as
5 indicated by 412. With the use of physical models 416 estimated wind
characteristics 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑, as shown by 420, are obtained through the physical
models 416 in function of the state of the wind turbine 408 and the parameters
412 of the wind turbine. From the measured wind characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑,
indicated by 406, an expected value, e.g. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) of the estimated
10 wind characteristics is obtained based on one or more relationships, i.e.
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 as indicated schematically by 414. The expected
value 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) is indicated by 418. A comparison 422 is carried out
between the expected estimated wind characteristics𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑)
indicated by 418 and the estimated wind characteristics 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 indicated by
15 420. The wind turbine is finally operated or a shutdown is carried out, as indicated
by 424, based at least in part on the comparison 422. The wind turbine operation
or shutdown 424 may be based on the comparison 422 between
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) and 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 and in particular the wind turbine
operation may further depend on the values of 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) and/or
20 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 and/or on a selection of one of the values e.g. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑),
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,the selection being based on the result of the comparison 422.
[0104] Methods of the present disclosure are directed to a calibration of the wind
characteristics sensor, i.e. of a wind measurement device, under the consideration
of physical models from which 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 is obtained during normal operation
25 times. The wind characteristics sensor, i.e. the wind measurement device at the
wind turbine location, is used to detect an under-performance of the wind turbine
and/or a misbehavior of the wind turbine, e.g. due to a significant stall condition
or a disturbed condition of the wind turbine in particular when the estimate
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 becomes inaccurate, i.e. when the physical model-based wind speed
30 estimation algorithms do not work properly anymore.
33
[0105] Methods of the present disclosure are beneficial in particular for ice
detection, stall detection, the possibility to perform seasonal calibration phases.
Due to the precision and accuracy of 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) it is furthermore
possible to obtain a reliable power curve measurement with e.g. a nacelle
5 anemometer forming a wind characteristics sensor of the wind turbine.
[0106] In some embodiments of the present disclosure, a method for operating a
wind turbine is described wherein, when the comparison between e.g.
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) and 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 shows that the estimated wind
characteristics 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 differ significantly from the expected wind
10 characteristics value, e.g. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), determined from the measured
wind characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑, the wind turbine is operated according to the
expected wind characteristics 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) determined from the
measured wind characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 or is shut down.
[0107] In some embodiments the comparison 306, 422 may include obtaining a
15 difference ∆ between the estimated wind characteristics 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 and the
expected wind characteristics 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) based on the one or more
relationships i.e. on 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄. It is intended that ∆ may be ∆=
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) where 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 may be the relationship
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or be based on the sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄. In some embodiments
20 based on a sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄, with 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 =
(𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,1, 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,2, … , 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝜈), the difference ∆ may be ∆=
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 −
𝐴𝑉𝐸𝑅𝐴𝐺𝐸((𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,1(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑),𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,2(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), … , 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝜈(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑)))
where 𝐴𝑉𝐸𝑅𝐴𝐺𝐸 may indicate any convenient average. Relationships 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑖
25 in the sequence 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄 may be obtained considering the wind turbine or a
wind turbine of the same type or by simulation. For uniformity of notation, it is
still written:
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) =
𝐴𝑉𝐸𝑅𝐴𝐺𝐸((𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,1(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑),𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,2(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), … , 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝜈(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑)))
34
. In some embodiments, the wind turbine is operated based at least on part on a
magnitude of the difference ∆.
[0108] In some embodiments the comparison 422 may correspond to the
comparison 306 and includes obtaining a difference ∆, e.g. a difference
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 5 ), between the estimated wind characteristics
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 and the expected wind characteristics, e.g. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), and
operating the wind turbine is based at least in part on a magnitude of the
difference ∆.
[0109] The difference ∆ may be a scalar or a vector and the magnitude of the
difference, e.g. of the difference 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 10 ), may be
measured by any suitable metric or norm, in particular by e.g. a Euclidean norm, a
maximum norm, etc. In particular the magnitude of the difference ∆, e.g.
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), is intended to be a non-negative real number
and the magnitude of said difference ∆ is zero if, and only if, the scalar or
vectorial operands, e.g. 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 and 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 15 ), are equal.
[0110] In some embodiments of the present disclosure, when the magnitude of
the difference ∆, e.g. of the difference 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), is
below a first threshold, for example below 2 m/s or below 1 m/s, the wind turbine
is operated based on the estimated wind characteristics 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑.
20 [0111] For example, in some embodiments, when 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 is close to
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), the magnitude of the difference 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 −
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) comes close to zero and therefore the magnitude of the
difference 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) is below a first threshold
when 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 is close to 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑). In such a condition a
25 significant stall condition or a disturbed condition of the wind turbine is not
expected, and therefore the wind turbine is operated according to 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 in
particular when 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 may be more accurate than 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑)
and/or 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑.
35
[0112] In some embodiments, when the magnitude of the difference ∆, for
example of the difference 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), is above the first
threshold, the wind turbine is operated based on the expected wind characteristic
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑). For example, when 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 differs significantly from
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑) the magnitude of 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 5 )
increases above the first threshold, and the wind turbine is likely in a stall or
disturbed condition and therefore likely the value of 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 becomes
unreliable and inaccurate. Therefore it is beneficial to operate the wind turbine
according to the expected wind characteristic value, e.g. according to
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 10 ), for a wind turbine operation and/or for a safe operation of
the wind turbine in order to prevent damage and/or for a shutdown of the wind
turbine.
[0113] In some embodiments, the turbine is switched to a safe mode of operation
when the magnitude of the difference ∆, e.g. the difference 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 −
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 15 ), is above a second threshold, for example above 3 m/s..
[0114] The safe mode may be related to a control of one or more pitch angles of
one or more blades of the wind turbine in order to prevent a significant stall
condition or the safe mode may include shutting down the wind turbine
completely.
20 [0115] In some embodiments, a message is transmitted to an operator when the
magnitude of the difference ∆, e.g. the difference 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 −
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), is above the first and/or the second threshold.
[0116] The transmission may be fully automated and the operator may be one or
more human operators and/or one or more computers or fully or partially
25 automated systems configured to control the wind turbine. The operator or the one
or more computers or the fully or partially automated systems configured to
control the wind turbine may be located in a wind park, or in a remote location or
even located at the wind turbine location or in the wind turbine itself. The
36
message may be transmitted by any suitable means, such as digital packets on a
network, or as modulated radio wave signals, or on a cable or optical waveguide.
The message may contain any additional information that is beneficial for the
control of the wind turbine and/or for obtaining information about the wind
5 turbine state or conditions.
[0117] In some embodiments, the magnitude of the difference ∆, e.g. the
difference 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), is memorized at different time
instants forming a sequence, and based on said sequence a normal condition or a
significant stall or a disturbed condition is determined, wherein in case of a
10 significant stall or disturbed condition a type of fault is determined from the
sequence , and the wind turbine is operated according to the determined type of
fault.
[0118] Memorizing the difference ∆, e.g. the difference 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 −
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), at different time instants, e.g. periodically sampling and
15 storing said difference, produces a sequence of values forming a history of the
difference. Based on said history, it is e.g. possible to record how the magnitude
of the difference ∆, e.g. the difference 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑),
increases when, for example, a significant stall condition or a disturbed condition
occurs. From the history of the difference ∆, information on the type of fault can
20 be obtained, where a type of fault may e.g. specify if a significant stall condition
is occurring, or what fault is occurring among different possible cases, e.g. with a
specification stating if blades are iced, or if dirt or aging is probably affecting the
wind turbine operation.
[0119] Other sources of information may as well be used in determining a type of
25 fault, e.g. information obtained from thermometers and/or other sensors placed
e.g. at or around the wind turbine location. Sources of information may as well
include weather forecasts or observations and wind forecasts or measurements at
different locations including the wind turbine location.
37
[0120] In some embodiments the wind turbine is shut down or operated in order
to control a pitch angle to avoid a stalling of the wind turbine, based on the
comparison of the estimated wind characteristics 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 to the expected wind
characteristic value, e.g. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), determined from the measured
5 wind characteristics 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑.
[0121]
[0122]In some embodiments, operating or shutting down the wind turbine
includes adjusting a pitch angle to avoid a significant stall condition of the wind
turbine.
10 [0123] The calibration phase may include repeated measurements of 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑
and of 𝑠𝑡𝑢𝑟𝑏𝑖𝑛𝑒 such that a sufficient number of ordered pairs is obtained in order
to obtain a sufficiently precise and accurate value for e.g. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝜔), for each
possible output value 𝜔 of the wind characteristic sensor. A sufficiently precise
and accurate value may be present when, for example, a narrow enough
confidence interval related to an average value of the sequence 𝑆𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 15 [𝜔]
obtained from 𝑆[𝜔] as described previously can be determined, e.g. considering
values in the sequence 𝑆𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑[𝜔] as samples of a Monte Carlo experiment for
which a desired width of the confidence interval is demanded for a desired
confidence level.
20 [0124] A calibration phase for determining or adjusting one or more wind
characteristics relationships may be carried out when it is known that a significant
stall condition or a disturbed condition is not present. Said determination may be
fully automated, e.g. automatically checking temperature and wind and other
conditions at the wind turbine location, like the presence of dirt or said
25 determination may be partially automated or be the result of a human supervision.
A calibration phase may include the use of measurement instruments that may be
removed after the calibration phase is completed. A human supervision may be
38
present during a calibration phase and absent afterwards or the calibration may be
fully automated.
[0125] Calibration phases and operation phases may alternate, e.g. periodically
alternating, in order to recalibrate, i.e. to adjust, e.g. 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 or 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑,𝑆𝐸𝑄
5 in order to account for e.g. an aging of the wind turbine or other time varying
properties of the wind turbine and/or to account for modifications in the
aerodynamic properties of the wind turbine location.
[0126] In some embodiments, the calibration phase is repeated until the expected
wind characteristics value approximates the estimated wind characteristics with
10 sufficient accuracy and precision.
[0127] In some embodiments during the calibration phase, the measured wind
characteristics further comprise measured data from one or more wind
measurement masts positioned at some distance from the wind turbine.
[0128] Therefore, in some embodiments, 𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 may be a vector including
15 values obtained from at least one local anemometer and/or at least one
measurement mast for measuring a wind condition at some distance from the wind
turbine.
[0129] In some embodiments, a wind turbine is described, the wind turbine
including: at least one wind measurement sensor, a wind turbine state sensor to
20 measure a state of the wind turbine for estimating wind characteristics at the wind
turbine location, a control system configured to control the wind turbine based at
least in part on inputs formed by measured wind characteristics measured by the
wind measurement sensor, and by measured wind turbine states measured by the
wind turbine state sensor, wherein the control system is configured to operate the
25 wind turbine according to methods described in the present disclosure. It is
assumed that wind turbine parameters are known by the control system.
39
[0130] In some embodiments, the wind characteristic may be a wind speed or a
magnitude of a wind speed and the wind characteristics sensor measures a
magnitude of the wind speed or the wind speed. In some embodiments, the wind
characteristics sensor may measure a magnitude and a direction of the wind speed.
5 In some embodiments the wind characteristics sensor may measure a vector
describing the wind speed. In some embodiments the magnitude of the wind speed
may be measured in m/s.
[0131]In some embodiments, the wind turbine further includes an information
processing system and at least one communication channel configured to transmit
10 information about the comparison of the estimated wind characteristics to the
expected wind characteristics during the operation phase.
[0132]The transmitted information transmitted, for instance over the
communication channel, may be used to control or monitor the wind turbine
operation.
15 [0133] Operating a wind turbine according to methods described herein or as
illustrated in FIG. 4 is beneficial in particular for detecting a significant stall
condition or a disturbed condition and allows an operation of the wind turbine that
minimizes the risks of damage to the wind turbine and/or that maximizes a
performance of the wind turbine in particular during a significant stall condition
20 or a disturbed condition, such as an increased magnitude of the difference ∆, e.g.
the difference 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), may e.g. be related to dirt
deposited on the blades of the wind turbine or to icing or to any other disturbed
condition and/or to a significant stall condition. Methods of the present disclosure
may allow to detect the disturbed condition and the significant stall condition,
25 according to e.g. a history of the magnitude of the difference ∆, e.g. the difference
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), memorized at different time instants in a
sequence. An increased magnitude of the difference ∆, e.g. the difference
𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 − 𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), may also be related to e.g. a significant stall
condition that may also be detected e.g. according to a sequence of values of the
40
magnitude of the difference ∆, e.g the difference 𝑤𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 −
𝐸𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑(𝑤𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑), memorized at different time instants. In this case, i.e. if a
significant stall condition is detected, it is beneficial to adjust a pitch angle of one
or more blades of the wind turbine or to shut down the wind turbine in order to
5 prevent e.g. a damage to the wind turbine.

WE CLAIM
1. Method for operating a wind turbine, the wind turbine including a wind
characteristics sensor for measuring a wind characteristic and at least one wind
5 turbine state sensor for measuring a state of the wind turbine, the method
comprising:
determining or adjusting one or more wind characteristics relationships;
and,
performing an operation phase, the operation phase including:
10 measuring the wind characteristics with the wind characteristics
sensor, thereby obtaining measured wind characteristics;
measuring the state of the wind turbine with the at least one wind
turbine state sensor and determining an estimated wind characteristic
from the measured state of the wind turbine and parameters of the
15 wind turbine;
comparing the estimated wind characteristics to an expected wind
characteristics determined from the measured wind characteristics,
wherein the expected wind characteristics is determined based on the
one or more wind characteristics relationships; and,
20 operating or shutting down the wind turbine based at least in part on
the comparison result.
2. The method of claim 1, wherein determining or adjusting one or more wind
characteristics relationships is performed when the wind turbine is not in a
25 significant stall condition and not in a disturbed condition and comprises:
42
measuring the wind characteristics of the wind turbine with the wind
characteristics sensor of the wind turbine, thereby obtaining measured wind
characteristics of the wind turbine;
measuring the state of the wind turbine with the at least one wind turbine
5 state sensor and determining an estimated wind characteristics of the wind
turbine from the measured state of the wind turbine and parameters of the
wind turbine,
determining or adjusting a relationship between the measured wind
characteristics of the wind turbine and the estimated wind characteristics of
10 the wind turbine; and
adjusting the one or more wind characteristics relationships to include the
relationship between the measured wind characteristics of the wind turbine
and the estimated wind characteristics of the wind turbine.
15 3. The method of claim 1 or 2, wherein determining or adjusting one or more
wind characteristics relationships comprises:
operating a wind turbine of the same type as the wind turbine when the wind
turbine of the same type is not in a significant stall condition and not in a
disturbed condition, the wind turbine of the same type including a wind
20 characteristic sensor and at least one wind turbine state sensor; and, during
the operation of the wind turbine of the same type, the method further
including:
measuring wind characteristics of the wind turbine of the same type
with the wind characteristics sensor of the wind turbine of the same
25 type, thereby obtaining measured wind characteristics of the wind
turbine of the same type; and
43
measuring the state of the wind turbine of the same type with the at
least one wind turbine state sensor of the wind turbine of the same
type and determining an estimated wind characteristics of the wind
turbine of the same type from the measured state of the wind turbine
5 of the same type and parameters of the wind turbine of the same type;
determining or adjusting a relationship between the measured wind
characteristics of the wind turbine of the same type and the estimated
wind characteristics of the wind turbine of the same type; and,
adjusting the one or more wind characteristics relationships to include
10 the relationship between the measured wind characteristics of the
wind turbine of the same type and the estimated wind characteristics
of the wind turbine of the same type.
4. The method of any of the preceding claims, wherein determining or adjusting
15 one or more wind characteristics relationships comprises:
simulating a wind and a wind turbine operation for the wind turbine
without a significant stall condition and without a disturbed condition of
the wind turbine, the simulation being based at least in part on a model of
the wind turbine;
20 obtaining simulated wind characteristics, simulated state and simulated
parameters of the wind turbine, determining simulated estimated wind
characteristics from the simulated state of the wind turbine and the
simulated parameters of the wind turbine;
determining or adjusting a relationship between the simulated wind
25 characteristics and the simulated estimated wind characteristics; and,
44
adjusting the one or more wind characteristics relationships to include the
relationship between the simulated wind characteristics and the simulated
estimated wind characteristics
5 5. The method of any of claims 1 to 4 wherein the one or more wind
characteristics relationships are further combined into a single combined
relationship, and wherein the expected wind characteristics is based on the single
combined relationship.
10 6. The method of any of the preceding claims, wherein, when the comparison
shows that the estimated wind characteristics differs significantly from the
expected wind characteristics determined from the measured wind characteristics,
the wind turbine is operated according to the expected wind characteristics
determined from the measured wind characteristics or is shut down.
15
7. The method of any of the preceding claims, wherein the comparing includes
obtaining a difference between the estimated wind characteristics and the
expected wind characteristics and operating the wind turbine is based at least in
part on a magnitude of the difference.
20
8. The method of claim 7, wherein, when the magnitude of the difference is below
a first threshold, the wind turbine is operated based on the estimated wind
characteristics.
45
9. The method of claim 7 or 8, wherein, when the magnitude of the difference is
above the first threshold, the wind turbine is operated based on the expected wind
characteristics.
5 10. The method of any of claims 7 to 9, wherein the turbine is switched to a safe
mode of operation or is shut down when the magnitude of the difference is above
a second threshold.
11. The method of any of claims 7 to 10, wherein a message is transmitted to an
10 operator when the magnitude of the difference is above the first and/or the second
threshold.
12. The method of any of claims 7 to 11, wherein the magnitude of the difference
is memorized at different time instants forming a sequence, and wherein based on
15 said sequence a normal condition, or a significant stall or disturbed condition, is
determined, and wherein in case of a significant stall or disturbed condition a type
of fault is determined from the sequence, and the wind turbine is operated
according to the determined type of the fault.
20 13. The method of any of claims 1 to 12, wherein operating or shutting down the
wind turbine includes adjusting a pitch angle to avoid a significant stall condition
of the wind turbine.
46
14. The method of any of claims 1 to 13 wherein the wind characteristic is a wind
speed and the wind characteristics sensor measures a magnitude of the wind
speed.
5 15. A wind turbine comprising
at least one wind measurement sensor; and
a wind turbine state sensor to measure a state of the wind turbine for
estimating wind characteristics at the wind turbine location;
a control system configured to control the wind turbine based at least in part
10 on inputs formed by measured wind characteristics measured by the wind
measurement sensor, and by measured wind turbine states measured by the
wind turbine state sensor,
wherein the control system is configured to operate the wind turbine according to
the method of any of claims 1 to 14.