Detecting wind turbine performance change

Field of invention

The present invention relates to a method of detecting a per formance change of a wind turbine, relates to an arrangement for detecting a performance change of a wind turbine and fur ther relates to a wind turbine comprising the arrangement.

Art Background

A wind turbine produces electrical power which is supplied to a utility grid. Thereby, conventionally, the control strategy for a wind turbine is designed to maximize power production output to the utility grid while minimizing structural loads and acoustic noise emissions. To achieve this, the control strategy conventionally relies on the designed aerodynamic capabilities of the wind turbine.

However, during the lifetime of the wind turbine, the aerody namic capabilities may change due to interfering external in fluences. If the actual aerodynamic capabilities deviate sig nificantly from the designed (or original) aerodynamic capa bility (due to for example environmental conditions, model inaccuracy, fouling of turbine blades, etc.), the wind tur bine may suffer from a significant performance loss (for ex ample in terms of power production as well as load and noise emission) .

Conventionally, a performance loss may have been detected by comparing an average power production over a period of time to a predetermined or expected power. This conventional meth od may be relatively slow and conservative. Stall detection of a wind turbine blade has been described relying on a sen sor to detect vibrations in the blade. The conventional meth ods of detection of a performance loss are either relatively slow, conservative, unreliable or require special sensor equipment not readily available on a wind turbine.

Thus, there may be a need for a method and an arrangement of detecting a performance change of a wind turbine which can be easily implemented, is performed in a fast manner (such as to allow online application), is reliable and/or has a better accuracy of conventionally known methods and arrangements.

Summary of the Invention

This need may be met by the subject matter according to the independent claims. Advantageous embodiments of the present invention are described by the dependent claims.

According to an embodiment of the present invention it is provided a method of detecting or estimating a performance change or fault or erroneous condition of a wind turbine, the method comprising: measuring a value of a wind related quan tity (e.g. wind speed or output power or generator torque); estimating a value of the wind related quantity based on at least wind turbine operational data and applying a (wind tur bine performance) model (including a model of expected per formance) ; comparing the measured value with the estimated value to derive a residual value of the wind related quanti ty; and indicating a performance change based on the residual value, the method in particular being performed for plural consecutive time steps.

The method may detect whether a performance change, in par ticular a performance loss or degradation and/or a fault and/or an erroneous condition of the wind turbine, is present or not. Further, the method may according to a particular em bodiment of the present invention also estimate a performance change, in particular performance degradation of the wind turbine, in the sense that the degree of performance change may be determined. The degree of performance change or per formance degradation may for example be quantified in several categories such as low performance degradation, high perfor mance degradation and very high performance degradation or no performance degradation.

The wind related quantity is a quantity which is affected by wind characteristics, in particular wind speed. The wind re lated quantity may be dependent on the wind speed. The value of the wind related quantity may be measured by a respective sensor or measurement device installed at or close to the wind turbine.

The wind turbine operational data may relate to data defining an operational state of the wind turbine including environ mental influence factors, including wind condition.

The estimation may be performed by a processor which is ap propriately programmed or which has a program loaded which is executed by the processor. The model may comprise a wind tur bine performance model, for example relating the wind related quantity to the wind turbine operational data. Thus, given the wind turbine operational data, the value of the wind re lated quantity may be estimated or derived, for example using a mathematical formula or a look-up table or any other data structure relating for example the wind turbine operational data with the wind related quantity. The model may have been previously derived based for example on experimental data of an intact errorless wind turbine. The model may reflect the expected performance. The performance change may be a change of the performance from the expected performance to a current performance .

The measured value may for example be compared with the esti mated value by forming a difference between the measured val ue and the estimated value. If there is no performance change, the measured value may substantially be equal or cor respond to the estimated value, at least on average. However, if there is a performance change, the measured value may de viate from the estimated value, in particular the stronger the performance change is. The performance change may be in dicated by an acoustic, optical and/or haptic signal. For ex ample, an alarm (acoustic, optical) may be raised if the method detects a performance change, in particular a perfor mance change having a predetermined strength or level or de gree .

All method steps may be continuously performed for plural consecutive time steps. The method may be performed online during operation of the wind turbine or may be performed of fline, thereby for example processing data previously ob tained from one or more wind turbines and stored.

The estimation of the value of the wind related quantity may for example be performed as has been described in

US 2012/0078518 A1. Thereby, the wind speed of a wind driving a wind turbine is described to be estimated including measur ing a current power generated by the wind turbine, measuring a current rotor speed of a rotor of the wind turbine and measuring a current blade pitch angle of a rotor blade of the rotor. According to this embodiment, the wind related quanti ty is the wind speed. For example, a three-dimensional refer ence look-up table comprising three orthogonal axes may be pregiven, wherein the first axis denotes a rotor speed, the second axis denotes a wind speed and the third axis denotes the blade pitch angle. The values of the three-dimensional reference look-up table may denote the power which can be generated by the wind turbine depending on the wind speed, the rotor speed and the blade pitch angle. The reference look-up table may also be called or referred to as Cp look-up table. The look-up table may have been determined previously by monitoring operation of the intact wind turbine. In other embodiments, the model may be implemented in a manner differ ent from a look-up table.

According to an embodiment of the present invention, the wind related quantity comprises one of: a wind speed; a wind tur bine output power; a wind turbine generator torque. In par- ticular, the wind related quantity may be either one of a wind speed, a wind turbine output power, a wind turbine gen erator torque. Any of these examples of wind related quanti ties may be measured as well as may be estimated. To each of these examples of wind related quantities, respective corre sponding wind turbine operational data may be provided based on which the respective example of the wind related quantity may be derived or estimated. It may be advantageous to use the wind speed as the wind related quantity, since for the wind speed, a reliable estimation algorithm may be known, as has been described above.

According to an embodiment of the present invention, the wind turbine operational data comprise at least one of: a wind turbine output power; a rotational speed of the generator or rotor; a rotor blade pitch angle; a wind speed. If the wind related quantity is the wind speed, for example the opera tional data may include values for the power output, the ro tational speed and the pitch angle. In other cases of the wind related quantity, for example being the wind turbine power or the generator torque, also the wind speed may be in cluded in the operational data. Thereby, these wind turbine operational data may be easily available in a conventional wind turbine and thus enable to implement the method in a simple manner.

According to an embodiment of the present invention, the re sidual value of the wind related quantity is determined by subtracting the estimated value from the measured value.

Thereby, a difference between the estimated value and the measured value of the wind related quantity may be derived. The difference (or the residual value) may be indicative of the presence or the absence of a performance change.

The performance change may for example in other embodiments be a performance gain in which case the residual value may have an opposite sign to the case where the performance change is a performance loss. Thus, the sign of the residual value may be indicative of whether the performance change is a performance gain or a performance loss.

According to an embodiment of the present invention, the method further comprises filtering, in particular low pass filtering and/or applying a moving average filter, the resid ual value over several time steps to obtain, in particular for each time step, a filtered residual value, wherein indi cating a performance change is based on the filtered residual value .

By filtering, high frequency components of the residual value may be decreased in amplitude or even be completely removed. Thereby, measurement errors or very rapid changes of the per formance change may be disregarded, thereby increasing the accuracy and/or reliability of the method. Further, the num ber of false positives may be reduced by the filtering.

According to an embodiment of the present invention, the method further comprises deriving a statistical characteris tics of the residual value or the filtered residual value over plural time steps, in particular including a mean and/or a standard deviation; and comparing the derived statistical characteristics with an expected statistical characteristics, based on which the performance change is detected.

The statistical characteristics may include one or more sta tistical parameters which describe the development or evolve- ment or distribution of the residual value over several time points or time steps. Considering the statistical character istics of the residual value may be statistically more rele vant and indicative than considering the residual value as such even for different time points. In particular, fluctua tions due to measurement errors may thereby be eliminated.

The expected statistical characteristics may be derived from experimental data of an intact wind turbine under known envi ronmental conditions, in particular accurately known wind speed . According to an embodiment of the present invention, deriving the statistical characteristics includes deriving at least one residual cumulative sum by deriving at least one cumula tive sum of the residual value or the filtered residual val ue, wherein indicating a performance change is based on the at least one residual cumulative sum, wherein the at least one residual cumulative sum in particular comprises an upper residual cumulative sum and/or a lower residual cumulative sum.

The cumulative sum (CUSUM) is a conventionally known sequen tial analysis technique which is used for monitoring change detection. CUSUM was developed to determine changes in the mean of a value. The cumulative sum may be easily implementa- ble. Furthermore, the cumulative sum may reliably indicate a significant change in the residual value which may in turn allow to reliably determine or detect whether a performance change occurred.

According to an embodiment of the present invention, the method further comprises comparing the at least one residual cumulative sum, in particular for at least one time step, to a threshold to obtain a comparison result, wherein indicating a performance change is based on the comparison result.

For example, the upper cumulative sum may be compared to a high threshold and/or the lower cumulative sum may be com pared to a low threshold. If either or both of the upper cu mulative sum and the lower cumulative sum is above or below the high threshold or the low threshold, respectively, the presence of a significant performance change may be indicat ed. Furthermore, several high thresholds and several low thresholds may be predefined. If for example the upper cumu lative sum exceeds a first high threshold, it may be indicat ed that there is a low performance change, if the cumulative sum exceeds a second high threshold, it may be indicated that the performance change is of intermediate degree and if the cumulative sum exceeds a third high threshold, it may be in dicated that the performance change is of very high degree.

In other embodiments, a binary decision may be taken in the sense that only a single threshold is defined relating to the upper cumulative sum. If the upper cumulative sum for example exceeds the single threshold, it may be indicated that there is a significant performance change.

According to an embodiment of the present invention, the wind related quantity is a wind speed, wherein the wind turbine operational data comprise a wind turbine output power and a rotational speed of a rotor or a generator and a rotor blade pitch angle, wherein the performance model relates values of a wind turbine output power and a rotational speed of genera tor or a rotor and a rotor blade pitch angle to estimated values of the wind speed. Thereby, the method may apply the conventionally known wind speed detection method as is de scribed in US 2012/0078518 A1.

According to an embodiment of the present invention, the wind related quantity is a wind turbine output power, wherein the wind turbine operational data comprise a wind speed and a ro tational speed of a rotor or a generator and a rotor blade pitch angle, wherein the performance model relates values of a wind speed and a rotational speed of generator or a rotor and a rotor blade pitch angle to estimated values of the wind turbine output power.

According to a further embodiment of the present invention, the wind related quantity is a generator torque, wherein the wind turbine operational data comprise a wind speed and a ro tational speed of a rotor or a generator and a rotor blade pitch angle, wherein the performance model relates values of a wind speed and a rotational speed of generator or a rotor and a rotor blade pitch angle to estimated values of the gen erator torque. The wind turbine output power or the generator torque may be alternatives which may on one hand be measured and on the other hand be estimated, in particular using a same or simi lar relationship between the four quantities wind speed, wind turbine output power (or torque) , rotational speed and rotor blade pitch angle which is also used for estimating the value of the wind speed based on the output power, the rotational speed and the blade pitch angle. As power is equal to torque times rotational speed, either the output power or the gener ator torque may be utilized in the calculation.

According to an embodiment of the present invention, the method is performed offline or online and/or the performance change is due to rotor blade stalling and/or rotor blade ic ing and/or a rotor blade contamination and/or a control error and/or an actuation error and/or a configuration error and/or a sensor error. Thereby, the method is enabled to detect a performance change which may be due to different reasons, thereby extending the applicability of the method.

According to an embodiment of the present invention it is provided a method of controlling a wind turbine, comprising: performing a method according to one of the preceding embodi ments and taking an action based on the estimated performance change. The action taken may for example involve controlling the wind turbine and/or inspecting a rotor blade for icing and/or contamination and/or inspecting one or more measure ment sensors and/or inspecting a control algorithm and/or configuration data and/or changing a control algorithm.

It should be understood, that features, individually or in any combination, disclosed, described or explained with re spect to a method of detecting a performance change may also be applied, individually or in any combination, to an ar rangement for detecting a performance change of a wind tur bine according to an embodiment of the present invention and vice versa. According to an embodiment of the present invention it is provided an arrangement for detecting a performance change of a wind turbine, the method comprising: an input module adapted to receive a measured value of a wind related quanti ty; a processor adapted: to estimate a value of the wind re lated quantity based on at least wind turbine operational da ta and applying a model; to compare the measured value with the estimated value to derive a residual value of the wind related quantity; and to output a signal indicating a perfor mance change based on the residual value.

The arrangement may comprise software and/or hardware. The arrangement may be configured to carry out the method accord ing to one of the above-described embodiments. The arrange ment may for example be part of a wind turbine controller.

According to an embodiment of the present invention it is provided a wind turbine, comprising: a rotor at which plural rotor blades are mounted; a generator coupled to the rotor; an anemometer; and an arrangement according to the preceding embodiment .

The aspects defined above and further aspects of the present invention are apparent from the examples of embodiment to be described hereinafter and are explained with reference to the examples of embodiment. The invention will be described in more detail hereinafter with reference to examples of embodi ment but to which the invention is not limited.

Brief Description of the Drawings

Embodiments of the present invention are now described with reference to the accompanying drawings. The invention is not limited to the described or illustrated embodiments.

Fig. 1 schematically illustrates a wind turbine according to an embodiment of the present invention comprising an arrange- ment for detecting a performance change according to an em bodiment of the present invention;

Figs. 2, 3 and 4 schematically illustrate portions of the wind turbine illustrated in Fig. 1 or the arrangement for de tecting a performance change according to an embodiment of the present invention.

Detailed Description

The wind turbine 1 according to an embodiment of the present invention illustrated in Fig. 1 in a schematic form comprises a rotor 3 having a coupled hub 5 at which plural rotor blades 7 are mounted. The rotor blades 7 are rotatable around a pitch angle axis 9 by using a pitching system 11 which com prises an actuator for actually adjusting the pitch angle and further may also comprise a pitch angle measurement portion for measuring the adjusted pitch angle. The wind turbine 1 further comprises a generator 13 which is coupled to the ro tor 3. The generator 13 outputs a three-phase power stream 15 and supplies it to a converter 17 which may comprise a AC-DC portion, a DC link and a DC-AC portion for converting the variable frequency power stream 15 to a substantially fixed frequency power stream 19 which is supplied to a wind turbine transformer 21 to transform the output stream to a higher voltage output stream 41.

The wind turbine 1 further comprises an arrangement 23 for detecting a performance change of the wind turbine according to an embodiment of the present invention. Thereby, the ar rangement 23 (e.g. being part of a wind turbine controller) comprises an input portion comprising input terminals 25 which are adapted to receive a value 27 of a wind related quantity, in particular a wind speed Vwind,meas as is meas ured by an anemometer 29. The anemometer 29 is in the illus trated embodiment mounted at a nacelle 31 which harbours the rotor 3, the generator 13, the converter 17, the transformer 21 as well as the arrangement 23. The input section comprising the input terminals 25 of the arrangement 23 is further configured to receive a blade pitch measurement signal 33 from the pitching system 11, to receive a rotor speed measurement signal 35 from a rotational speed measurement device 37 and further receive an output power measurement signal 39 of the output power 41 from an electri cal measurement system 43 arranged downstream the transformer 21. The nacelle 21 is mounted on top of a wind turbine tower 45.

The arrangement 23 is configured to estimate a value of the wind speed 47 based on the pitch angle measurement signal 33, the rotational speed measurement signal 35 and the power out put signal 39. Thereby, the arrangement 23 comprises a pro cessor 24 which estimates and outputs the estimated value 47 of the wind speed. Further, the processor 24 is configured to compare the measured value 27 of the wind speed with the es timated value 47 of the wind speed to derive a residual value of the wind speed. Further, the arrangement 23 indicates a performance change based on the residual value and thereby outputs a respective indicating signal 49 for example to a display device 51.

In the embodiment illustrated in Fig. 1, the wind related quantity is the wind speed, however, in other embodiments, the wind related quantity may be the wind turbine output pow er or the wind turbine generator torque, i.e. the torque of the generator 13. The measured pitch angle signal 33, the measured rotational speed signal 35 and the measured output power 39 represent an example of wind turbine operational da ta collectively referred to as signal 53 (see Fig. 2) .

Fig. 2 schematically illustrates a portion of the wind tur bine including a portion of the arrangement 23. The arrange ment 23 comprises a wind estimation module 55 which receives the wind turbine operational data 53, in particular the out put power signal 39, the rotational speed measurement signal 35 and the pitch angle measurement system 33. Via the input terminal 25, the arrangement 23 receives the wind speed meas urement signal 27 from the anemometer 29. Using a subtraction element 57 of the arrangement 23, the estimated wind speed 47 is subtracted from the measured wind speed 27 to obtain a re sidual value 59 of the wind speed.

The arrangement 23 further comprises a filter 61 which is configured for low pass filtering and/or applying a moving average of the residual value 59 over several time steps to obtain a filtered residual value 63, in particular for every time step i. It should be noted, that the wind speed measure ment signal 27 as well as the wind turbine operational data 53 are continuously determined for several time steps as la belled with the index i in the following.

Fig. 3 shows another portion of the arrangement 23 which may be comprised in the arrangement 23 as is illustrated in Fig. 1. The residual value 59 of the wind speed or the filtered residual value 63 of the wind speed is supplied to a cumula tive sum calculation model 65 which may be (for example in software) comprised in the arrangement 23. The cumulative sum calculation model 65 calculates, in particular for every time step, an upper cumulative sum as well as a lower cumulative sum C ⁺ (i), Ch(i), for example as follows:

Deriving the cumulative sum is one example of deriving a sta tistical characteristics of the residual value 59 of the wind speed or the filtered residual value 63 of the wind speed.

The statistical characteristics is collectively labelled with reference sign 67. The upper cumulative sum 69 as well as the lower cumulative sum 71 are output by the cumulative sum de termination model 65.

Fig. 4 schematically illustrates a further portion of the ar rangement 23 which further processes the upper cumulative sum 69 as well as the lower cumulative sum 71. The portion of the arrangement as illustrated in Fig. 4 comprises a comparison module 73 for comparing at least one residual cumulative sum 69, 71 to a threshold 75, 77 to obtain a comparison result which is output from the comparison module 73 and labelled with reference sign 79. The comparison result 79 is one exam ple of an indicating signal 49 which indicates whether a per formance change is present and also whether a performance degradation or a performance loss is present or not. The thresholds 75, 77 may be adjusted depending on the applica tion. Further, according to different embodiments, different thresholds for different levels of performance changes may be predefined allowing to at least coarsely quantify the perfor mance change .

The arrangement further comprises a control corrective action module 81 which receives the comparison result 79 and derives values of control parameters collectively referred to as ref erence sign 83. The control parameters may for example com prise a reference of a speed set point 85, a reference of a power/torque set point 87 and/or a reference for a pitch an gle 89.

The comparison result 79 is further supplied to a user noti fication module 91 which may for example comprise the display 51 illustrated in Fig. 1. The user notification module 91 may output a warning 93 or an alarm 95 which may for example be displayed on the display 51 illustrated in Fig. 1.

In the following, one particular embodiment is described in further detail also referring to Figs. 2, 3 and 4, without limiting the invention: The detection algorithm may consist of three steps:

1. Calculate residual between measured and estimated wind speed. The residual of the measured and estimated pow

er/torque can be used as an alternative.

2. Evaluate changes to the statistical properties (mean, var iance, etc.) of the residual over time.

3. Detect if the change in the statistical properties (mean, variance, etc.) of the residual exceeds a certain threshold and take corrective control action and/or provide relevant notifications .

The three steps are described in more details in the follow ing .

Step 1 :

In the first step, the residual between the measured and es timated wind speed is calculated.

The wind estimation in the turbine may be done based on a Cp curve as describe in US 2012/0078518.

The Wind Estimator functions as a digital twin for the tur bine and the mapping from power, rotor speed and pitch angle to wind speed through Cp curves represents the expected per formance of the turbine. Due to measurement noise and the fact that the anemometer provides a point measurement while the Wind Estimator provides a wind speed averaged across the rotor plane, there will always be some deviations between the measured and estimated wind speed. To increase robustness to wards single outliers in the residual caused by e.g. brief sensor outage, communication delays or similar, a low pass filter or moving average filter can be introduced and applied to the calculated residual.

A block diagram for Step 1 is illustrated in Fig. 2. Step 2 :

Under normal operating conditions, the wind speed residual (Vres) will have certain statistical characteristics ex pressed through e.g. mean and standard deviation. A signifi cant shift in these characteristics indicates that something has changed in the relationship between the estimated and measured performance (in this case expected performance is expressed through estimated wind speed but could also be based on other measures such as power/torque ) .

Changes of the statistical properties of the residual can be detected by use of the cumulative sum (CUSUM) approach, but other methods can be applied in this step as well, such as the exponentially weighted moving average (EWMA) . If using the CUSUM approach for detecting changes to the mean of the residual, the upper cumulative sum is calculated as:

where is the target (expected) mean of Vres under normal operation, is the target (expected) standard deviation of

Vres under normal operation, while ^® is the minimum amount of shift from the target mean that should be detected ex pressed as a multiple of the target standard deviation. The lower cumulative sum can be calculated in a similar way:

A block diagram for Step 2 is illustrated in Fig. 3.

Step 3: The upper and lower cumulative sums are compared to a thresh old limit indicating when performance of the turbine is devi ating significantly from what is expected. This can be ex pressed in a multiple of standard deviations from the target mean.

When the limit is exceeded, it is possible to take corrective actions when conditions permit this. This could be by chang ing the operating point of the turbine by modifying the speed, power (torque) or pitch set-point. This can be supple mented by user notification through a warning or an alarm.

For offline application of this method, only the user notifi cation may be relevant. A block diagram for Step 3 is illustrated in Fig. 4.

It should be noted that the term "comprising" does not ex clude other elements or steps and "a" or "an" does not ex clude a plurality. Also elements described in association with different embodiments may be combined. It should also be noted that reference signs in the claims should not be con strued as limiting the scope of the claims.