CONTROL OF A WIND POWER PLANT

Technical field

The invention relates to a control system for a wind turbine power plant, and to a method of controlling such a power plant.

Background to the invention

A common type of wind turbine generator is the three-bladed upwind horizontal-axis wind turbine generator (HAWT), in which the turbine rotor is at the front of the nacelle and facing the wind upstream of its supporting turbine tower. The blades capture the energy of the wind which is converted to mechanical torque at the rotor which is then transferred through a drive train to a generator. The generator converts the mechanical power into electrical power which is then injected into the electrical grid, which may be by way of a power electronics frequency converter which takes into account grid requirements.

Typically, the control characteristic of a wind turbine generator is expressed as a power curve, as shown in Figure 1 , which partitions the operation of the wind turbine generator into a number of regions. Region A covers operation of the wind turbine where the wind speed is too low to drive the blades to generate power. Region A extends to the 'cut-in' wind speed V _C i at which point the wind flow is energetic enough for the wind turbine generator to be activated in order to start generating power. The operation then moves into region 'B' in which the wind speed is above the cut-in wind speed V _C i but is too low for the generator to produce maximum or 'rated power'. Thus, in region 'B', which is also known as below-rated operation, the wind turbine is controlled to maximise the power captured from the wind. Usually, generator torque control regulates the rotor speed to optimise the blade tip speed ratio, whilst blade pitch is held substantially constant.

Where the wind speed increases into region 'C, it is considered to be at or above rated wind speed V _R , such that the wind turbine generator is able to produce its rated power P _R . As the skilled person will understand, the rated power is the maximum power output which the generator is designed to produce on a continuous basis. Therefore, in this above rated power region, the control objective is to maintain the power output so that it is substantially constant, and this is typically achieved by blade pitch control. During blade pitch control, the l generator torque is held substantially constant, although generator torque control may be used to regulate relatively high frequency speed variations of the rotor.

The use of blade pitch control and generator torque control can maintain rated power for a range of wind speeds. However, a point will be reached at which the wind speed is too high for continued safe operation of the wind turbine generator at which point it must be shut down to prevent damage to the generator and other components. This point is referred to as the max rated wind speed V _M R. To provide the power output profile represented in Figure 1 , a wind turbine control system is operable to produce power in accordance with an external power reference. The power reference is usually fed to the control system by an external system, for example from a power plant controller in accordance with a grid set point which it receives from a grid or transmission system operator (TSO). The power plant controller is responsible for coordinating the control of the wind turbines within the power plant so that the plant behaves as a single generating unit.

The power plant controller has authority to vary the power reference to take account of various operational conditions whilst ensuring that the power demanded by the TSO is satisfied, as close as possible. The procedure of determining appropriate power reference values from each of the wind turbines in a power plant based on the TSO set point, whilst taking into account the operational status and local environmental conditions of each wind turbine, is often referred to as dispatching. Similarly, the function of the power plant controller that is responsible for this procedure is referred to as the 'dispatcher'.

It has been observed that during the dispatching of power references to the wind turbines in a power plant, the dispatcher may behave erratically. For, example the dispatcher may dispatch a power reference to a wind turbine which varies at a high rate of change, and which through large absolute swings in value. In certain circumstances, such behaviour may induce undesirable stresses on the wind turbine structure, which may shorten the life of components of the wind turbine. Therefore, there is a need to identify this erratic behaviour of a power plant controller and to take mitigating action, but in a way which is reliable such that false positives are avoided. It is against this background that the invention has been devised. Summary of the invention

In a first aspect, the invention provides a method for controlling a wind turbine, comprising: receiving a signal indicative of a power reference for the wind turbine; determining a variability parameter based on the signal; determining a structural oscillation parameter associated with the wind turbine; and correlating the variability parameter and the structural oscillation parameter; and generating an action signal in dependence on the result of the correlation. The invention extends to and therefore embraces a system for controlling a wind turbine including a wind turbine controller that is configured to receive a signal indicative of an external power reference for setting the power generation level of the wind turbine, wherein the controller analyses the signal thereby to determine a variability parameter of the signal, determines a structural oscillation parameter associated with the wind turbine, and correlates the variability parameter and the structural oscillation parameter, and generates an action signal in dependence on the result of the correlation.

Beneficially, the embodiment of the invention provide a reliable and dependable approach to analysing the incoming power reference signal from the power plant controller, for example, to determine if the signal has characteristics that are likely to cause damage to the wind turbine in which the invention is implemented. Moreover, the embodiments are generally software based which make their implementation more straightforward and also enables them to be implemented on in-service wind turbines with few, if any, hardware modifications. In some embodiments, the variability parameter may be determined by a statistical data analysis process. Examples of such process are a weighted counter algorithm and a rainflow count algorithm. Such processes are known generally for use in fatigue analysis, but are not believed to be known for use in systems analysis/diagnosis to determine, via analysis of a system's output, whether that system is in some way faulty or not behaving as expected.

The structural oscillation parameter may be indicative of tower motion, and accordingly may be embodied as a tower loading signal or acceleration signal. Where data is captured over a predetermined time frame or window, the structural oscillation parameter may be an average value of tower motion over a predetermined data capture period.

The action signal may include providing a notification to external control equipment, so that appropriate action can be taken, although the action signal may trigger other responses, for example modification an internal power reference signal local to the wind turbine, e.g. by reducing the ramp rate of the internal power reference signal.

The invention may also be expressed as: a computer program product downloadable from a communication network and/or stored on a machine readable medium, comprising program code instructions for implementing a method as defined above, and also as a machine readable medium having stored thereon a computer program product as defined above.

Brief description of the drawings

Embodiments of the invention will now be described by way of example only to the following drawings, in which:

Figure 1 is an exemplary power curve for a wind turbine;

Figure 2 is a system diagram of a wind power plant in accordance with an embodiment of the invention;

Figure 3 is a schematic system view of a wind turbine that forms part of the wind power plant in Figure 2;

Figure 4 is a block diagram of a controller in accordance with an embodiment of the invention; Figures 5a to 5d are a series of graphs illustrating a rainflow counting statistical analysis technique;

Figure 6 is a block diagram illustrating a weighted zero-crossing counter statistical analysis technique;

Figure 7 is a flow chart in accordance with an embodiment of the invention; and

Figures 8a to 8c are plots illustrating the variability of external and internal power reference values, tower acceleration, and an excitation indicator.

Detailed description of embodiments of the invention With reference to Figure 2, a wind turbine power plant 2 includes a plurality of wind turbines 4. Each of the wind turbine 4 in Figure 2 is a horizontal axis wind turbine (HAWT), although it should be appreciated that the invention is also applicable to other types of wind turbine generation systems.

As is conventional, the power outputs of the wind turbine 4 are interconnected at a point of common coupling 5, which feeds an external power grid 10. Note that the point of common coupling may also be referred to herein as 'PCC for brevity. The power grid 10 therefore receives power from the wind power plant 2 that is a combination of the outputs from all of the wind turbine 4 in the power plant 2. It will be appreciated that the electrical power generated by the wind turbine modules 4 depends on the wind energy available in the locality of each of those wind turbines 4, since the wind speed typically varies from location to location across the power plant 2. The wind turbines 4 form part of an industrial control system and, as such are connected together by a communications network 14.

The communications network 14 enables each of the wind turbines 4 to communicate bi- directionally with a power plant controller 16. Such a communications network is

conventional in wind power plants featuring modern utility-scale wind turbines so a detailed explanation will not be given here. However, it should be understood that the network 14 enables the wind turbines 4 to feed operational data to the power plant controller 16, and enables the power plant controller 16 to transmit control commands and other data to each of the wind turbines 4, either on a basis of general broadcast transmissions (same data for all wind turbines) or directed transmissions (commands/data tailored to specific wind turbines).

In the embodiment of Fig. 2, the network 14 is depicted as lines, which suggests a cable- based infrastructure. Although a cable-based infrastructure is acceptable, it should be noted that this is not to be considered limiting and, as such, the network 14 may be embodied as a wireless system. Typically, communications protocols for such control systems are standardised by equipment vendors, examples of which are governed by IEC 60870-5-101 or 104, IEC 61850 and DNP3. However, alternative protocols such as TCP/IP may also be used.

An important function of the power plant controller 14 is to provide each of the wind turbine modules 4 with power reference values which dictates the maximum power level or 'power limit' at which the wind turbine modules 4 should operate. The power reference values are illustrated in Figure 2 by the single variable P _REF , although it should be appreciated that the power plant controller 16 may also provide a reactive power reference Q _RE F, though not shown here for clarity.

The power plant controller 16 determines P _RE F based on several factors, including: the operational status of the wind turbine modules 4; the grid status, and the power demand PDEM as provided by a transmission system operator (TSO), labelled here as 18. Overall, the power plant controller 16 will endeavour to operate the wind turbine modules 4 so that the power produced by the power plant 2 as a whole meets the power demanded by the TSO 18. In doing so, the power plant controller 16 determines a specific value of P _RE F in respect of each of the wind turbine modules in the power plant 2. This ensures that power production from the plant 2 can be optimised to take account of varying conditions across the plant 2. Having described the overall arrangement of the wind power plant 2, the following description will focus on the more detailed structure of the wind turbine modules 4. With this in mind, Figure 3 illustrates an example of a power generation system architecture which gives context to the invention, as will become apparent. Represented schematically as a system diagram the wind turbine module or 'system' includes features that are significant for this discussion, but it should be appreciated that many other conventional features are not shown here for brevity, for example yaw control equipment, control network, local power distribution network and so on. However, the skilled person would understand that these features would be present in a practical implementation, and so their presence is implied. Also it should be noted that the specific architecture discussed here is used as an example to illustrate the technical functionality of the invention, and so the invention may be implemented by a system having a different specific architecture.

With reference to Figure 3, the wind turbine module 4 includes a bladed rotor 20, which drives a transmission 22 by way of an input drive shaft 24. Although the transmission 20 is shown here in the form of a gearbox, it is also known for wind turbines to have a direct-drive architecture which do not include a gearbox. The transmission 22 has an output shaft 26 which drives a generator 28 for generating electrical power. Three phase electrical power generation is usual in utility scale wind turbine systems, but this is not essential for the purpose of this discussion.

The generator 28 is connected to a frequency converter 32 by a suitable three-phase electrical connector such as a cable or bus 34. The frequency converter 30 may be of a conventional architecture and, as is known, converts the output frequency of the generator 28 to a suitable output voltage level and frequency that is suitable for supplying to an internal electrical grid via a transformer 36 and filter 37. In this figure, the internal electrical grid is represented by the PCC 5. It will be appreciated that the specific architecture described here is a two-level back-to-back voltage source full scale frequency converter (FSC) system, which includes a generator-side converter 38 and a grid-side converter 40 which are coupled via a DC link 42. The general architecture of such a system is conventional and will not be described in more detail. Furthermore, the skilled person will understand that other architectures are known, such as doubly-fed induction generator-based systems (DFIG).

To put the invention into context, a brief discussion of the control strategy of the wind turbine will now be provided by way of example. As is known, variable-speed wind turbines typically operate under two main control strategies: below-rated power and above-rated power. As is known, the term 'rated power' is used here in its accepted sense to mean the power output at which the wind turbine system is rated or certified to produce under continuous operation. Similarly, the use of the term 'rated wind speed' should be understood to mean the lowest wind speed at which the rated power of a wind turbine is produced.

Below rated power occurs at wind speeds between the cut-in speed and rated wind speed which, typically, is between 10 and 17m/s, but may be different depending on the size of the wind turbine. In this operating region, the wind turbine module 4 is operable to control the rotor speed so as to maximise the energy captured from the wind. This is achieved by controlling the rotor speed so that the tip speed ratio is at an optimum value, namely between 6 and 7. To control the rotor speed, the wind turbine module 4 controls the generator torque so as to track an internal power reference, as will be described.

Above-rated power occurs when the wind speed has increased to, or exceeds, the rated wind speed. In this operating condition, the objective of the wind turbine module 4 is to maintain a constant output power. This is achieved by controlling the generator torque to be substantially constant, so as to track a constant internal power reference, but varying the pitch angle of the blades which adjusts the resulting lift and drag force of the blade in the rotor plane. This will control the torque transferred to the rotor shaft so that the rotational speed, and also the generated power of the system, is kept constant below a set threshold.

In order to achieve the below-rated power and above-rated power control objectives, the wind turbine module 4 is equipped with a converter control module 50. The converter control module 50 is operable to control the frequency converter 32 via torque reference T _REF to influence the torque exerted on the rotor 10 by the generator 26, and also to control the pitch of the blades via pitch reference P0 _REF , and thereby the speed of the rotor 10, through a blade pitch adjustment system 52. The converter control module 50 receives a plurality of control inputs, but two control inputs are shown specifically here: a rotor speed reference parameter N _RE F and an internal power reference parameter PINT_REF _. which are provided by a higher level controller, such as an operational sequence controller 53 either directly to the converter control module 50 or through a data distribution network based on a suitable protocol, such as Ethernet. The converter control module 50 also receives monitoring inputs so that it can determine the correct operation of the components under its control. Specifically, the converter control module 50 receives a machine speed parameter N _S which may come from speed sensors 54 associated with the rotor, the transmission, or the generator, and a power output parameter P _S from the frequency converter 32.

During below-rated conditions, the converter control module 50 is operable primarily to control the generator torque, which is associated with, and calculated from, the internal power reference PINT_REF. by outputting a demanded torque signal T _RE F to the frequency converter 32 in order to track PINT_REF- Similarly, at operating conditions above-rated power, the converter control module 50 is operable to hold the generator torque substantially constant (and, therefore, to track the constant internal power reference P| _N T_REF) but to provide control input P0 _REF to the pitch control system 52 to modulate, collectively, the pitch angles of all three blades of the rotor 10. At this point it should be noted that the behaviour of the converter control module 50 and the operational frequency controller 53 is to ensure that the power generation of the wind turbine is maximised without exceeding the power limit governed by the power plant controller 22 power reference signal P _REF . From the above explanation, it will be appreciated that the power plant controller 14 is responsible for providing a power reference parameter to the individual wind turbines of the power plant so that the power produced by the power plant is maximised without exceeding the power limit dictated by the power reference parameter. Since the nature of gird power demand is influenced by consumer behaviour, the power demand set point as set by the TSO 18 will vary over time. Accordingly, the power plant controller 16 is operable to configure the values of P _REF to change gradually so as to track the power demand set point. However, there are circumstances where the power plant controller 16 may need to change P _REF quickly, Examples are during spinning reserve operation when a grid anomaly is detected, for example a grid frequency drop; and when one turbine in a wind park stops due to an error and one or more other turbines needs to ramp their power output up to compensate.

Problems can result, however, when the power plant controller 16 exhibits behaviour in which the power reference P _RE F provided to the wind turbine controller becomes erratic. This can be caused by instabilities in the power plant control loop induced by e.g. communication delays or faulty available power feedback from the turbines due to e.g. ice on blades.

In the invention, the operational controller 53 is operable to monitor the variability of the external power reference value and to make a determination if the variability of the external power reference value is considered to be excessive, so as to present a risk of damage to the wind turbine structure. The controller 53 monitors the external power reference P _RE F by way of a statistical analysis process in order to identify whether the variability in the parameter may cause unacceptable loading on the structure of the wind turbine. It should be noted at this point that various statistical analysis techniques may be used to analyse the variability of P _REF . Two specific examples are provided here, and will be described in more detail below. The first statistical technique is a rainflow counting method, and the second technique is a zero crossing counter, as will be explained later in more detail.

An embodiment illustrating the above approach will now be described with reference to Figures 4 to 6.

Figure 4 illustrates the operational controller 53 in more detail. It should be appreciated at this point that the functional blocks illustrated in Figure 4 illustrate a specific functionality of the operational controller 53 which is not intended to represent its entire functionality. In practice, however, the operational controller 53 would carry out many other functions, but as these functions are not directly relevant to the invention they are not depicted nor described here so as not to obscure understanding of the invention. Furthermore, it should be understood that as the functional blocks in Figure 4 represent functionality, they may be implemented on hardware, software or firmware, either within the same processing environment or on a distributed processing architecture. That is to say, the functional architecture illustrated in Figure 4 is not intended to limit the invention to a specific hardware or software architecture, platform or processing environment. In overview, the operational controller 53 includes a power reference module 100 and an analysis unit, labelled generally as 102. As has been mentioned above, a function of the operational controller 53 is to convert the external power reference PREF received from the power plant controller 16 into an internal power reference to control the operation of the power converter 32. For this reason, the power reference module 100 receives as a first input an external reference signal P _RE F from the power plant controller 16 and, as a second input, operation data 104 from the wind turbine components. It will be appreciated that the operational data 104 may include generator speed and power information, wind speed, pitch references and so on. Based on these inputs, the power reference module 100 provides an internal power reference signal PINT_REF to the converter control module 50. The process through which the power reference module 100 converts the external power reference signal to the internal power reference signal is conventional and will not be described here in further detail so as not to obscure the invention. The analysis unit 102 provides functionality that runs alongside the power reference module 100 and analyses the external power reference P _RE F to detect any anomalous behaviour which could indicate a malfunctioning power plant controller 16. Importantly, the analysis unit 102 is able to identify that behaviour that has a high risk of causing damage to the wind turbine through extreme loading and, moreover, to take appropriate action to mitigate that damage by correcting the anomalous behaviour of the power reference signal P _REF .

In summary, the analysis unit 102 comprises an analysis module 104, an excitation indicator module 106, hereinafter ΈΙ module', and a limit check module 108. The analysis unit 102 is operable to analyse the power reference signal P _RE F and to provide an output indicative of its variability, which is shown here as variability signal 109.

Examples of two techniques that may be implemented by the analysis module 104 will now be described. Rainflow counting technique

Rainflow counting algorithms are known generally in the field of fatigue analysis of structural components. However, their usefulness in analysing controller behaviour has heretofore not been appreciated. The rainflow counting technique is illustrated in Figures 5a to 5d, which illustrate the conversion of the raw power reference signal PREF into a series of uniform reversals, as seen in Figure 5d. The raw signal P _RE F is shown in Figure 5 and it will be noted that it is quite noisy. A filtering process, for example using a known racetrack filter, removes oscillations below a predetermined threshold, resulting in a P _RE F profile illustrated in Figure 5b. Valleys and peaks from the filtered P _RE F signal are identified, as shown by the nodes on Figure 5c, and the ranges of oscillations are found, as indicated by the double-headed arrows. The ranges are then weighed and summed using a suitable cumulative damage model, such as the Palmgren-Miner's rule that is suitably tuned using a low Wohler coefficient, for instance of approximately 0.3, although this is merely exemplary.

The resulting equivalent load is shown in Figure 5d, which denotes a oscillating signal which yields the same rainflow count as the raw P _RE F signal. The amplitude of the signal, denoted by the vertically oriented double headed arrow, is used to gauge the excitation of the P _REF signal, and corresponds to the variability signal 109 mentioned above.

Weighted zero-crossing technique

An alternative to the rainflow counting technique is to implement a weighted zero crossing technique, an example of which will be described with reference to Figure 6.

The PREF is first differentiated 150 and is then run through a low pass filter 152 in order to filter out any unwanted oscillations. The filtered derivative of P _REF is then processed by a zero crossing detector 154, which outputs a positive signal each time the input signal crosses zero. The output of the zero crossing detector 154 is input into a suitable data storage structure 156, such as a ring buffer in this example, so that at each time interval at Ό' is stored in the buffer if no zero crossing is detected and a '1 ' is stored if a zero crossing was detected in that time interval. The size of the ring buffer should be suitably configured, for example as the size of the analysis time window (30 second for example) divided by the sample time.

In order to compute the number of zero crossings occurring in the timeframe of the analysis window, the sum of all the stored values in the ring buffer is computed at summing block 158. The output of the summing block 158 is input into a multiplier 160, which multiplies the summed output with a modified standard deviation value 162. In this embodiment, the modified standard deviation value is derived by passing raw P _REF signal through a standard deviation function 164 and subsequently through a root function, whereby the root index corresponds to the inverse of the Wohler coefficient mentioned above for the rainflow counting technique (e.g. 3). The output of the multiplier provides the variability signal 109.

Having described two statistical techniques for analysing the P _RE F signal, the discussion will now return to the functionality of the controller 53.

The output of the analysis module 104, that is to say the variability signal 109, is fed to the El module 106, which is operable to correlate the variability signal 109 with data relating to the structural oscillation of the wind turbine. On its own, the variability signal 109 that is output from the analysis module 104 may not be sufficient to make a robust determination as to whether the variability of the power reference signal P _RE F is such that structural damage to the wind turbine is likely to occur. Therefore, the El module 106 interprets the variability signal 109 to provide an excitation indicator 110.

In this embodiment, the El module 106 interprets the variability signal by making reference to a tower acceleration signal ACC. The tower acceleration signal may be obtained from suitable acceleration sensors mounted in appropriate positions in the tower, and averaged over a predetermined time period. More broadly, signals from other components may also be used, such as suitably filtered generator speed signal, collective pitch angle, blade load measurements, tower load sensors. At present it is believed that the tower acceleration signal is useful as it is widely available on wind turbines, and blade pitch changes are readily identified as changes in the tower motion.

In more detail, the El module 106 correlates the variability signal 109 and the tower acceleration signal ACC in order to output a compound signal that is indicative of the excitation that is affecting the structural components of the wind turbine. Here, the output signal is referred to as an excitation indicator 1 10. The generation of the excitation indicator 110 may be achieved in various ways. However, one way is to multiply the variability signal 109 with the standard deviation of the acceleration, such that the excitation indicator 110 has a large value when both the variability signal is large and the tower oscillation level is large.

The excitation indicator 110 is input into the limit check module 108 that determines whether the excitation indicator 110 exceeds a predetermined threshold, which would indicate that structural damage to the wind turbine is likely to occur. In this embodiment, two outcomes are possible in the event that the excitation indicator 1 10 is determined to be too high. Firstly, the limit check module 108 may issue an alert signal 112 which will indicate to a suitable supervisory safety system that there is a problem so that appropriate response can be taken. Such a response might be that the wind turbine is shut down, if certain other conditions are satisfied. However, in this embodiment the limit check module 108 also generates an action signal to a rate limit module 1 14 that interrupts the output of the power reference module 100 by applying or decreasing the rate limit to the output signal PINT_REF. Several rate limits can be applied successively in order to mitigate the excitation of the wind turbine. The default rate limit on PINT_REF might be 0.25pu/s (per-unit) which upon detection of an excitation is decreased to 0.1 pu/s. In case an excitation is still detected after decreasing the rate limit it could be further decreased to 0.05 pu/s or even lower.

Having described an embodiment of the functionality of the controller 53, it will now be described with reference to the flow chart in Figure 7. The process 200 initiates at step 202, which may be when the wind turbine is run up to operational speed and, furthermore, the process 200 may be run periodically in order to provide a continuous monitoring function of the operation of the power plant controller 16.

The process 200 analyses P _REF data on a rolling window basis, such that a data window of a given width may be analysed every predetermined time interval. It is envisaged that an appropriate window width will be between 30 seconds and 2 minutes, at a sample rate of between 10Hz and 100Hz .

Therefore, at step 204, P _RE F data for the duration of the moving time window is captured and is processed by the analysis module 104 to provide the variability signal 108. Once the data captured in the time window has been analysed in the below process, the time internal is incremented and the process is implemented again for the data set captured by the next time window. So, the process is continuous. The variability signal 108 is then correlated with the motion of the tower in step 206, as described above with reference to the El module 106. The output of the El module 106, that is to say, the excitation indicator 1 10 is then limit checked at decision step 208. If the excitation indicator 1 10 is less than a predetermined threshold T, then the process will terminate at step 210. However, if the excitation indicator 110 exceeds the predetermined threshold T, then appropriate action is taken at step 212. As mentioned above, at this point an alert action may be issued to a supervisory control system so that a determination may be made as to whether the wind turbine should be shut down. However, in this embodiment, direct action is also taken to modify the ramp rate of the power reference signal P| _N T_REF which acts to limit the structural excitation of the wind turbine, thereby reducing the risk of damage. Finally, the process terminates at step 214. The above process 200 is illustrated visually in Figures 8a-8c, which show the various signals introduced above running along a common timeline.

Figure 8a shows the two power reference signals, P _RE F as set by the power plant controller 16, and PINT_REF as set by the wind turbine controller 53. As can be seen, in an initial time region, labelled as TV, PINT_REF follows PREF exactly, and in region A the wind turbine can be considered to be operating at rated power.

It will be noted, however, that at time T1 onwards, PREF is reduced sharply by the power plant controller 16. This may be in order to implement some 'spinning reserve' in the power plant or in response to a corresponding reduction in demand from the TSO 18. It will be noted that the internal power reference signal PINT_REF follows the reduction in PREF although it does so at a more gradual rate of change. P| _N T_REF then follows P _RE F as P _RE F rises and falls at a relative low rate up until time period T2. Between time period T1 and T2, labelled as region 'Β', it can be considered that the power plant controller is operating normally. So, P _RE F is reduced sharply at time period T1 in response to valid operational circumstances and then proceeds to rise and fall gradually to regulate the power output from the wind turbine up to time period T2.

Remaining within region B, but referring to Figures 8b and 8c, it will be seen that the tower acceleration, illustrated in Figure 8b, has variability that is within acceptable limits from TO up to T2, and drops in magnitude at time period T1 which corresponds with the change in P _RE F value. Similarly, with reference to Figure 8c, it can be seen that the excitation indicator 1 10 is also relatively low in value between time periods TO and T2. Returning to Figure 8a, at time period T2, and up to T3, it will be seen that PREF becomes erratic at this point, and undergoes large step changes in value. In response, the wind turbine controller 53, more specifically the power reference module 104, attempts to control the wind turbine so as to accord with the changing power limit imposed by the power plant controller 16, albeit that PINT_REF lags PREF slightly due to the inherent ramp rate implemented by the module 104. The result of the large swings in power reference values causes extreme loading on the wind turbine, and this is depicted in Figure 8b, in which it can be seen that the tower acceleration signal ACC begins to fluctuate significantly. Also, with reference to Figure 8c, it can be seen that the excitation signal 1 10 increases to a level which exceeds the predetermined threshold for this value, labelled as T on Figure 8c.

The period between T2 and T3, labelled as region 'C, represents a single snapshot of one data capture window implemented by the process 200 described above with reference to Figure 7, although it should be appreciated that in practice the data capture window would move slowly along the x-axis tracking the generated P _RE F data. Thus, once a data capture process has been completed and the captured data has been analysed which identifies an unacceptably high variability signal, the process 200 takes action to implement a reduced ramp rate via rate limit module 1 14. With reference to Figure 8a, it can be seen that, although the external power reference signal PREF is still undergoing large step changes, the internal power reference value PINT_REF varies in a much more gradual way. This has a direct impact on reducing the structural oscillation of the wind turbine, as can be appreciated by reference to Figure 8b from T3 onwards, and also acts to reduce the excitation indicator 1 10 as shown in Figure 8c.

It will be appreciate that various modifications may be made to the specific embodiments described above without departing from the inventive concept, as defined by the claims. In the above embodiments, it has been described that the internal power reference signal may be rate limited in order to mitigate the effects of the fluctuating P _REF signal or, alternatively, than an alert may be issued such that an operator, or the PCC, may take appropriate action. It is also possible, however, that the wind turbine could disregard the power reference signal completely and revert to some other internally-regulated power generation regime.

Also, instead of implementing a rate limiter to control the rate at which the internal power reference signal is able to increase and decrease, alternative options include controlling the thrust variability via the pitch control mechanism, or controlling an active tower damping mechanism,