Claims:
1. A method for controlling a wind farm comprising a plurality of wind turbines electrically coupled to a grid, the method comprising:
receiving operational data of the wind farm;
determining a subset of wind turbines from the plurality of wind turbines subjected to a change from a first wind condition to a second wind condition;
determining a plurality of operational parameters of the subset of wind turbines based on the operational data;
determining a dynamic forecasting model based on the plurality of operational parameters;
determining a power estimate value based on the dynamic forecasting model and the operational data, wherein the power estimate value is representative of an estimate of power generated by the wind farm at a future instant of time; and
controlling at least one wind turbine of the wind farm based on the power estimate value.
2. The method of claim 1, wherein determining the subset of wind turbines comprises selecting the subset of wind turbines from the plurality of wind turbines, located at a periphery of the wind farm and subjected to the change from the first wind condition to the second wind condition.
3. The method of claim 2, wherein determining the plurality of operational parameters comprises measuring the plurality of operational parameters of the subset of wind turbines during a training period.
4. The method of claim 1, wherein determining the dynamic forecasting model comprises:
determining a parametric equation for estimating a power output from the subset of wind turbines based on the plurality of operational parameters; and
modifying one or more parameters of the parametric equation based on an error correction technique, using the operational data.
5. The method of claim 1, wherein determining the dynamic forecasting model comprises determining one among an artificial neural network model, a support vector machine model, and a regression model.
6. The method of claim 1, wherein determining the power estimate value comprises applying the dynamic forecasting model to the plurality of wind turbines of the wind farm.
7. The method of claim 1, wherein controlling the at least one wind turbine comprises:
comparing the power estimate value with an upper power limit value and a lower power limit value;
increasing a wind farm power output towards the upper power limit value if the power estimate value is less than the upper power limit value; and
decreasing the wind farm power output to a value less than the upper power limit value if the power estimate value is greater than the upper power limit value.
8. The method of claim 1, wherein the operational data comprises at least one of an environmental parameter, an operating state, and a power output of a wind turbine among the plurality of wind turbines.
9. The method of claim 1, wherein receiving the operational data comprises acquiring data from at least one of a weather station, a met mast, and another wind farm located outside the wind farm.
10. A system for controlling a wind farm comprising a plurality of wind turbines electrically coupled to a grid, the system comprising:
a data acquisition unit communicatively coupled to the wind farm and configured to receive operational data of the wind farm;
a model generator unit communicatively coupled to the data acquisition unit and configured to:
determine a subset of wind turbines from the plurality of wind turbines, subjected to a change from a first wind condition to a second wind condition;
determine a plurality of operational parameters of the subset of wind turbines, based on the operational data;
determine a dynamic forecasting model based on the plurality of operational parameters;
a power controller unit communicatively coupled to the data acquisition unit and the model generator unit and configured to:
determine a power estimate value based on the dynamic forecasting model and the operational data, wherein the power estimate value is representative of an estimate of power generated by the wind farm at a future instant of time; and
control at least one turbine of the wind farm based on the power estimate value.
11. The system of claim 10, wherein the model generator unit is further configured to select the subset of wind turbines from the plurality of wind turbines, located at a periphery of the wind farm and subjected to the change from the first wind condition to the second wind condition.
12. The system of claim 11, wherein the model generator unit is further configured to measure the plurality of operational parameters of the subset of wind turbines during a training period.
13. The system of claim 10, wherein the model generator unit is further configured to determine the dynamic forecasting model by:
determining a parametric equation for estimating a power output from the subset of wind turbines based on the plurality of operational parameters; and
modifying one or more parameters of the parametric equation based on an error correction technique, using the operational data.
14. The system of claim 10, wherein the model generator unit is further configured to determine the dynamic forecasting model by determining one of an artificial neural network model, a support vector machine model, and a regression model.
15. The system of claim 10, wherein the model generator unit is further configured to determine the power estimate value by applying the dynamic forecasting model to the plurality of wind turbines of the wind farm.
16. The system of claim 10, wherein power controller unit is further configured to control at least one wind turbine by:
comparing the power estimate value with an upper power limit value and a lower power limit value;
increasing a wind farm power output towards the upper power limit value if the power estimate value is less than the lower power limit value; and
decreasing the wind farm power output to a value less than the upper power limit value if the power estimate value is greater than the upper power limit value.
17. The system of claim 10, wherein the operational data comprises at least one of an environmental parameter, an operating state, and a power output of a wind turbine among the plurality of wind turbines.
18. The system of claim 10, wherein the data acquisition unit is further configured to acquire the operational data from at least one of an anemometer, a weather station, a met mast located outside the wind farm, and another wind farm.
, Description:BACKGROUND
[0001] Embodiments of the present invention relate to wind farms and, in particular, to controlling power production from a wind farm.
[0002] A wind farm is a group of wind turbines used for production of electric power. Individual turbines are typically interconnected via a medium voltage power collection system which in turn is connected to a high voltage transmission system or ‘a grid’. A large wind farm may include about 50 to 100 individual wind turbines and span an extended area of hundreds of square kilometers.
[0003] Electricity generated from wind turbines can be highly variable due to different wind conditions. Wind farms are often operated based on the “ramp rates” or rate of change in power (expressed in kilowatts/second) of the particular wind turbines. The wind farm as a whole may have a total power output ramp rate referred to herein as the farm ramp rate (RRfarm) which is the aggregate of all the power change provided to a power collection system. Utility companies in general and entities connected to a wind farm in particular have requirements to prevent fluctuations of the power grid caused by high variations in wind conditions. It is therefore desirable to control power ramp rates of wind farms taking into consideration the maximum prescribed power ramp rates of such power resources to prevent occurrence of higher than acceptable ramp rates.
BRIEF DESCRIPTION
[0004] In accordance with one aspect of the present invention, a method for controlling a wind farm having a plurality of wind turbines electrically coupled to a grid, is disclosed. The method includes receiving operational data of the wind farm and determining a subset of wind turbines from the plurality of wind turbines subjected to a change from a first wind condition to a second wind condition. The method also includes determining a plurality of operational parameters of the subset of wind turbines based on the operational data. Further, the method includes determining a dynamic forecasting model based on the plurality of operational parameters. The method further includes determining a power estimate value based on the dynamic forecasting model and the operational data. The power estimate value is representative of an estimate of power generated by the wind farm at a future instant of time. The method also includes controlling at least one wind turbine of the wind farm based on the power estimate value.
[0005] In accordance with another aspect of the present invention, a system for controlling a wind farm comprising a plurality of wind turbines electrically coupled to a grid is disclosed. The system includes a data acquisition unit communicatively coupled to the wind farm and configured to receive operational data of the wind farm. The system further includes a model generator unit communicatively coupled to the data acquisition unit and configured to determine a subset of wind turbines from the plurality of wind turbines, subjected to a change from a first wind condition to a second wind condition. The model generator is also configured to determine a plurality of operational parameters of the subset of wind turbines, based on the operational data. The model generator is also configured to determine a dynamic forecasting model based on the plurality of operational parameters. The system further includes a power controller unit communicatively coupled to the data acquisition unit and the model generator unit and configured to determine a power estimate value based on the dynamic forecasting model and the operational data. The power estimate value is representative of an estimate of power generated by the wind farm at a future instant of time. The power controller unit is also configured to control at least one turbine of the wind farm based on the power estimate value.
DRAWINGS
[0006] These and other features and aspects of embodiments of the present invention will become better understood when the following detailed description is read with reference to the accompanying drawings in which like characters represent like parts throughout the drawings, wherein:
[0007] FIG. 1 is a block diagram of a system for controlling power production from a wind farm in accordance with an exemplary embodiment;
[0008] FIG. 2 is a signal flow diagram of a system for controlling power production from the wind farm in accordance with an exemplary embodiment;
[0009] FIG. 3 is a schematic illustration of a plurality of wind turbines experiencing a wind front in accordance with an exemplary embodiment;
[0010] FIG. 4 is a graphical representation illustrating a quality of short term estimation of power produced by the wind farm in accordance with an exemplary embodiment; and
[0011] FIG. 5 is a flow chart of a method for controlling power production from a wind farm in accordance with an exemplary embodiment.
DETAILED DESCRIPTION
[0012] Embodiments disclosed herein relate to forecasting of variable short term and medium term power production from a power generation unit such as a wind farm. Specifically, a plurality of operational parameters of the power generation unit is used to determine a dynamic forecasting model for forecasting a short term power production. An estimate of the short term power production is used to control one or more wind turbines of the wind farm.
[0013] FIG. 1 is a diagrammatic illustration of a system 100 for controlling power production from a wind farm 102 in accordance with an exemplary embodiment. The wind farm 102 includes a plurality of wind turbines 104 coupled to an electric grid 106 and configured to generate electric power for the electric grid 106. The system 100 is configured to receive operational data 108 from the wind farm 102 and generate a control signal 122 for controlling power production matching the requirement of the electric grid 106. The system 100 includes a data acquisition unit 112, a model generator unit 114, a power controller unit 116, a processor unit 124, and a memory unit 126 communicatively interconnected via a communication bus 128.
[0014] The data acquisition unit 112 is communicatively coupled to the wind farm 102 and configured to receive the operational data 108 from the wind farm 102. The operational data 108 includes wind farm data of the plurality of wind turbines 104 and neighborhood data of the surroundings of the wind farm 102. The wind farm data may include, for example, wind conditions, blade tip speed ratios, turbine operating states, blade pitch angles, and power generated by each wind turbine 104. The term ‘blade tip speed ratio’ as used herein is a ratio between a tangential speed of a tip of a blade and an actual velocity of wind. The term ‘blade pitch angle’ as used herein is a turning of an angle of attack of a blade during a wind. The wind conditions may be representative of a plurality of wind parameters such as wind speed and wind direction measured at a nacelle of a wind turbine. The wind conditions may also be representative of one or more wind parameters at a periphery of the wind farm 102 or at another wind farm. In some embodiments, one or more parameters of the wind farm data are estimated from other measured parameters. The neighborhood data includes a plurality of parameters measured from the surroundings of the wind farm 102. In one embodiment, the neighborhood data is measured using meteorological instruments disposed on a measurement tower (also referred as “met masts”). In one embodiment, the neighborhood data includes wind speed and wind direction at a location away from a center location of the wind farm 102, which may be representative of an early indication of wind speed variations.
[0015] In one embodiment, the data acquisition unit 112 is configured to acquire at least one of an environmental parameter such as but not limited to a temperature value, a wind parameter value, and a humidity value at a location in the wind farm corresponding to a wind turbine among the plurality of wind turbines 104. The data acquisition unit 112 may further be configured to acquire the operational data 108 from at least one of an anemometer (not shown in FIG.1), a weather station or a met mast 132 located outside the wind farm 102 and another wind farm 130.
[0016] The model generator unit 114 is communicatively coupled to the data acquisition unit 112. The model generator unit 114 is configured to receive the operational data 108 from the data acquisition unit 112 and determine a dynamic forecasting model 120. Further, the model generator unit 114 is configured to determine a subset of wind turbines from the plurality of wind turbines 104 subjected to a change from a first wind condition to a second wind condition (or a ‘wind front’). As used herein, the term ‘wind front’ is representative of a changing wind condition experienced by the wind farm 102. A wind front is representative of wind conditions including at least one of a change in the wind speed and a change in the wind direction that alters the power output of the wind farm 102 either for a short time duration or for a medium time duration. The term “short term” may be representative of a time duration ranging from a few minutes to about 30 minutes. The term “medium term” may be representative of a time duration ranging from about one hour to few hours. The model generator unit 114 is further configured to determine a plurality of operational parameters of the subset of wind turbines experiencing the wind front from the operational data 108. In one specific embodiment, the model generator unit 114 is further configured to perform a data filtering of the operational data 108 to identify one or more wind turbines among the plurality of wind turbines 104 that are operating in a power curtailed mode in order to meet regulatory requirements and to exclude operational data of the wind turbines 104 operating in the power curtailed mode when determining the dynamic forecasting model 120.
[0017] In one embodiment, a probability distributional technique is used to generate the dynamic forecasting model 120. The dynamic forecasting model 120 is indicated by symbol f and represented by the equation:

P_t=f(P_(t-1),P_(i,t-1):i?F) (1)
where Pt is representative of power generated by the wind farm 102 at time instant t, Pt-1 is representative of power generated by the wind farm 102 at time instant t-1, and Pi,t-1 is representative of power generated by ith wind turbine at time instant t-1. The term F is indicative of a subset of wind turbines among the plurality of wind turbines 104 that are considered for generation of the dynamic forecasting model 120. The plurality of wind parameters over a training period T from a present time instant is considered. In one embodiment, the training period T is a pre-defined time period for monitoring a determined set of wind parameters. In another embodiment, the training period T is selected such that the wind conditions are generally observed to be stable.
[0018] In one embodiment, the model generator unit 114 is further configured to estimate the training period T during which the plurality of wind parameters is stable. In another embodiment, the model generator unit 114 is configured to determine a change in the training period T when one or more wind parameters are not stable during the training period T. A longer time period T’ which is greater than the period T, is selected as the training period for generating the dynamic forecasting model 120. The time period T is represented by a time interval T=[0, -m], where in one example, m is an integer representative of time in minutes. A root mean squared error e(T) is determined for the dynamic forecasting model f based on the plurality of wind parameters measured during the time period T. The time period T’ is represented by a time interval [0, -m-s], where s is a positive integer representative of time in minutes. A root mean squared error e(T’) is determined for a dynamic forecasting model f’, using the wind parameters measured during the time period T’. If e(T’)>e(T)+? for a positive threshold ?, the dynamic forecasting model f is continued to be used. Otherwise, the period T’ is used as the training period and the new dynamic forecasting model f’ is considered for the prediction of power production from the plurality of wind turbines 104.
[0019] In another embodiment, the model generator unit 114 is further configured to determine the training period T based on a statistical significance technique. For such a technique, a maximum training period or a subset of the maximum training period is considered. In one embodiment, the maximum training period is represented as N which is equal to fourteen days. Further, the operational data 108 is divided into two sets. The first set of operational data corresponds to most recent k days and the second set of operational data corresponds to last N-k days. A first set of principal components of a vector representing power generated by individual wind turbines 104 based on the first set of operational data, is determined. Further, a first principal component vector is generated based on the first set of principal components. Then, a second set of principal components of a vector representing power generated by individual wind turbines 104 based on the second set of operational data, is determined. Further, a second principal component vector is generated based on the second set of principal components. A statistical significance test is performed for the first principal component vector and the second principal component vector to determine if the first principal component vector and the second principal component vector are distinguishable. Alternate values of k are chosen till the difference between the first principal component vector and the second principal component vector are statistically significant. A value of k=m is selected for building the dynamic forecasting model f when a null hypothesis (difference between the first principal component vector and the second principal component vector is not statistically significant) is rejected.
[0020] In an alternate embodiment, principal component loadings are determined for the first set of operational data. Further, the principal component loadings are stored in the memory unit 126 for future use. The principal component loadings are used to generate the first set of principal components. The principal component loadings corresponding to the second set of operational data which is stored previously in the memory unit 126 are retrieved and used to generate the second set of principal components. The first set of principal components and the second set of principal components are tested using a Hoteling’s T test for statistical difference. The principal component loadings refers to a plurality of left singular vectors obtained from a singular vale decomposition (SVD) multiplied by corresponding singular values.
[0021] In one embodiment, the model generator unit 114 is configured to determine the subset of the plurality of wind turbines 104 subjected to the change from the wind front. Initially, the dynamic forecasting model f is determined as represented by equation (1), considering all of the plurality of wind turbines of the wind farm 102. The dynamic forecasting model f is modified by excluding one or more of the plurality of wind turbines 104 to generate a modified version of the dynamic forecasting model g. The one or more of the plurality of wind turbines 104 to be excluded from the modelling are selected such that the accuracy of the modified dynamic forecasting model is within a specified range with reference to the accuracy of the dynamic forecasting model f. A root mean squared error (RMSE) value may be used to evaluate the accuracy of the dynamic forecasting model f with its modified version g. In alternate embodiments, a minimum absolute deviation value or any other appropriate metric for quantifying the difference between the dynamic forecasting model f and the modified dynamic forecasting model may be used.
[0022] In one embodiment, an F-test is used to select one or more wind turbines 104 to be excluded from the modelling to generate the modified version of the dynamic forecasting model. It may be noted herein that F-test is any statistical test having a test statistic in the form of an F-distribution or T-distribution with reference to the null hypothesis that one or more predictive features in the form of turbine level power is insignificant for predicting farm level power. The modification of the dynamic forecasting model f may be continued by excluding additional wind turbines 104 until the number of wind turbines 104 contributing to the dynamic forecasting model f is less than a pre-determined integer value. In one embodiment, the modification of the dynamic forecasting model f is performed by excluding wind turbines 104 in incremental steps. A greedy algorithm or exhaustive searching based technique may be used to identify the wind turbines 104 to be excluded. In another embodiment, the modification of the dynamic forecasting model f is performed by reintroducing one or more of the excluded wind turbines 104 to the dynamic forecasting model f.
[0023] In another embodiment, a wake model is used to model the wind speed corresponding to a particular wind turbine 104 and estimate short term power production from the wind farm 102. The wake model is representative of ‘wake of a wind turbine characterized by increased turbulence and decreased wind speed.’ The wake model is also used to determine wind speed as function of distance travelled by wind within the wind farm 102. In one example, the wake model comprises a Jensen-Park model. In some embodiments, when an upward ramp in wind speed is observed, the wake model enables the model generator unit 114 to identify a first portion of the wind farm 102 that experiences higher wind speeds. Further, the wake model also enables identification of portions of the wind farm 102 which do not experience changes in the wind speed. In such an embodiment, the model generator unit 114 generates a first dynamic forecasting model for the first portion of the wind farm 102 and a second dynamic forecasting model for other portions of the wind farm 102.
[0024] In one embodiment, the model generator unit 114 is configured to determine a parametric equation for estimating a power output from the subset of wind turbines based on the plurality of operational parameters. A parametric equation is representative of a surface as a function of several variables or parameters and often used to model the power output from a wind farm or a wind turbine. Further, the model generator unit 114 is also configured to determine or modify one or more parameters of the parametric equation based on an error correction technique, using the operational data 108. A training data set having a plurality of operational parameters and corresponding power output is used to determine or modify one or more parameters of the parametric equation. The parameters of the parametric equation may be determined based on a squared error minimization technique. In some embodiments, the model generator unit 114 is configured to determine one of an artificial neural network (ANN) model, a support vector machine (SVM) model, and a regression model such as auto regression (AR) model and auto regressive integrate moving average (ARIMA) model.
[0025] The power controller unit 116 is communicatively coupled to the data acquisition unit 112 and the model generator unit 114 and configured to determine a power estimate value representative of an estimate of power generated by the wind farm 102 at a future instant of time based on the dynamic forecasting model f and the plurality of operational parameters 118. The power controller unit 116 is further configured to control one or more of the plurality of wind turbines 104 of the wind farm 102 based on the power estimate value.
[0026] In one embodiment, the power controller unit 116 is further configured to compare the power estimate value to an upper power limit value and a lower power limit value. The upper power limit value may be representative of a regulatory ceiling for power supply by the wind farm 102 to the electric grid 106. In one embodiment, the lower power limit value may be required to avoid inefficient use of wind farm resources. It may be noted that, the upper power limit value and the lower power limit value may be dynamically set based on the operational data 108 and/or data from the electric grid 106. Specifically, the upper power limit value and lower power limit value may be determined based on power produced from the wind farm 102 at present time, power estimate value indicative of power produced by the wind farm 102 at a future instant of time, and the power required by the electric grid 106. The power controller unit 116 is further configured to increase a wind farm power output towards the upper power limit value if the power estimate value is less than the upper power limit value. In another embodiment, the power controller unit 116 is further configured to decrease the wind farm power output to a value less than the upper power limit value if the power estimate value is greater than the upper power limit value. The power controller unit 116 is configured to change the wind farm power output in steps. The steps may be determined based on at least one of the prescribed ramp rate limits of the wind farm 102. The power controller unit 116 is configured to control the power output of the wind farm 102 by controlling power output of one or more of the plurality of wind turbines 104. In one embodiment, the one or more wind turbines are selected among the plurality of wind turbines 104 based on wind speed profile experienced by the wind turbines 104 and the power estimate value.
[0027] The memory unit 126 is communicatively coupled to the communication bus 128 and may be accessed by one or more of the data acquisition unit 112, the model generator unit 114, and the power controller unit 116. In one exemplary embodiment, the memory unit 126 includes one or more memory modules. The memory unit 126 may be a non-transitory storage medium. For example, the memory unit 126 may be a dynamic random access memory (DRAM) device, a static random access memory (SRAM) device, a digital versatile disc read only memory (DVD-ROM) device, a digital versatile rewritable (DVD-RW) device, a flash memory device, or other non-volatile storage devices. In another embodiment, a non-transitory computer readable medium may be encoded with a program having instructions to instruct the processor unit 124 to perform functions of one or more of the data acquisition unit 112, the model generator unit 114, and the power controller unit 116.
[0028] The processor unit 124 is communicatively coupled to the communications bus 128 and may include at least one or an arithmetic logic unit, a microprocessor, a general purpose controller, and a processor array to perform the desired computations or run the computer programs. In one embodiment, the processor unit 124 may be configured to aid the data acquisition unit 112, the model generator unit 114, and the power controller unit 116 to perform the associated tasks. It may be noted that while the embodiment of FIG. 1 depicts the processor unit 124 as a separate unit, in certain embodiments, one or more of the data acquisition unit 112, the model generator unit 114, and the power controller unit 116 many include at least one processor unit.
[0029] FIG. 2 is a block diagram of a system 200 for controlling power production from the wind farm in accordance with an exemplary embodiment of FIG. 1. The system 200 is a schematic representation of data transformation and processing flow in the system 100. The operational data 108 includes but not limited to an environmental parameter 202, an operating state 204 of one or more of the plurality of wind turbines 104, and a power output 206 of one or more of the plurality of wind turbines 104. The environmental parameter 202 of a particular wind turbine 104 may be a vector having a plurality of scalar parameters such as but not limited to a temperature value, a wind speed value, a wind direction, and a humidity value. The operating state 204 of a particular wind turbine 104 may be in the form of a vector indexed by a time parameter. The power output 206 may be in the form of a time series with suitable sampling intervals. The operational data 108 may further include neighborhood data 208 generated external to the wind farm. In one embodiment, the neighborhood data 208 includes but is not limited to data acquired from the anemometer, the weather station or a met mast 132, and another wind farm 130.
[0030] The operational data 108 is processed by the data acquisition unit 112 to generate a plurality of operational parameters 118, using a dynamic parameter selection processing technique 212. The dynamic parameter selection technique includes selection of a subset of wind turbines 104 of the wind farm 102 and acquisition of the plurality of operational parameters 118 corresponding to the subset of wind turbines 104. The dynamic parameter selection is performed by data acquisition unit 112 and is explained in with reference to a subsequent figure. The dynamic forecasting model 120 is generated by a model estimator 216, based on the plurality of operational parameters 118. In one embodiment, the dynamic forecasting model 120 is determined by estimating one of a regression model 220, a support vector machine (SVM) model 218, or an artificial neural network (ANN) model 222. In one embodiment, the regression model 220 may be one of an auto regressive (AR) model and an auto regressive integrated moving average (ARIMA) model. Further, the model estimator 216 further refines the dynamic forecasting model 120 based on an error correction signal 232 generated by a model error corrector 230. The model error corrector 230 receives a reference data 210 and a corresponding estimate 228 generated by the dynamic forecasting model 120 and generates an error correction signal 232 using the error correction technique. In one embodiment, the reference data 210 is the operational data 108 stored previously in the memory unit 126 and retrieved to be used by the model error corrector 230. A power estimate generator 226 receives the operational data 108, the dynamic forecasting model 120, and generates one or more power estimate values 234. In one embodiment, a subset of the plurality of power estimate values 234 is used for generation of the error correction signal 232.
[0031] FIG. 3 is a schematic diagram 300 illustrating selection of a subset of wind turbines 302 in the wind farm 102 in accordance with an exemplary embodiment. The schematic diagram 300 shows a plurality of wind turbines 104 dispersed within the boundary of the wind farm 102. The schematic diagram 300 also shows a wind front 308 experienced by the wind farm 102. A subset of wind turbines 302 of the plurality of wind turbines 104 experiences the wind front 308 before the remaining wind turbines of the wind farm 102. In the illustrated embodiment, the wind front 308 has a velocity profile 310. In one embodiment, the wind front 308 has a higher initial velocity of six meters per second and a lower velocity of four meters per second at a tail part 306 of the wind front 308. The wind front 308 is experienced by the remaining of the wind turbines among the plurality of wind turbines 104 with additional time delays.
[0032] FIG. 4 is a graph 400 illustrating a quality of short term estimation of power produced by a wind farm in accordance with an exemplary embodiment. The graph 400 includes an x-axis 402 representative of time and a y-axis 404 representative of amplitude of power production. The graph includes three curves 406, 408 and 410 representative of actual power output from the wind farm, power production values obtained from conventional persistence model, and power production values obtained from a dynamic forecasting model respectively. It should be observed herein that the curve 410 corresponding to the dynamic forecasting model is more proximate to the curve 406 corresponding to the actual power output compared to the curve 408 corresponding to the conventional persistence model.
[0033] FIG. 5 is a flow chart 500 of a method for controlling power production from a wind farm in accordance with an exemplary embodiment. The method is performed by the system of FIG.1. The method incudes receiving operational data of a wind farm having a plurality of wind turbines electrically coupled to a grid in step 502. The operational data is measured while the wind farm is generating power and supplying it to the grid. The plurality of wind turbines is provided with a plurality of sensors for measuring a plurality of parameters of the wind turbines. The step of receiving the operational data includes measuring at least one of an environmental parameter, an operating state of a wind turbine, and a power output of the wind turbine among the plurality of wind turbines. The step of receiving the operational data also includes acquiring data from at least one of an anemometer, a weather station, a met mast, and another wind farm located outside the wind farm.
[0034] The method also includes determining a subset of wind turbines from the plurality of wind turbines, subjected to a change from a first wind condition to a second wind condition as in step 504. The method also includes determining a plurality of operational parameters from the operational data, corresponding to a subset of wind turbines among the plurality of wind turbines in step 506. The determining of the plurality of operational parameters includes selecting a subset of wind turbines from the plurality of turbines, typically located at a periphery of the wind farm and experiencing the change in the wind condition. The determining of the plurality of operational parameters also includes measuring one or more parameters of the subset of wind turbines for a pre-specified time period.
[0035] The subset of wind turbines is selected based on the location of the subset of wind turbines which receives the wind front at the earliest time instant. The subset of wind turbines is further selected based on the operational data. Further, the method includes determining a dynamic forecasting model based on the plurality of operational parameters in step 508. The method of determining the dynamic forecasting model includes determining one of an artificial neural network model, a support vector machine model, and a regression model. The dynamic forecasting model is generated by determining a parametric equation for estimating a power output from the subset of wind turbines based on the plurality of operational parameters. The generation of the dynamic forecasting model includes modifying one or more parameters of the parametric equation based on an error correction technique, using the operational data.
[0036] The method also includes determining a power estimate value at a future instant of time from the wind farm based on the dynamic forecasting model and the plurality of operational parameters in step 510. The method also includes controlling one or more wind turbines of the wind farm based on the power estimate value in step 512.
[0037] Specifically, the controlling step includes comparing the power estimate value with an upper power limit value and a lower power limit value. If the power estimate value is less than the lower power limit value, a wind farm power output is increased towards the upper power limit value. If the power estimate value is greater than the upper power limit value, the wind farm power output is decreased to a value less than the upper power limit value.
[0038] Embodiments of the present invention enable the wind farm to operate efficiently and maximize supply of power to the electric grid. As a result, the power production from the wind farm is less dependent on variations in wind speed and direction. The fluctuations in power produced by the wind farm are effectively estimated even during ramp conditions and wind gusts. Further, forced power curtailments are avoided and optimal power production is ensured for a particular wind condition.
[0039] It is to be understood that not necessarily all such objects or advantages described above may be achieved in accordance with any particular embodiment. Thus, for example, those skilled in the art will recognize that the systems and techniques described herein may be embodied or carried out in a manner that achieves or improves one advantage or group of advantages as taught herein without necessarily achieving other objects or advantages as may be taught or suggested herein.
[0040] While the technology has been described in detail in connection with only a limited number of embodiments, it should be readily understood that the specification is not limited to such disclosed embodiments. Rather, the technology can be modified to incorporate any number of variations, alterations, substitutions or equivalent arrangements not heretofore described, but which are commensurate with the spirit and scope of the claims. Additionally, while various embodiments of the technology have been described, it is to be understood that aspects of the specification may include only some of the described embodiments. Accordingly, the specification is not to be seen as limited by the foregoing description, but is only limited by the scope of the appended claims.