A METHOD FOR COMPUTER-IMPLEMENTED ANALYSIS OF A WIND FARM COMPRISING A NUMBER OF WIND TURBINES

Abstract

Provided is a computer-implemented analysis of a wind farm including wind turbines, including applying an object detection algorithm to a digital image showing the current state of the earth's ground in a surrounding area of the wind farm which extracts and localizes detected objects not belonging to the wind farm within the image; identifying specific objects out of detected objects, each specific object having an influence on the air flow occurring at the rotor of at least one wind turbine; determining whether there is a change with respect to the number of specific objects in comparison to an earlier state of the earth's ground in the surrounding area of the wind farm; and performing one or more actions if there is a change with respect to the number of specific objects in comparison to the earlier state of the earth's ground in the surrounding area of the wind farm.

[00001]$\mathrm{Figure}1$

Claims

1. A method for computer-implemented analysis of a wind farm comprising a plurality of wind turbines with respective rotors, wherein at each time point of one or more time points the following steps are performed: i) applying an object detection algorithm to a digital image showing a current state of the earth's ground in a surrounding area of the wind farm, resulting in an extraction and localization of a number of detected objects not belonging to the wind farm within the digital image; ii) identifying a number of specific objects out of the number of detected objects, each specific object having an influence on an air flow occurring at the rotor of at least one wind turbine of the wind farm; iii) determining whether there is a change with respect to the number of specific objects in comparison to an earlier state of the earth's ground in the surrounding area of the wind farm; and iv) performing one or more actions in case that there is a change with respect to the number of specific objects in comparison to the earlier state of the earth's ground in the surrounding area of the wind farm.

2. The method according to claim 1, wherein the one or more actions comprise: generating an output via a user interface, the output informing a user about the change with respect to the number of specific objects; and/or adapting a control procedure which controls an operation of the wind farm.

3. The method according to claim 1, wherein the number of specific objects comprises: one or more further wind turbines of another wind farm and/or one or more trees and/or one or more ground inclinations and/or one or more ground depressions and/or one or more buildings.

4. The method according to claim 1, wherein the air flow change caused by the change of the number of specific objects is determined, where one or more effects on the wind farm caused by an air flow change are derived, thus resulting in an adaptation of the control procedure of the wind farm based on the one or more effects.

5. The method according to claim 4, wherein one effect is a change of the mechanical load acting on one or more wind turbines of the wind farm.

6. The method according to claim 5, wherein the control procedure is adapted such that a life-time of one or more wind turbines of the wind farm is not decreased or is increased.

7. The method according to claim 4, wherein one effect is a change of the generated electric power of one or more wind turbines of the wind farm.

8. The method according to claim 7, wherein the control procedure is adapted such that an electric energy production of the wind farm within a predetermined time interval is increased.

9. The method according to claim 1, wherein steps i) to iv) are repeated in case that an updated digital image is available.

10. The method according to claim 1, wherein the object detection algorithm is based on a trained data driven model.

11. The method according to claim 10, wherein the data driven model is a neural network.

12. The method according to claim 1, wherein the digital image is an image taken by a flying object over ground, the flying object being a satellite or a plane or a drone.

13. A system for computer-implemented analysis of a wind farm comprising a plurality of wind turbines with respective rotors, where the system comprises a processor configured to carry out a method, in which at each time point of one or more time points the following steps are performed: i) applying an object detection algorithm to a digital image showing a current state of the earth's ground in a surrounding area of the wind farm, resulting in an extraction and localization of a number of detected objects not belonging to the wind farm within the digital image; ii) identifying a number of specific objects out of the number of detected objects, each specific object having an influence on an air flow occurring at the rotor of at least one wind turbine of the wind farm; iii) determining whether there is a change with respect to the number of specific objects in comparison to an earlier state of the earth's ground in the surrounding area of the wind farm; and iv) performing one or more actions in case that there is a change with respect to the number of specific objects in comparison to the earlier state of the earth's ground in the surrounding area of the wind farm.

14. (canceled)

15. A computer program product, comprising a computer readable hardware storage device having computer readable program code stored therein, said program code executable by a processor of a computer system to implement the method according to claim 1.

16. A computer program with program code for carrying out the method according to claim 1 when the program code is executed on a computer.

Description

BRIEF DESCRIPTION

[0029] Some of the embodiments will be described in detail, with reference to the following figures, wherein like designations denote like members, wherein:

[0030] FIG. 1 shows a flow chart illustrating the steps according to a method based on an embodiment of the invention; and

[0031] FIG. 2 shows a system for performing the method as shown in FIG. 1.

DETAILED DESCRIPTION

[0032] Embodiments of the invention as described in the following refers to the analysis of a wind farm already in operation. The control procedure of the wind farm was initially specified when launching the wind farm. At this time, obstacles around the location of the wind farm were taken into account in order to define a control procedure resulting in a high energy production of the wind farm and a low mechanical load on the wind turbines of the farm.

[0033] In order to update this control procedure, digital satellite images IM provided from the earth's ground in the surrounding area of the wind farm are checked in regular intervals. To do so, the method of FIG. 1 has access to a database comprising such digital satellite images. E.g., the satellite images IM are taken from a central database which is publicly available. An example of a satellite image IM is shown schematically in FIG. 1. The digital image comprises the wind farm 1 having five wind turbines 2. Furthermore, two wind turbines 3 of another wind farm 4 as well as a building 3′ are located in the surrounding area of the wind farm 1. The wind turbines 2 are shown schematically by white horizontal bars in the image IM of FIG. 1. Analogously, the wind turbines 3 are shown schematically by hatched horizontal bars. Moreover, the building 3′ is shown schematically by a hatched square. When the wind farm 1 was launched, the turbines 3 and the building 3′ where not present and, thus, not taken into account when determining the control procedure of the wind farm.

[0034] In the method of FIG. 1, the image IM is analyzed in step S1 by an object detection algorithm ODA. As a result of step S1, the wind turbines 3 as well as the building 3′ are extracted from the image IM. In the embodiment described herein, a well-known algorithm based on a Convolutional Neural Network CNN is used for detecting objects within the image. The Convolutional Neural Network is trained based on training images having known objects of known object types shown therein. As known from the prior art, a Convolutional Neural Network comprises convolutional layers followed by pooling layers as well as fully connected layers in order to extract and classify the objects within an image.

[0035] In step S2 of the method of FIG. 1, it is determined which of the objects 3, 3′ have an influence on the air flow occurring at the rotor of at least one turbine 2 of the wind farm 1. In the scenario of FIG. 1, it is determined that the wind turbines 3 of wind farm 4 have an influence on this air flow due to the wake of the rotors of these wind turbines. Contrary to that, the building 3′ does not affect the air flow occurring at the rotors of the turbines 2. Hence, as a result of step S2, it is determined that the wind turbines 3 are specific objects having an influence on the air flow occurring at the rotor of at least one turbine of the wind farm 1.

[0036] In a next step S3, the image IM is compared with a digital satellite image taken from the surrounding area of the wind farm 1 at an earlier time than the image IM, e.g. at the time when the wind farm was commissioned. Alternatively, the image IM may be compared with other digital data describing the objects in the surrounding area of the wind farm at an earlier time, e.g. with digital data from the planning phase of the wind farm 1. In the embodiment described herein, it is determined that the wind turbines 3 were not present in the digital satellite image taken at an earlier time from the surrounding area of the wind farm 1.

[0037] As a result of step S3, is it determined that there is a change with respect to the number of specific objects in comparison to an earlier state of the earth's ground in a surrounding area of the wind farm. As a result, appropriate actions AC are performed in step S4. Particularly, an output OU informing about this change is provided via a user interface UI, e.g. via a visual user interface in the form of display. The user interface UI may be located at the site of the wind farm operator so that the operator is informed about this change. As a consequence, the operator can initiate appropriate steps for changing the control strategy of the wind farm, if necessary.

[0038] In the embodiment described herein, an automatic adaption of the control procedure CP (i.e. the control program) of the wind farm is performed as another action AC in step S4 in order to optimize the electric energy production and/or the life time of the wind farm taking into account that additional wind turbines 3 are located adjacent to the wind farm 1. In other words, the control procedure CP is changed so that a higher energy production and/or a longer life time is achieved in comparison to the non-changed control procedure.

[0039] FIG. 2 shows an embodiment of a system in order to perform the method as described with respect to FIG. 1. The system is implemented as a platform based on a central server SE comprising a processor PR which is used in order to perform the above described method steps S1 to S4. To do so, the server SE has access to up-to-date satellite images IM where the server can process the data of a plurality of different wind farms.

[0040] After having analyzed an image IM, the corresponding actions AC are automatically performed in case of a change with respect to the specific objects. I.e., the output OU informing about this change is provided via the user interface UI and the control procedure CP used for controlling the wind farm is optimized. Moreover, further outputs may be generated by the user interface UI, e.g. an output providing information about the adaption of the control procedure.

[0041] Embodiments of the invention as described in the foregoing has several advantages. Particularly, the conditions with respect to the operation of a wind farm can be checked regularly based on updated images of the wind farm. In one embodiment, those images are satellite images. Nevertheless, the images may also refer to images taken by other flying objects (plane or drone) or by a camera installed on ground at the location of the wind farm. In a preferred embodiment, the control procedure of the wind farm is adapted automatically in case that the operation conditions of the wind farm have changed due to changes in its environment.

[0042] Although the present invention has been disclosed in the form of preferred embodiments and variations thereon, it will be understood that numerous additional modifications and variations could be made thereto without departing from the scope of the invention.

[0043] For the sake of clarity, it is to be understood that the use of “a” or “an” throughout this application does not exclude a plurality, and “comprising” does not exclude other steps or elements.