METHOD FOR CONTROLLING WIND TURBINES OF A WIND PARK USING A TRAINED AI MODEL

Abstract

A method for controlling wind turbines. Incident signal data is obtained from wind turbines and fed to an artificial intelligence (AI) model in order to identify patterns in the incident signals generated by the wind turbines. One or more actions are associated to the identified patterns, based on identified actions performed by the wind turbines in response to the generated incident signals. During operation of the wind turbines, one or more incident signals from one or more wind turbines are detected and compared to patterns identified by the AI model. In the case that the detected incident signal(s) match(es) at least one of the identified patterns, the wind turbine(s) are controlled by performing the action(s) associated with the matching pattern(s).

Claims

1. A method for controlling wind turbines of a wind park, the wind park comprising a plurality of wind turbines, the method comprising: obtaining incident signal data from a plurality of data providing wind turbines, the incident signal data including incident signals generated by the data providing wind turbines; feeding the incident signal data to an artificial intelligence (AI) model and training the AI model by means of the incident signal data in order to identify patterns in the incident signals generated by the data providing wind turbines; identifying actions performed by the data providing wind turbines in response to the generated incident signals, and, based thereon, associating one or more actions to the identified patterns; during operation of the wind turbines of the wind park, detecting one or more incident signals from one or more wind turbines of the wind park; comparing the detected incident signal(s) to patterns identified by the AI model; and in the case that the detected incident signal(s) match(es) at least one of the identified patterns, controlling the wind turbine(s) of the wind park by performing the action(s) associated with the matching pattern(s).

2. The method of claim 1, further comprising monitoring wind conditions at a site of the wind park, and wherein controlling the wind turbine(s) of the wind park is further based on the monitored wind conditions.

3. The method of claim 1, wherein the incident signals include warnings and/or alarms.

4. The method of claim 1, wherein the actions being associated with the identified patterns include shutting down one or more wind turbines.

5. The method of claim 1, wherein controlling the wind turbines of the wind park comprises shutting down one or more wind turbines of the wind park, before shut down procedures at the one or more wind turbines have been initiated based on warnings or alarms generated by the wind turbines themselves.

6. The method of claim 1, further comprising: obtaining incident signal data from the wind turbines of the wind park, during operation of the wind park; feeding the obtained incident signal data to the AI model; and retraining the AI model based on the incident signal data, thereby improving the AI model.

7. The method of claim 1, further comprising determining that an extreme weather event is occurring, based on comparing the detected incident signal(s) to patterns identified by the AI model.

8. The method of claim 1, wherein the patterns identified by the AI model include patterns in which identical or similar incident signals have been generated by two or more data providing wind turbines arranged in the same wind park.

9. The method of claim 1, further comprising controlling at least one of the wind turbines of the wind park by performing an action, in response to a match between detected incident signal(s) of another wind turbine of the wind park and at least one of the identified patterns.

10. The method of claim 1, further comprising monitoring wind conditions at a site of the wind park, and wherein controlling the wind turbine(s) of the wind park is further based on the monitored wind conditions; and wherein the actions being associated with the identified patterns include shutting down one or more wind turbines.

11. The method of claim 10, wherein controlling the wind turbines of the wind park comprises shutting down one or more wind turbines of the wind park, before shut down procedures at the one or more wind turbines have been initiated based on warnings or alarms generated by the wind turbines themselves.

12. A method for controlling wind turbines of a wind park, the wind park comprising a plurality of wind turbines, the method comprising: obtaining first incident signal data from a plurality of wind turbines, the first incident signal data including incident signals generated by the wind turbines; feeding the first incident signal data to an artificial intelligence (AI) model and training the AI model by means of the first incident signal data in order to identify patterns in the incident signals generated by the wind turbines; identifying actions performed by the wind turbines in response to the generated incident signals, and, based thereon, associating one or more actions to the identified patterns; during operation of the wind turbines of the wind park, detecting one or more incident signals from one or more wind turbines of the wind park; comparing the detected one or more incident signals to patterns identified by the AI model; in the case that the detected one or more incident signals match at least one of the identified patterns, controlling respective one or more wind turbines by performing the action associated with the matching pattern; obtaining second incident signal data from the wind turbines, during operation of the wind park; feeding the obtained second incident signal data to the AI model; and retraining the AI model based on the incident signal data, thereby improving the AI model; and determining that an extreme weather event is occurring, based on comparing the detected incident signal(s) to patterns identified by the retrained AI model.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The invention will now be described in further detail with reference to the accompanying drawings in which

[0054] FIG. 1 illustrates an extreme weather event, in the form of a tornado, travelling through a wind park, and

[0055] FIG. 2 illustrates control of wind turbines of a wind park in accordance with a method according to an embodiment of the invention.

DETAILED DESCRIPTION OF THE DRAWINGS

[0056] FIG. 1 illustrates a wind park 1 comprising a plurality of wind turbines 2. An extreme weather event, in the form of a tornado 3, is travelling through the wind park 1 along a direction indicated by arrow 4.

[0057] Wind turbines 2a and 2b have already been passed by the tornado 3. In response thereto, the wind turbines 2a and 2b have generated one or more incident signals, e.g. in the form of one or more warnings and/or one or more alarms. In response to the generated incident signals, the wind turbines 2a and 2b have performed one or more relevant actions in order to protect the wind turbines 2a, 2b from the impact of the tornado 3. For instance, the wind turbines 2a, 2b may have been shut down.

[0058] Since wind turbine 2a and wind turbine 2b have both been passed by the tornado 3, the sequence or pattern of incident signals generated by the two wind turbines 2a, 2b are most likely identical or at least very similar. However, since the tornado 3 passed wind turbine 2a before it passed wind turbine 2b, there may be a time delay between the incident signals generated by wind turbine 2a and the corresponding incident signals generated by wind turbine 2b.

[0059] Wind turbine 2c is about to be passed by the tornado 3. Accordingly, wind turbine 2c has probably already generated at least some of the incident signals which were generated by wind turbines 2a and 2b. The incident signals generated by wind turbine 2c are compared to patterns of incident signals identified by means of an AI model, e.g. in a manner which has been described previously and/or in a manner which will be described in further detail below with reference to FIG. 2.

[0060] The comparison may reveal that there is a match between one of the identified patterns and the incident signals generated by wind turbine 2c. It may further be revealed that the matching pattern also matches the incident signals generated by wind turbines 2a and 2b, and that the generated incident signals are compatible with the passage of a tornado 3. Thus, as soon as such a match is identified, wind turbine 2c is controlled in accordance therewith, e.g. by shutting down the wind turbine 2c, even if the warnings and/or alarms generated by the wind turbine 2c itself have not yet caused the wind turbine 2c to shut down.

[0061] Furthermore, since the tornado 3 is heading towards wind turbine 2d, wind turbine 2d may also have generated, e.g., the first of a sequence of incident signals, corresponding to the incident signals generated by wind turbines 2a, 2b and 2c. As described above, it has already been established that wind turbines 2a, 2b and 2c are generating identical or similar sequences of incident signals, and that this pattern is compatible with a tornado 3 passing through the wind park 2. The fact that the three wind turbines 2a, 2b and 2c have generated the same or similar sequences of incident signals may also be regarded as a pattern of incident signals. Accordingly, as soon as wind turbine 2d generates the first incident signal of the pattern of incident signals, it may be concluded that it is very likely that the tornado 3 will reach wind turbine 2d very soon. Thus, already at this stage, relevant actions, such as shutting down wind turbine 2d, may be performed.

[0062] FIG. 2 illustrates control of wind turbines 2 of a wind park 1 in accordance with a method according to an embodiment of the invention.

[0063] A plurality of data providing wind turbines 5, three of which are shown, arranged in a plurality of data providing wind parks 6, one of which is shown, collect data during operation of the data providing wind turbines 5. The collected data comprises incident signal data, including incident signals, such as warnings and alarms, generated by the data providing wind turbines 5. The incident signal data may further comprise information regarding actions performed by the data providing wind turbines in response to the generated incident signals.

[0064] The data providing wind turbines 5 provide the incident signal data to a local data hub 7 of the data providing wind park 6. The incident signal data from the local data hubs 7 of each of the data providing wind parks 6 is provided to a central data hub 8, and the central data hub 8 feeds the incident signal data to an AI model 9.

[0065] The wind turbines 2 of wind park 1 may also act as data providing wind turbines in the manner described above. In this case incident signal data generated by wind turbines 2 are also provided to the AI model 9, in the same manner as the incident signal data generated by the data providing wind turbines 5 of wind park 6.

[0066] The AI model is trained based on the incident signal data. Thereby patterns in the incident signals generated by the data providing wind turbines 5 are identified. Furthermore, the patterns are labelled by associating actions performed by the data providing wind turbines 5, in response to the generated incident signals, to the patterns. Thus, the labelled patterns indicate which combinations of incident signals were generated by the data providing wind turbines 5, as well as which actions the data providing wind turbines 5 performed in response thereto.

[0067] The trained model is provided to a local detector 10 of the wind park 1 to be controlled. During operation of the wind turbines 2 of the wind park 1, the wind turbines 2 collect data, including incident signals generated by the wind turbines 2, and provide this to a local data hub 11 of the wind park 1. The generated incident signals are then compared to the patterns identified by the AI model. In the case that there is a match between the incident signals generated by at least one of the wind turbines 2 and at least one of the identified patterns, the detector 10 instructs the relevant wind turbine(s) 2 to perform the action(s) associated with the matching pattern(s), e.g. shutting down one or more of the wind turbines 2.

[0068] Furthermore, the local data hub 11 provides the incident signals generated by wind turbines 2 to the central data hub 8, which in turn feeds this data to the AI model. The AI model is then re-trained, based on the additional incident signals, thereby obtaining an updated and improved model.

[0069] The trained data model may further be distributed to other wind parks, such as the data providing wind park 6. In this case the trained model is provided to a detector 12 of the data providing wind park 6, and the data providing wind turbines 5 can be controlled in the same manner as described above with reference to the wind turbines 2 of the wind park 1.