CN-121993349-A - Wind driven generator control method, system and server

CN121993349ACN 121993349 ACN121993349 ACN 121993349ACN-121993349-A

Abstract

The invention provides a control method, a system and a server of a wind driven generator, which relate to the field of safety control of the wind driven generator, the method takes a clearance radar as a safety hard constraint, obviously reduces the risk of tower sweeping, can keep the control reliability under the conditions of rain fog, clutter and the like, reduces unnecessary power limit and shutdown occurrence probability, greatly reduces the misjudgment rate of the wind driven generator in the clearance control process, and can couple wind measurement load shedding with the clearance safety in the control process of the wind driven generator, and can carry out feedforward control under the working conditions of gust, shearing, yaw error and the like, thereby reducing the load of the wind driven generator.

Inventors

GUO XIPENG
He Liedong
TIAN QILIANG
ZHANG XINBIN
WANG HANCHEN
Ye Xingfei
LI XINGWEI
QIU GUOLIANG
LIN HANGBING
WU SIHONG
LI QIAN
LI DALIN
ZHANG QING
ZHANG DU
MA BIN

Assignees

华能江西清洁能源有限责任公司
中国华能集团清洁能源技术研究院有限公司

Dates

Publication Date: 20260508
Application Date: 20260324

Claims (10)

1. A method of controlling a wind turbine, the method comprising: Acquiring wind field data acquired by a wind lidar, clearance distance data acquired by a clearance radar and running state data of a wind generating set; Based on the wind field data, the clearance distance data and the running state data, predicting the clearance distance of the wind generating set in a future time window to obtain a clearance predicted value; and when judging that the clearance risk exists according to the clearance predicted value, outputting a control instruction to the wind generating set.
2. The method according to claim 1, wherein the step of predicting a clearance distance of the wind turbine generator set in a future time window based on the wind farm data, the clearance distance data, and the operation state data comprises: aligning the wind field data, the clearance distance data and the unit operation state data according to a time stamp to generate a fusion time sequence data set; And inputting the fusion time sequence data set into a pre-constructed headroom prediction model, wherein the headroom prediction model is used for learning the mapping relation among wind field change, unit response and headroom dynamic evolution, and outputting the headroom prediction value.
3. The method according to claim 2, wherein the headroom prediction model is a deep learning model constructed based on a time series neural network, the method further comprising: Acquiring wind field data, clearance data and unit state data under historical operation conditions as training samples; and training the time sequence neural network to learn a nonlinear mapping from the historical time sequence data to the future clearance distance by taking the actual clearance distance at the future moment as a supervision tag.
4. The method of controlling a wind turbine of claim 1, further comprising: Performing signal quality evaluation on the wind field data and/or the clearance distance data to generate corresponding quality evaluation parameters; And dynamically determining the fusion weight of the wind field data and/or the clearance distance data when predicting the clearance distance according to the quality evaluation parameter.
5. The method of controlling a wind turbine of claim 1, further comprising: Acquiring prediction confidence corresponding to the headroom predicted value; And when the headroom predicted value is lower than a preset safety headroom threshold and the predicted confidence is higher than a preset confidence threshold, judging that headroom risk exists.
6. The method of controlling a wind turbine of claim 1, further comprising: According to the data availability or credibility of the anemometry laser radar and the clearance radar, dynamically adjusting a judgment strategy of clearance risk; when the unavailability or credibility of the anemometry laser radar data is lower than a first threshold value, switching to a first judging mode taking the clearance radar data as priority; when the unavailability or credibility of the clearance radar data is lower than a second threshold value, switching to a second judging mode based on the priority of the wind-measuring laser radar data; and outputting a preset power limiting or stopping instruction when the unavailability or credibility of both sensor data is lower than the corresponding threshold value.
7. The method according to claim 1, wherein the step of outputting a control command to the wind turbine generator set when it is determined that there is a headroom risk based on the headroom predicted value, comprises: acquiring a clearance safety threshold corresponding to the clearance risk, and calculating a difference value between the clearance safety threshold and the clearance predicted value; Determining a constraint condition corresponding to the control instruction based on the difference value, and acquiring the control instruction when the structural load of the wind generating set is the minimum value under the constraint condition; And sending the control instruction to the wind generating set.
8. The wind turbine control method of claim 7, wherein the control command includes at least one of a yaw correction command, a pitch adjustment command, a torque adjustment command, and a power adjustment command; The structural load includes at least one of a tower load, a blade load, a yaw system load, and a drive train load.
9. A wind turbine control system, the system comprising: the data acquisition module is used for acquiring wind field data acquired by the anemometry laser radar, clearance distance data acquired by the clearance radar and running state data of the wind generating set; The clearance prediction value determining module is used for predicting the clearance distance of the wind generating set in a future time window based on the wind field data, the clearance distance data and the running state data to obtain a clearance prediction value; And the control execution module is used for outputting a control instruction to the wind generating set when the clearance risk is judged to exist according to the clearance predicted value.
10. A server comprising a processor and a memory, the memory storing computer executable instructions executable by the processor, the processor executing the computer executable instructions to implement the steps of the wind turbine control method of any one of claims 1 to 8.

Description

Wind driven generator control method, system and server Technical Field The invention relates to the field of wind driven generator control, in particular to a wind driven generator control method, a wind driven generator control system and a wind driven generator control server. Background The risk detection of the blade tower sweeping in the existing wind driven generator is mainly realized through a blade tip clearance monitoring device, and when the blade tip of the blade enters a threshold area, the wind driven generator is triggered to execute protection strategies such as speed reduction, blade collection or shutdown. The method mainly comprises a clearance detection and protection scheme based on a laser radar/camera/IMU, a scheme for calculating clearance based on laser radar ranging and transmitting the clearance into a main control to control, a method for inverting tip vacuum net amount values aiming at three-wire clearance radar ranging, and the like. In the prior art, wind-measuring lidar is widely used for yaw/control optimization, so that additional load is reduced, the generating capacity of a wind driven generator is improved, but the reliability of the wind driven generator is obviously reduced under the conditions of rain and fog and the like, and the problem of higher misjudgment rate of the wind driven generator in the clearance control process under severe weather is caused. Disclosure of Invention In view of the above, the invention aims to provide a control method, a system and a server of a wind driven generator, wherein the method takes a clearance radar as a safety hard constraint, remarkably reduces the risk of tower sweeping, can maintain control reliability under the conditions of rain fog, clutter and the like, reduces unnecessary power limitation and shutdown occurrence probability, greatly reduces the misjudgment rate of the wind driven generator in the clearance control process, and can couple wind measurement load shedding with the clearance safety in the control process of the wind driven generator, perform feedforward control under the working conditions of gust, shearing, yaw error and the like, and reduce the load of the wind driven generator. In a first aspect, an embodiment of the present invention provides a method for controlling a wind turbine, including: Acquiring wind field data acquired by a wind lidar, clearance distance data acquired by a clearance radar and running state data of a wind generating set; Based on wind field data, clearance distance data and running state data, predicting the clearance distance of the wind generating set in a future time window to obtain a clearance predicted value; and when judging that the clearance risk exists according to the clearance predicted value, outputting a control instruction to the wind generating set. Optionally, the step of predicting the clearance distance of the wind generating set in the future time window based on the wind farm data, the clearance distance data and the running state data comprises: aligning the wind field data, the clearance distance data and the running state data according to the time stamp to generate a fusion time sequence data set; And inputting the fusion time sequence data set into a pre-constructed headroom prediction model, wherein the headroom prediction model is used for learning the mapping relation between wind field change, unit response and headroom dynamic evolution and outputting a headroom predicted value. Optionally, the headroom prediction model is a deep learning model constructed based on a time sequence neural network, and the method further comprises the following steps: Acquiring wind field data, clearance data and unit state data under historical operation conditions as training samples; The training time sequence neural network learns a nonlinear mapping from the historical time sequence data to the future clearance distance by taking the actual clearance distance at the future time as a supervision tag. Optionally, the method further comprises: Carrying out signal quality evaluation on wind field data and/or clearance distance data to generate corresponding quality evaluation parameters; and dynamically determining fusion weights of the wind field data and/or the clearance distance data when predicting the clearance distance according to the quality evaluation parameters. Optionally, the method further comprises: Obtaining a prediction confidence corresponding to the headroom predicted value; And when the headroom predicted value is lower than a preset safety headroom threshold and the predicted confidence is higher than a preset confidence threshold, judging that the headroom risk exists. Optionally, the method further comprises: According to the availability or credibility of the data of the wind-measuring laser radar and the clearance radar, dynamically adjusting a judgment strategy of the clearance risk; when the unavailability or credibility of the anemometry