CN-122017287-A - Fault diagnosis method and system for anemometer of wind turbine generator

CN122017287ACN 122017287 ACN122017287 ACN 122017287ACN-122017287-A

Abstract

The invention relates to the technical field of fault diagnosis, in particular to a fault diagnosis method and system of a wind turbine anemometer. The fault diagnosis method can adapt to real-time wind condition changes, ensure that a unit in the community and a unit to be tested keep higher wind speed correlation, improve wind speed fitting accuracy, increase data dimension by fusing environment data and unit health data, effectively offset interference of non-wind speed factors on diagnosis results, reduce misjudgment rate caused by instantaneous air flow fluctuation by adopting a double-threshold judgment mechanism and combining instantaneous residual error and sliding window residual error characteristics, and can realize fault type subdivision and root positioning, provide accurate guidance for operation and maintenance and reduce operation and maintenance cost.

Inventors

GUO RENHONG
Ruan Shiying
ZHANG GEN
Sun Panrong
LI JIANPING
Wu guangke
Liang Zaizhong
DENG ZIQIAO
LIU HESHENG
DENG QIRONG
ZHANG YANG
LI LINYAN

Assignees

广东粤电珠海海上风电有限公司
西安热工研究院有限公司

Dates

Publication Date: 20260512
Application Date: 20260204

Claims (10)

1. A method for fault diagnosis of an anemometer of a wind turbine, comprising: S1, collecting first wind speed data of a wind turbine to be tested and other wind turbines in a wind power plant where the wind turbine to be tested is located, and calculating wind speed correlation coefficients of the wind turbine to be tested and the other wind turbines; S2, incorporating the wind turbine to be tested and the wind turbine with the wind speed correlation coefficient larger than a preset threshold value into a diagnostic community; S3, collecting second wind speed data and active power data of all wind turbines in the diagnostic community, collecting environment data and turbine health data, and fusing to form a multi-dimensional characteristic data set after standardized pretreatment; S4, inputting the multi-dimensional characteristic data set into a pre-trained machine learning model, and calculating a wind speed fitting value of the wind turbine to be tested; s5, calculating an instantaneous residual error between the wind speed fitting value and an anemometer actual measurement value equipped by the wind turbine generator to be tested, and calculating a residual error root mean square in a preset time window; S6, when the instantaneous residual error is larger than a first threshold value and the root mean square of the residual error is larger than a second threshold value, judging that the anemometer has a fault; And S7, extracting residual sequence characteristics in a fault state, determining a fault type based on a preset fault characteristic library, and matching corresponding fault sources.
2. The method for diagnosing faults of the anemometer of the wind turbine according to claim 1, wherein the wind speed correlation coefficient r in the step S1 is calculated by Pearson correlation coefficient, and the calculation formula is as follows: ; In the formula, For the wind speed sequence of the wind turbine to be tested, For other wind speed sequences of the wind turbines, As the average value of the wind speed sequence of the wind turbine to be tested, And the average value of the wind speed sequences of other wind turbines.
3. The method for diagnosing a fault of an anemometer of a wind turbine according to claim 1, wherein the step S1 further comprises: When the wind direction change of the wind power plant exceeds a preset angle and/or the time-lapse flow intensity exceeds a preset value, recalculating the wind speed correlation coefficient; and (3) re-screening the neighbor ratio units meeting the conditions to form a new diagnosis community according to the newly calculated wind speed correlation coefficient, eliminating the neighbor ratio units with the wind speed correlation reduced due to wind direction change, and incorporating the new neighbor ratio units with high wind speed correlation.
4. The method for diagnosing faults of the anemometer of the wind turbine generator according to claim 1, wherein the step of fusing all the data into the multi-dimensional characteristic data set after the standardized preprocessing in the step S3 includes: filling the missing data by adopting a linear interpolation method, and triggering data acquisition abnormal alarm when the continuous missing data exceeds a preset value; The environmental data in the step S3 comprise atmospheric temperature data, relative humidity data and atmospheric pressure data, and the unit health data comprise cabin vibration acceleration data, anemometer power supply voltage data and gearbox oil temperature data.
5. The fault diagnosis method of the wind turbine anemometer according to claim 1, wherein the pre-trained machine learning model in the step S4 is model-trained by adopting a random forest regression model, and the number of decision trees of the random forest regression model is set to be 50-200 by 5-fold cross validation of the training parameters of the optimization model.
6. The method for diagnosing faults of an anemometer of a wind turbine according to claim 1, wherein the step S5 includes: calculating the instantaneous residual The formula is calculated as follows: ; In the formula, For the measured value of the anemometer, Fitting values for the wind speed; setting the preset time window as a sliding time window of 30-120min, and calculating the residual root mean square in the sliding time window The formula is calculated as follows: ; and N is the number of data points of the sliding time window.
7. The fault diagnosis method for the wind turbine anemometer according to claim 1, wherein the first threshold in the step S6 is determined based on a 3 sigma principle of historical normal operation data, and the second threshold is 90% -95% of a fractional number of residual root mean square in a historical normal state.
8. The method for diagnosing faults of a wind turbine anemometer according to claim 1, wherein the step of extracting the residual sequence feature in the fault state in the step S7 includes: Performing wavelet transform decomposition on the residual sequence in the fault state, wherein the residual sequence feature extraction comprises residual average value Residual fluctuation frequency Number of residual mutation ; The fault types comprise measurement drift, fluctuation abnormality, partial jamming and complete jamming.
9. A fault diagnosis system for an anemometer of a wind turbine, comprising: The correlation coefficient calculation module is used for collecting first wind speed data of the wind turbine to be measured and other wind turbine in the wind power plant where the wind turbine to be measured is located and calculating wind speed correlation coefficients of the wind turbine to be measured and other wind turbine; the diagnostic community establishment module is used for bringing the wind turbine to be tested and the wind turbine with the adjacent ratio with the wind speed correlation coefficient larger than a preset threshold value into a diagnostic community; The data set acquisition module is used for acquiring second wind speed data and active power data of all wind turbines in the diagnostic community, acquiring environment data and turbine health data, and fusing the environment data and the turbine health data after standardized pretreatment to form a multi-dimensional characteristic data set; The wind speed fitting calculation module is used for inputting the multi-dimensional characteristic data set into a pre-trained machine learning model and calculating a wind speed fitting value of the wind turbine to be tested; The fault judging module is used for calculating the instantaneous residual error between the wind speed fitting value and the anemometer actual measurement value equipped by the wind turbine generator to be tested and calculating the residual error root mean square in a preset time window; the fault judging module is used for judging that the anemometer has a fault when the instantaneous residual error is larger than a first threshold value and the root mean square of the residual error is larger than a second threshold value; The fault type analysis module is used for extracting residual sequence characteristics in a fault state, determining the fault type based on a preset fault characteristic library and matching corresponding fault sources.
10. A computer readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements a method for fault diagnosis of a wind turbine anemometer according to any one of claims 1 to 8.

Description

Fault diagnosis method and system for anemometer of wind turbine generator Technical Field The invention belongs to the technical field of fault diagnosis, and particularly relates to a fault diagnosis method and system for an anemometer of a wind turbine generator. Background The wind speed data are important variables for evaluating the contents such as the power characteristic, the output performance, the running state and the like of the whole wind turbine in the wind turbine, the wind speed is usually measured by an anemometer arranged in the wind turbine at present, and the accuracy of the wind speed data measured by the anemometer directly influences the evaluation index of the whole wind turbine, so that fault diagnosis is required to be carried out on the anemometer, and the stability and the reliability of the anemometer are ensured. The diagnosis method for the anemometer in the prior art mostly adopts a fault mutual diagnosis community formed by a fixed number of adjacent units, does not consider the reduction of the wind speed correlation among the units caused by the change of real-time wind conditions, reduces fitting accuracy, only depends on basic data such as wind speed, active power and the like, does not merge environment interference with health data of the wind turbine, cannot offset interference of non-wind speed factors on a diagnosis result, and in addition, judges faults through only a single instantaneous residual error threshold, is easy to misjudge due to instantaneous airflow fluctuation, and cannot subdivide fault types and positioning sources. Problems in the existing diagnosis method can lead to low accuracy and reliability of anemometer fault diagnosis, and misjudgment or missed judgment is easy to cause, so that operation and maintenance cost is increased or potential safety hazards are caused. Disclosure of Invention Based on the problems existing in the prior art, the embodiment of the invention aims to provide a fault diagnosis method and system for an anemometer of a wind turbine generator. In order to achieve the above purpose, the present invention adopts the following technical scheme: A fault diagnosis method of a wind turbine anemometer comprises the following steps: S1, collecting first wind speed data of a wind turbine to be tested and other wind turbines in a wind power plant where the wind turbine to be tested is located, and calculating wind speed correlation coefficients of the wind turbine to be tested and the other wind turbines; S2, incorporating the wind turbine to be tested and the wind turbine with the wind speed correlation coefficient larger than a preset threshold value into a diagnostic community; S3, collecting second wind speed data and active power data of all wind turbines in the diagnostic community, collecting environment data and turbine health data, and fusing to form a multi-dimensional characteristic data set after standardized pretreatment; S4, inputting the multi-dimensional characteristic data set into a pre-trained machine learning model, and calculating a wind speed fitting value of the wind turbine to be tested; s5, calculating an instantaneous residual error between the wind speed fitting value and an anemometer actual measurement value equipped by the wind turbine generator to be tested, and calculating a residual error root mean square in a preset time window; S6, when the instantaneous residual error is larger than a first threshold value and the root mean square of the residual error is larger than a second threshold value, judging that the anemometer has a fault; And S7, extracting residual sequence characteristics in a fault state, determining a fault type based on a preset fault characteristic library, and matching corresponding fault sources. The invention is further improved in that the wind speed correlation coefficient r in the step S1 is calculated by a Pearson correlation coefficient, and the calculation formula is as follows: ; In the formula, For the wind speed sequence of the wind turbine to be tested,For other wind speed sequences of the wind turbines,As the average value of the wind speed sequence of the wind turbine to be tested,And the average value of the wind speed sequences of other wind turbines. The invention is further improved in that the step S1 further comprises: When the wind direction change of the wind power plant exceeds a preset angle and/or the time-lapse flow intensity exceeds a preset value, recalculating the wind speed correlation coefficient; and (3) re-screening the neighbor ratio units meeting the conditions to form a new diagnosis community according to the newly calculated wind speed correlation coefficient, eliminating the neighbor ratio units with the wind speed correlation reduced due to wind direction change, and incorporating the new neighbor ratio units with high wind speed correlation. The invention further improves that the step of fusing all the data to form the multi-dimensional