1. A data anomaly detection method of a wind generating set is characterized by comprising the following steps:

acquiring trend change data of the wind generating set in a preset time period;

performing transverse detection and longitudinal detection on the acquired trend change data, wherein the transverse detection is used for determining whether the trend change data has continuous abnormality, and the longitudinal detection is used for determining whether the trend change data has transient mutation;

and determining whether the wind generating set has data abnormity or not based on the detection results of the transverse detection and the longitudinal detection.

2. The data anomaly detection method according to claim 1, wherein the step of obtaining trend change data of the wind turbine generator set within a predetermined time period comprises:

acquiring instantaneous data of the wind generating set,

determining statistics of the acquired instantaneous data at each statistical cycle, and determining each of the statistics determined within a predetermined period of time as trend change data,

the preset time period comprises a plurality of statistical cycles, and the duration of each statistical cycle is consistent with the duration of the data acquisition cycle of the wind power plant where the wind generating set is located and the data acquisition cycle of the monitoring control system.

3. The data anomaly detection method according to claim 2, characterized in that the statistical value of the acquired instantaneous data at any statistical period is determined by:

determining whether the number of the acquired instant data reaches a counting number, wherein the counting number is determined according to the duration of any one statistical period and the duration of a sampling period of the instant data;

if the number of the acquired transient data reaches the counting number, determining the statistic value of the transient data in any statistic period based on the transient data of the number.

4. The data abnormality detection method according to claim 2 or 3, characterized in that the statistical value of the instantaneous data in any one statistical period includes an average value of the instantaneous data acquired in said any one statistical period or a maximum value among the instantaneous data acquired in said any one statistical period.

5. The data anomaly detection method according to claim 2, further comprising preprocessing the acquired transient data as follows:

comparing the value of the acquired instantaneous data with a data rating;

if the value of the acquired instantaneous data is not greater than the data rated value, keeping the value of the instantaneous data unchanged;

if the value of the acquired instantaneous data is greater than the data rating, replacing the value of the instantaneous data with the data rating,

wherein the statistical value is determined based on the pre-processed transient data.

6. The data abnormality detection method according to claim 1, characterized in that the step of performing lateral detection and longitudinal detection on the acquired trend change data includes:

determining a lateral tendency index of the acquired trend change data, the lateral tendency index being used to characterize a fluctuation duration of abnormal data in the trend change data,

determining a longitudinal trend index of the acquired trend change data, the longitudinal trend index being used to characterize a fluctuating rate of change of the trend change data,

wherein it is determined whether the wind turbine generator set has data anomaly based on the determined lateral tendency index and the longitudinal tendency index.

7. The data anomaly detection method according to claim 6, characterized in that the step of determining a lateral trend index of the acquired trend change data comprises:

sequentially performing the following steps for each adjacent first trend change data and second trend change data in the predetermined time period according to the time sequence:

comparing the value of the first trend change data with a set threshold,

if the value of the first trend change data is not less than the set threshold, no counting statistics are carried out,

if the value of the first trend change data is less than the set threshold, comparing the value of the second trend change data with the set threshold,

if the value of the second trend change data is smaller than the set threshold value, the counting value of the counting statistics is increased by one,

if the value of the second trend change data is not less than the set threshold value, stopping counting statistics, comparing the current count value with the stored count value,

if the current count value is greater than the stored count value, updating the stored count value with the current count value,

if the current count value is not greater than the stored count value, not updating the stored count value;

the stored final count value is determined as a lateral trend index of the trend change data.

8. The data abnormality detection method according to claim 6, characterized in that the step of determining a longitudinal trend index of the acquired trend change data includes:

determining the minimum value and the maximum value of the trend change data in the preset time period, taking the determined minimum value as a first endpoint value of a numerical range, taking the determined maximum value as a second endpoint value of the numerical range, and equally dividing the numerical range to obtain a plurality of numerical intervals;

determining a numerical value interval to which each trend change data in the preset time period belongs;

acquiring a first number of trend change data contained in a numerical value interval of the first number close to the first endpoint value;

acquiring a second number of trend change data contained in a numerical value interval of a second number close to the second endpoint value;

and calculating the sum of the first number and the second number, calculating the ratio of the sum to the total number of the trend change data in the preset time period, and determining the ratio as a longitudinal trend index of the trend change data.

9. The data anomaly detection method according to claim 6, wherein the step of determining whether the wind turbine generator set has the data anomaly based on the detection results of the transverse detection and the longitudinal detection comprises the following steps:

comparing the determined transverse trend index with a first set value and comparing the determined longitudinal trend index with a second set value;

if the determined transverse trend index is not smaller than a first set value and the determined longitudinal trend index is not smaller than a second set value, determining that the wind generating set has data abnormity;

and if the determined transverse trend index is smaller than a first set value and/or the determined longitudinal trend index is smaller than a second set value, determining that the wind generating set has no data abnormality.

10. A data anomaly detection device of a wind generating set is characterized by comprising:

the trend data acquisition module is used for acquiring trend change data of the wind generating set in a preset time period;

the data detection module is used for carrying out transverse detection and longitudinal detection on the acquired trend change data, wherein the transverse detection is used for determining whether the trend change data has continuous abnormality, and the longitudinal detection is used for determining whether the trend change data has transient mutation;

and the data anomaly identification module is used for determining whether the wind generating set has data anomaly or not based on the detection results of the transverse detection and the longitudinal detection.

11. The data anomaly detection device according to claim 10, wherein the trend data acquisition module includes:

the instantaneous data acquisition submodule acquires instantaneous data of the wind generating set,

a statistic determination sub-module that determines statistics of the acquired instantaneous data at each statistic cycle and determines each of the statistics determined within a predetermined time period as trend change data,

the preset time period comprises a plurality of statistical cycles, and the duration of each statistical cycle is consistent with the duration of the data acquisition cycle of the wind power plant where the wind generating set is located and the data acquisition cycle of the monitoring control system.

12. The data anomaly detection device of claim 11, wherein the statistics determination sub-module determines statistics of the acquired transient data at any statistical period by:

determining whether the number of the acquired instant data reaches a counting number, wherein the counting number is determined according to the duration of any one statistical period and the duration of a sampling period of the instant data;

if the number of the acquired transient data reaches the counting number, determining the statistic value of the transient data in any statistic period based on the transient data of the number.

13. The data abnormality detection apparatus according to claim 11 or 12, characterized in that the statistical value of the instantaneous data in any one statistical period includes an average value of the instantaneous data acquired in said any one statistical period or a maximum value among the instantaneous data acquired in said any one statistical period.

14. The data anomaly detection device according to claim 11, wherein the trend data acquisition module further comprises: the data preprocessing submodule is used for preprocessing the acquired instantaneous data as follows:

comparing the value of the acquired instantaneous data with a data rating;

if the value of the acquired instantaneous data is not greater than the data rated value, keeping the value of the instantaneous data unchanged;

if the value of the acquired instantaneous data is greater than the data rating, replacing the value of the instantaneous data with the data rating,

wherein the statistics determination sub-module determines the statistics based on the pre-processed transient data.

15. The data anomaly detection device according to claim 10, wherein the data detection module comprises:

a transverse detection sub-module for determining a transverse trend index of the acquired trend change data, the transverse trend index being used for representing the fluctuation duration of abnormal data in the trend change data,

a longitudinal detection sub-module for determining a longitudinal trend index of the acquired trend change data, the longitudinal trend index being used for characterizing the fluctuation change rate of the trend change data,

wherein the data anomaly identification module determines whether the wind generating set has data anomalies based on the determined transverse trend index and the longitudinal trend index.

16. The data anomaly detection device of claim 15, wherein the lateral detection sub-module determines the lateral trend index by:

sequentially performing the following steps for each adjacent first trend change data and second trend change data in the predetermined time period according to the time sequence:

comparing the value of the first trend change data with a set threshold,

if the value of the first trend change data is not less than the set threshold, no counting statistics are carried out,

if the value of the first trend change data is less than the set threshold, comparing the value of the second trend change data with the set threshold,

if the value of the second trend change data is smaller than the set threshold value, the counting value of the counting statistics is increased by one,

if the value of the second trend change data is not less than the set threshold value, stopping counting statistics, comparing the current count value with the stored count value,

if the current count value is greater than the stored count value, updating the stored count value with the current count value,

if the current count value is not greater than the stored count value, not updating the stored count value;

the stored final count value is determined as a lateral trend index of the trend change data.

17. The data anomaly detection device of claim 15, wherein the longitudinal detection sub-module determines the longitudinal trend index by:

determining the minimum value and the maximum value of the trend change data in the preset time period, taking the determined minimum value as a first endpoint value of a numerical range, taking the determined maximum value as a second endpoint value of the numerical range, and equally dividing the numerical range to obtain a plurality of numerical intervals;

determining a numerical value interval to which each trend change data in the preset time period belongs;

acquiring a first number of trend change data contained in a numerical value interval of the first number close to the first endpoint value;

acquiring a second number of trend change data contained in a numerical value interval of a second number close to the second endpoint value;

and calculating the sum of the first number and the second number, calculating the ratio of the sum to the total number of the trend change data in the preset time period, and determining the ratio as a longitudinal trend index of the trend change data.

18. The data anomaly detection device of claim 15, wherein the data anomaly identification module compares the determined lateral trend index with a first set value and compares the determined longitudinal trend index with a second set value; if the determined transverse trend index is not smaller than a first set value and the determined longitudinal trend index is not smaller than a second set value, determining that the wind generating set has data abnormity; and if the determined transverse trend index is smaller than a first set value and/or the determined longitudinal trend index is smaller than a second set value, determining that the wind generating set has no data abnormality.

19. A controller, comprising:

a processor;

a memory for storing a computer program which, when executed by the processor, implements a method of data anomaly detection for a wind park according to any one of claims 1 to 9.

20. A computer-readable storage medium, in which a computer program is stored, which, when being executed by a processor, carries out a method of data anomaly detection for a wind park according to any one of claims 1 to 9.

Background

The wind generating set is a set of relatively complex system, and at present, a MW-level permanent magnet wind generator highly integrates comprehensive disciplines such as aerodynamics, structural mechanics, electromechanics, material science, power electronic technology, power system analysis, relay protection technology, automatic control technology, modern communication and the like, and becomes a set of complex energy conversion system. Therefore, the same fault may occur due to different reasons, for example, a "three-axis angle inconsistency" fault of the pitch system is taken as an example, the fault may occur due to a fault of the encoder itself, a fault of the power supply of the encoder, a fault of the pitch system, a fault of the drive, and a fault of the acquisition module of the controller.

In order to analyze the fault reason of the wind generating set, fault data when the wind generating set is in fault needs to be recorded, the function can automatically, accurately and timely record the change conditions of various electrical quantities in the processes before and after the fault occurs at the moment of the fault, and the function plays an important role in analyzing and processing the accident and judging whether the protection acts correctly or not through analyzing and comparing the various electrical quantities and playing an important role in safe and reliable operation of the wind generating set.

The technology is a means for embodying the value of big data and an advancing foundation stone, and the practice is the final value embodying of the big data. For wind power generation, batch acquisition, storage and operation of the operating data of the wind generating set can be realized through the big data platform, and data support is provided for technicians.

However, when data analysis is actually performed, it is found that the used fault file is recorded and generated only after the wind turbine generator system fails (for example, operation data from 30 seconds before the failure to 30 seconds after the failure is recorded), so that the operation data of the wind turbine generator system does not generate the fault file in a month or longer, and the operation data available for analysis cannot be obtained, which causes a great degree of condition influence on fault early warning. And because the number of the related wind generating sets is huge and the workload is huge, the data acquisition mode is difficult to realize.

In addition, from the industrial technology line, the working method of the data analysis and fault early warning system is generally as follows: acquiring mass historical data of all measuring points of equipment to establish a model; calculating a real-time data pre-estimated value; and comparing the deviation between the actual value and the estimated value, and generating early warning when the deviation is abnormal. Therefore, data acquisition is a prerequisite condition for realizing fault early warning and is a basis for realizing fault early warning.

From the technical route of data analysis, generally, the shorter the sampling period is, the larger the amount of recoverable information is, so that the fault file mainly generated by the wind generating set is currently applied to data analysis, the sampling period of the fault file is 20 milliseconds, and the original data and the data change trend can be obtained to the greatest extent. However, the fault file can be generated only when the wind generating set fails and stops, and real-time monitoring and fault early warning of the wind generating set are inconvenient.

At present, instantaneous Data files for Data analysis are derived by an SCADA (Supervisory Control And Data Acquisition), And the derived Data is continuous Data, but the sampling period of the SCADA is long, And generally 5 to 7 seconds are used for recording a value, And the sampling period is too long, so that the Data change process within 5 to 7 seconds cannot be recorded, And the Data analysis result is not accurate enough.

Taking the change situation of the capacitance voltage of the pitch system of the wind generating set as an example, fig. 1 shows a waveform curve of the capacitance voltage recorded by a fault file, and fig. 2 shows a waveform curve of the capacitance voltage recorded by SCADA. As can be seen from fig. 1 and 2, although the data recorded by the fault file with the sampling period of 20 milliseconds greatly fluctuates, in the SCADA data with the sampling period of 7 seconds, only one abnormal data value shown in the red box can be acquired, i.e., in the SCADA-derived data, it is difficult to restore the variation trend of the data.

For example, assuming that the data obtained with a 20 ms sampling period is 2.1, 2.6, 2.4, 2.0, 7.1, 3.1 … … 2.8.8, since the sampling period of the SCADA is long, only the 1 st data 2.1 and the last data 2.8 may be acquired, so that the middle 7.1 jump cannot be acquired, and therefore, it cannot be accurately determined whether the data is abnormal. And the data recorded by the fault file is recorded after the data is greater than the fault threshold, and if the fault threshold is 8.0, the fault file can only record the data greater than 8.0, but the data of 7.1 which occurs before cannot be recorded.

Disclosure of Invention

An object of an exemplary embodiment of the present invention is to provide a method and an apparatus for detecting data anomaly of a wind turbine generator system, so as to overcome at least one of the above-mentioned disadvantages.

In one general aspect, there is provided a data anomaly detection method of a wind turbine generator system, the data anomaly detection method including: acquiring trend change data of the wind generating set in a preset time period; performing transverse detection and longitudinal detection on the acquired trend change data, wherein the transverse detection is used for determining whether the trend change data has continuous abnormality, and the longitudinal detection is used for determining whether the trend change data has transient mutation; and determining whether the wind generating set has data abnormity or not based on the detection results of the transverse detection and the longitudinal detection.

Optionally, the step of obtaining trend change data of the wind generating set in a predetermined time period may comprise: the method comprises the steps of obtaining instantaneous data of the wind generating set, determining a statistic value of the obtained instantaneous data in each statistic period, and determining each statistic value determined in a preset time period as trend change data, wherein the preset time period can comprise a plurality of statistic periods, and the duration of each statistic period is consistent with the duration of a data acquisition period of a data acquisition and monitoring control system of a wind power plant where the wind generating set is located.

Alternatively, the statistical value of the acquired instantaneous data at any statistical period can be determined by: determining whether the number of the acquired instant data reaches a counting number, wherein the counting number is determined according to the duration of any one statistical period and the duration of a sampling period of the instant data; if the number of the acquired transient data reaches the counting number, determining the statistic value of the transient data in any statistic period based on the transient data of the number.

Alternatively, the statistical value of the instantaneous data in any statistical period may include an average value of the instantaneous data acquired in any statistical period or a maximum value of the instantaneous data acquired in any statistical period.

Optionally, the data anomaly detection method may further include preprocessing the acquired transient data as follows: comparing the value of the acquired instantaneous data with a data rating; if the value of the acquired instantaneous data is not greater than the data rated value, keeping the value of the instantaneous data unchanged; if the value of the acquired transient data is greater than the data rating, replacing the value of the transient data with the data rating, wherein the statistical value is determined based on the preprocessed transient data.

Optionally, the step of performing horizontal detection and vertical detection on the acquired trend change data may include: determining a transverse trend index of the acquired trend change data, wherein the transverse trend index is used for representing the fluctuation duration of abnormal data in the trend change data, and determining a longitudinal trend index of the acquired trend change data, wherein the longitudinal trend index is used for representing the fluctuation change rate of the trend change data, and whether the wind generating set has data abnormality or not can be determined based on the determined transverse trend index and the longitudinal trend index.

Optionally, the step of determining a lateral trend index of the acquired trend change data may comprise: sequentially performing the following steps for each adjacent first trend change data and second trend change data in the predetermined time period according to the time sequence: comparing the value of the first trend change data with a set threshold, if the value of the first trend change data is not less than the set threshold, not counting, if the value of the first trend change data is less than the set threshold, comparing the value of the second trend change data with the set threshold, if the value of the second trend change data is less than the set threshold, adding one to the counting value of the counting, if the value of the second trend change data is not less than the set threshold, stopping counting, comparing the current counting value with the stored counting value, if the current counting value is greater than the stored counting value, updating the stored counting value with the current counting value, and if the current counting value is not greater than the stored counting value, not updating the stored counting value; the stored final count value is determined as a lateral trend index of the trend change data.

Optionally, the step of determining a longitudinal trend index of the acquired trend change data may comprise: determining the minimum value and the maximum value of the trend change data in the preset time period, taking the determined minimum value as a first endpoint value of a numerical range, taking the determined maximum value as a second endpoint value of the numerical range, and equally dividing the numerical range to obtain a plurality of numerical intervals; determining a numerical value interval to which each trend change data in the preset time period belongs; acquiring a first number of trend change data contained in a numerical value interval of the first number close to the first endpoint value; acquiring a second number of trend change data contained in a numerical value interval of a second number close to the second endpoint value; and calculating the sum of the first number and the second number, calculating the ratio of the sum to the total number of the trend change data in the preset time period, and determining the ratio as a longitudinal trend index of the trend change data.

Optionally, the step of determining whether the wind turbine generator set has data abnormality or not based on the detection results of the transverse detection and the longitudinal detection may include: comparing the determined transverse trend index with a first set value and comparing the determined longitudinal trend index with a second set value; if the determined transverse trend index is not smaller than a first set value and the determined longitudinal trend index is not smaller than a second set value, determining that the wind generating set has data abnormity; and if the determined transverse trend index is smaller than a first set value and/or the determined longitudinal trend index is smaller than a second set value, determining that the wind generating set has no data abnormality.

In another general aspect, there is provided a data abnormality detection apparatus of a wind turbine generator system, the data abnormality detection apparatus including: the trend data acquisition module is used for acquiring trend change data of the wind generating set in a preset time period; the data detection module is used for carrying out transverse detection and longitudinal detection on the acquired trend change data, wherein the transverse detection is used for determining whether the trend change data has continuous abnormality, and the longitudinal detection is used for determining whether the trend change data has transient mutation; and the data anomaly identification module is used for determining whether the wind generating set has data anomaly or not based on the detection results of the transverse detection and the longitudinal detection.

Optionally, the trend data acquisition module may include: the wind generating set data acquisition and monitoring system comprises an instantaneous data acquisition submodule and a statistic value determination submodule, wherein the instantaneous data acquisition submodule acquires instantaneous data of the wind generating set, the statistic value determination submodule determines the statistic value of the acquired instantaneous data in each statistic period and determines each statistic value determined in a preset time period as trend change data, the preset time period can comprise a plurality of statistic periods, and the duration of each statistic period is consistent with the duration of a data acquisition period of a data acquisition and monitoring control system of a wind power plant where the wind generating set is located.

Alternatively, the statistics determination sub-module may determine statistics of the acquired transient data at any statistical period by: determining whether the number of the acquired instant data reaches a counting number, if so, determining the counting number according to the duration of any one statistical period and the duration of a sampling period of the instant data; a statistical value of the instantaneous data at any one statistical period is determined based on the number of instantaneous data.

Alternatively, the statistical value of the instantaneous data in any statistical period may include an average value of the instantaneous data acquired in any statistical period or a maximum value of the instantaneous data acquired in any statistical period.

Optionally, the trend data acquiring module may further include: the data preprocessing submodule is used for preprocessing the acquired instantaneous data as follows: comparing the value of the acquired instantaneous data with a data rating; if the value of the acquired instantaneous data is not greater than the data rated value, keeping the value of the instantaneous data unchanged; if the value of the acquired transient data is greater than the data rating, replacing the value of the transient data with the data rating, wherein the statistics determination sub-module may determine the statistics based on the pre-processed transient data.

Optionally, the data detection module may include: the transverse detection submodule determines a transverse trend index of the acquired trend change data, the transverse trend index is used for representing the fluctuation duration of abnormal data in the trend change data, the longitudinal detection submodule determines a longitudinal trend index of the acquired trend change data, the longitudinal trend index is used for representing the fluctuation change rate of the trend change data, and the data abnormality identification module can determine whether the wind generating set has data abnormality or not based on the determined transverse trend index and the longitudinal trend index.

Alternatively, the lateral detection sub-module may determine the lateral trend index by: sequentially performing the following steps for each adjacent first trend change data and second trend change data in the predetermined time period according to the time sequence: comparing the value of the first trend change data with a set threshold, if the value of the first trend change data is not less than the set threshold, not counting, if the value of the first trend change data is less than the set threshold, comparing the value of the second trend change data with the set threshold, if the value of the second trend change data is less than the set threshold, adding one to the counting value of the counting, if the value of the second trend change data is not less than the set threshold, stopping counting, comparing the current counting value with the stored counting value, if the current counting value is greater than the stored counting value, updating the stored counting value with the current counting value, and if the current counting value is not greater than the stored counting value, not updating the stored counting value; the stored final count value is determined as a lateral trend index of the trend change data.

Alternatively, the longitudinal detection sub-module may determine the longitudinal trend index by: determining the minimum value and the maximum value of the trend change data in the preset time period, taking the determined minimum value as a first endpoint value of a numerical range, taking the determined maximum value as a second endpoint value of the numerical range, and equally dividing the numerical range to obtain a plurality of numerical intervals; determining a numerical value interval to which each trend change data in the preset time period belongs; acquiring a first number of trend change data contained in a numerical value interval of the first number close to the first endpoint value; acquiring a second number of trend change data contained in a numerical value interval of a second number close to the second endpoint value; and calculating the sum of the first number and the second number, calculating the ratio of the sum to the total number of the trend change data in the preset time period, and determining the ratio as a longitudinal trend index of the trend change data.

Optionally, the data anomaly identification module may compare the determined lateral trend index with a first set value and compare the determined longitudinal trend index with a second set value; if the determined transverse trend index is not smaller than a first set value and the determined longitudinal trend index is not smaller than a second set value, determining that the wind generating set has data abnormity; and if the determined transverse trend index is smaller than a first set value and/or the determined longitudinal trend index is smaller than a second set value, determining that the wind generating set has no data abnormality.

In another general aspect, there is provided a controller comprising: a processor; a memory for storing a computer program which, when executed by the processor, implements the above-described method of data anomaly detection for a wind turbine generator set.

In another general aspect, there is provided a computer readable storage medium having stored thereon a computer program which, when being executed by a processor, implements the above-mentioned method of data anomaly detection for a wind park.

By adopting the data anomaly detection method and device of the wind generating set, the calculated amount of data can be effectively reduced, and the data statistics precision can be improved.

Drawings

The above and other objects, features and advantages of exemplary embodiments of the present invention will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings which illustrate exemplary embodiments.

FIG. 1 shows a waveform of instantaneous data of a capacitive voltage of a pitch system recorded by a wind generating set;

FIG. 2 illustrates an instantaneous data waveform of a capacitance voltage of a pitch system recorded by a data acquisition and monitoring control System (SCADA);

FIG. 3 shows a flow chart of a method of data anomaly detection of a wind park according to an exemplary embodiment of the present invention;

FIG. 4 shows a flowchart of the steps of determining statistics of transient data at each statistical period according to an exemplary embodiment of the present invention;

FIG. 5 illustrates a waveform of instantaneous data of the capacitive voltage of the pitch system recorded by a wind park according to an exemplary embodiment of the invention;

FIG. 6 illustrates a trend data waveform of a capacitive voltage of a pitch system according to an exemplary embodiment of the present invention;

FIG. 7 illustrates a waveform of instantaneous data of the capacitive voltage of the pitch system recorded by a wind park according to another exemplary embodiment of the invention;

FIG. 8 illustrates a trend data waveform of a capacitive voltage of a pitch system according to another exemplary embodiment of the present invention;

FIG. 9 shows a flowchart of the steps of determining a lateral trend index for trend change data according to an exemplary embodiment of the present invention;

FIG. 10 is a flowchart illustrating steps for determining a longitudinal trend index of trend change data according to an exemplary embodiment of the present invention;

FIG. 11 shows a block diagram of a data anomaly detection device of a wind park according to an exemplary embodiment of the present invention;

FIG. 12 illustrates a block diagram of a trend data acquisition module according to an exemplary embodiment of the present invention;

FIG. 13 illustrates a communication topology of a monitoring center with various wind turbine generator sets within a wind farm in accordance with an exemplary embodiment of the present invention;

fig. 14 illustrates a block diagram of a controller according to an exemplary embodiment of the present invention.

Detailed Description

Various example embodiments will now be described more fully with reference to the accompanying drawings, in which some example embodiments are shown.

Fig. 3 shows a flow chart of a data anomaly detection method of a wind park according to an exemplary embodiment of the present invention.

Referring to fig. 3, in step S10, trend change data of the wind turbine generator set over a predetermined period of time is acquired.

For example, transient data of the wind turbine generator set may be acquired, statistics of the acquired transient data at each statistical cycle may be determined, and the statistics of the determined transient data over a predetermined period of time may be determined as trend change data.

Here, the transient data of the wind turbine generator set may include various operating parameters of the wind turbine generator set collected by a controller of the wind turbine generator set, and the above-described data abnormality detection method may be performed for any one of the operating parameters. As an example, the predetermined period of time may comprise a plurality of statistical cycles, preferably the duration of the statistical cycles is identical to the duration of the data acquisition cycle of the wind park in which the wind generating set is located and of the supervisory control SCADA system.

Taking a controller of a wind generating set as an example of a PLC (programmable logic controller), the time of a sampling period for the PLC to acquire various operating parameters is generally 20 milliseconds, an SCADA system of a wind power plant acquires data from each wind generating set according to the data sampling period, and the time of the data sampling period of the SCADA system is generally 5-7 seconds. In an exemplary embodiment of the invention, within the controller of the wind park, statistics of the instantaneous data of the wind park within the same time window as the data acquisition period of the SCADA system are calculated to upload the calculated statistics to the SCADA system at each data sampling period of the SCADA system.

In an example of the present invention, the statistical value of the instantaneous data at each statistical period may be determined based on a trend averaging method. Here, the trend average method is a method of calculating a moving average of trend values and then obtaining trend change data by calculating a moving average of the trend values based on a plurality of time-series moving averages calculated by the moving average method.

The step of determining the statistical value of the instantaneous data at each statistical period is described below with reference to fig. 4. It should be understood that the method for determining the statistical value of the instantaneous data in each statistical period shown in fig. 4 is only an example, and the present invention is not limited thereto, and the statistical value may be determined in other manners.

Fig. 4 shows a flowchart of the steps of determining statistics of transient data at each statistical period according to an exemplary embodiment of the present invention.

Referring to fig. 4, in step S101, transient data of the wind turbine generator set is collected.

In step S102, the acquired instantaneous data are stored in the array in order, and counting is started.

For example, the collected instantaneous data can be sequentially stored in an array according to the time sequence, and the number of the instantaneous data stored in the array can be counted.

In step S103, it is determined whether the number of transient data within the array reaches the count number. Here, the duration of the statistical period is characterized by the number of counts.

For example, the number of counts may be determined based on the duration of any one of the statistical periods and the duration of the sampling period of the instantaneous data. As an example, the ratio of the duration of the arbitrary statistical period to the duration of the sampling period of the instantaneous data may be determined as the number of counts.

In an example, assuming that the sampling period of the PLC of the wind turbine generator system is 20 milliseconds, the duration of the statistical period is 5 seconds (i.e., the data acquisition period of the SCADA system), and at this time, the number of counts is: 5 seconds × 1000 ÷ 20 milliseconds ═ 250.

If the number of the transient data in the array does not reach (is less than) the count number, the step S101 is executed to continue to collect transient data.

If the number of transient data in the array reaches (equals) the count number, step S104 is performed: and determining the statistical value of the instantaneous data in the statistical period based on the instantaneous data in the array, clearing the count value, and returning to execute the step S101 to continue to calculate the statistical value of the next statistical period.

As an example, the statistical value of the instantaneous data in any statistical period may include, but is not limited to, an average value of each instantaneous data acquired in any statistical period, or a maximum value among each instantaneous data acquired in any statistical period. In addition, the statistical value may include a variance or a residual of each instantaneous data acquired in any statistical period.

The existing continuous calculation method adopting the sliding average value still has a large data calculation amount because the data calculation updating period is also the using period (such as 20 milliseconds) of the controller of the wind generating set, and the method cannot be well matched with the SCADA system.

After the processing, the data acquired by the SCADA system in each data acquisition period is not data of a time point any more, but is a statistical value of instantaneous data in a statistical period, the statistical value can reflect the variation trend of the data in the statistical period, and data abnormity can be detected subsequently based on the trend variation data acquired by the SCADA system, so that fault early warning judgment is carried out.

FIG. 5 illustrates a waveform of instantaneous data of the capacitive voltage of the pitch system recorded by a wind park according to an exemplary embodiment of the invention. FIG. 6 illustrates a trend data waveform of a capacitive voltage of a pitch system according to an exemplary embodiment of the invention.

In this example, by taking the instantaneous data of the wind generating set as the capacitor voltage of the pitch system as an example, fig. 5 shows a waveform diagram of the operation data of the capacitor voltage in the fault file collected on site, where the sampling period is 20 milliseconds, and it can be seen from fig. 5 that the capacitor voltage has great fluctuation and the fluctuation value is asymmetric. Fig. 6 shows trend data of the wind turbine generator system, namely a waveform diagram of data collected by the SCADA system, and it can be seen from fig. 6 that the trend data obtained based on the method can well restore the data change trend.

FIG. 7 illustrates a waveform of instantaneous data of the capacitive voltage of the pitch system recorded by a wind park according to another exemplary embodiment of the invention. FIG. 8 illustrates a trend-change data waveform of a capacitive voltage of a pitch system according to another exemplary embodiment of the invention.

In this example, by taking the instantaneous data of the wind generating set as the capacitor voltage of the pitch system as an example, fig. 7 shows a waveform diagram of the operation data of the capacitor voltage in the fault file collected on site, where the sampling period is 20 milliseconds, and it can be seen from fig. 7 that the capacitor voltage has great fluctuation and the fluctuation values are symmetrical. Fig. 8 shows trend data of the wind turbine generator system, namely a waveform diagram of data collected by the SCADA system, and it can be seen from fig. 8 that the trend data obtained based on the method can well restore the data change trend.

In a preferred example, the data abnormality detection method of a wind turbine generator system according to an exemplary embodiment of the present invention may further include: the acquired transient data is preprocessed to determine a statistical value at each statistical period based on the preprocessed transient data.

For example, the process of pre-processing the acquired transient data may be as follows: comparing the value of the acquired instantaneous data with a data rating; if the value of the acquired instantaneous data is not greater than (less than or equal to) the data nominal value, keeping the value of the instantaneous data unchanged; if the value of the acquired instantaneous data is greater than the data rating, the value of the instantaneous data is replaced with the data rating. Here, the data nominal values for different instantaneous data differ, and when the comparison is performed, the value of the acquired instantaneous data is compared with the corresponding data nominal value.

Taking the instantaneous data of the wind generating set as the capacitance voltage of the pitch system as an example, the acquired instantaneous data can be preprocessed in a pitch controller or a controller of the wind generating set, assuming that the rated value of the capacitance voltage is 85 volts, if the acquired value of the capacitance voltage is not more than 85 volts, the value is kept unchanged, and if the acquired value of the capacitance voltage is more than 85 volts, the value is calculated as a statistical value according to 85 volts. Through the preprocessing, the error influence caused by the fact that the average value of the upper fluctuation value and the lower fluctuation value is a normal value can be eliminated.

The purpose of the pretreatment is as follows: since instantaneous data fluctuates up and down as shown in fig. 7 when it fluctuates, it is difficult to detect data abnormality if the average value is directly found. For example, assuming that the value of the normal data is 4 and the acquired data series are 4, 6, 2, and 4, and here, the values 6 and 2 are abnormal, but if the average value is directly obtained, the result is (4+6+2+4) ÷ 4 ═ 4, and the statistical value obtained at this time cannot reflect the abnormal condition of the data.

In one example, for a case where the statistical value of the transient data in any statistical period is a maximum value among the transient data acquired in any statistical period, the maximum value may be determined by an envelope detection method.

Here, by using software to perform envelope detection on transient data, whether the data fluctuates or not can be detected more accurately, and the accuracy of the fault warning can be improved.

Returning to fig. 3, in step S20, the acquired trend change data is subjected to lateral detection and longitudinal detection. Here, the horizontal direction detection is used to determine whether or not there is a persistent abnormality in the trend change data, and the vertical direction detection is used to determine whether or not there is a transient sudden change in the trend change data.

In order to improve the accuracy of detecting the trend change data, in the exemplary embodiment of the present invention, the data trend is detected and determined in a dual and complementary manner by means of transverse and longitudinal cross detection.

Compared with the traditional slope detection method (calculating the slope of the curve), the method disclosed by the embodiment of the invention has the advantages that the horizontal detection and the longitudinal detection are carried out on the trend change data, the curve digitization is realized, the more accurate and reliable detection precision is realized, and the slope detection method cannot reflect the duration time of the data fluctuation.

The purpose of the transverse detection is to detect whether a continuous section of data in the trend change data is smaller than a set threshold, and the detection method can automatically eliminate single data jump on one hand and can detect the fluctuation time of the original data on the other hand.

For example, a lateral trend index and a longitudinal trend index of the acquired trend change data may be determined to determine whether a data anomaly exists for the wind turbine generator set based on the determined lateral trend index and longitudinal trend index.

As an example, the lateral trend index is used to characterize the duration of fluctuation of abnormal data in trend change data. The process of determining the lateral trend index of the trend change data is described below with reference to fig. 9. It should be understood that the manner of determining the lateral tendency index shown in fig. 9 is merely an example, and the present invention is not limited thereto, and the lateral tendency index may be determined in other manners.

FIG. 9 illustrates a flowchart of the steps of determining a lateral trend index for trend change data according to an exemplary embodiment of the present invention.

Referring to fig. 9, in step S201, the value of the ith trend data (e.g., the first trend data) is compared with a set threshold, i.e., it is determined whether the value of the ith trend data is less than the set threshold.

Before this, trend change data over a predetermined period of time may be acquired, and the length of the trend change data (i.e., the total number of trend change data) may be determined.

As an example, the set threshold may be half of the sum of the maximum statistical value and the minimum statistical value in the trend change data over a predetermined period of time. However, the present invention is not limited to this, and the size of the set threshold may be determined in other ways.

If it is determined that the value of the ith tendency change data is not less than (greater than or equal to) the set threshold value, step S208 is performed.

If the value of the ith trend data is determined to be smaller than the set threshold, executing step S202: and comparing the value of the (i + 1) th trend change data with a set threshold value, namely, determining whether the value of the (i + 1) th trend change data is smaller than the set threshold value.

If it is determined that the value of the (i + 1) th trend data (e.g., the second trend data) is smaller than the set threshold, step S203 is executed: the count value of the count statistic is incremented by one.

If the value of the (i + 1) th trend data is determined to be not less than the set threshold, the data is considered to be normal, at this time, counting statistics is stopped, and step S204 is executed: the current count value is determined.

In step S205, the current count value is compared with the stored count value, i.e., it is determined whether the current count value is greater than the stored count value.

If it is determined that the current count value is greater than the stored count value, step S206 is performed: the stored count value is updated with the current count value.

If it is determined that the current count value is not greater than (less than or equal to) the stored count value, step S207 is performed: the stored count value is not updated.

In the case where the value of the (i + 1) th trend change data is not less than the set threshold, after step S206 or step S207 is executed, the current count value needs to be cleared, and in the case where the value of the (i + 1) th trend change data is less than the set threshold, the count value does not need to be cleared.

In step S208, it is determined whether i +1 is equal to m, where the initial value of i is 1, m represents the total number of trend change data in a predetermined period of time, and m is a natural number greater than zero.

If it is determined that i +1 is not equal to m, step S209 is performed: so that i is i +1, and returns to execute step S201.

If it is determined that i +1 is equal to m, step S210 is performed: the stored final count value is determined as a lateral trend index of the trend change data.

The maximum value (i.e., the count value) of the consecutive number of data having a downward trend in the trend change data can be obtained by the above-described detection process shown in fig. 9.

In the above-described embodiment of performing lateral detection on trend change data, the width of the abnormal data fluctuation in the trend change data can be accurately detected, but the depth of the trend change data fluctuation cannot be detected.

In the exemplary embodiment of the present invention, by performing longitudinal detection on the trend change data, on one hand, whether an instantaneous abrupt change (linear descent) state occurs in the trend change data can be detected, and on the other hand, the proportion of a larger value and a smaller value in the trend change data can be detected, so as to distinguish the data from small data fluctuation, thereby avoiding misjudgment. In the exemplary embodiment of the invention, the trend change data is transversely and longitudinally detected in a crossed manner, so that the defect of a single detection method is overcome, and the comprehensive data detection condition is covered.

As an example, the longitudinal trend index is used to characterize the fluctuating rate of change of trend change data. The process of determining the longitudinal trend index of the trend change data is described below with reference to fig. 10. It should be understood that the manner of determining the longitudinal tendency index shown in fig. 10 is merely an example, and the present invention is not limited thereto, and the longitudinal tendency index may be determined in other manners.

FIG. 10 shows a flowchart of the steps of determining a longitudinal trend index of trend change data according to an exemplary embodiment of the present invention.

Referring to fig. 10, in step S220, a numerical range to which each of the trend change data within a predetermined period of time belongs is determined.

For example, a plurality of numerical intervals may be obtained by: the method includes the steps of determining a minimum value and a maximum value of trend change data in a predetermined period of time, taking the determined minimum value as a first endpoint value of a numerical range, taking the determined maximum value as a second endpoint value of the numerical range, and equally dividing the numerical range to obtain a plurality of numerical intervals.

In step S230, a first number of trend change data included in a numerical range of a first number near the first endpoint value is acquired.

In step S240, a second number of trend change data included in the numerical range of the second number close to the second endpoint value is acquired.

In one example, assuming that a numerical range from the maximum value to the minimum value is 10-equally divided to obtain 10 numerical intervals, the number of the trend change data included in the two numerical intervals near the maximum value may be determined as a first number, and the number of the trend change data included in the two numerical intervals near the minimum value may be determined as a second number, for example, the number of the 1 st and 2 nd numerical intervals, and the number of the trend change data in the 9 th and 10 th numerical intervals among the divided 10 numerical intervals.

In step S250, a sum of the first number and the second number is calculated, and a ratio of the sum to the total number of the trend change data in a predetermined period of time is calculated, and the calculated ratio is determined as a longitudinal trend index of the trend change data.

Here, the ratio of the sum of the numbers of the trend change data included in the four value ranges to the total number reflects the ratio of a larger value to a smaller value in the trend change data, and a larger ratio indicates that the trend change data in the middle portion is smaller, the data is closer to a right angle in a sudden change (e.g., a drop) or the probability of a slight fluctuation in the surface trend change data is smaller.

Returning to fig. 3, in step S30, it is determined whether there is a data abnormality in the wind turbine generator set based on the detection results of the lateral detection and the longitudinal detection.

For example, the determined lateral tendency index is compared with a first set value, and the determined longitudinal tendency index is compared with a second set value; if the determined transverse trend index is not smaller than a first set value and the determined longitudinal trend index is not smaller than a second set value, determining that the wind generating set has data abnormity; and if the determined transverse trend index is smaller than a first set value and/or the determined longitudinal trend index is smaller than a second set value, determining that the wind generating set has no data abnormality. Here, the first setting value and the second setting value may be set empirically, but the present invention is not limited thereto, and may be set by other means, and a person skilled in the art may set the magnitudes of the first setting value and the second setting value according to actual needs (detection accuracy).

The data anomaly detection method can be used for calculating relevant statistics on the operation data of the wind generating set, and the statistical accuracy is improved. For example, for the collection of the current and the voltage of the variable pitch motor, the current and the voltage of the variable pitch motor are transient in the operation process of the wind generating set, if the collected instantaneous data are transmitted to the SCADA system, the result is extremely inaccurate if the data such as the motor efficiency, the motor power value, the motor torque and the like are calculated according to the feathering data because the sampling period of the SCADA system is too long, and the data anomaly detection method can reduce the calculated amount of the data and improve the accuracy of data statistics.

Fig. 11 shows a block diagram of a data anomaly detection apparatus of a wind park according to an exemplary embodiment of the present invention.

As shown in fig. 11, the data abnormality detecting apparatus 100 of the wind turbine generator set according to the exemplary embodiment of the present invention includes: a trend data acquisition module 101, a data detection module 102 and a data anomaly identification module 103.

Specifically, the trend data acquisition module 101 acquires trend change data of the wind turbine generator set in a predetermined time period.

Fig. 12 shows a block diagram of the trend data acquisition module 101 according to an exemplary embodiment of the present invention.

As shown in fig. 12, the trend data acquisition module 101 according to an exemplary embodiment of the present invention may include: an instantaneous data acquisition sub-module 11 and a statistics determination sub-module 12.

Specifically, the transient data acquisition submodule 11 acquires transient data of the wind turbine generator set. Here, the transient data of the wind turbine may refer to various operating parameters of the wind turbine collected by a controller of the wind turbine.

The statistical value determination sub-module 12 determines the statistical value of the acquired instantaneous data at each statistical cycle, and determines the statistical value of each instantaneous data determined within a predetermined time period as the trend change data.

Here, the predetermined period of time may include a plurality of statistical cycles, the duration of the statistical cycles being consistent with the duration of the data acquisition cycle of the wind park in which the wind generating set is located and the data acquisition cycle of the supervisory control system.

In a preferred example, the statistics determination sub-module 12 may determine statistics of the acquired transient data at any statistical period in the following manner.

Determining the number of counts according to the duration of any statistical period and the duration of the sampling period of the instantaneous data; and determining whether the number of the acquired instantaneous data reaches the counting number, and if the number of the acquired instantaneous data reaches the counting number, determining a statistical value of the instantaneous data in any statistical period based on the instantaneous data of the number. And if the number of the transient data in the array does not reach the counting number, continuously acquiring the transient data.

As an example, the statistical value of the instantaneous data in any statistical period may include, but is not limited to, an average value of each instantaneous data acquired in any statistical period, or a maximum value among each instantaneous data acquired in any statistical period.

Preferably, the trend data acquisition module 101 according to an exemplary embodiment of the present invention may further include: a data preprocessing sub-module (not shown) that preprocesses the acquired transient data, at which time the statistics determination sub-module 12 determines statistics based on the preprocessed transient data.

The data preprocessing submodule carries out preprocessing on the acquired instantaneous data in the following steps: comparing the value of the acquired instantaneous data with a data rating; if the value of the acquired instantaneous data is not greater than the data rated value, keeping the value of the instantaneous data unchanged; if the value of the acquired instantaneous data is greater than the data rating, the value of the instantaneous data is replaced with the data rating.

Returning to fig. 11, the data detection module 102 performs lateral detection and longitudinal detection on the acquired trend change data. Here, the horizontal direction detection is used to determine whether or not there is a persistent abnormality in the trend change data, and the vertical direction detection is used to determine whether or not there is a transient sudden change in the trend change data.

For example, the data detection module 102 may include a lateral detection sub-module and a longitudinal detection sub-module (not shown).

Specifically, the transverse detection sub-module determines a transverse trend index of the acquired trend change data, and the longitudinal detection sub-module determines a longitudinal trend index of the acquired trend change data, at which time, the data anomaly identification module 103 may determine whether the wind turbine generator set has a data anomaly based on the determined transverse trend index and the longitudinal trend index.

As an example, the lateral trend index may be used to characterize the duration of fluctuation of anomalous data in the trend change data. Alternatively, the lateral detection sub-module may determine the lateral trend index in the following manner.

The following steps are sequentially executed for each adjacent first trend change data and second trend change data according to the time sequence: respectively comparing the first trend change data and the second trend change data with a set threshold, if both the first trend change data and the second trend change data are smaller than the set threshold, adding one to the counting value of the counting statistics, if both the first trend change data and the second trend change data are not smaller than the set threshold, or not performing the counting statistics, if the first trend change data and the second trend change data are not smaller than the set threshold, stopping the counting statistics, comparing the current counting value with the stored counting value, if the current counting value is larger than the stored counting value, updating the stored counting value by using the current counting value, and if the current counting value is not larger than the stored counting value, not updating the stored counting value; the stored final count value is determined as a lateral trend index of the trend change data. As an example, the set threshold may be half of the sum of the maximum statistical value and the minimum statistical value in the trend change data over a predetermined period of time.

As an example, the longitudinal trend index may be used to characterize a fluctuating rate of change of trend change data. For example, the longitudinal detection sub-module may determine the longitudinal trend index in the following manner.

Determining the minimum value and the maximum value of the trend change data in a preset time period, taking the determined minimum value as a first endpoint value of a numerical range, taking the determined maximum value as a second endpoint value of the numerical range, and equally dividing the numerical range to obtain a plurality of numerical intervals; determining a numerical value interval to which each trend change data in a preset time period belongs; acquiring a first number of trend change data contained in a numerical interval of a first number close to the first endpoint value, and acquiring a second number of trend change data contained in a numerical interval of a second number close to the second endpoint value; and calculating the sum of the first number and the second number, calculating the ratio of the obtained sum to the total number of the trend change data in the preset time period, and determining the ratio as a longitudinal trend index of the trend change data.

The data anomaly identification module 103 determines whether the wind generating set has data anomaly or not based on the detection results of the transverse detection and the longitudinal detection.

For example, the data anomaly identification module 103 may compare the determined lateral trend index to a first set value and the determined longitudinal trend index to a second set value; if the determined transverse trend index is not smaller than a first set value and the determined longitudinal trend index is not smaller than a second set value, determining that the wind generating set has data abnormity; and if the determined transverse trend index is smaller than a first set value and/or the determined longitudinal trend index is smaller than a second set value, determining that the wind generating set has no data abnormality.

FIG. 13 illustrates a communication topology of a monitoring center with various wind turbine generator sets within a wind farm according to an exemplary embodiment of the present invention.

As shown in fig. 13, a programmable logic controller PLC may be selected to control the operation of the wind turbine generator system and collect the operation data of the wind turbine generator system, and as an example, an HMI touch screen may be used for field control and monitoring. The Programmable Logic Controller (PLC) is connected with an optical fiber switch (OSM) on the optical fiber ring network through an industrial Ethernet communication module, the monitoring center is connected into the industrial Ethernet redundant optical fiber ring network through the OSM, and a Personal Computer (PC) industrial personal computer is used for carrying out remote control and monitoring on the site.

According to the data link shown in fig. 13, data analysis and operation for realizing fault early warning can be obtained, and can be realized in a PLC, a touch screen, and a monitoring center, specifically as follows:

data analysis and operation are carried out in the PLC, namely, the data abnormity detection method is executed in the PLC, data can be collected in real time and operated, but because the fault early warning model needs to be optimized continuously, programs may need to be upgraded for many times subsequently, for a large batch of wind generating sets, larger and repeated workload can be caused, and the generated energy can be reduced because the wind generating sets are stopped.

Data analysis and operation are carried out in the HMI touch screen, the HMI touch screen obtains trend change data from the SCADA system, namely, the data abnormity detection method is executed in the HMI touch screen, shutdown is not needed during upgrading, but the requirements on the function and the performance of the touch screen are high, and the cost is increased.

Data analysis and operation are carried out in the monitoring center, the monitoring center obtains the trend change data from the SCADA system, workload related to program upgrading is small, power generation loss caused by stopping is avoided, maintenance is convenient, and the trend change data obtained from the SCADA system can well restore the original data change trend.

Fig. 14 illustrates a block diagram of a controller according to an exemplary embodiment of the present invention.

As shown in fig. 14, the controller 200 according to an exemplary embodiment of the present invention includes: a processor 201 and a memory 202.

In particular, the memory 202 is used for storing a computer program which, when being executed by the processor 201, implements the above-mentioned data anomaly detection method of a wind park.

Here, the data abnormality detecting method of the wind turbine generator system shown in fig. 3 may be executed in the processor 201 shown in fig. 14. That is, each module shown in fig. 11 may be implemented by a general-purpose hardware processor such as a digital signal processor or a field programmable gate array, may be implemented by a special-purpose hardware processor such as a special chip, or may be implemented entirely in software by a computer program, for example, may be implemented as each module in the processor 201 shown in fig. 14.

There is also provided, in accordance with an exemplary embodiment of the present invention, a computer-readable storage medium storing a computer program. The computer readable storage medium stores a computer program which, when executed by a processor, causes the processor to perform the above-described data anomaly detection method of a wind turbine generator set. The computer readable recording medium is any data storage device that can store data read by a computer system. Examples of the computer-readable recording medium include: read-only memory, random access memory, read-only optical disks, magnetic tapes, floppy disks, optical data storage devices, and carrier waves (such as data transmission through the internet via wired or wireless transmission paths).

According to the data anomaly detection method and device for the wind generating set, disclosed by the exemplary embodiment of the invention, data can be acquired without triggering faults of the wind generating set, data acquisition is not required manually, detection on the fluctuation of the operating data can be realized only by detecting the data change trend, the data anomaly of the wind generating set can be recognized as early as possible, and the method and device have important significance for realizing fault early warning and operation monitoring of the wind generating set.

In addition, in the data anomaly detection method and device for the wind generating set according to the exemplary embodiment of the invention, compared with the method for directly detecting the lower limit value, the related horizontal and vertical detection methods can obtain more accurate and more information quantity of trend change data, instead of simply obtaining a TRUE or FALSE judgment result, so that a data basis can be provided for further analysis, and meanwhile, voltage values of different voltage levels can be directly detected without adjusting related parameters according to the voltage levels.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present invention as defined by the following claims.