CN117307403A - Early warning method and system for torque fluctuation recognition - Google Patents

Abstract

The invention provides a torque fluctuation identification early warning method and system, and belongs to the technical field of wind power generation intelligence; the method comprises the following steps: continuously acquiring real-time data of main state parameters of the fan and torque of the fan, and integrating the acquired real-time data into time data with a specific length for torque fluctuation identification analysis; empirical mode decomposition is carried out on the data with specific length to obtain mode components, the mode components are filtered and recombined to obtain recombined component data, and then envelope spectrogram decomposition is carried out to obtain frequency and amplitude data corresponding to the fan at the moment; screening out characteristic frequency and frequency multiplication data from the frequency and amplitude data, carrying out cluster analysis on the screened characteristic frequency and frequency multiplication data on the basis of torque fluctuation experience data, obtaining an analysis result and pushing the analysis result; the invention is easy to realize and has economic cost, better real-time early warning effect can be obtained, follow-up measures are timely taken on the basis of early warning, and the safety of the transmission chain and the whole wind turbine generator is improved.

Description

Early warning method and system for torque fluctuation recognition

Technical Field

The invention belongs to the technical field of wind power generation intelligence, and is applied to an intelligent early warning process of a wind turbine generator, in particular to an early warning method and system for torque fluctuation identification.

Background

With the technical development of offshore wind turbine generators, the capacity of the wind turbine generators is larger and larger, the length of fan blades is longer and longer, and the tower is higher and higher, so that the requirements of the safety of the wind turbine generators are also rapidly improved. The transmission chain is a core mechanical component of the wind turbine generator, and the important degree of the transmission chain is self-evident, and plays an important role in converting wind energy into electric energy in the wind turbine generator. In the industrial application process, once the wind turbine generator frequently generates torque fluctuation, the wind turbine generator can be damaged, and meanwhile, the damage of a transmission chain and the like can be caused.

The negative effects of torque ripple on the generator mainly include the following:

1. vibration and noise can be generated when the generator normally operates, and the vibration and noise can be aggravated due to large torque fluctuation, so that the service life and the normal operating state of equipment are affected;

2. when the torque fluctuation of the generator is large, the frequency of a power grid is unstable, and the frequency deviates from a standard value to cause the loss of electric energy;

3. the large torque fluctuation can cause stress concentration of parts in the generator, increase failure rate and influence operation reliability.

In the prior art, in order to realize monitoring, distinguishing and early warning functions aiming at torque fluctuation phenomenon, an acceleration sensor is usually installed in a transmission chain and a generator component, the vibration condition of the transmission chain is sensed in real time through the sensor, and vibration data of the transmission chain are collected in real time. However, the collected data is subjected to remote offline analysis in a periodic offline transmission mode, whether vibration data are abnormal or not is judged, whether a transmission chain is abnormal or not is further judged, and the torque fluctuation condition is analyzed temporarily; the method only has acceleration data, the analysis process is offline and the real-time rate is low, the abnormal reasons cannot be confirmed, and when an early warning phenomenon occurs in the actual application process, boarding is needed for further inspection, so that the defects and limitations are large.

Disclosure of Invention

The invention aims to make up the defects and the shortcomings of the prior art, and reduces the problems of a transmission chain and a generator caused by torque fluctuation by timely finding out the abnormal condition of the torque fluctuation of the wind turbine generator. The invention provides an early warning method and system for torque fluctuation identification, which are easy to realize and have economic cost, and has a better real-time early warning effect; on the basis of early warning, follow-up measures are timely taken, abnormal vibration of the transmission chain and the generator caused by torque fluctuation can be reduced, and safety of the transmission chain and the whole wind turbine generator is improved.

The invention adopts the following technical scheme to achieve the purpose:

a method of early warning of torque ripple identification, the method comprising the steps of:

s1, continuously acquiring real-time data of main state parameters of a fan and torque of the fan;

s2, integrating the acquired real-time data into time data with a specific length for torque fluctuation identification analysis in a subsequent process;

s3, performing empirical mode decomposition on the time data with the specific length to obtain a mode component; screening modal components and recombining the modal components to obtain recombined component data;

s4, carrying out envelope spectrum decomposition on the weight data of the heavy components to obtain frequency and amplitude data corresponding to the fan at the moment;

s5, screening out characteristic frequency and frequency multiplication data from the frequency and amplitude data;

s6, carrying out cluster analysis on the screened characteristic frequency and the frequency multiplication data on the basis of torque fluctuation empirical data to obtain an analysis result; the analysis results include the type of torque ripple and the cause of the abnormality.

Further, in step S3, the basis for screening and reorganizing the modal components is the correlation between the modal components and the time data with a specific length.

Further, in step S5, the characteristic frequency and the frequency multiplication data are screened out by determining the reference frequency in the frequency and amplitude data; the reference frequency is the frequency corresponding to the 1 st amplitude in the frequency and amplitude data.

Further, in step S6, the characteristic frequency and the frequency multiplication data are clustered to obtain a clustering result; obtaining a corresponding analysis result according to the clustering result in the region of the torque fluctuation empirical data; and (3) corresponding different analysis results to different early warning trigger conditions, and transmitting the corresponding early warning trigger conditions to a wind turbine generator control system in real time when the analysis results are generated, so as to change the running state of the wind turbine generator.

The invention also provides a torque fluctuation identification early warning system, which comprises an edge equipment platform, a data integration module, an empirical mode decomposition module, an envelope spectrum analysis module, a characteristic frequency screening module and a cluster analysis and result pushing module;

the edge equipment platform is used for continuously acquiring real-time data of main state parameters of the fan and torque of the fan to obtain a section of continuous multiple 10ms data points;

the data integration module is used for integrating a plurality of 10ms data points into data with specific length time and providing a data basis for torque fluctuation identification analysis of the subsequent module;

the empirical mode decomposition module is used for performing empirical mode decomposition on the data with the specific length to obtain a mode component; the method is also used for screening and reorganizing according to the correlation between the modal component and the time data with the specific length to obtain reorganized component data;

the envelope spectrum analysis module is used for carrying out envelope spectrum decomposition on the recombination component data to obtain frequency and amplitude data corresponding to the fan at the moment;

the characteristic frequency screening module is used for screening characteristic frequency and frequency multiplication data from the frequency and amplitude data; the screening process is based on the determined reference frequency, wherein the reference frequency is the frequency corresponding to the 1 st amplitude in the frequency and amplitude data;

the cluster analysis and result pushing module is used for carrying out cluster analysis on the characteristic frequency and the frequency multiplication data, and pushing analysis results corresponding to different early warning triggering conditions to the wind turbine generator control system.

In summary, by adopting the technical scheme, the invention has the following beneficial effects:

1. according to the method, the process acquires and processes the data of the wind turbine generator in real time, and the torque fluctuation condition can be effectively identified; the method can help to carry out specific optimization on the corresponding torque fluctuation condition in the unit control system in the development and design stage of the wind turbine so as to improve the safety of the unit;

2. in the operation process of the later actual wind field, the method can timely transmit the early warning trigger result to the unit control system to control the wind turbine to stop or run in a power-down mode as soon as possible, so that the condition of overrun of the vibration of the wind turbine caused by torque fluctuation is avoided, and the safe and stable operation of the wind turbine is ensured.

Drawings

FIG. 1 is a schematic flow chart of the method of the present invention;

FIG. 2 is a flow chart of data integration in the present invention;

FIG. 3 is a schematic diagram of empirical mode decomposition in the present invention;

FIG. 4 is a schematic diagram of the envelope spectrum after decomposition and recombination in the present invention.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.

Thus, the following detailed description of the embodiments of the invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

Example 1

As shown in fig. 1, a method for early warning torque fluctuation recognition includes the following steps:

s1, continuously acquiring real-time data of main state parameters of a fan and torque of the fan;

s2, integrating the acquired real-time data into time data with a specific length for torque fluctuation identification analysis in a subsequent process;

s3, performing empirical mode decomposition on the time data with the specific length to obtain a mode component; screening modal components and recombining the modal components to obtain recombined component data;

s4, carrying out envelope spectrum decomposition on the weight data of the heavy components to obtain frequency and amplitude data corresponding to the fan at the moment;

s5, screening out characteristic frequency and frequency multiplication data from the frequency and amplitude data;

s6, carrying out cluster analysis on the screened characteristic frequency and the frequency multiplication data on the basis of torque fluctuation empirical data to obtain an analysis result; the analysis results include the type of torque ripple and the cause of the abnormality.

The details of each step in the method will be described in detail in this embodiment. Firstly, in step S1, real-time data of a fan is obtained through an edge device platform. The real-time data acquired in this embodiment is a continuous length of multiple 10ms data points, and the sampling frequency is 100Hz, that is, 100 continuous 10ms data points are acquired every second.

On this basis, reference can be made to the schematic of fig. 2; in step S2, a plurality of 10ms data points are integrated into a specific length of time data. The duration of the specific length of time data of this embodiment is 30 seconds, i.e. 30 consecutive time-per-second data will be acquired and integrated in total. In the process, all 10ms data points in each second time are continuously acquired one by one in a data length judging mode, and data deduplication operation is performed after each second of data acquisition; and outputting data when the total length of the acquired data per second meets a preset time threshold, namely, the total length of 30 seconds is reached, and completing the data integration process.

The empirical mode decomposition process of step S3 is followed, as illustrated by the decomposition of fig. 3. In this step, the basis for screening and reorganizing the modal components is that the correlation between the modal components and the data of a specific length time, that is, the components with large correlation are screened and reorganized.

In this embodiment, the specific procedure here is as follows:

s31, performing empirical mode decomposition on the integrated data with specific length to obtain a continuous series of intrinsic mode components;

s32, presetting a correlation threshold, performing correlation comparison on the intrinsic mode component and the original content of the time data with the specific length, and screening out the mode component larger than the correlation threshold;

s33, reorganizing the screened modal components to obtain reorganized component data.

In step S4, the envelope spectrum analysis process is performed on the recombinant component data in a conventional manner, so as to complete the decomposition of the envelope spectrum, as shown in fig. 4, and thus the corresponding frequency and amplitude data is directly obtained.

In step S5, the characteristic frequency and the frequency multiplication data are screened out by determining the reference frequency in the frequency and amplitude data; the reference frequency in this embodiment is the frequency corresponding to the 1 st amplitude in the frequency and amplitude data; the specific process is as follows:

s51, integrating the frequency and amplitude data into a dictionary, wherein the key value of the dictionary is frequency, the attribute value of the dictionary is amplitude, the corresponding relation of the frequency-amplitude is obtained, and the frequency-amplitude data point is determined;

s52, arranging the magnitudes in a descending order to obtain ordered frequency-magnitude data points;

s53, presetting a specific screening value n, where n=500 in this embodiment; screening out the first 500 frequency-amplitude data points of the sorting queue;

s54, taking the frequency corresponding to the 1 st amplitude in the first 500 frequency-amplitude data points of the sequencing queue as a reference frequency;

and S55, further obtaining the amplitude corresponding to the frequency multiplication according to the determined reference frequency, thereby completing the screening of the characteristic frequency and the frequency multiplication data.

In this embodiment, the specific process of step S55 is as follows:

s551, constructing a reference list according to the reference frequency, wherein a plurality of frequency reference values corresponding to different integer multiples of the reference frequency are stored in the reference list; the reference list in this embodiment is: [1×fundamental, 2×fundamental, 3×fundamental, 4×fundamental, 5×fundamental ];

s552, comparing each amplitude data in the first 500 frequency-amplitude data points with the amplitude data adjacent to the amplitude data before and after the amplitude data; when the amplitude data of a certain point is larger than the amplitude data adjacent to the certain point, determining the certain point as a peak point, and storing the frequency-amplitude data point corresponding to the peak point in a screening dictionary; after the comparison screening of all n frequency-amplitude data points is completed, a complete screening dictionary is obtained;

s553, traversing the frequency value data in the screening dictionary; when the deviation between the frequency value data and any frequency reference value in the reference list is within +/-0.01, storing the frequency value data in a frequency multiplication list; and by analogy, after traversing is completed, all the characteristic frequency and frequency multiplication data containing amplitude information are obtained.

Finally, in step S6, clustering is carried out on the characteristic frequency and the frequency multiplication data to obtain a clustering result; obtaining a corresponding analysis result according to the clustering result in the region of the torque fluctuation empirical data; and (3) corresponding different analysis results to different early warning trigger conditions, and transmitting the corresponding early warning trigger conditions to a wind turbine generator control system in real time when the analysis results are generated, so as to change the running state of the wind turbine generator.

In this embodiment, for different early warning trigger conditions, an early warning threshold may be preset; when the early warning trigger condition is generated but is smaller than the early warning threshold value, the capacity of the wind turbine generator is reduced; when the early warning trigger condition is generated and is larger than the early warning threshold value, stopping the wind turbine, and timely notifying relevant operation and maintenance personnel to perform field inspection on the wind turbine.

Example 2

On the basis of embodiment 1, this embodiment provides an early warning system for torque fluctuation recognition corresponding to the method of embodiment 1. The system comprises an edge equipment platform, a data integration module, an empirical mode decomposition module, an envelope spectrum analysis module, a characteristic frequency screening module and a cluster analysis and result pushing module.

The edge equipment platform is used for continuously acquiring real-time data of main state parameters of the fan and torque of the fan to obtain a section of continuous multiple 10ms data points;

the data integration module is used for integrating a plurality of 10ms data points into data with a specific length, and providing a data basis for torque fluctuation identification analysis of the subsequent module;

the empirical mode decomposition module is used for performing empirical mode decomposition on the data with the specific length to obtain a mode component; the method is also used for screening and reorganizing according to the correlation between the modal component and the time data with the specific length to obtain reorganized component data;

the envelope spectrum analysis module is used for carrying out envelope spectrum decomposition on the recombination component data to obtain frequency and amplitude data corresponding to the fan at the moment;

the characteristic frequency screening module is used for screening characteristic frequency and frequency multiplication data from the frequency and amplitude data; the screening process is based on the determined reference frequency, wherein the reference frequency is the frequency corresponding to the 1 st amplitude in the frequency and amplitude data;

and the cluster analysis and result pushing module is used for carrying out cluster analysis on the characteristic frequency and the frequency multiplication data, and pushing the analysis result to a wind turbine generator control system corresponding to different early warning triggering conditions.

Claims (10)

1. The early warning method for torque fluctuation identification is characterized by comprising the following steps:

s1, continuously acquiring real-time data of main state parameters of a fan and torque of the fan;

s2, integrating the acquired real-time data into time data with a specific length for torque fluctuation identification analysis in a subsequent process;

s3, performing empirical mode decomposition on the time data with the specific length to obtain a mode component; screening modal components and recombining the modal components to obtain recombined component data;

s4, carrying out envelope spectrum decomposition on the weight data of the heavy components to obtain frequency and amplitude data corresponding to the fan at the moment;

s5, screening out characteristic frequency and frequency multiplication data from the frequency and amplitude data;

s6, carrying out cluster analysis on the screened characteristic frequency and the frequency multiplication data on the basis of torque fluctuation empirical data to obtain an analysis result; the analysis results include the type of torque ripple and the cause of the abnormality.

2. The early warning method for torque fluctuation identification according to claim 1, characterized in that: in step S1, acquiring real-time data of a fan through an edge equipment platform; the real-time data is a continuous segment of a plurality of 10ms data points, and the sampling frequency is 100Hz.

3. The early warning method for torque fluctuation identification according to claim 2, characterized in that: in step S2, integrating a plurality of 10ms data points into a specific length of time data; continuously acquiring all 10ms data points in each second time one by one in a data length judging mode, and executing data deduplication operation after each second data acquisition; when the total length of the acquired data per second meets a preset time threshold, outputting the data to complete the data integration process.

4. The early warning method for torque fluctuation identification according to claim 1, characterized in that: in step S3, the basis for screening and reorganizing the modal components is the correlation between the modal components and the time data with a specific length.

5. The method for early warning torque ripple recognition according to claim 4, wherein the specific process of step S3 is as follows:

s31, performing empirical mode decomposition on the integrated data with specific length to obtain a continuous series of intrinsic mode components;

s32, presetting a correlation threshold, performing correlation comparison on the intrinsic mode component and the original content of the time data with the specific length, and screening out the mode component larger than the correlation threshold;

s33, reorganizing the screened modal components to obtain reorganized component data.

6. The early warning method for torque fluctuation identification according to claim 1, characterized in that: in step S5, the characteristic frequency and the frequency multiplication data are screened out by determining the reference frequency in the frequency and amplitude data; the reference frequency is the frequency corresponding to the 1 st amplitude in the frequency and amplitude data.

7. The early warning method for torque fluctuation recognition according to claim 6, wherein the specific process of step S5 is as follows:

s51, integrating the frequency and amplitude data into a dictionary, wherein the key value of the dictionary is frequency, the attribute value of the dictionary is amplitude, the corresponding relation of the frequency-amplitude is obtained, and the frequency-amplitude data point is determined;

s52, arranging the magnitudes in a descending order to obtain ordered frequency-magnitude data points;

s53, presetting a specific screening value n, and screening out first n frequency-amplitude data points of a sequencing queue;

s54, taking the frequency corresponding to the 1 st amplitude in the first n frequency-amplitude data points of the sequencing queue as a reference frequency;

and S55, further obtaining the amplitude corresponding to the frequency multiplication according to the determined reference frequency, thereby completing the screening of the characteristic frequency and the frequency multiplication data.

8. The method for early warning torque ripple recognition according to claim 7, wherein the specific process of step S55 is as follows:

s551, constructing a reference list according to the reference frequency; storing a plurality of frequency reference values corresponding to different integer multiples of a reference frequency in a reference list;

s552, comparing each amplitude data in the first n frequency-amplitude data points with the amplitude data adjacent to the first n frequency-amplitude data points; when the amplitude data of a certain point is larger than the amplitude data adjacent to the certain point, determining the certain point as a peak point, and storing the frequency-amplitude data point corresponding to the peak point in a screening dictionary; after the comparison screening of all n frequency-amplitude data points is completed, a complete screening dictionary is obtained;

s553, traversing the frequency value data in the screening dictionary; when the deviation between the frequency value data and any frequency reference value in the reference list is within +/-0.01, storing the frequency value data in a frequency multiplication list; after the traversal is completed, all the characteristic frequency and frequency multiplication data containing the amplitude information are obtained.

9. The early warning method for torque fluctuation identification according to claim 1, characterized in that: in step S6, clustering is carried out on the characteristic frequency and the frequency multiplication data to obtain a clustering result; obtaining a corresponding analysis result according to the clustering result in the region of the torque fluctuation empirical data; and (3) corresponding different analysis results to different early warning trigger conditions, and transmitting the corresponding early warning trigger conditions to a wind turbine generator control system in real time when the analysis results are generated, so as to change the running state of the wind turbine generator.

10. A torque fluctuation identification early warning system is characterized in that: the system comprises an edge equipment platform, a data integration module, an empirical mode decomposition module, an envelope spectrum analysis module, a characteristic frequency screening module and a cluster analysis and result pushing module;

the edge equipment platform is used for continuously acquiring real-time data of main state parameters of the fan and torque of the fan to obtain a section of continuous multiple 10ms data points;

the data integration module is used for integrating a plurality of 10ms data points into data with specific length time and providing a data basis for torque fluctuation identification analysis of the subsequent module;

the empirical mode decomposition module is used for performing empirical mode decomposition on the data with the specific length to obtain a mode component; the method is also used for screening and reorganizing according to the correlation between the modal component and the time data with the specific length to obtain reorganized component data;

the envelope spectrum analysis module is used for carrying out envelope spectrum decomposition on the recombination component data to obtain frequency and amplitude data corresponding to the fan at the moment;

the characteristic frequency screening module is used for screening characteristic frequency and frequency multiplication data from the frequency and amplitude data; the screening process is based on the determined reference frequency, wherein the reference frequency is the frequency corresponding to the 1 st amplitude in the frequency and amplitude data;

the cluster analysis and result pushing module is used for carrying out cluster analysis on the characteristic frequency and the frequency multiplication data, and pushing analysis results corresponding to different early warning triggering conditions to the wind turbine generator control system.