CN113982850A - Fan comprehensive health analysis method and system fusing high-frequency and low-frequency signals - Google Patents
Fan comprehensive health analysis method and system fusing high-frequency and low-frequency signals Download PDFInfo
- Publication number
- CN113982850A CN113982850A CN202111060718.6A CN202111060718A CN113982850A CN 113982850 A CN113982850 A CN 113982850A CN 202111060718 A CN202111060718 A CN 202111060718A CN 113982850 A CN113982850 A CN 113982850A
- Authority
- CN
- China
- Prior art keywords
- monitoring data
- fan
- frequency
- frequency monitoring
- low
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000036541 health Effects 0.000 title claims abstract description 64
- 238000004458 analytical method Methods 0.000 title claims abstract description 18
- 238000012544 monitoring process Methods 0.000 claims abstract description 202
- 238000000034 method Methods 0.000 claims abstract description 41
- 238000001514 detection method Methods 0.000 claims abstract description 27
- 238000012502 risk assessment Methods 0.000 claims abstract description 25
- 238000013441 quality evaluation Methods 0.000 claims abstract description 23
- 238000012549 training Methods 0.000 claims abstract description 12
- 238000012545 processing Methods 0.000 claims abstract description 10
- 238000012216 screening Methods 0.000 claims abstract description 9
- 238000000605 extraction Methods 0.000 claims abstract description 7
- 230000007246 mechanism Effects 0.000 claims description 19
- 230000000007 visual effect Effects 0.000 claims description 11
- 238000010998 test method Methods 0.000 claims description 8
- 238000012423 maintenance Methods 0.000 claims description 6
- 238000010586 diagram Methods 0.000 claims description 5
- 238000012847 principal component analysis method Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 230000005477 standard model Effects 0.000 claims description 2
- 230000005856 abnormality Effects 0.000 description 11
- 230000002159 abnormal effect Effects 0.000 description 10
- 238000003745 diagnosis Methods 0.000 description 6
- 230000005540 biological transmission Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000005802 health problem Effects 0.000 description 3
- 238000002360 preparation method Methods 0.000 description 3
- 238000009826 distribution Methods 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000005461 lubrication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F03—MACHINES OR ENGINES FOR LIQUIDS; WIND, SPRING, OR WEIGHT MOTORS; PRODUCING MECHANICAL POWER OR A REACTIVE PROPULSIVE THRUST, NOT OTHERWISE PROVIDED FOR
- F03D—WIND MOTORS
- F03D17/00—Monitoring or testing of wind motors, e.g. diagnostics
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02B—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO BUILDINGS, e.g. HOUSING, HOUSE APPLIANCES OR RELATED END-USER APPLICATIONS
- Y02B10/00—Integration of renewable energy sources in buildings
- Y02B10/30—Wind power
Abstract
The invention relates to a method and a system for analyzing the comprehensive health of a fan by fusing high and low frequency signals, wherein the method comprises the following steps: collecting low-frequency monitoring data and high-frequency monitoring data, performing time alignment and quality evaluation, and removing data which do not accord with quality indexes; screening out monitoring data under a preset working condition; performing windowing processing on the monitoring data; performing feature extraction on the monitoring data under each time window, and combining the extracted features to obtain a feature subset of the fan under the corresponding time window to form a feature set; performing model training according to the feature set to obtain a fan abnormity monitoring model; inputting the feature set corresponding to the current monitoring data of the fan into the model to obtain the current health risk assessment value of the fan, and determining whether the fan breaks down currently according to the health risk assessment value. The method makes full use of the possessed data base, builds the model and carries out the abnormity detection, and meets the requirements of users on the realization of real system-level health analysis of the fan.
Description
Technical Field
The invention relates to the technical field of fan fault detection, in particular to a fan comprehensive health analysis method and system fusing high and low frequency signals.
Background
With the reduction of the cost of the sensor and the continuous improvement of the computing power, the health state monitoring and the operation and maintenance of the wind driven generator are also developed from simple digitization to intellectualization. The wind turbine generator fault early warning system monitors blades, a transmission mechanism, a generator, an electric control system and the like of a fan by utilizing an SCADA low-frequency signal (the sampling rate of the SCADA low-frequency signal is usually 1Hz at most) and a CMS high-frequency vibration signal (the frequency of the CMS high-frequency vibration signal is at least 10KHz) which are widely adopted in the wind power industry at present. However, the functions provided by the two types of signal monitoring systems are relatively independent, the existing fan fault early warning system rarely combines the two types of signal monitoring systems into a system fan health assessment function provided by a user, and the existing fan fault early warning system cannot completely meet the requirement of the user on realizing real system-level intelligent health analysis of the wind driven generator.
Disclosure of Invention
In order to solve the technical problems or at least partially solve the technical problems, the invention provides a method and a system for analyzing the comprehensive health of a fan by fusing high and low frequency signals.
In a first aspect, the invention provides a comprehensive health analysis method for a fan fusing high and low frequency signals, which comprises the following steps:
collecting low-frequency monitoring data and high-frequency monitoring data of a fan in a historical time period, aligning the low-frequency monitoring data with the high-frequency monitoring data in time, evaluating the quality of the low-frequency monitoring data and the high-frequency monitoring data, and removing data which do not meet quality indexes;
screening out monitoring data under a preset working condition from the low-frequency monitoring data and the high-frequency monitoring data which are reserved after the quality evaluation; performing windowing processing on the monitoring data to obtain monitoring data under a plurality of time windows; respectively extracting characteristics of the low-frequency monitoring data and the high-frequency monitoring data under each time window, combining the extracted characteristics to obtain characteristic subsets of the fan under the corresponding time window, and forming characteristic sets of the characteristic subsets under each time window according to a time sequence;
performing model training according to the feature set to obtain a fan abnormity monitoring model; inputting the feature set corresponding to the current low-frequency monitoring data and the high-frequency monitoring data of the fan into the fan abnormity monitoring model to obtain a current health risk assessment value of the fan, and determining whether the fan breaks down currently according to the health risk assessment value.
In a second aspect, the present invention provides a comprehensive health analysis system for a wind turbine with high and low frequency signals fused, including:
the data acquisition module is used for acquiring low-frequency monitoring data and high-frequency monitoring data of the fan in a historical time period, performing time alignment on the low-frequency monitoring data and the high-frequency monitoring data, performing quality evaluation on the low-frequency monitoring data and the high-frequency monitoring data, and removing data which do not meet quality indexes;
the characteristic extraction module is used for screening out monitoring data under a preset working condition from the low-frequency monitoring data and the high-frequency monitoring data which are reserved after the quality evaluation; performing windowing processing on the monitoring data to obtain monitoring data under a plurality of time windows; respectively extracting characteristics of the low-frequency monitoring data and the high-frequency monitoring data under each time window, combining the extracted characteristics to obtain characteristic subsets of the fan under the corresponding time window, and forming characteristic sets of the characteristic subsets under each time window according to a time sequence;
the health analysis module is used for carrying out model training according to the feature set to obtain a fan abnormity monitoring model; inputting the feature set corresponding to the current low-frequency monitoring data and the high-frequency monitoring data of the fan into the fan abnormity monitoring model to obtain a current health risk assessment value of the fan, and determining whether the fan breaks down currently according to the health risk assessment value.
The method and the system for analyzing the comprehensive health of the fan fusing the high-frequency and low-frequency signals, provided by the embodiment, are used for collecting low-frequency monitoring data and high-frequency monitoring data of the fan in a historical time period, performing time alignment and quality evaluation on the monitoring data, screening out the monitoring data under a preset working condition, performing windowing processing on the monitoring data, performing feature extraction on the monitoring data under each time window, combining the extracted features to obtain feature subsets of the fan under the corresponding time window, and forming the feature subsets under each time window into feature sets according to a time sequence; and then, training a fan abnormity monitoring model according to the characteristic set, and further performing fault identification and monitoring by using the trained fan abnormity monitoring model. According to the method, data quality evaluation is introduced, so that the modeling availability of data can be ensured, and after low-frequency monitoring data and high-frequency monitoring data are fused, non-supervised learning is performed, so that a fault sample is not needed, and only definite and sufficient fan state data are needed. The wind driven generator abnormal detection and early warning method has the advantages that the data base which is possibly possessed by the wind driven generator at present is fully utilized, abnormal detection and early warning are carried out aiming at the health problem of the wind driven generator, and the requirements of users on the realization of real system-level intelligent health analysis of the wind driven generator can be met.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description, serve to explain the principles of the invention.
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without inventive exercise.
FIGS. 1 to 3 are schematic flow charts of a comprehensive health analysis method for a fan with high and low frequency signals fused according to the present invention;
FIG. 4 is an architecture diagram of a comprehensive health evaluation system of a fan of a monitoring center of a wind farm booster station according to the present invention;
FIG. 5 is an architecture diagram of a comprehensive health assessment system for fans in a centralized control center of a group or provincial level according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
In a first aspect, the invention provides a method for analyzing comprehensive health of a wind turbine by fusing high and low frequency signals, as shown in fig. 1 to 3, the method comprising:
s110, collecting low-frequency monitoring data and high-frequency monitoring data of a fan in a historical time period, aligning the low-frequency monitoring data with the high-frequency monitoring data in time, evaluating the quality of the low-frequency monitoring data and the high-frequency monitoring data, and removing data which do not meet quality indexes;
in specific implementation, the acquiring low-frequency monitoring data and high-frequency monitoring data of the fan in a historical time period may include: acquiring low-frequency monitoring data of the fan from a data acquisition and monitoring control System (SCADA), wherein the low-frequency monitoring data comprises at least one of fan bearing temperature, fan rotating speed, cabin temperature, wind speed and wind power; acquiring high-frequency monitoring data of the fan from a Condition Monitoring System (CMS), wherein the high-frequency monitoring data comprises a bearing high-frequency vibration signal of the fan.
The sampling frequency of the low-frequency monitoring data cannot exceed 1HZ, and the frequency of the high-frequency monitoring data cannot be lower than 10 KHZ.
The monitoring data acquisition module is mainly used for acquiring, accessing, transmitting and storing the low-frequency monitoring data and the high-frequency monitoring data of the wind driven generator. Because the low frequency monitoring data and the high frequency monitoring data are mainly collected in an independent mode in the fan, the timestamps are not aligned, and the data collection module also needs to align the time of the low frequency monitoring data and the time of the high frequency monitoring data.
The quality evaluation of the low-frequency monitoring data and the high-frequency monitoring data aims to detect whether data which do not meet the quality requirement exist in the detection data, and then the data which do not meet the quality requirement are deleted. The quality evaluation of the monitoring data can be realized by a data quality evaluation module, and the data quality evaluation module is mainly responsible for respectively judging and evaluating the quality of the high-frequency monitoring data and the low-frequency monitoring data to ensure that the modeling data is available. For low-frequency monitoring data, a general data quality check term for time series is generally adopted. For high-frequency monitoring data, because such data generally has a strong mechanism analysis value, such as vibration, audio frequency, etc., a mechanism data quality detection item adapted to such signal characteristics is generally adopted.
In a specific implementation, the method for evaluating the quality of the low-frequency monitoring data may include: at least one of a singular value detection method, a missing point proportion detection method and a data overrun detection method. The singular value detection method can detect and eliminate the singular value in the low-frequency detection data. The missing point proportion detection method can detect the proportion of missing data, the whole data is available when the proportion of missing data is not higher than a threshold value, and the whole data is unavailable when the proportion of missing data is higher than the threshold value. The data overrun detection method is to detect whether data in the detected data exceed an upper limit value and a lower limit value, and eliminate the data exceeding the upper limit value and the lower limit value. The purpose of this is to eliminate some data that often fluctuate because of external interference factors, and avoid these data influence the fault identification rate of accuracy.
In a specific implementation, the method for evaluating the quality of the high-frequency detection data may include: at least one of an effectiveness test method, a mean shift test method, a positive-negative point number proportion test method, and a continuous value proportion test method. The inspection methods also detect and delete abnormal fluctuation data in the monitoring data, which is not described herein.
S120, screening out monitoring data under a preset working condition from the low-frequency monitoring data and the high-frequency monitoring data which are reserved after the quality evaluation; performing windowing processing on the monitoring data to obtain monitoring data under a plurality of time windows; respectively extracting characteristics of the low-frequency monitoring data and the high-frequency monitoring data under each time window, combining the extracted characteristics to obtain characteristic subsets of the fan under the corresponding time window, and forming characteristic sets of the characteristic subsets under each time window according to a time sequence;
the preset working condition can be a working condition of the fan except a shutdown working condition and an electricity limiting working condition. That is, the preset operating condition is a non-jamming operating condition. Under the preset working condition, the fan can run healthily and also can break down. The monitoring data under the shutdown working condition and the electricity limiting working condition are deleted, and only the monitoring data under the preset working condition are reserved for subsequent abnormal detection. Monitoring data under preset working conditions can be screened out specifically according to conditions such as wind speed and generator rotating speed, and the data are suitable for modeling.
In particular, data may be flushed prior to windowing the data. During windowing, the low-frequency monitoring data and the high-frequency monitoring data may be respectively subjected to windowing, for example, data segmentation is performed every 1 hour to realize windowing. Statistical features of the SCADA data are calculated in the data every 1 hour time window. In the data of the time window every 1 hour, the time-frequency domain characteristics of the CMS data are calculated.
That is, the features extracted from the low frequency monitoring data in each time window may include: and extracting statistic characteristics of the low-frequency monitoring data under each time window, wherein the statistic characteristics comprise at least one of a mean value, a variance, a maximum value, a minimum value and a peak value. The feature of extracting the high-frequency monitoring data under each time window comprises the following steps: and extracting time-frequency domain mechanism characteristics of the high-frequency monitoring data under each time window, wherein the time-frequency domain mechanism characteristics comprise at least one of frequency domain energy size, pass frequency effective value and kurtosis corresponding to fault characteristic frequency. The feature extraction may specifically be implemented by a feature preparation module.
And combining the statistic characteristics and the time-frequency domain mechanism characteristics under each time window to form a characteristic subset under one time window, and forming a characteristic set by the characteristic subsets under each time window according to a time sequence.
It can be understood that the statistical characteristics are statistical characteristics, and the time-frequency domain mechanism characteristics are mechanism characteristics, and have both statistical characteristics and mechanism characteristics in each time window.
S130, performing model training according to the feature set to obtain a fan abnormity monitoring model; inputting the feature set corresponding to the current low-frequency monitoring data and the high-frequency monitoring data of the fan into the fan abnormity monitoring model to obtain a current health risk assessment value of the fan, and determining whether the fan breaks down currently according to the health risk assessment value.
The feature set corresponding to the current monitoring data, for example, the low-frequency monitoring data and the high-frequency monitoring data of the current day, is subjected to windowing processing by taking each hour of the monitoring data of the current day as a time window, feature subsets in each time window are extracted, and the feature subsets in each time window are combined to form the current corresponding feature set. The current feature set is the same as the formation of the feature set over the historical time period.
As shown in fig. 2 and 3, in S130, an abnormality detection module may be specifically used to detect an abnormality of the fan. The function of the abnormity detection module is to evaluate the abnormity risk of the fan and push the abnormity risk of the fan to a visual interface for display. The basic idea of anomaly detection is as follows: based on the baseline model of the health state, the degree of deviation between the current feature set and the baseline model is compared, and therefore changes of the current fan health state are measured. Such a degree of offset may be based on methods that can quantify the differences between the distributions represented by the data (e.g., residual methods, similarity calculations, etc.). The health risk assessment value can be a statistic which can represent the overall health level of the fan, and can also be a value between 0 and 100 after being normalized.
The feature set corresponding to the historical time period is used for training a fan abnormity monitoring model, and the fan abnormity monitoring model can be constructed by adopting a principal component analysis method.
In particular implementation, the health risk assessment process of the wind turbine anomaly monitoring model may include:
and the fan abnormity monitoring model determines the deviation degree between the current characteristic set and a baseline model, and determines the current health risk assessment value of the fan according to the deviation degree.
The baseline model is a health standard model constructed by adopting a health characteristic set of the fan and adopting a principal component analysis method.
It can be understood that both the fan anomaly monitoring model and the baseline model can be constructed using principal component analysis methods. The training data adopted by the fan abnormity monitoring model is obtained by training the monitoring data in a historical time period, and the fan may break down in the time period. The baseline model is obtained by training health data in a time period, and provides a healthy standard line. Therefore, the current feature set is compared with the healthy standard line to obtain the offset degree of the current feature set relative to the healthy standard line, the current health risk assessment value of the fan can be determined according to the offset degree, and the fan is considered to be in fault when the health risk assessment value is higher than a preset threshold value.
When the wind turbine anomaly monitoring model runs on line, the model projects the current feature set and a baseline model in a principal component space, and Hotelling's T2 (namely Hotelling statistic) or SPE (namely q statistic) is used as an anomaly risk assessment value.
In specific implementation, the method provided by the invention can further comprise: when the current fault of the fan is determined, calculating the contribution degree of each feature in the feature set to the residual error, taking the feature with the largest contribution degree as the root cause of the current health risk of the fan, and displaying the health risk assessment value and the root cause on a visual interface.
The residual errors are data representing the degree of offset between the current feature set and the healthy standard line, the contribution of some features to the generation of the residual errors is large, the contribution of some features to the generation of the residual errors is small, the contribution degree of each feature in the feature set to the residual errors is calculated, then the components with faults can be judged according to the sequence in a mode that the contribution degree is from large to small, for example, if the relevant features of the generator bearing are ranked at the top, the fact that the abnormity is probably caused by the generator bearing is shown, and the features are used as the root cause of the fan fault. In order to enable a user to know the current condition of the fan more intuitively, the health risk assessment value and the root cause of the fan can be displayed on a visual interface.
In specific implementation, the method provided by the invention can further comprise: and comparing the mechanism characteristics corresponding to the characteristics with the maximum contribution degree with a plurality of preset fault level thresholds, determining the current fault level of the fan, and providing a maintenance decision according to the fault level.
That is to say, when a fault occurs, the mechanism characteristic (for example, the outer ring fault frequency of the bearing) corresponding to the characteristic with the largest contribution degree (for example, the related characteristic of the generator bearing) is screened out, the fault level corresponding to the mechanism characteristic, that is, the fault level of the current bearing abnormality of the wind turbine, is determined by comparing the mechanism characteristic with a plurality of preset fault level thresholds which are set in advance, then a corresponding maintenance decision (for example, lubrication or replacement) can be provided according to the fault level, and the maintenance decision is displayed on a visual interface.
In specific implementation, the method provided by the invention can further comprise: when the fan is determined to be in a current fault, displaying the low-frequency monitoring data and the high-frequency monitoring data in a time window corresponding to the current fault on the visual interface, and displaying a contrast scatter diagram of the projection of the characteristic set input in the principal component space and the projection of the health characteristic set in the principal component space on the visual interface.
Referring to fig. 2 and 3, the diagnosis of the abnormality may be implemented using an abnormality diagnosis module. The abnormity diagnosis module is mainly responsible for analyzing root causes aiming at abnormal conditions, helping a user to quickly locate the possible reasons of the abnormity, judging the fault level and supporting maintenance and repair decisions. The root cause analysis is mainly determined by calculating the degree of contribution to the generation of an abnormality. The main idea for calculating such contribution degree is generally: the degree of abnormality (i.e., the degree of distribution shift) is measured by calculating the weight of each type of feature. The definition of the fault grade needs to screen out the characteristic with large abnormal contribution degree, and then define the fault grade according to the degree of the corresponding mechanism characteristic, such as vibration intensity, bearing vibration signal, harmonic wave size of a certain fault frequency, and the like. And judging the integral fault severity of the fan according to a set threshold rule.
The system comprises a data acquisition module, a data quality evaluation module, a feature preparation module, an abnormality detection module and an abnormality diagnosis module, wherein the data acquisition module, the data quality evaluation module, the feature preparation module, the abnormality detection module and the abnormality diagnosis module can be integrated and borne on a special server, and the server can be flexibly and selectively deployed in a wind farm booster station monitoring center, a provincial level centralized control center or a group monitoring center. See fig. 4 and 5. For example, the modules are carried on a special server of a monitoring center of a wind farm booster station, and the server performs data transmission with a low-frequency Scada data server and a CMS system data server of a transmission chain through a switch and a Modbus protocol.
That is to say, when a fault occurs, the monitoring data, namely the original data, in the current fault time window is displayed on the visual interface, and then the comparative scatter diagram of the current feature set and the projection of the health feature set on the principal component space is displayed on the visual interface, so that a user can more intuitively know the difference between the current condition and the health condition of the fan.
The main innovation points of the invention are as follows: (1) data quality evaluation is introduced into the method, so that the modeling availability of data can be ensured; (2) after the low-frequency monitoring data and the high-frequency monitoring data are fused, non-supervised learning is carried out, a fault sample is not required, and only definite and sufficient fan state data are required; (3) when the fan is abnormal, determining root cause by using the contribution degree of the characteristics to the residual error; (4) and (4) assisting in judging the fault level according to the degree of the mechanism characteristics.
The invention aims to provide a fan health assessment method fusing low-frequency monitoring data and high-frequency monitoring data of a wind driven generator, which makes full use of the data base possibly possessed by the wind driven generator at present, carries out abnormity detection and early warning aiming at fan health problems, and carries out preliminary diagnosis on abnormal root causes by means of integrating mechanism characteristics and statistical characteristics, thereby rapidly positioning the problems.
The invention provides a comprehensive health analysis method of a fan fusing high and low frequency signals, which comprises the steps of collecting low frequency monitoring data and high frequency monitoring data of the fan in a historical time period, carrying out time alignment and quality evaluation on the monitoring data, screening out the monitoring data under a preset working condition, carrying out windowing processing on the monitoring data, carrying out feature extraction on the monitoring data under each time window, combining the extracted features to obtain feature subsets of the fan under corresponding time windows, and forming the feature subsets under each time window into feature sets according to a time sequence; and then, training a fan abnormity monitoring model according to the characteristic set, and further performing fault identification and monitoring by using the trained fan abnormity monitoring model. According to the method, data quality evaluation is introduced, so that the modeling availability of data can be ensured, and after low-frequency monitoring data and high-frequency monitoring data are fused, non-supervised learning is performed, so that a fault sample is not needed, and only definite and sufficient fan state data are needed. The wind driven generator abnormal detection and early warning method has the advantages that the data base which is possibly possessed by the wind driven generator at present is fully utilized, abnormal detection and early warning are carried out aiming at the health problem of the wind driven generator, and the requirements of users on the realization of real system-level intelligent health analysis of the wind driven generator can be met.
In a second aspect, the present invention provides a comprehensive health analysis system for a wind turbine with high and low frequency signals fused, including:
the data acquisition module is used for acquiring low-frequency monitoring data and high-frequency monitoring data of the fan in a historical time period, performing time alignment on the low-frequency monitoring data and the high-frequency monitoring data, performing quality evaluation on the low-frequency monitoring data and the high-frequency monitoring data, and removing data which do not meet quality indexes;
the characteristic extraction module is used for screening out monitoring data under a preset working condition from the low-frequency monitoring data and the high-frequency monitoring data which are reserved after the quality evaluation; performing windowing processing on the monitoring data to obtain monitoring data under a plurality of time windows; respectively extracting characteristics of the low-frequency monitoring data and the high-frequency monitoring data under each time window, combining the extracted characteristics to obtain characteristic subsets of the fan under the corresponding time window, and forming characteristic sets of the characteristic subsets under each time window according to a time sequence;
and the health analysis module is used for inputting the feature set into a fan abnormity monitoring model to obtain a current health risk assessment value of the fan, and determining whether the fan breaks down currently according to the health risk assessment value.
It is understood that reference may be made to corresponding parts of the first aspect for examples, embodiments, advantages, etc. in the system provided by the second aspect.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal (such as a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.
While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (10)
1. A fan comprehensive health analysis method fusing high and low frequency signals is characterized by comprising the following steps:
collecting low-frequency monitoring data and high-frequency monitoring data of a fan in a historical time period, aligning the low-frequency monitoring data with the high-frequency monitoring data in time, evaluating the quality of the low-frequency monitoring data and the high-frequency monitoring data, and removing data which do not meet quality indexes;
screening out monitoring data under a preset working condition from the low-frequency monitoring data and the high-frequency monitoring data which are reserved after the quality evaluation; performing windowing processing on the monitoring data to obtain monitoring data under a plurality of time windows; respectively extracting characteristics of the low-frequency monitoring data and the high-frequency monitoring data under each time window, combining the extracted characteristics to obtain characteristic subsets of the fan under the corresponding time window, and forming characteristic sets of the characteristic subsets under each time window according to a time sequence;
performing model training according to the feature set to obtain a fan abnormity monitoring model; inputting the feature set corresponding to the current low-frequency monitoring data and the high-frequency monitoring data of the fan into the fan abnormity monitoring model to obtain a current health risk assessment value of the fan, and determining whether the fan breaks down currently according to the health risk assessment value.
2. The method of claim 1, further comprising:
when the current fault of the fan is determined, calculating the contribution degree of each feature in the feature set to the residual error, taking the feature with the maximum contribution degree as the root cause of the current health risk of the fan, and displaying the health risk assessment value and the root cause on a visual interface.
3. The method of claim 2, further comprising:
and comparing the mechanism characteristics corresponding to the characteristics with the maximum contribution degree with a plurality of preset fault level thresholds, determining the current fault level of the fan, and providing a maintenance decision according to the fault level.
4. The method of claim 1, wherein the health risk assessment process of the wind turbine anomaly monitoring model comprises:
the fan abnormity monitoring model determines the deviation degree between the current characteristic set and a baseline model, and determines the current health risk assessment value of the fan according to the deviation degree; the baseline model is a health standard model constructed by adopting a health characteristic set of the fan and adopting a principal component analysis method, and the fan abnormity monitoring model is constructed by adopting the principal component analysis method.
5. The method of claim 4, further comprising:
when the fan is determined to be in a current fault, displaying the low-frequency monitoring data and the high-frequency monitoring data in a time window corresponding to the current fault on the visual interface, and displaying a contrast scatter diagram of the projection of the current characteristic set in the principal component space and the projection of the health characteristic set in the principal component space on the visual interface.
6. The method of claim 1,
the quality evaluation method for the low-frequency monitoring data comprises the following steps: at least one of a singular value detection method, a missing point proportion detection method and a data overrun detection method;
the quality evaluation method for the high-frequency detection data comprises the following steps: at least one of an effectiveness test method, a mean shift test method, a positive-negative point number proportion test method, and a continuous value proportion test method.
7. The method of claim 1, wherein the predetermined operating condition is an operating condition of the wind turbine other than a shutdown condition and a power limit condition.
8. The method of claim 1,
the features extracted from the low frequency monitoring data for each time window include: extracting statistical characteristics of the low-frequency monitoring data under each time window, wherein the statistical characteristics comprise at least one of a mean value, a variance, a maximum value, a minimum value and a peak value;
the feature of extracting the high-frequency monitoring data under each time window comprises the following steps: and extracting time-frequency domain mechanism characteristics of the high-frequency monitoring data under each time window, wherein the time-frequency domain mechanism characteristics comprise at least one of frequency domain energy size, pass frequency effective value and kurtosis corresponding to fault characteristic frequency.
9. The method of claim 1, wherein collecting low frequency and high frequency monitoring data of the wind turbine over a historical period of time comprises:
acquiring low-frequency monitoring data of the fan from a data acquisition and monitoring control system, wherein the low-frequency monitoring data comprises at least one of fan bearing temperature, fan rotating speed, cabin temperature, wind speed and wind power; and acquiring high-frequency monitoring data of the fan from a state monitoring system, wherein the high-frequency monitoring data comprises a bearing high-frequency vibration signal of the fan.
10. The utility model provides a fuse fan of high-low frequency signal synthesizes health analytic system which characterized in that includes:
the data acquisition module is used for acquiring low-frequency monitoring data and high-frequency monitoring data of the fan in a historical time period, performing time alignment on the low-frequency monitoring data and the high-frequency monitoring data, performing quality evaluation on the low-frequency monitoring data and the high-frequency monitoring data, and removing data which do not meet quality indexes;
the characteristic extraction module is used for screening out monitoring data under a preset working condition from the low-frequency monitoring data and the high-frequency monitoring data which are reserved after the quality evaluation; performing windowing processing on the monitoring data to obtain monitoring data under a plurality of time windows; respectively extracting characteristics of the low-frequency monitoring data and the high-frequency monitoring data under each time window, combining the extracted characteristics to obtain characteristic subsets of the fan under the corresponding time window, and forming characteristic sets of the characteristic subsets under each time window according to a time sequence;
the health analysis module is used for carrying out model training according to the feature set to obtain a fan abnormity monitoring model; inputting the feature set corresponding to the current low-frequency monitoring data and the high-frequency monitoring data of the fan into the fan abnormity monitoring model to obtain a current health risk assessment value of the fan, and determining whether the fan breaks down currently according to the health risk assessment value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111060718.6A CN113982850B (en) | 2021-09-10 | 2021-09-10 | Fan comprehensive health analysis method and system integrating high-low frequency signals |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111060718.6A CN113982850B (en) | 2021-09-10 | 2021-09-10 | Fan comprehensive health analysis method and system integrating high-low frequency signals |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113982850A true CN113982850A (en) | 2022-01-28 |
| CN113982850B CN113982850B (en) | 2023-11-10 |
Family
ID=79735594
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111060718.6A Active CN113982850B (en) | 2021-09-10 | 2021-09-10 | Fan comprehensive health analysis method and system integrating high-low frequency signals |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113982850B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115013261A (en) * | 2022-08-08 | 2022-09-06 | 国网浙江省电力有限公司舟山供电公司 | State monitoring method and system for offshore wind farm |
| WO2023226213A1 (en) * | 2022-05-24 | 2023-11-30 | 山东瑞美油气装备技术创新中心有限公司 | Device fault detection method based on baseline data space |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102768115A (en) * | 2012-06-27 | 2012-11-07 | 华北电力大学 | Method for dynamically monitoring health status of wind turbine gearbox in real time |
| CN104503399A (en) * | 2014-12-09 | 2015-04-08 | 华电电力科学研究院 | Group stage wind power generator set state monitoring and fault diagnosis platform |
| WO2016086360A1 (en) * | 2014-12-02 | 2016-06-09 | Abb Technology Ltd | Wind farm condition monitoring method and system |
| CN106662072A (en) * | 2014-11-18 | 2017-05-10 | Abb瑞士股份有限公司 | Wind turbine condition monitoring method and system |
| CN110219816A (en) * | 2018-03-02 | 2019-09-10 | 国家能源投资集团有限责任公司 | Method and system for Fault Diagnosis of Fan |
| CN110688617A (en) * | 2019-10-09 | 2020-01-14 | 北京天泽智云科技有限公司 | Fan vibration abnormity detection method and device |
| CN112177863A (en) * | 2020-09-07 | 2021-01-05 | 上海电气风电集团股份有限公司 | Vibration monitoring system, wind power generation system and wind power plant |
| CN112727702A (en) * | 2020-12-11 | 2021-04-30 | 中国大唐集团科学技术研究院有限公司火力发电技术研究院 | Health management and fault early warning method for wind turbine generator |
| CN112945531A (en) * | 2021-02-03 | 2021-06-11 | 西人马(西安)测控科技有限公司 | Method, device and equipment for detecting cracks of fan blade and computer storage medium |
-
2021
- 2021-09-10 CN CN202111060718.6A patent/CN113982850B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102768115A (en) * | 2012-06-27 | 2012-11-07 | 华北电力大学 | Method for dynamically monitoring health status of wind turbine gearbox in real time |
| CN106662072A (en) * | 2014-11-18 | 2017-05-10 | Abb瑞士股份有限公司 | Wind turbine condition monitoring method and system |
| WO2016086360A1 (en) * | 2014-12-02 | 2016-06-09 | Abb Technology Ltd | Wind farm condition monitoring method and system |
| CN104503399A (en) * | 2014-12-09 | 2015-04-08 | 华电电力科学研究院 | Group stage wind power generator set state monitoring and fault diagnosis platform |
| CN110219816A (en) * | 2018-03-02 | 2019-09-10 | 国家能源投资集团有限责任公司 | Method and system for Fault Diagnosis of Fan |
| CN110688617A (en) * | 2019-10-09 | 2020-01-14 | 北京天泽智云科技有限公司 | Fan vibration abnormity detection method and device |
| CN112177863A (en) * | 2020-09-07 | 2021-01-05 | 上海电气风电集团股份有限公司 | Vibration monitoring system, wind power generation system and wind power plant |
| CN112727702A (en) * | 2020-12-11 | 2021-04-30 | 中国大唐集团科学技术研究院有限公司火力发电技术研究院 | Health management and fault early warning method for wind turbine generator |
| CN112945531A (en) * | 2021-02-03 | 2021-06-11 | 西人马(西安)测控科技有限公司 | Method, device and equipment for detecting cracks of fan blade and computer storage medium |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023226213A1 (en) * | 2022-05-24 | 2023-11-30 | 山东瑞美油气装备技术创新中心有限公司 | Device fault detection method based on baseline data space |
| CN115013261A (en) * | 2022-08-08 | 2022-09-06 | 国网浙江省电力有限公司舟山供电公司 | State monitoring method and system for offshore wind farm |
| CN115013261B (en) * | 2022-08-08 | 2022-12-06 | 国网浙江省电力有限公司舟山供电公司 | State monitoring method and system for offshore wind farm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113982850B (en) | 2023-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113982850B (en) | Fan comprehensive health analysis method and system integrating high-low frequency signals | |
| EP2478423A1 (en) | Supervised fault learning using rule-generated samples for machine condition monitoring | |
| CN112660745A (en) | Intelligent diagnosis method and system for carrier roller fault and readable storage medium | |
| CN112504673A (en) | Carrier roller fault diagnosis method and system based on machine learning and storage medium | |
| CN116826958A (en) | Intelligent safety inspection method for digital transmission channel | |
| CN115437358A (en) | Intelligent state monitoring and fault diagnosis system and fault diagnosis method for industrial robot | |
| CN116383633A (en) | Method and system for detecting faults of machine-made sand vibrating screen through multi-factor comprehensive analysis | |
| KR20210006832A (en) | Method and apparatus for machine fault diagnosis | |
| CN113849774A (en) | Intelligent equipment maintenance platform based on life analysis and health monitoring | |
| CN110107461B (en) | Fan fault early warning method, device, equipment and computer readable storage medium | |
| CN113193616A (en) | Health state evaluation method for power transmission channel monitoring equipment | |
| CN108613820A (en) | A kind of online allophone monitoring algorithm for GIS bulk mechanicals defect diagonsis and positioning | |
| CN106228248B (en) | A kind of system automatic trouble diagnosis method based on fuzzy FMEA analysis | |
| CN111784865A (en) | Hydrogenation station inspection data processing method and hydrogenation station inspection device | |
| CN116881661A (en) | Performance automatic analysis method and system based on low-voltage power capacitor | |
| CN117074852A (en) | Power distribution network electric energy monitoring and early warning management method and system | |
| CN112660746B (en) | Roller fault diagnosis method and system based on big data technology and storage medium | |
| CN115328986A (en) | Power plant safety early warning data analysis processing method and system | |
| CN111224468B (en) | Photovoltaic equipment safe operation and maintenance platform based on cloud computing | |
| CN108199482B (en) | Maintenance prompting method and device | |
| CN116453437B (en) | Display screen module testing method, device, equipment and storage medium | |
| CN116484263B (en) | Intelligent self-service machine fault detection system and method | |
| US11047833B2 (en) | Method for automatic determination of trend in graphic analysis of turbomachines | |
| CN114638060B (en) | Fault prediction method, system and electronic equipment | |
| KR101967637B1 (en) | Signal Data Processing Apparatus For Prediction And Diagnosis Of Nuclear Power Plant By Augmented Reality |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |