CN115828732A - Fan fault identification method based on data driving - Google Patents
Fan fault identification method based on data driving Download PDFInfo
- Publication number
- CN115828732A CN115828732A CN202211369803.5A CN202211369803A CN115828732A CN 115828732 A CN115828732 A CN 115828732A CN 202211369803 A CN202211369803 A CN 202211369803A CN 115828732 A CN115828732 A CN 115828732A
- Authority
- CN
- China
- Prior art keywords
- data
- fan
- identification method
- fault identification
- characteristic parameters
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000010801 machine learning Methods 0.000 claims abstract description 16
- 238000012549 training Methods 0.000 claims abstract description 16
- 239000011159 matrix material Substances 0.000 claims abstract description 10
- 238000012545 processing Methods 0.000 claims abstract description 10
- 230000000007 visual effect Effects 0.000 claims abstract description 5
- 230000009467 reduction Effects 0.000 claims abstract description 4
- 238000013480 data collection Methods 0.000 claims description 12
- 238000011156 evaluation Methods 0.000 claims description 12
- 230000001133 acceleration Effects 0.000 claims description 11
- 238000012795 verification Methods 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 238000012217 deletion Methods 0.000 claims description 3
- 230000037430 deletion Effects 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 239000013598 vector Substances 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 abstract description 4
- 238000012423 maintenance Methods 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 230000003449 preventive effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Landscapes
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a data-driven fan fault identification method, which comprises the following steps: s1, modeling fan data by using a high-dimensional statistical matrix model; s2, analyzing fan data and performing feature dimension reduction; and S3, performing visual processing on the fan data, and judging the running state of the fan through a machine learning model. The method for learning and distinguishing the state by using the collected fan related data does not depend on a physical mechanism model; searching internal relation between the characteristic parameters and the state labels in a training set by utilizing a machine learning model through analyzing historical data of the fan; based on the high-dimensional statistical characteristic of big data, the application range is wide, the robustness is high, and the safety is reliable.
Description
Technical Field
The invention relates to the field of wind driven generators, in particular to a fan fault identification method based on data driving.
Background
Renewable energy sources play an important role in energy source layout, and due to the influence of factors such as environment and weather, power generation equipment such as a wind driven generator is easy to damage, and the timely fault finding, fault identification or equipment state prejudgment is the basis for guaranteeing the power supply reliability and the power quality of users. The existing method for identifying the fan fault has the following defects: 1. the frequent overhaul causes the waste of manpower and material resources; 2. the power failure caused by maintenance can influence the power consumption requirement of a user; 3. the overhaul itself may affect equipment health; 4. the traditional mechanism modeling is difficult to adapt to different scenes and the method is complex.
The online fan fault identification aims at analyzing the running state of fan equipment based on running data before and after a fan fault occurs, then carrying out fault identification on the fan and finding potential problems of the equipment in time. On-line evaluation has the following advantages over planned maintenance: 1. daily operation data is taken as a main part, complex mechanism modeling is not needed, and the generalization performance is strong; 2. no extra power failure maintenance arrangement is needed; 2. the state of the fan can be identified in real time to form a visual platform. The pain point of the online fan fault identification is that it is difficult to find a universal state evaluation index effectively. The traditional fan fault identification is usually based on single or multiple self-quantity measurement, is in a lower dimensionality, and is difficult to utilize equipment multi/high-dimensionality data to form an effective (namely, small standard deviation/fluctuation) equipment fault state identification index.
The basis of wind turbine fault identification is state detection and state analysis. The state detection means acquiring data, and the state analysis means analyzing data. Common methods for condition monitoring include: inspection tour, preventive test, live test and online monitoring. Thus, online monitoring is not equal to condition monitoring, but it provides only one of the important sources of data. Besides the reference of monitoring parameters, the equipment state evaluation also needs to consider a plurality of factors such as the condition of the equipment maintenance test in the past, the performance and the failure rate of the same equipment of the same manufacturer and the like, and can be obtained by a comprehensive analysis method of the factors.
In conclusion, the development of a novel data-driven fan fault identification method becomes a current urgent need.
Disclosure of Invention
In view of the above defects in the prior art, the present invention provides a data-driven fan fault identification method, including:
s1, modeling fan data by using a high-dimensional statistical matrix model;
s2, analyzing fan data and performing feature dimension reduction;
and S3, performing visual processing on the fan data, and judging the running state of the fan through a machine learning model.
In a preferred embodiment of the present invention, the step S1 includes: collecting characteristic parameters of the fan during operation, establishing a relation between an observed value of fan data and the change of a fan operation state, and splicing and modeling the characteristic parameters as vectors by using a high-dimensional statistical matrix model.
Further, the characteristic parameters include: the method comprises the following steps of pitch angle of variable pitch, acceleration peak value, acceleration effective value, fault allowable yaw level, net side active power and wind speed.
In a preferred embodiment of the present invention, the step S2 includes:
s21, preprocessing the historical data of the characteristic parameters;
and S22, screening the preprocessed historical data of the characteristic parameters through a machine learning model.
Further, the step S21 includes: and carrying out abnormal value deletion and normalization processing on the collected historical data of the characteristic parameters, wherein the normalization processing comprises the steps of subtracting the statistical mean value of the whole historical data from each parameter data, and dividing the statistical mean value by the variance of the whole historical data.
Further, the step S21 includes establishing a data matrix divided by a sampling time window.
Further, the step S22 is implemented by an XGBoost model.
Further, the step S22 includes: the historical data of the preprocessed characteristic parameters are randomly and averagely divided into N parts through a python built-in function, each part of data collection comprises the characteristic parameters in a time period, N-1 parts of the data collection are alternately used as a training set, the other part of data collection is used as a verification set, and the average value of N times of evaluation results is used as the evaluation of the algorithm precision.
Further, the evaluating specifically comprises: the XGboost model is trained through a training set, and then the feature importance calculated from the training data set is evaluated through a verification set; the XGboost model is packaged in a SelectFromModel example, the features on the training set are selected, the model is trained by using the selected feature subset, and then the verification set is evaluated under the same feature scheme.
In a preferred embodiment of the present invention, the step S3 includes: graphically representing the trend of each characteristic value of the historical data along with the change of time so as to display a data curve in a normal state and a fault state; the data processed in step S2 is input to the machine learning model, and the machine learning model outputs and displays a 0/1 identifier indicating normal or failure on the data curve after learning.
Compared with the prior art, the invention has the following beneficial effects: 1. the method for learning and distinguishing the state by using the collected fan related data does not depend on a physical mechanism model; and searching the internal relation between the characteristic parameters and the state labels in the training set by analyzing the historical data of the fan by using a machine learning model. 2. The method is based on the high-dimensional statistical characteristic of the big data, and has the advantages of wide application range, high robustness and reliable safety. 3. The invention has very important practical value for vigorously developing renewable energy sources and ensuring efficient operation and maintenance and stable operation of fan equipment.
The conception, the specific structure and the technical effects of the present invention will be further described with reference to the accompanying drawings to fully understand the objects, the features and the effects of the present invention.
Drawings
FIG. 1 is a flow chart of a data driven based fan fault identification method of the present invention;
FIG. 2 is a flow chart of feature screening in accordance with a preferred embodiment of the present invention.
Detailed Description
The technical contents of the preferred embodiments of the present invention will be more clearly and easily understood by referring to the drawings attached to the specification. The present invention may be embodied in many different forms of embodiments and the scope of the invention is not limited to the embodiments set forth herein.
In the drawings, structurally identical elements are represented by like reference numerals, and structurally or functionally similar elements are represented by like reference numerals throughout the several views. The size and thickness of each component shown in the drawings are arbitrarily illustrated, and the present invention is not limited to the size and thickness of each component. The thickness of the components may be exaggerated where appropriate in the figures to improve clarity.
As shown in fig. 1, the method for identifying a fault of a fan based on data driving according to the present invention includes:
s1, modeling fan data by using a high-dimensional statistical matrix model;
s2, analyzing fan data and performing feature dimension reduction;
and S3, performing visual processing on the fan data, and judging the running state of the fan through a machine learning model.
The step S1 may specifically include: collecting characteristic parameters of the fan during operation, establishing a relation between an observed value of fan data and the change of a fan operation state, and splicing and modeling the characteristic parameters as vectors by using a high-dimensional statistical matrix model. The characteristic parameters may include: the method comprises the following steps of pitch angle of variable pitch, acceleration peak value, acceleration effective value, fault allowable yaw level, net side active power and wind speed.
The step S2 may specifically include:
s21, preprocessing historical data of the characteristic parameters;
and S22, screening the preprocessed historical data of the characteristic parameters through a machine learning model.
Further, step S21 may include: and carrying out abnormal value deletion and normalization processing on the collected historical data of the characteristic parameters, wherein the normalization processing comprises the steps of subtracting the statistical mean value of the whole historical data from each parameter data, and dividing the statistical mean value by the variance of the whole historical data. Preferably, a matrix of data divided by a sampling time window may be established.
Further, step S22 may include: the historical data of the preprocessed characteristic parameters are randomly and averagely divided into N parts through a python built-in function, each part of data collection comprises the characteristic parameters in a time period, N-1 parts of the data collection are alternately used as a training set, the other part of data collection is used as a verification set, and the average value of N times of evaluation results is used as the evaluation of the algorithm precision. For example, historical data of the preprocessed characteristic parameters can be randomly and averagely divided into 5 parts through a python built-in function, each part of data collection comprises the characteristic parameters in a time period, 4 parts of the data collection are alternately used as a training set, the other part of data collection is used as a verification set, and the average value of 5 times of evaluation results is used as the evaluation of the algorithm precision.
According to an embodiment of the present invention, the above evaluation specifically includes: training the XGboost model through a training set, and evaluating the feature importance calculated from the training data set through a verification set; the XGboost model is packaged in a SelectFromModel example, the features on the training set are selected, the model is trained by using the selected feature subset, and then the verification set is evaluated under the same feature scheme.
According to the embodiment of the invention, the features are screened through a machine learning model, the XGboost calculates which feature is selected as a separation point according to the gain condition of the structure score, and the importance of a certain feature can be judged according to the average gain of the feature in all trees, namely, the average gain of the feature in all trees is calculated, and the greater the gain is, the more important the feature is. As shown in fig. 2, the feature screening process may include:
-entering a day history data;
-the selectfrommermadel model outputs thresholds for feature selection;
-outputting an index value of the selection feature;
-outputting the name of the selected feature.
The step S3 may specifically include: graphically representing the trend of each characteristic value of the historical data along with the change of time so as to display a data curve in a normal state and a fault state; the data processed in step S2 is input to a machine learning model, and the machine learning model outputs and displays a 0/1 identifier indicating normal or failure on the data curve after learning. Specifically, a trend graph of the characteristic parameter changing with time is generated, that is, a trend graph of the characteristic parameter value is generated by using the horizontal axis as time and the vertical axis as the characteristic parameter, and by taking the acceleration as the characteristic parameter as an example, the trend graph can display that the amplitude of the acceleration peak value, the acceleration effective value, the X-axis acceleration and the Y-axis acceleration before and after the fault occurs obviously changes, and can display a straight line which extends along the horizontal axis and has a vertical axis value of 0 in the fault-free time period and a straight line which extends along the horizontal axis and has a vertical axis value of 1 in the fault-occurring time period, so as to make the contrast before and after the fault occurs obvious.
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations could be devised by those skilled in the art in light of the present teachings without departing from the inventive concepts. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.
Claims (10)
1. A fan fault identification method based on data driving is characterized by comprising the following steps:
s1, modeling fan data by using a high-dimensional statistical matrix model;
s2, analyzing fan data and performing feature dimension reduction;
and S3, performing visual processing on the fan data, and judging the running state of the fan through a machine learning model.
2. The data-driven-based fan fault identification method according to claim 1, wherein the step S1 comprises: collecting characteristic parameters of the fan during operation, establishing a relation between an observed value of fan data and changes of a fan operation state, and performing splicing modeling by using the characteristic parameters as vectors by using a high-dimensional statistical matrix model.
3. The data-driven-based fan fault identification method of claim 2, wherein the characteristic parameters include: the method comprises the following steps of pitch angle of variable pitch, acceleration peak value, acceleration effective value, fault allowable yaw level, net side active power and wind speed.
4. The data-driven-based fan fault identification method according to claim 1, wherein the step S2 comprises:
s21, preprocessing the historical data of the characteristic parameters;
and S22, screening the preprocessed historical data of the characteristic parameters through a machine learning model.
5. The data-driven-based fan fault identification method according to claim 4, wherein the step S21 comprises: and carrying out abnormal value deletion and normalization processing on the collected historical data of the characteristic parameters, wherein the normalization processing comprises the steps of subtracting the statistical mean value of the whole historical data from each parameter data, and dividing the statistical mean value by the variance of the whole historical data.
6. The data drive-based fan fault identification method of claim 5, wherein the step S21 comprises establishing a data matrix divided by a sampling time window.
7. The data-drive-based wind turbine fault identification method according to claim 4, wherein the step S22 is implemented by an XGboost model.
8. The data-driven-based fan fault identification method according to claim 7, wherein the step S22 includes: the historical data of the preprocessed characteristic parameters are randomly and averagely divided into N parts through a python built-in function, each part of data collection comprises the characteristic parameters in a time period, N-1 parts of the data collection are alternately used as a training set, the other part of data collection is used as a verification set, and the average value of N times of evaluation results is used as the evaluation of the algorithm precision.
9. The data-driven-based fan fault identification method of claim 8, wherein the evaluating specifically comprises: the XGboost model is trained through a training set, and then the feature importance calculated from the training data set is evaluated through a verification set; the XGboost model is packaged in a SelectFromModel example, the features on the training set are selected, the model is trained by using the selected feature subset, and then the verification set is evaluated under the same feature scheme.
10. The data-driven-fan-failure-based identification method according to any one of claims 1-9, wherein the step S3 comprises: graphically representing the trend of each characteristic value of the historical data along with the change of time so as to display a data curve in a normal state and a fault state; the data processed in step S2 is input to a machine learning model, and the machine learning model outputs and displays a 0/1 identifier indicating normal or failure on the data curve after learning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211369803.5A CN115828732A (en) | 2022-11-03 | 2022-11-03 | Fan fault identification method based on data driving |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211369803.5A CN115828732A (en) | 2022-11-03 | 2022-11-03 | Fan fault identification method based on data driving |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115828732A true CN115828732A (en) | 2023-03-21 |
Family
ID=85526385
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211369803.5A Pending CN115828732A (en) | 2022-11-03 | 2022-11-03 | Fan fault identification method based on data driving |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115828732A (en) |
-
2022
- 2022-11-03 CN CN202211369803.5A patent/CN115828732A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108897954A (en) | Wind turbines temperature pre-warning method and its system based on BootStrap confidence calculations | |
| CN109492777A (en) | A kind of Wind turbines health control method based on machine learning algorithm platform | |
| CN109324604A (en) | A kind of intelligent train resultant fault analysis method based on source signal | |
| CN111141517B (en) | Fan fault diagnosis method and system | |
| CN110059775A (en) | Rotary-type mechanical equipment method for detecting abnormality and device | |
| CN101738998B (en) | System and method for monitoring industrial process based on local discriminatory analysis | |
| CN111666978B (en) | Intelligent fault early warning system for IT system operation and maintenance big data | |
| CN113177646A (en) | Power distribution equipment online monitoring method and system based on self-adaptive edge proxy | |
| CN115018178A (en) | Power station fan fault early warning method based on deep learning | |
| CN108506171A (en) | A kind of large-scale half direct-drive unit cooling system for gear box fault early warning method | |
| CN113313365A (en) | Degradation early warning method and device for primary air fan | |
| CN117708637A (en) | Wind turbine generator blade fault diagnosis method based on improved k-means clustering analysis | |
| CN117703690A (en) | Wind generating set health state assessment method and system | |
| CN112257224A (en) | Method, system and terminal for overhauling state of steam turbine generator | |
| CN115828732A (en) | Fan fault identification method based on data driving | |
| CN116502043A (en) | Finish rolling motor state analysis method based on isolated forest algorithm | |
| CN110794683A (en) | Wind power gear box state evaluation method based on deep neural network and kurtosis characteristics | |
| CN113671287B (en) | Intelligent detection method, system and readable storage medium for power grid automation terminal | |
| US11339763B2 (en) | Method for windmill farm monitoring | |
| CN113268552B (en) | Generator equipment hidden danger early warning method based on locality sensitive hashing | |
| CN111524336A (en) | Generator set early warning method and system | |
| CN117993562A (en) | Wind turbine generator system fault prediction method and system based on artificial intelligent big data analysis | |
| Yang et al. | Condition Monitoring for Wind Turbine Pitch System Using Multi-parameter Health Indicator | |
| Cao | Big Data Technology Application in Mechanical Intelligent Fault Diagnosis | |
| Fu et al. | Machine Learning-Based Adaptive Fault Detection Method for Wind Turbine Gearboxes with Imbalanced Data through An IIoT Platform |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20240616 Address after: No.15, Jianshe West Street, Qiaoxi District, Zhangjiakou City, Hebei Province Applicant after: State Grid Jibei Zhangjiakou Fengguang storage and transmission new energy Co.,Ltd. Country or region after: China Applicant after: STATE GRID CORPORATION OF CHINA Applicant after: STATE GRID JIBEI ELECTRIC POWER CO., LTD. Research Institute Address before: No.15, Jianshe West Street, Qiaoxi District, Zhangjiakou City, Hebei Province Applicant before: State Grid Jibei Zhangjiakou Fengguang storage and transmission new energy Co.,Ltd. Country or region before: China Applicant before: STATE GRID CORPORATION OF CHINA Applicant before: SHANGHAI JIAO TONG University Applicant before: STATE GRID JIBEI ELECTRIC POWER CO., LTD. Research Institute |
|
| TA01 | Transfer of patent application right |