CN110674842A - Wind turbine generator main shaft bearing fault prediction method - Google Patents
Wind turbine generator main shaft bearing fault prediction method Download PDFInfo
- Publication number
- CN110674842A CN110674842A CN201910788602.0A CN201910788602A CN110674842A CN 110674842 A CN110674842 A CN 110674842A CN 201910788602 A CN201910788602 A CN 201910788602A CN 110674842 A CN110674842 A CN 110674842A
- Authority
- CN
- China
- Prior art keywords
- data
- value
- main shaft
- shaft bearing
- analysis
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2148—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the process organisation or structure, e.g. boosting cascade
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a method for predicting the fault of a main shaft bearing of a wind turbine generator, which is based on historical fault maintenance data of a main shaft bearing of a fan and combines statistics and machine learning methods, takes a plurality of monitoring indexes of the fan as input variables, takes the state of the main shaft bearing as a predicted output variable, performs statistical analysis on the predicted value of the output variable, and sets a threshold value for fault prediction. The method has higher accuracy and stability, can realize the prediction of the main shaft bearing fault one circle in advance, and can discover early abnormity as soon as possible.
Description
Technical Field
The invention relates to the technical field of wind power generation, in particular to a method for predicting a main shaft bearing fault of a wind turbine generator.
Background
With the continuous development of wind power generation technology, in recent years, wind generating sets of large megawatt and low wind speed models are on the rise, and the research and development and production of offshore large megawatt compact units are substantially advanced.
The main shaft bearing is used as a key component of a transmission system of the wind generating set, and is influenced by random natural wind, so that the main shaft bearing bears huge random impact force, and various types of faults are generated. Once the main shaft bearing breaks down, if the main shaft bearing cannot be maintained in time, the unit is forced to be shut down to replace expensive components, and the whole unit is damaged to cause huge loss if the main shaft bearing is heavy.
At present, the wind field is mainly used for performing routine maintenance and repair, and the evaluation of the state of the fan and the diagnosis of the fault are too dependent on the experience of operation and maintenance personnel; on the other hand, the existing fan monitoring system can give an alarm for the overrun index, but because the wider threshold value is mainly adopted, the triggering early warning time is too late, and the fault cannot be found earlier. The risk of the health deterioration of the fan can be reduced by timely and accurate early warning, so that the loss caused by scrapping of parts and overlong shutdown is reduced.
At present, a plurality of researches on early warning of a wind turbine generator exist, wherein the early warning is performed by combining a wind speed based on a power abnormity method; on one hand, the method is not targeted enough and cannot locate the fault to a specific part; on the other hand, the method is not enough for the utilization of the existing data and the fan indexes thereof, and is easy to cause false alarm.
Therefore, in order to improve the purpose, accuracy and timeliness of fan maintenance, good fan early warning is very urgent need.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, provides a method for predicting the fault of a main shaft bearing of a wind turbine generator, has high accuracy and stability, can predict the fault of the main shaft bearing one circle in advance, and can discover early abnormality as soon as possible.
In order to achieve the purpose, the technical scheme provided by the invention is as follows: a wind turbine main shaft bearing fault prediction method is based on historical fault maintenance data of a fan main shaft bearing, and combines statistics and machine learning methods, a plurality of monitoring indexes of a fan are used as input variables, a main shaft bearing state is used as a prediction output variable, a prediction value of the output variable is subjected to statistical analysis, and a threshold value is set for fault prediction; which comprises the following steps:
1) data exploration
1.1) data quality analysis
The data quality analysis is the basis of validity and accuracy of data mining analysis conclusion, the main task of the data quality analysis is to check whether dirty data exists in original data, the dirty data refers to data which does not meet requirements and cannot be directly used for analysis, and the dirty data comprises missing values, abnormal values, inconsistent values, repeated data and data containing special symbols;
1.2) data feature analysis
After the data is subjected to quality analysis, data characteristic analysis is performed by means of drawing a chart and calculating certain characteristic quantities, and the data characteristic analysis comprises the following steps:
distribution analysis: the distribution characteristics and the distribution types of the data can be revealed;
and (3) comparative analysis: comparing two or more interrelated indicators to quantitatively display and account for the relationship between the indicators;
statistical analysis: carrying out statistical description on the quantitative data by using the statistics;
2) data cleansing
The data cleaning mainly comprises the steps of deleting irrelevant data and repeated data in an original data set, smoothing noise data, screening data irrelevant to a mining theme, and processing a missing value and an abnormal value;
there are 3 methods for handling missing data values: deleting records, interpolating data and not processing;
the method for processing the abnormal value comprises the following steps: deleting records, processing according to missing values, correcting the average value and not processing when the records are regarded as the missing values;
3) data normalization and imbalance processing
Carrying out standardization processing on the data, and converting the data into a proper form so as to be suitable for the requirement of an algorithm;
in order to eliminate the influence of dimension and value range difference between indexes, standard processing is needed, and data is scaled according to a proportion so as to fall into a specific area, thereby facilitating comprehensive analysis; the data standardization method comprises the following steps: min-max normalization and zero-mean normalization;
the problem of unbalance of positive and negative samples in an original data set can cause that a model cannot be correctly classified; the data balance processing method comprises the following steps: an ADASYNN sampling method and a SMOTE sampling method; the ADASYNN sampling method is a self-adaptive synthesis sampling method, and a certain mechanism is adopted to automatically determine the number of synthesized samples required to be generated by each few samples; the SMOTE sampling method interpolates between a few classes of samples to produce additional samples.
4) Data correlation analysis
Carrying out correlation analysis on the data processed in the step 3): firstly, a random forest method is used for feature selection, variables with large correlation with the state of a main shaft bearing are selected, then the fault mechanism of the main shaft bearing is analyzed, the variables with large correlation with the state of the main shaft bearing are selected, the correlation variables obtained by combining the two methods are combined, and the variables with large correlation among the variables are deleted by a Pearson correlation coefficient calculation method, so that data dimensionality is reduced, and model training is prevented from being over-fitted; finally, the input variables comprise a wind speed real-time value, a generator power real-time value, a generator rotating speed real-time value, an X-direction vibration value, a Y-direction vibration value, an engine room temperature, an outdoor temperature, a hub temperature, a non-driving end temperature of a gear box, a gear box oil temperature, a main shaft bearing A temperature, a cable twisting angle, an annual energy production amount and a power grid current L1; the output variable is the main shaft bearing state value: '0' indicates normal, '1' indicates failure;
5) data modeling
Adopting an XGboost algorithm, inputting processed data into an XGboost classification model for training, adding cross validation to prevent over-fitting of training, importing test data to be analyzed after training, and performing fault prediction; the XGboost algorithm is an extreme gradient boosting algorithm, is an ensemble learning method, and is excellent in most regression and classification problems.
6) Model evaluation and failure prediction
The model evaluation mainly comprises the steps of calculating indexes such as the accuracy P, the recall ratio R, F1 value, the AUC, the ROC curve and the like of the model, and if the accuracy, the recall ratio, the F1 value and the AUC are all larger than 0.9, and the ROC curve is close to the upper left corner, the training requirement is met; wherein, the F1 value is a comprehensive evaluation index of accuracy and recall rate, the F1 is 2P R/(P + R), and the larger the F1 value is, the more ideal the model is; the ROC curve is a receiver operation characteristic curve and is used for evaluating the prediction capability of the model, and the larger the area under the curve (AUC) is, or the curve is closer to the upper left corner, the more ideal the model is; the AUC is the area enclosed by the ROC curve and the coordinate axis, the numerical value is less than or equal to 1, and the higher the AUC is, the more ideal the model is;
and a fault prediction stage for performing real-time diagnosis for a specified time, and if the proportion of '1' in the prediction data of a certain day is more than 0.9, judging that the main shaft bearing of the day has a fault.
Compared with the prior art, the invention has the following advantages and beneficial effects:
1. the method is based on big data, all historical fault maintenance data of the fan main shaft bearing are used, and the data set is large enough.
2. The method adopts the XGboost algorithm for modeling, has short time consumption for training a large amount of data and high classification accuracy, and effectively divides the state of the main shaft bearing into fault and normal states.
3. The training data of the method is the data of one circle before the fault of the main shaft bearing, so that the model can predict the fault of the main shaft bearing one circle in advance.
4. The method of the invention utilizes the random forest algorithm and combines with engineering experience to fully screen out the characteristic quantity related to the main shaft bearing fault and ensure that the input variable can represent the state of the main shaft bearing.
5. The method is dynamic, the recorded fault data can be dynamically included in the next calculation, and the accuracy can be further improved along with the increasing accumulation of the data.
Drawings
FIG. 1a is a wind speed data distribution diagram.
FIG. 1b is a graph of power data distribution.
FIG. 1c is a graph of generator speed data distribution.
FIG. 1d is a plot of cabin temperature data.
FIG. 2 is a graph of the results of normalization of some of the monitoring metrics.
FIG. 3 is a graph of model test ROC.
Fig. 4 is a diagram showing the prediction result of the failure of the main shaft bearing.
Detailed Description
The present invention will be further described with reference to the following specific examples.
The method for predicting the failure of the main shaft bearing of the wind turbine generator system is based on historical failure maintenance data of the main shaft bearing of a fan, combines statistics and machine learning methods, takes a plurality of monitoring indexes of the fan as input variables, takes the state of the main shaft bearing as a prediction output variable, performs statistical analysis on a predicted value of the output variable, and sets a threshold value for performing failure prediction; which comprises the following steps:
1) data exploration: selecting data of a circle before a main shaft bearing abrasion fault occurs in the wind turbine generator, performing quality analysis and characteristic analysis on all monitoring indexes, checking data quality, data distribution and the like, referring to the attached drawings 1a, 1b, 1c and 1d, wherein the data distribution is part of the monitoring index data distribution, and the table 1 is part of the monitoring index statistical analysis. Wherein grWindSpeed is a wind speed real-time value, grGenPowerForProcess _1sec is a generator power real-time value, grGenSpeedForProcess is a generator rotational speed real-time value, grTempRotorBearA _1sec is a main shaft bearing A temperature, and grIL1 is a grid current L1.
TABLE 1
Data quality analysis is the basis for the validity and accuracy of data mining analysis conclusions. The main task of data quality analysis is to check the raw data for the presence of dirty data, which is generally data that is not satisfactory and cannot be used directly for analysis. The dirty data mainly includes: missing values, outliers, inconsistent values, duplicate data, and data containing special symbols.
After the data is subjected to quality analysis, data characteristic analysis can be performed by means of drawing a chart, calculating certain characteristic quantities and the like. The data characteristic analysis mainly comprises the following steps:
distribution analysis: the distribution characteristics and the distribution types of the data can be revealed;
and (3) comparative analysis: two or more interrelated indicators are compared to quantitatively display and account for the relationship between the indicators.
Statistical analysis: the quantitative data is statistically described by the statistics.
2) Data cleansing
The data cleaning is mainly to delete irrelevant data and repeated data in an original data set, smooth noise data, screen data irrelevant to a mining theme, process missing values, abnormal values and the like. Outliers include: error values generated during data transmission, and the like (for example, the wind speed is a negative value).
Methods for processing missing data values can be classified into 3 types: delete records, data interpolation, and do not process. The present embodiment uses a method of deleting a record.
The method for processing the abnormal value comprises the following steps: delete record, treat as missing value (treat according to missing value), average value correct, do not process. The present embodiment selects a method of deleting a record after analyzing the cause of the occurrence of an abnormal value.
3) Data normalization and imbalance processing
The data is standardized and converted into a suitable form to suit the requirements of the algorithm.
In order to eliminate the influence of dimension and value range difference between indexes, standard processing is required, and data is scaled according to a proportion so as to fall into a specific area, thereby facilitating comprehensive analysis. The data standardization method mainly comprises the following steps: min-max normalization, zero-mean normalization, etc. The present embodiment prefers min-max normalization.
The problem of unbalance of positive and negative samples in the original data set can cause that the model cannot be classified correctly. The data balance processing method mainly comprises the following steps: ADASYN sampling method, SMOTE sampling method, and the like.
In the embodiment, the input data are mapped into [0,1] by adopting a minimum-maximum standardization method, so that the influence of dimension on model training is reduced. By using the SMOTE sampling method, the problem that positive and negative samples in an original data set are unbalanced is solved, the classification accuracy of the model is improved, and the result is a part of monitoring index standardization result shown in table 2.
TABLE 2
4) Data correlation analysis
Performing correlation analysis on the clean data prepared in the step 3), and performing feature selection by using a random forest method, as shown in the attached figure 2: the greater the importance of the feature, the stronger the correlation with the state of the main shaft bearing. Selecting variables with large correlation with the state of the main shaft bearing from the graph shown in fig. 2, wherein ikwhtismonth is monthly power generation amount, grIL1 is grid current L1, ikwhtisyear is annual power generation amount, grIL2 is grid current L2, grwingtimeenergyoutputtotal is total power generation amount, gravailabilistallytotal is total unit availability, grcabletwist total is cable torsion total, grnacelepopitiontositetotal is total cabin direction angle, iOperationHoursOverall is total running time, grwingtimeavilabetableraterateraterateratedata is current availability, grtemproretorarib _1 is main shaft bearing B temperature, grtemproretorarea _1 is main shaft bearing a temperature, grtemp1gear box _1 oil temperature, and temporaderblade 2 blade temperature _ 2 motor temperature; then analyzing the failure mechanism of the main shaft bearing, and selecting a variable with larger relevance to the state of the main shaft bearing; and (3) combining the correlation variables obtained by the two methods, and deleting the variables with larger correlation among the variables by a Pearson correlation coefficient calculation method, so that the data dimensionality is reduced, and the model is prevented from being over-fitted by training. Finally, the input variables comprise a wind speed real-time value, a generator power real-time value, a generator rotating speed real-time value, an X-direction vibration value, a Y-direction vibration value, an engine room temperature, an outdoor temperature, a hub temperature, a non-driving end temperature of a gear box, a gear box oil temperature, a main shaft bearing A temperature, a cable twisting angle, an annual energy production amount and a power grid current L1; the output variable is the main shaft bearing state value ('0' for normal, '1' for fault).
5) Data modeling
And adopting an XGboost algorithm, inputting the processed data into an XGboost classification model for training, and adding cross validation to prevent over-training and over-fitting. And after training, importing test data to be analyzed to predict faults.
6) Model evaluation and failure prediction
The model evaluation mainly comprises the steps of calculating indexes such as the accuracy P, the recall ratio R, F1 value, the AUC, the ROC curve and the like of the model, and if the accuracy, the recall ratio, the F1 value and the AUC are all larger than 0.9, and the ROC curve is close to the upper left corner, the training requirement is met; wherein, the F1 value is a comprehensive evaluation index of accuracy and recall rate, the F1 is 2P R/(P + R), and the larger the F1 value is, the more ideal the model is; the ROC curve is a receiver operation characteristic curve and is used for evaluating the prediction capability of the model, and the larger the area under the curve (AUC) is, or the curve is closer to the upper left corner, the more ideal the model is; the AUC is the area enclosed by the coordinate axes under the ROC curve, the numerical value is less than or equal to 1, and the higher the AUC is, the more ideal the model is.
And a fault prediction stage for performing real-time diagnosis for a specified time, and if the proportion of '1' in the prediction data of a certain day is more than 0.9, judging that the main shaft bearing of the day has a fault.
In this embodiment, the accuracy, recall, F1 value, and AUC are respectively: 0.9942, 0.9943, 0.9942 and 0.9943 are all larger than 0.9, and meet the training requirement. FIG. 3 is a model test ROC curve, and the ROC curve is observed and very close to the upper left corner, so that the training requirement is met. Inputting new test data to predict the main shaft bearing fault.
7) Viewing test set output values
As shown in fig. 4, the test set is data of one week before the wear failure of the main shaft bearing, and the wear failure of the main shaft bearing occurs in 11 months and 26 days. After model prediction, the proportion of '1' in the prediction data at 11 months and 20 days is larger than 0.9, so that the main shaft bearing can be judged to have a fault one week in advance.
In summary, the data is prepared in the early stage through the steps of data exploration, data cleaning, data standardization and unbalance processing and data correlation analysis. And then, the early-stage prepared data is input into the XGboost classification model for training, and after model evaluation and meeting requirements, the main shaft bearing fault can be predicted one week in advance, so that the method has practical application value and is worthy of popularization.
The above-mentioned embodiments are merely preferred embodiments of the present invention, and the scope of the present invention is not limited thereto, so that the changes in the shape and principle of the present invention should be covered within the protection scope of the present invention.
Claims (1)
1. A wind turbine generator system main shaft bearing fault prediction method is characterized by comprising the following steps: the method is based on historical fault maintenance data of a fan main shaft bearing, combines statistics and machine learning methods, takes a plurality of monitoring indexes of the fan as input variables, takes the state of the main shaft bearing as a prediction output variable, carries out statistical analysis on the prediction value of the output variable, and sets a threshold value for carrying out fault prediction; which comprises the following steps:
1) data exploration
1.1) data quality analysis
The data quality analysis is the basis of validity and accuracy of data mining analysis conclusion, the main task of the data quality analysis is to check whether dirty data exists in original data, the dirty data refers to data which does not meet requirements and cannot be directly used for analysis, and the dirty data comprises missing values, abnormal values, inconsistent values, repeated data and data containing special symbols;
1.2) data feature analysis
After the data is subjected to quality analysis, data characteristic analysis is performed by means of drawing a chart and calculating certain characteristic quantities, and the data characteristic analysis comprises the following steps:
distribution analysis: the distribution characteristics and the distribution types of the data can be revealed;
and (3) comparative analysis: comparing two or more interrelated indicators to quantitatively display and account for the relationship between the indicators;
statistical analysis: carrying out statistical description on the quantitative data by using the statistics;
2) data cleansing
The data cleaning mainly comprises the steps of deleting irrelevant data and repeated data in an original data set, smoothing noise data, screening data irrelevant to a mining theme, and processing a missing value and an abnormal value;
there are 3 methods for handling missing data values: deleting records, interpolating data and not processing;
the method for processing the abnormal value comprises the following steps: deleting records, processing according to missing values, correcting the average value and not processing when the records are regarded as the missing values;
3) data normalization and imbalance processing
Carrying out standardization processing on the data, and converting the data into a proper form so as to be suitable for the requirement of an algorithm;
in order to eliminate the influence of dimension and value range difference between indexes, standard processing is needed, and data is scaled according to a proportion so as to fall into a specific area, thereby facilitating comprehensive analysis; the data standardization method comprises the following steps: min-max normalization and zero-mean normalization;
the problem of unbalance of positive and negative samples in an original data set can cause that a model cannot be correctly classified; the data balance processing method comprises the following steps: an ADASYNN sampling method and a SMOTE sampling method; the ADASYNN sampling method is a self-adaptive synthesis sampling method, and a certain mechanism is adopted to automatically determine the number of synthesized samples required to be generated by each few samples; the SMOTE sampling method is to interpolate between a few classes of samples to produce additional samples;
4) data correlation analysis
Carrying out correlation analysis on the data processed in the step 3): firstly, a random forest method is used for feature selection, variables with large correlation with the state of a main shaft bearing are selected, then the fault mechanism of the main shaft bearing is analyzed, the variables with large correlation with the state of the main shaft bearing are selected, the correlation variables obtained by combining the two methods are combined, and the variables with large correlation among the variables are deleted by a Pearson correlation coefficient calculation method, so that data dimensionality is reduced, and model training is prevented from being over-fitted; finally, the input variables comprise a wind speed real-time value, a generator power real-time value, a generator rotating speed real-time value, an X-direction vibration value, a Y-direction vibration value, an engine room temperature, an outdoor temperature, a hub temperature, a non-driving end temperature of a gear box, a gear box oil temperature, a main shaft bearing A temperature, a cable twisting angle, an annual energy production amount and a power grid current L1; the output variable is the main shaft bearing state value: '0' indicates normal, '1' indicates failure;
5) data modeling
Adopting an XGboost algorithm, inputting processed data into an XGboost classification model for training, adding cross validation to prevent over-fitting of training, importing test data to be analyzed after training, and performing fault prediction; the XGboost algorithm is an extreme gradient boosting algorithm and is an integrated learning method;
6) model evaluation and failure prediction
The model evaluation mainly comprises the steps of calculating indexes of accuracy P, recall R, F1 value, AUC and ROC curve of the model, and if the accuracy, the recall, the F1 value and the AUC are all larger than 0.9, and the ROC curve is close to the upper left corner, the training requirement is met; wherein, the F1 value is a comprehensive evaluation index of accuracy and recall rate, the F1 is 2P R/(P + R), and the larger the F1 value is, the more ideal the model is; the ROC curve is a receiver operation characteristic curve and is used for evaluating the prediction capability of the model, and the area under the curve, namely the AUC, is larger, or the curve is closer to the upper left corner, so that the model is more ideal; the AUC is the area enclosed by the ROC curve and the coordinate axis, the numerical value is less than or equal to 1, and the higher the AUC is, the more ideal the model is;
and a fault prediction stage for performing real-time diagnosis for a specified time, and if the proportion of '1' in the prediction data of a certain day is more than 0.9, judging that the main shaft bearing of the day has a fault.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910788602.0A CN110674842A (en) | 2019-08-26 | 2019-08-26 | Wind turbine generator main shaft bearing fault prediction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910788602.0A CN110674842A (en) | 2019-08-26 | 2019-08-26 | Wind turbine generator main shaft bearing fault prediction method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110674842A true CN110674842A (en) | 2020-01-10 |
Family
ID=69075777
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910788602.0A Pending CN110674842A (en) | 2019-08-26 | 2019-08-26 | Wind turbine generator main shaft bearing fault prediction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110674842A (en) |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111337244A (en) * | 2020-03-13 | 2020-06-26 | 华风数据(深圳)有限公司 | Method and device for monitoring and diagnosing faults of input shaft of fan gearbox |
CN111380686A (en) * | 2020-04-26 | 2020-07-07 | 华风数据(深圳)有限公司 | XGboost algorithm model-based fan main bearing fault monitoring and diagnosis method |
CN111412948A (en) * | 2020-03-31 | 2020-07-14 | 南京云思顿环保科技有限公司 | Kitchen waste equipment fault diagnosis method based on deep forest |
CN111539553A (en) * | 2020-03-31 | 2020-08-14 | 华北电力大学 | Wind turbine generator fault early warning method based on SVR algorithm and skewness |
CN111562496A (en) * | 2020-05-15 | 2020-08-21 | 北京天工智造科技有限公司 | Motor running state judgment method based on data mining |
CN111639304A (en) * | 2020-06-02 | 2020-09-08 | 江南大学 | CSTR fault positioning method based on Xgboost regression model |
CN111754031A (en) * | 2020-06-11 | 2020-10-09 | 苏州量盾信息科技有限公司 | Criminal suspect prediction method based on machine learning technology |
CN111947928A (en) * | 2020-08-10 | 2020-11-17 | 山东大学 | Multi-source information fusion bearing fault prediction system and method |
CN112577739A (en) * | 2020-12-03 | 2021-03-30 | 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 | Over-temperature fault diagnosis and early warning method for bearing at driving end of wind turbine generator |
CN112576454A (en) * | 2020-12-11 | 2021-03-30 | 龙源(北京)风电工程技术有限公司 | Wind turbine generator main shaft temperature early warning method and device based on multi-dimensional early warning strategy |
CN112765878A (en) * | 2021-01-11 | 2021-05-07 | 龙源(北京)风电工程技术有限公司 | Intelligent diagnosis method and device for abnormity of generator bearing of wind turbine generator |
CN112966879A (en) * | 2021-04-02 | 2021-06-15 | 阳光电源股份有限公司 | Environmental test chamber fault prediction method and device, computer equipment and storage medium |
CN113129054A (en) * | 2021-03-30 | 2021-07-16 | 广州博冠信息科技有限公司 | User identification method and device |
CN113221453A (en) * | 2021-04-30 | 2021-08-06 | 华风数据(深圳)有限公司 | Fault monitoring and early warning method for output shaft of gearbox of wind turbine generator |
CN113505850A (en) * | 2021-07-27 | 2021-10-15 | 西安热工研究院有限公司 | Boiler fault prediction method based on deep learning |
CN113614758A (en) * | 2020-01-22 | 2021-11-05 | 京东方科技集团股份有限公司 | Equipment index goodness grade prediction model training method, monitoring system and method |
CN114357663A (en) * | 2022-03-15 | 2022-04-15 | 中国长江三峡集团有限公司 | Method for training gearbox fault diagnosis model and gearbox fault diagnosis method |
CN117032120A (en) * | 2023-08-16 | 2023-11-10 | 广东超级龙节能环保科技有限公司 | Integrated intelligent cloud control system and control method for air compression station |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106844826A (en) * | 2016-12-02 | 2017-06-13 | 上海电机学院 | A kind of method for the diagnosis of gearbox of wind turbine failure predication |
CN108363714A (en) * | 2017-12-21 | 2018-08-03 | 北京至信普林科技有限公司 | A kind of method and system for the ensemble machine learning for facilitating data analyst to use |
CN108954680A (en) * | 2018-07-13 | 2018-12-07 | 电子科技大学 | A kind of air-conditioning energy consumption prediction technique based on operation data |
CN109543210A (en) * | 2018-09-28 | 2019-03-29 | 国电电力宁夏新能源开发有限公司 | A kind of Wind turbines failure prediction system based on machine learning algorithm platform |
CN110070060A (en) * | 2019-04-26 | 2019-07-30 | 天津开发区精诺瀚海数据科技有限公司 | A kind of method for diagnosing faults of bearing apparatus |
-
2019
- 2019-08-26 CN CN201910788602.0A patent/CN110674842A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106844826A (en) * | 2016-12-02 | 2017-06-13 | 上海电机学院 | A kind of method for the diagnosis of gearbox of wind turbine failure predication |
CN108363714A (en) * | 2017-12-21 | 2018-08-03 | 北京至信普林科技有限公司 | A kind of method and system for the ensemble machine learning for facilitating data analyst to use |
CN108954680A (en) * | 2018-07-13 | 2018-12-07 | 电子科技大学 | A kind of air-conditioning energy consumption prediction technique based on operation data |
CN109543210A (en) * | 2018-09-28 | 2019-03-29 | 国电电力宁夏新能源开发有限公司 | A kind of Wind turbines failure prediction system based on machine learning algorithm platform |
CN110070060A (en) * | 2019-04-26 | 2019-07-30 | 天津开发区精诺瀚海数据科技有限公司 | A kind of method for diagnosing faults of bearing apparatus |
Non-Patent Citations (4)
Title |
---|
余小高 著: "《基于大数据的高风险学生预测研究》", 31 May 2019 * |
周永章 等著: "《地球科学大数据挖掘与机器学习》", 30 September 2018, 中山大学出版社 * |
张昆 著: "《专业学位研究生实验课程 荟萃篇》", 30 November 2017, 华中科技大学出版社 * |
王桂兰 等: "XGBoost 算法在风机主轴承故障预测中的应用", 《电力自动化设备》 * |
Cited By (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113614758A (en) * | 2020-01-22 | 2021-11-05 | 京东方科技集团股份有限公司 | Equipment index goodness grade prediction model training method, monitoring system and method |
CN111337244A (en) * | 2020-03-13 | 2020-06-26 | 华风数据(深圳)有限公司 | Method and device for monitoring and diagnosing faults of input shaft of fan gearbox |
CN111337244B (en) * | 2020-03-13 | 2022-04-01 | 华风数据(深圳)有限公司 | Method and device for monitoring and diagnosing faults of input shaft of fan gearbox |
CN111412948A (en) * | 2020-03-31 | 2020-07-14 | 南京云思顿环保科技有限公司 | Kitchen waste equipment fault diagnosis method based on deep forest |
CN111539553A (en) * | 2020-03-31 | 2020-08-14 | 华北电力大学 | Wind turbine generator fault early warning method based on SVR algorithm and skewness |
CN111539553B (en) * | 2020-03-31 | 2023-10-24 | 华北电力大学 | Wind turbine generator fault early warning method based on SVR algorithm and off-peak degree |
CN111380686A (en) * | 2020-04-26 | 2020-07-07 | 华风数据(深圳)有限公司 | XGboost algorithm model-based fan main bearing fault monitoring and diagnosis method |
CN111562496A (en) * | 2020-05-15 | 2020-08-21 | 北京天工智造科技有限公司 | Motor running state judgment method based on data mining |
CN111639304A (en) * | 2020-06-02 | 2020-09-08 | 江南大学 | CSTR fault positioning method based on Xgboost regression model |
CN111639304B (en) * | 2020-06-02 | 2023-02-21 | 江南大学 | CSTR fault positioning method based on Xgboost regression model |
CN111754031A (en) * | 2020-06-11 | 2020-10-09 | 苏州量盾信息科技有限公司 | Criminal suspect prediction method based on machine learning technology |
CN111947928B (en) * | 2020-08-10 | 2021-11-05 | 山东大学 | Multi-source information fusion bearing fault prediction system and method |
CN111947928A (en) * | 2020-08-10 | 2020-11-17 | 山东大学 | Multi-source information fusion bearing fault prediction system and method |
CN112577739A (en) * | 2020-12-03 | 2021-03-30 | 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 | Over-temperature fault diagnosis and early warning method for bearing at driving end of wind turbine generator |
CN112576454A (en) * | 2020-12-11 | 2021-03-30 | 龙源(北京)风电工程技术有限公司 | Wind turbine generator main shaft temperature early warning method and device based on multi-dimensional early warning strategy |
CN112765878A (en) * | 2021-01-11 | 2021-05-07 | 龙源(北京)风电工程技术有限公司 | Intelligent diagnosis method and device for abnormity of generator bearing of wind turbine generator |
CN113129054A (en) * | 2021-03-30 | 2021-07-16 | 广州博冠信息科技有限公司 | User identification method and device |
CN113129054B (en) * | 2021-03-30 | 2024-05-31 | 广州博冠信息科技有限公司 | User identification method and device |
CN112966879A (en) * | 2021-04-02 | 2021-06-15 | 阳光电源股份有限公司 | Environmental test chamber fault prediction method and device, computer equipment and storage medium |
CN113221453A (en) * | 2021-04-30 | 2021-08-06 | 华风数据(深圳)有限公司 | Fault monitoring and early warning method for output shaft of gearbox of wind turbine generator |
CN113505850A (en) * | 2021-07-27 | 2021-10-15 | 西安热工研究院有限公司 | Boiler fault prediction method based on deep learning |
CN114357663A (en) * | 2022-03-15 | 2022-04-15 | 中国长江三峡集团有限公司 | Method for training gearbox fault diagnosis model and gearbox fault diagnosis method |
CN117032120A (en) * | 2023-08-16 | 2023-11-10 | 广东超级龙节能环保科技有限公司 | Integrated intelligent cloud control system and control method for air compression station |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110674842A (en) | Wind turbine generator main shaft bearing fault prediction method | |
CN103645052B (en) | A kind of gearbox of wind turbine remote online status monitoring and lifetime estimation method | |
Zhi-Ling et al. | Expert system of fault diagnosis for gear box in wind turbine | |
CN111539553A (en) | Wind turbine generator fault early warning method based on SVR algorithm and skewness | |
JP4995134B2 (en) | Wind turbine monitoring device, method and program | |
CN112487910A (en) | Fault early warning method and system for nuclear turbine system | |
CN111878320A (en) | Monitoring method and system of wind generating set and computer readable storage medium | |
CN111896254A (en) | Fault prediction system and method for variable-speed variable-load large rolling bearing | |
CN107291991A (en) | A kind of Wind turbines early defect method for early warning based on dynamic network mark | |
Márquez et al. | An overview of wind turbine maintenance management | |
CN112001511A (en) | Equipment reliability and dynamic risk evaluation method, system and equipment based on data mining | |
CN115453356B (en) | Power equipment operation state monitoring and analyzing method, system, terminal and medium | |
Cevasco et al. | O&M cost-Based FMECA: Identification and ranking of the most critical components for 2-4 MW geared offshore wind turbines | |
CN115578084A (en) | Wind turbine generator set frequency converter fault early warning method based on deep convolution self-encoder | |
CN114687930A (en) | Wind turbine generator fault early warning closed-loop management and control system and method | |
Harzendorf et al. | Reducing cost uncertainty in the drivetrain design decision with a focus on the operational phase | |
CN109409758B (en) | Hydropower station equipment health state evaluation method and system | |
Joshuva et al. | Split-point and attribute-reduced classifier approach for fault diagnosis of wind turbine blade through vibration signals | |
CN116976191A (en) | Method for predicting shafting stability degradation trend of hydroelectric generating set | |
CN111308016A (en) | Gear box fault diagnosis method, system, equipment and storage medium | |
CN116596120A (en) | Variable working condition degradation trend prediction method and device for pumped storage unit | |
CN114320773B (en) | Wind turbine generator system fault early warning method based on power curve analysis and neural network | |
CN115906437A (en) | Fan state determination method, device, equipment and storage medium | |
CN114417514A (en) | Wind turbine generator gearbox fault diagnosis system and method, computer equipment and storage medium | |
Song et al. | Framework of designing an adaptive and multi-regime prognostics and health management for wind turbine reliability and efficiency improvement |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200110 |
|
RJ01 | Rejection of invention patent application after publication |