CN115693781A - Wind power output limitation estimation method under extreme weather conditions of low temperature and cold tide - Google Patents
Wind power output limitation estimation method under extreme weather conditions of low temperature and cold tide Download PDFInfo
- Publication number
- CN115693781A CN115693781A CN202211442870.5A CN202211442870A CN115693781A CN 115693781 A CN115693781 A CN 115693781A CN 202211442870 A CN202211442870 A CN 202211442870A CN 115693781 A CN115693781 A CN 115693781A
- Authority
- CN
- China
- Prior art keywords
- wind power
- data
- power output
- time period
- extreme weather
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a wind power output limitation estimation method under the extreme weather condition of low temperature and cold tide, which is characterized by dividing the data of the existing wind power plant for training and testing a prediction model, screening the training data by combining the definition index of the low temperature and cold tide to obtain the data of the wind power output limitation time period and the output normal time period under the extreme weather condition of low temperature and cold tide, and performing sample data combination expansion through timeGAN and cycleGAN; respectively constructing feature maps of the two types of data, and combining SVM algorithm training to obtain a wind power output limited discrimination model; respectively constructing a wind power output limited time period prediction model and a wind power output normal time period prediction model under the low-temperature cold tide extreme weather condition through an LSTM network model; and (3) a wind power output comprehensive prediction model under the extreme weather conditions of low temperature and cold tide is combined and constructed, test data are input into the comprehensive prediction model and then are subjected to prediction, and a result is output. The method realizes the early warning of the wind power output limitation under the extreme weather conditions of low temperature and cold tide, improves the wind power output prediction power precision, and has popularization value.
Description
Technical Field
The invention relates to a wind power output limitation estimation method under a low-temperature cold tide extreme weather condition, and belongs to the field of new energy power prediction.
Background
The wind power installation proportion of China is continuously improved, and wind power occupies an increasingly important position in the field of energy. However, wind power is often influenced by the operating environment conditions of the wind power plant, and different types of weather processes can influence wind power output to different degrees. In the low-temperature cold tide weather, as an extreme weather event, the phenomena of severe temperature drop, continuous strong wind, rime, frost and the like are often accompanied in the process, and the wind power output is often obviously adversely affected.
At present, wind power prediction is mainly carried out by constructing a prediction model according to a wind speed prediction result to predict wind power output, however, training data are often preprocessed when the prediction model is constructed, abnormal data concentrated in training are eliminated, so that the wind power prediction model does not consider the influence brought by extreme weather events represented by low-temperature cold weather, the wind power prediction result and actual wind power generate large deviation under the condition of low-temperature cold weather, and the scheduling application requirement of a power system is difficult to meet.
Disclosure of Invention
The invention aims to provide a wind power output limitation estimation method under the extreme weather condition of low temperature and cold tide, which aims to solve the problems in the background art.
The purpose of the invention is realized by the following technical measures:
a wind power output limitation estimation method under extreme weather conditions of low temperature and cold tide comprises the following steps:
s1: dividing the existing wind power plant data for training and testing a prediction model, screening the training data by combining with low-temperature cold-tide weather definition indexes to obtain wind power output limited Time period and output normal Time period data under the extreme weather condition of low-temperature cold tide, and performing sample data combination expansion by using a timeGAN (Time Series generated adaptive network) and a cycleGAN (Cycle generated adaptive network);
s2: respectively constructing feature maps of the two types of data of a wind power output limited time period and a wind power output normal time period based on the data subjected to S1 data screening and sample expansion, and training by combining with an SVM (support vector machines) algorithm to obtain a wind power output limited discrimination model under the extreme weather conditions of low temperature and cold tide;
s3: on the basis of data screened by the S1 and sample expanded, respectively constructing a wind power output limited time period and a wind power output normal time period prediction model under the extreme weather conditions of low temperature and cold tide through an LSTM network model;
s4: the method comprises the steps of constructing a wind power output comprehensive prediction model under the extreme weather conditions of low temperature and cold tide in a combined mode, inputting test data of S1 into the comprehensive prediction model to perform prediction by taking RMSE (Root Mean square error) and MAE (Mean absolute error) as performance evaluation standards of the comprehensive prediction model, and outputting results.
Further, the specific steps of S1 include: the existing wind farm data are divided, the acquired NWP (Numerical Weather Prediction) data set with the wind farm time resolution of 15 minutes and the corresponding power data are included, 70% of the data are used for training a Prediction model, and 30% of the data are used for testing the Prediction model. And screening the training data by combining with low-temperature cold tide weather definition indexes to obtain NWP data and wind power data of the wind power output limited time period and the output normal time period under the low-temperature cold tide extreme weather condition. And based on the obtained data, respectively adopting timeGAN and cycleGAN to carry out combined expansion on the sample data, respectively inputting the data of the limited output time period and the normal output time period into a timeGAN model to carry out corresponding type data expansion, and simultaneously inputting the data of the limited output time period and the normal output time period into a cycleGAN model to carry out two types of data common expansion.
Further, the specific step of S2 includes: and taking the wind power output limited time period data and the wind power output normal time period data obtained in the step S1 as original data, performing feature optimization on the original data based on correlation analysis to obtain an optimal feature subset, performing linear transformation on the data by combining a Min-Max standardization (Min-Max Normalization) method, mapping the result between [0 and 255], converting the obtained imaged data into a gray scale map, taking the gray scale map as a feature map of a corresponding time period, selecting an SVM (support vector machine) as a classifier model for wind power output limitation judgment, and training to obtain a wind power output limitation judgment model under the low-temperature cold tide extreme weather condition.
Further, the specific step of S3 includes:
s3.1: aiming at the normal wind power output time period under the extreme weather condition of low temperature and cold tide, selecting original NWP data expanded by an S1 sample as the input of a prediction model of the normal wind power output time period, wherein the original NWP data comprises wind speed, wind direction, air temperature and air pressure, and obtaining the prediction model of the normal wind power output time period under the extreme weather condition of low temperature and cold tide by combining with the training of an LSTM (Long Short-Term Memory) algorithm;
s3.2: aiming at the wind power output limited time period under the extreme weather condition of low temperature and cold tide, feature mining is carried out by combining a VMD (spatial Mode Decomposition) signal Decomposition algorithm on the basis of original NWP data expanded by an S1 sample, the NWP data are decomposed into components with different frequencies, high-dimensional meteorological features are constructed, feature selection is carried out by combining a random forest algorithm, typical meteorological features of the wind power output limited time period under the low temperature and cold tide are obtained, the typical meteorological features comprise a wind speed high-frequency component, a temperature low-frequency component and a wind direction low-frequency component and serve as input of a wind power output limited time period prediction model, and the wind power output limited time period prediction model under the extreme weather condition of low temperature and cold tide is obtained by combining LSTM algorithm training.
Further, the step of S4 includes: and (3) combining the wind power output limited discrimination model under the low-temperature cold tide extreme weather condition obtained in the step (S2) with the wind power output limited time period and output normal time period prediction model under the low-temperature cold tide extreme weather condition obtained in the step (S3), and constructing to obtain a wind power output comprehensive prediction model under the low-temperature cold tide extreme weather condition. Judging whether the wind power output is limited by an output limited judging model, and predicting according to a judgment result and a corresponding wind power output limited time period prediction model or a wind power output normal time period prediction model to obtain a prediction result; and (3) taking RMSE and MAE as the performance evaluation standards of the comprehensive prediction model, using 30% of test data of S1 for the comprehensive prediction model to enter prediction, and outputting a result.
The invention achieves the following beneficial effects: the method disclosed by the invention realizes the expansion of the low-temperature cold tide weather small sample data set through timeGAN, meets the data scale requirement of model training, efficiently eliminates redundant data through a signal decomposition technology VMD and a random forest selection algorithm to improve the prediction precision of the wind power output loss time period under the low-temperature cold tide weather condition, improves the final prediction precision by constructing a wind power output comprehensive prediction model under the low-temperature cold tide extreme weather condition in a combined manner, and plays an important role in wind power prediction under the low-temperature cold tide extreme weather condition.
Drawings
Fig. 1 is a block diagram of the overall flow structure of the present invention.
Detailed Description
The following detailed description is made with reference to the accompanying drawings in the embodiment of the invention, and the purpose of the invention is to provide a wind power output limitation estimation method under the extreme weather conditions of low temperature and cold tide.
As shown in FIG. 1, a method for estimating wind power output limitation under extreme weather conditions of low temperature and cold tide is characterized by comprising the following steps:
s1: dividing the existing wind power plant data for training and testing a prediction model, screening the training data by combining with low-temperature cold tide weather definition indexes to obtain data of a limited wind power output period and a normal output period under the extreme weather condition of low-temperature cold tide, and performing sample data combination expansion through timeGAN and cycleGAN; the method comprises the following specific steps:
the method comprises the steps of dividing the existing wind power plant data, wherein the data comprise an acquired NWP data set with the wind power plant time resolution of 15 minutes and corresponding power data, 70% of the data are used for training a prediction model, and 30% of the data are used for testing the prediction model. And screening the training data by combining with low-temperature cold tide weather definition indexes to obtain NWP data and wind power data of the wind power output limited time period and the output normal time period under the low-temperature cold tide extreme weather condition. And based on the obtained data, performing combined expansion on the sample data by adopting a timeGAN and a cycleGAN respectively, inputting the data of the limited output time period and the normal output time period into a timeGAN model respectively to perform corresponding type data expansion, and inputting the data of the limited output time period and the normal output time period into a cycleGAN model simultaneously to perform common expansion on the two types of data.
S2: respectively constructing feature maps of the two types of data of the limited wind power output time period and the normal wind power output time period based on the data subjected to S1 data screening and sample expansion, and combining SVM algorithm training to obtain a wind power output limited discrimination model under the extreme weather conditions of low temperature and cold tide; the method comprises the following specific steps:
and (2) taking the data of the limited wind power output time period and the normal wind power output time period under the extreme weather condition of low temperature and cold tide obtained in the step (S1) as original data, performing feature optimization on the original data based on correlation analysis to obtain an optimal feature subset, performing linear transformation on the data by combining a Min-Max standardization method to map the result to the range between [0 and 255], converting the obtained imaged data into a gray graph to be used as a feature map of a corresponding time period, selecting an SVM (support vector machine) as a classifier model for judging the limited wind power output, and training to obtain a judgment model for judging the limited wind power output under the extreme weather condition of low temperature and cold tide.
S3: based on data screened by S1 and sample expanded, a wind power output limited time period and output normal time period prediction model under the low-temperature cold tide extreme weather condition is respectively established through an LSTM network model, and the method specifically comprises the following steps:
s3.1: aiming at the normal wind power output time period under the extreme weather condition of low temperature cold tide, selecting original NWP data expanded by the S1 sample as the input of a prediction model of the normal wind power output time period, wherein the original NWP data comprise wind speed, wind direction, air temperature and air pressure, and combining LSTM algorithm training to obtain the prediction model of the normal wind power output time period under the extreme weather condition of low temperature cold tide;
s3.2: aiming at the wind power output limited time period under the extreme weather condition of low temperature and cold tide, based on original NWP data expanded by an S1 sample, combining a VMD signal decomposition algorithm to perform feature mining, decomposing the NWP data into components with different frequencies, constructing high-dimensional meteorological features, then combining a random forest algorithm to perform feature selection, obtaining typical meteorological features of the wind power output limited time period under the low temperature and cold tide, wherein the typical meteorological features comprise a wind speed high-frequency component, a temperature low-frequency component and a wind direction low-frequency component, and are used as the input of a wind power output limited time period prediction model, and combining LSTM algorithm training to obtain a wind power output limited time period prediction model under the extreme weather condition of low temperature and cold tide;
s4: the method comprises the following steps of constructing a wind power output comprehensive prediction model under the extreme weather conditions of low temperature and cold tide in a combined mode, inputting test data of S1 into the comprehensive prediction model by taking RMSE and MAE as performance evaluation standards of the comprehensive prediction model, then performing prediction, and outputting results, wherein the method specifically comprises the following steps:
combining the wind power output limited distinguishing model obtained in the step S2 under the low-temperature cold tide extreme weather condition with the wind power output limited time period and output normal time period prediction model obtained in the step S3 under the low-temperature cold tide extreme weather condition, and constructing to obtain a wind power output comprehensive prediction model under the low-temperature cold tide extreme weather condition; judging whether the wind power output is limited by an output limited judging model, and predicting according to a judgment result and a corresponding wind power output limited time period prediction model or a wind power output normal time period prediction model to obtain a prediction result; and (3) taking RMSE and MAE as the performance evaluation standards of the comprehensive prediction model, using 30% of test data of S1 for the comprehensive prediction model to enter prediction, and outputting a result.
Claims (5)
1. A wind power output limitation estimation method under extreme weather conditions of low temperature and cold tide is characterized by comprising the following steps:
s1: dividing the existing wind power plant data for training and testing a prediction model, screening the training data by combining with low-temperature cold tide weather definition indexes to obtain wind power output limited time period data and wind power output normal time period data under the extreme weather condition of low-temperature cold tide, and performing sample data combination expansion through timeGAN and cycleGAN;
s2: respectively constructing feature maps of the two types of data of the limited wind power output time period and the normal wind power output time period based on the data subjected to S1 data screening and sample expansion, and combining SVM algorithm training to obtain a wind power output limited discrimination model under the extreme weather conditions of low temperature and cold tide;
s3: based on data screened by the S1 and sample expanded, respectively constructing a wind power output limited time period and output normal time period prediction model under the low-temperature cold tide extreme weather condition through an LSTM network model;
s4: and (3) a wind power output comprehensive prediction model under the extreme weather conditions of low temperature and cold tide is combined and constructed, RMSE and MAE are used as performance judgment standards of the comprehensive prediction model, the test data of S1 are input into the comprehensive prediction model and then are subjected to prediction, and a result is output.
2. The method for estimating the limited wind power output under the extreme weather conditions of low temperature, cold and tide as claimed in claim 1, wherein in the step S1, the existing wind farm data is divided, including the acquired NWP data set with the wind farm time resolution of 15 minutes and the corresponding power data thereof, 70% of the data is used for training the prediction model, and 30% of the data is used for testing the prediction model; screening the training data by combining with low-temperature cold tide weather definition indexes to obtain NWP data and wind power data of the wind power output limited time period and the output normal time period under the extreme weather condition of the low-temperature cold tide; and based on the obtained data, respectively adopting timeGAN and cycleGAN to carry out combined expansion on the sample data, respectively inputting the data of the limited output time period and the normal output time period into a timeGAN model to carry out corresponding type data expansion, and simultaneously inputting the data of the limited output time period and the normal output time period into a cycleGAN model to carry out two types of data common expansion.
3. The method for estimating the wind power output limitation under the extreme weather condition of the low temperature cold tide according to claim 1, wherein in the step S2, the data of the wind power output limitation time period and the output normal time period under the extreme weather condition of the low temperature cold tide obtained in the step S1 are used as original data, feature optimization is performed on the original data based on correlation analysis to obtain an optimal feature subset, linear transformation is performed on the data by combining a Min-Max standardization method to map the result between [0,255], the obtained imaging data is converted into a gray scale map to be used as a feature map of a corresponding time period, an SVM is selected to be used as a classifier model for wind power output limitation judgment, and a wind power limitation judgment model under the extreme weather condition of the low temperature cold tide is obtained through training.
4. The method for estimating the wind power output limitation under the extreme weather condition of the low temperature cold tide according to claim 1, wherein in the step S3, a wind power output limitation time period and output normal time period prediction model under the extreme weather condition of the low temperature cold tide is constructed, and the specific steps include:
s3.1: aiming at the normal wind power output time period under the extreme weather condition of low temperature and cold tide, selecting original NWP data expanded by an S1 sample as the input of a prediction model of the normal wind power output time period, wherein the original NWP data comprises wind speed, wind direction, air temperature and air pressure, and obtaining the prediction model of the normal wind power output time period under the extreme weather condition of low temperature and cold tide by combining with LSTM algorithm training;
s3.2: aiming at the wind power output limited time period under the extreme weather condition of the low-temperature cold tide, feature mining is carried out on the basis of original NWP data expanded by an S1 sample by combining a VMD signal decomposition algorithm, the NWP data are decomposed into components with different frequencies, high-dimensional meteorological features are constructed, feature selection is carried out by combining a random forest algorithm, typical meteorological features of the wind power output limited time period under the low-temperature cold tide are obtained, the typical meteorological features comprise a wind speed high-frequency component, a wind temperature low-frequency component and a wind direction low-frequency component and serve as input of a wind power output limited time period prediction model, and the wind power output limited time period prediction model under the extreme weather condition of the low-temperature cold tide is obtained by combining LSTM algorithm training.
5. The method for estimating the limitation of the wind power output under the extreme weather condition of the low-temperature cold tide according to claim 1, wherein in the step S4, the wind power output limitation discrimination model under the extreme weather condition of the low-temperature cold tide obtained in the step S2 is combined with the wind power output limitation time period and output normal time period prediction model under the extreme weather condition of the low-temperature cold tide obtained in the step S3, and a wind power output comprehensive prediction model under the extreme weather condition of the low-temperature cold tide is constructed; judging whether the wind power output is limited by the output limited judging model, and predicting according to a judgment result and a corresponding wind power output limited time period prediction model or a wind power output normal time period prediction model to obtain a prediction result; and (3) taking RMSE and MAE as the performance evaluation standards of the comprehensive prediction model, using 30% of test data of S1 for the comprehensive prediction model to enter prediction, and outputting a result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211442870.5A CN115693781A (en) | 2022-11-17 | 2022-11-17 | Wind power output limitation estimation method under extreme weather conditions of low temperature and cold tide |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211442870.5A CN115693781A (en) | 2022-11-17 | 2022-11-17 | Wind power output limitation estimation method under extreme weather conditions of low temperature and cold tide |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115693781A true CN115693781A (en) | 2023-02-03 |
Family
ID=85053143
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211442870.5A Pending CN115693781A (en) | 2022-11-17 | 2022-11-17 | Wind power output limitation estimation method under extreme weather conditions of low temperature and cold tide |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115693781A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117631090A (en) * | 2024-01-25 | 2024-03-01 | 南京信息工程大学 | Cold tide identification method and device |
-
2022
- 2022-11-17 CN CN202211442870.5A patent/CN115693781A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117631090A (en) * | 2024-01-25 | 2024-03-01 | 南京信息工程大学 | Cold tide identification method and device |
| CN117631090B (en) * | 2024-01-25 | 2024-05-14 | 南京信息工程大学 | Cold tide identification method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110261109B (en) | Rolling bearing fault diagnosis method based on bidirectional memory cyclic neural network | |
| CN111861023B (en) | Mixed wind power prediction method and device based on statistics | |
| CN111878322B (en) | Wind power generator device | |
| CN110659672B (en) | Method and device for predicting step-by-step uncertainty of output of wind turbine generator | |
| CN115693781A (en) | Wind power output limitation estimation method under extreme weather conditions of low temperature and cold tide | |
| CN112507479B (en) | Oil drilling machine health state assessment method based on manifold learning and softmax | |
| CN112462736A (en) | Wind turbine generator fault diagnosis method based on data analysis | |
| CN111695724A (en) | Wind speed prediction method based on hybrid neural network model | |
| CN112633632A (en) | Integrated short-term wind power cluster power prediction method based on signal decomposition technology | |
| CN114372344B (en) | Quantitative identification method for subsynchronous oscillation damping characteristic influence factors | |
| CN115827411A (en) | Online monitoring and operation and maintenance evaluation system and method for automation equipment | |
| CN117192371B (en) | Test method and system for motor driver of new energy automobile | |
| CN112215281A (en) | Fan blade icing fault detection method | |
| Deng et al. | A fractional hilbert transform order optimization algorithm based DE for bearing health monitoring | |
| CN114034375A (en) | System and method for measuring noise of ultra-high voltage transmission line | |
| CN113722977B (en) | Gas turbine rotor fault early warning method based on hybrid prediction | |
| CN116720616A (en) | Short-term power prediction method for extremely-large wind field Jing Feng electric field | |
| Xin et al. | Automatic separation of temperature effects from bridge cable force data based on the VMD-PE-KLD algorithm | |
| CN104791233B (en) | Based on the reciprocating compressor method for diagnosing faults for improving the solution of ball vector machine closure ball | |
| CN118228897B (en) | CEEMDAN-SE-TCN-GRU model-based power load prediction method | |
| CN118656272A (en) | Equipment operation process monitoring system | |
| CN114881073B (en) | Sectional fault diagnosis method and system for variable working condition rolling bearing | |
| CN111780149B (en) | Remote diagnosis method for equipment state of pulverizing system of thermal power plant | |
| Yu et al. | Multifractal analysis and stacked autoencoder-based feature learning method for multivariate processes monitoring | |
| CN118152871A (en) | Digital twinning-based power distribution network fault diagnosis system and method |
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 |