CN116050504A - Wind power short-term prediction model based on deep learning - Google Patents
Wind power short-term prediction model based on deep learning Download PDFInfo
- Publication number
- CN116050504A CN116050504A CN202310088915.1A CN202310088915A CN116050504A CN 116050504 A CN116050504 A CN 116050504A CN 202310088915 A CN202310088915 A CN 202310088915A CN 116050504 A CN116050504 A CN 116050504A
- Authority
- CN
- China
- Prior art keywords
- data
- wind power
- term prediction
- deep learning
- prediction model
- 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
- 238000013135 deep learning Methods 0.000 title claims abstract description 26
- 238000010248 power generation Methods 0.000 claims abstract description 6
- 238000010220 Pearson correlation analysis Methods 0.000 claims abstract description 5
- 230000006870 function Effects 0.000 claims description 31
- 238000013528 artificial neural network Methods 0.000 claims description 16
- 238000003062 neural network model Methods 0.000 claims description 16
- 238000000034 method Methods 0.000 claims description 13
- 230000007246 mechanism Effects 0.000 claims description 12
- 230000003044 adaptive effect Effects 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 8
- 230000009466 transformation Effects 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 5
- 238000013459 approach Methods 0.000 claims description 4
- 230000002457 bidirectional effect Effects 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 4
- 238000012886 linear function Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 230000003399 chemotactic effect Effects 0.000 claims description 3
- 238000011161 development Methods 0.000 abstract description 6
- 238000007619 statistical method Methods 0.000 abstract description 6
- 238000000605 extraction Methods 0.000 abstract description 5
- 230000035605 chemotaxis Effects 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000003475 lamination Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2462—Approximate or statistical queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2474—Sequence data queries, e.g. querying versioned data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Strategic Management (AREA)
- Evolutionary Computation (AREA)
- Marketing (AREA)
- Molecular Biology (AREA)
- Fuzzy Systems (AREA)
- Databases & Information Systems (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- General Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Tourism & Hospitality (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Wind Motors (AREA)
Abstract
The invention discloses a wind power short-term prediction model based on deep learning, which comprises the following steps: acquiring wind power generation power and NWP original data; filling up the data missing value by using the average value; reducing the data by pearson correlation analysis, and eliminating unnecessary features; according to the invention, wind power is subjected to short-term prediction by adopting a wind power short-term prediction model based on deep learning, on one hand, continuity of object development is acknowledged, and past time sequence data is used for statistical analysis to estimate the development trend of objects; on the other hand, the randomness caused by the influence of accidental factors is fully considered, in order to eliminate the influence caused by random fluctuation, the historical data is utilized for carrying out statistical analysis, the data is properly processed for trend prediction, the problem of insufficient information extraction among data features is solved, and the prediction precision of the model is further improved.
Description
Technical Field
The invention belongs to the technical field of wind power prediction, and particularly relates to a wind power short-term prediction model based on deep learning.
Background
Wind power prediction techniques refer to predicting the amount of power that a wind farm can output over a period of time in the future in order to schedule a dispatch plan. This is because wind energy belongs to an unstable energy source with random fluctuation, and large-scale wind power is integrated into a system, so that new challenges are brought to the stability of the system. The power generation scheduling mechanism needs to know the wind power output power of several hours in the future. Dividing according to a wind farm output prediction time scale, wherein the method comprises the following steps of: long-term prediction, medium-term prediction, short-term prediction, and ultra-short term prediction.
At present, the traditional short-term prediction mode of wind power generally depends on feature engineering, and generally only a single time point in the future can be predicted basically aiming at a single time sequence, the information extraction among data features is insufficient, and serious limitations still exist during use.
Disclosure of Invention
The invention aims to overcome the existing defects, provides a wind power short-term prediction model based on deep learning, and solves the problems that the traditional wind power short-term prediction mode proposed in the background art generally depends on characteristic engineering, and generally only predicts a single time sequence in the future, information extraction among data features is insufficient, and serious limitation still exists when the wind power short-term prediction model is used.
In order to achieve the above purpose, the present invention provides the following technical solutions: a wind power short-term prediction model based on deep learning comprises the following steps:
step one: acquiring wind power generation power and NWP original data;
step two: filling up the data missing value by using the average value;
step three: reducing the data by pearson correlation analysis, and eliminating unnecessary features;
step four: normalizing the data to obtain normalized data;
step five: modeling the multivariate time series data by using a graph neural network, namely constructing an adaptive adjacency matrix to learn hidden space dependency relations among variables;
step six: extracting time features by using a bidirectional gating circulating unit;
step seven: weighting the features extracted in the previous step through an attention mechanism, and outputting through a full connection layer;
step eight: and (5) carrying out inverse normalization on the predicted data to obtain a final result.
Preferably, in the first step, before predicting the wind power, the characteristics of the wind energy are first known so as to reasonably select parameters, and the wind energy is most remarkable in variability, and the magnitude and direction of the wind power at each time point and space point are instantaneously changed due to the influence of weather, time and geographical environment factors.
Preferably, in the second step, the variable (attribute) having no missing value in the data set is referred to as complete data, the variable having a missing value in the data set is referred to as missing data, and the method of filling the data is generally to fill in by using larger-scale and variable data.
Preferably, in the second step, attention is paid to interpretation of sequence points with low coverage when using these data for data padding.
Preferably, in the third step, a pearson product moment correlation coefficient (PPMCC/PCCs) is generally used in statistics for measuring a statistical indicator of the degree of closeness of the correlation between data.
Preferably, in the fourth step, when the data is normalized, the data is processed in a data isotactics manner, so that the comprehensive results of different acting forces cannot be correctly reflected by directly adding the data indexes with different properties, and the reverse index data properties are considered to be changed first, so that all the indexes are chemotactic to the acting forces of the evaluation scheme, and then the correct results can be obtained by adding the data indexes.
Preferably, in the fifth step, the built RBF neural network model is composed of three layers, namely an input layer, a hidden layer and an output layer, and is a novel and effective feedforward neural network.
Preferably, the input layer of the RBF neural network model is used for transmitting an input signal to the hidden layer; the hidden layer is described by a radial basis function, the basis function of the hidden layer mostly adopts a Gaussian basis function, and nonlinear transformation is carried out; the output layer is characterized by a simple linear function, linear transformation is performed, in the RBF neural network model, the connection weight of the input layer to the hidden layer is 1, the connection weight of the hidden layer to the output layer is adjustable, the function of the hidden layer locally responds to the input signal, and when the input signal approaches to the central range of the function, the hidden layer node can generate larger output, so that the radial basis function network has the advantages of excellent local approximation capability and high learning speed.
Preferably, in the fifth step, the graph neural network is used, so that the graph neural network model does not need an explicit graph structure when processing data.
Preferably, in the seventh step, the attention mechanism includes a first arithmetic logic unit, an activation function circuit, a scaling circuit, and a second arithmetic logic unit.
Compared with the prior art, the invention provides a wind power short-term prediction model based on deep learning, which has the following beneficial effects:
1. according to the invention, wind power is subjected to short-term prediction by adopting a wind power short-term prediction model based on deep learning, on one hand, continuity of object development is acknowledged, and past time sequence data is used for statistical analysis to estimate the development trend of objects; on the other hand, the randomness caused by the influence of accidental factors is fully considered, in order to eliminate the influence caused by random fluctuation, historical data is utilized for statistical analysis, the data is properly processed for trend prediction, the problem of insufficient information extraction among data features is solved, and the prediction precision of a model is further improved;
2. compared with the traditional prediction method, the method has the following advantages: weak dependence or independent characteristic engineering, and directly extracting the relation from the data; the traditional method is basically aimed at a single time sequence, and deep learning can simultaneously predict a plurality of time sequences, and in addition, the correlation between the time sequences can be captured; future time points can be predicted simultaneously; the open source of the back propagation framework, one only needs to focus on the composition of the model and the design of the loss function.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate and together with the embodiments of the invention and do not constitute a limitation to the invention, and in which:
FIG. 1 is a flow chart of steps of a wind power short-term prediction model based on deep learning;
fig. 2 is a schematic structural diagram of a wind power short-term prediction model based on deep learning.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1-2, the present invention provides a technical solution: a wind power short-term prediction model based on deep learning comprises the following steps:
step one: acquiring wind power generation power and NWP original data;
step two: filling up the data missing value by using the average value;
step three: reducing the data by pearson correlation analysis, and eliminating unnecessary features;
step four: normalizing the data to obtain normalized data;
step five: modeling the multivariate time series data by using a graph neural network, namely constructing an adaptive adjacency matrix to learn hidden space dependency relations among variables;
step six: extracting time features by using a bidirectional gating circulating unit;
step seven: weighting the features extracted in the previous step through an attention mechanism, and outputting through a full connection layer;
step eight: and (5) carrying out inverse normalization on the predicted data to obtain a final result.
In the invention, preferably, in the first step, before the wind power is predicted, the characteristics of the wind energy are known at first so as to reasonably select parameters, the most remarkable characteristic of the wind energy is variability, and the magnitude and the direction of the wind power at each time point and space point are instantaneously changed due to the influence of climate, time and geographic environment factors.
In the present invention, in the second step, it is preferable that the variable (attribute) having no missing value in the data set is referred to as complete data, the variable having a missing value in the data set is referred to as missing data, and the method of filling the data is generally to fill the data by using larger-scale and variable data.
In the present invention, it is preferable that in the second step, attention should be paid to interpretation of sequence points containing low coverage when data is padded using these data.
In the present invention, it is preferable that in the third step, a pearson product moment correlation coefficient (PPMCC/PCCs) is generally used in statistics for measuring a statistical index of the degree of correlation closeness between data.
In the invention, preferably, in the fourth step, when the data is normalized, the data chemotaxis processing mode is adopted, the comprehensive results of different acting forces cannot be correctly reflected by directly adding the data indexes with different properties, and the inverse index data properties are considered to be changed first, so that all the indexes have the same chemotaxis on acting forces of an evaluation scheme, and then the correct results can be obtained by adding the data indexes.
In the invention, preferably, in the fifth step, the established RBF neural network model is composed of three layers, namely an input layer, a hidden layer and an output layer, and is a novel and effective feedforward neural network.
In the present invention, preferably, the input layer of the RBF neural network model is used to transfer the input signal to the hidden layer; the hidden layer is described by a radial basis function, the basis function of the hidden layer mostly adopts a Gaussian basis function, and nonlinear transformation is carried out; the output layer is characterized by a simple linear function, linear transformation is performed, in the RBF neural network model, the connection weight of the input layer to the hidden layer is 1, the connection weight of the hidden layer to the output layer is adjustable, the function of the hidden layer locally responds to the input signal, and when the input signal approaches to the central range of the function, the hidden layer node can generate larger output, so that the radial basis function network has the advantages of excellent local approximation capability and high learning speed.
In the present invention, preferably, in the fifth step, the graph neural network is used, so that the graph neural network model does not need an explicit graph structure when processing data.
In the present invention, preferably, in step seven, the attention mechanism includes a first arithmetic logic unit, an activation function circuit, a scaling circuit, and a second arithmetic logic unit.
Examples
Referring to fig. 1-2, the present invention provides a technical solution: a wind power short-term prediction model based on deep learning comprises the following steps:
step one: acquiring wind power generation power and NWP original data;
step two: filling up the data missing value by using the average value;
step three: reducing the data by pearson correlation analysis, and eliminating unnecessary features;
step four: normalizing the data to obtain normalized data;
step five: modeling the multivariate time series data by using a graph neural network, namely constructing an adaptive adjacency matrix to learn hidden space dependency relations among variables;
step six: extracting time features by using a bidirectional gating circulating unit;
step seven: weighting the features extracted in the previous step through an attention mechanism, and outputting through a full connection layer;
step eight: and (5) carrying out inverse normalization on the predicted data to obtain a final result.
In this embodiment, preferably, in the first step, before the wind power is predicted, the characteristics of the wind energy are first known so as to reasonably select parameters, the most significant characteristic of the wind energy is variability, and due to the influence of weather, time and geographical environmental factors, the magnitude and direction of the wind power at each time point and space point are instantaneously changed, and the graph neural network shows a very high capability in terms of processing relationship dependence. But the graph neural networks require well-defined graph structures for information propagation, which means that they cannot be directly applied to multivariate time sequences where correlations are not known in advance.
In this embodiment, preferably, in the second step, the variables (attributes) in the data set that do not contain the missing values are referred to as complete data, the variables in the data set that contain the missing values are referred to as missing data, and the method of filling the data by filling the variables in the multivariate time sequence with larger-scale and variable data can be regarded as nodes in the graph, and they are connected to each other by hidden dependency relationships. Therefore, modeling the multivariate time series data using the graph neural network can preserve the time trace of the multivariate time series while fully exploiting the correlation between the time series.
In this embodiment, it is preferable that in the second step, attention should be paid to explanation of sequence points containing low coverage when data padding is performed using these data.
In the embodiment, preferably, in the third step, a pearson product moment correlation coefficient (PPMCC/PCCs) is generally used in statistics to measure a statistical index of the degree of closeness of the correlation between the data, and an adaptive graph neural network is constructed, that is, an adaptive adjacency matrix is constructed to learn the hidden spatial dependency between the variables.
In the embodiment, preferably, in the fourth step, when the data is normalized, the data co-chemotaxis processing mode is adopted, and the comprehensive results of different acting forces cannot be correctly reflected by the direct addition of the data indexes with different properties, the inverse index data properties need to be considered and changed, so that all the indexes have the same chemotaxis to acting forces of an evaluation scheme, and then the correct result GCN+BIGRU can be obtained by the addition of the multiple time series prediction: firstly, constructing an adaptive GCN network to extract spatial characteristics among wind farm variables, and then using BIGRU to extract time characteristics and weighting by using an attention mechanism.
In the embodiment, preferably, in the fifth step, the established RBF neural network model is composed of three layers, namely an input layer, a hidden layer and an output layer, and is a novel and effective feedforward neural network.
In this embodiment, preferably, the input layer of the RBF neural network model is used to transfer the input signal to the hidden layer; the hidden layer is described by a radial basis function, the basis function of the hidden layer mostly adopts a Gaussian basis function, and nonlinear transformation is carried out; the output layer is characterized by a simple linear function, linear transformation is performed, in the RBF neural network model, the connection weight of the input layer to the hidden layer is 1, the connection weight of the hidden layer to the output layer is adjustable, the function of the hidden layer locally responds to the input signal, and when the input signal approaches to the central range of the function, the hidden layer node can generate larger output, so that the radial basis function network has the advantages of excellent local approximation capability and high learning speed.
In this embodiment, preferably, in the fifth step, the graph neural network is used, so that the graph neural network model does not need an explicit graph structure when processing data.
In this embodiment, preferably, in step seven, the attention mechanism includes a first arithmetic logic unit, an activation function circuit, a scaling circuit, and a second arithmetic logic unit.
In this embodiment, finally, a linear regression layer is used to synthesize the space-time characteristics and generate a prediction result of the electric power, which specifically includes the following steps:
1. constructing an adaptive adjacency matrix (learning the relationship of edges through model training);
2. inputting data, and performing two graph roll lamination to obtain characteristic data;
graph convolution layer:
(1) Inputting data to perform full connection layer operation;
(2) Multiplying the output result of the full-connection layer with the self-adaptive adjacent matrix, and obtaining an output characteristic result through an activation function;
then, the characteristic data is accessed into BiGRU to extract time sequence characteristics and weighted through an attention mechanism, and the model implementation steps are as follows:
1) Carrying out correlation screening on the historical data;
2) Normalizing the screened data;
3) Excavating the coupling relation among the features by utilizing the GCN depth;
4) Inputting the processed data into BIGUR to extract the dependency relationship between the characteristics and time;
5) The feature vectors processed in the previous step are endowed with different weights through an attention mechanism, and finally are output through a full connection layer;
6) And (5) inversely normalizing the predicted data to obtain a final result.
According to the invention, wind power is subjected to short-term prediction by adopting a wind power short-term prediction model based on deep learning, on one hand, continuity of object development is acknowledged, and past time sequence data is used for statistical analysis to estimate the development trend of objects; on the other hand, the randomness caused by the influence of accidental factors is fully considered, in order to eliminate the influence caused by random fluctuation, historical data is utilized for statistical analysis, the data is properly processed for trend prediction, the problem of insufficient information extraction among data features is solved, and the prediction precision of a model is further improved;
compared with the traditional prediction method, the method has the following advantages: weak dependence or independent characteristic engineering, and directly extracting the relation from the data; the traditional method is basically aimed at a single time sequence, and deep learning can simultaneously predict a plurality of time sequences, and in addition, the correlation between the time sequences can be captured; future time points can be predicted simultaneously; the open source of the back propagation framework, one only needs to focus on the composition of the model and the design of the loss function.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (10)
1. A wind power short-term prediction model based on deep learning is characterized in that: the method comprises the following steps:
step one: acquiring wind power generation power and NWP original data;
step two: filling up the data missing value by using the average value;
step three: reducing the data by pearson correlation analysis, and eliminating unnecessary features;
step four: normalizing the data to obtain normalized data;
step five: modeling the multivariate time series data by using a graph neural network, namely constructing an adaptive adjacency matrix to learn hidden space dependency relations among variables;
step six: extracting time features by using a bidirectional gating circulating unit;
step seven: weighting the features extracted in the previous step through an attention mechanism, and outputting through a full connection layer;
step eight: and (5) carrying out inverse normalization on the predicted data to obtain a final result.
2. The deep learning-based short-term prediction model for wind power according to claim 1, wherein: in the first step, before wind power is predicted, the characteristics of wind energy are known at first so as to reasonably select parameters, the most remarkable characteristic of wind energy is variability, and the magnitude and direction of wind power at each time point and space point are instantaneously changed due to the influence of climate, time and geographic environment factors.
3. The deep learning-based short-term prediction model for wind power according to claim 1, wherein: in the second step, the variable (attribute) without missing value in the data set is called complete data, the variable with missing value in the data set is called missing data, and the method of filling the data is usually performed by using larger-scale and variable data.
4. The deep learning-based short-term prediction model for wind power according to claim 1, wherein: in the second step, attention is paid to interpretation of sequence points containing low coverage when using these data for data padding.
5. The deep learning-based short-term prediction model for wind power according to claim 1, wherein: in the third step, pearson product moment correlation coefficient (PPMCC/PCCs) is generally used in statistics for measuring a statistical index of the degree of correlation between data.
6. The deep learning-based short-term prediction model for wind power according to claim 1, wherein: in the fourth step, when the data is normalized, the data indexes with different properties are directly added together in a data chemotactic processing mode, so that the comprehensive results of different acting forces cannot be correctly reflected, the inverse index data properties are considered to be changed, the acting forces of all indexes on an evaluation scheme are chemotactic, and then the correct results can be obtained after the adding together.
7. The deep learning-based short-term prediction model for wind power according to claim 1, wherein: in the fifth step, the built RBF neural network model consists of three layers, namely an input layer, a hidden layer and an output layer, and is a novel and effective feedforward neural network.
8. The deep learning-based short-term prediction model for wind power of claim 7, wherein: the input layer of the RBF neural network model is used for transmitting an input signal to the hidden layer; the hidden layer is described by a radial basis function, the basis function of the hidden layer mostly adopts a Gaussian basis function, and nonlinear transformation is carried out; the output layer is characterized by a simple linear function, linear transformation is performed, in the RBF neural network model, the connection weight of the input layer to the hidden layer is 1, the connection weight of the hidden layer to the output layer is adjustable, the function of the hidden layer locally responds to the input signal, and when the input signal approaches to the central range of the function, the hidden layer node can generate larger output, so that the radial basis function network has the advantages of excellent local approximation capability and high learning speed.
9. The deep learning-based short-term prediction model for wind power according to claim 1, wherein: in the fifth step, the graph neural network is used, so that the graph neural network model does not need an explicit graph structure when processing data.
10. The deep learning-based short-term prediction model for wind power according to claim 1, wherein: in the seventh step, the attention mechanism includes a first arithmetic logic unit, an activation function circuit, a scaling circuit, and a second arithmetic logic unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310088915.1A CN116050504A (en) | 2023-02-03 | 2023-02-03 | Wind power short-term prediction model based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310088915.1A CN116050504A (en) | 2023-02-03 | 2023-02-03 | Wind power short-term prediction model based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116050504A true CN116050504A (en) | 2023-05-02 |
Family
ID=86119897
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310088915.1A Pending CN116050504A (en) | 2023-02-03 | 2023-02-03 | Wind power short-term prediction model based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116050504A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117411078A (en) * | 2023-09-22 | 2024-01-16 | 华中科技大学 | New energy grid-connected system output prediction method considering privacy protection |
| CN117852729A (en) * | 2024-03-08 | 2024-04-09 | 西安邮电大学 | Weather prediction method and system based on sequential decomposition composition and attention mechanism |
-
2023
- 2023-02-03 CN CN202310088915.1A patent/CN116050504A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117411078A (en) * | 2023-09-22 | 2024-01-16 | 华中科技大学 | New energy grid-connected system output prediction method considering privacy protection |
| CN117852729A (en) * | 2024-03-08 | 2024-04-09 | 西安邮电大学 | Weather prediction method and system based on sequential decomposition composition and attention mechanism |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tian et al. | Short-term wind speed prediction based on improved PSO algorithm optimized EM-ELM | |
| Liang et al. | A novel wind speed prediction strategy based on Bi-LSTM, MOOFADA and transfer learning for centralized control centers | |
| CN116050504A (en) | Wind power short-term prediction model based on deep learning | |
| CN105868853B (en) | Method for predicting short-term wind power combination probability | |
| CN112561058B (en) | Short-term photovoltaic power prediction method based on Stacking-integrated learning | |
| CN113128113B (en) | Lean information building load prediction method based on deep learning and transfer learning | |
| CN113255210B (en) | Method and system for diagnosing federal fault of wind turbine generator | |
| CN110059867B (en) | Wind speed prediction method combining SWLSTM and GPR | |
| CN112653198A (en) | Wind power output scene generation method and system based on prediction box | |
| CN105931134A (en) | Method for evaluating reliability of system with grid-connected photovoltaic power station | |
| Guan | Wave height prediction based on CNN-LSTM | |
| CN114462718A (en) | CNN-GRU wind power prediction method based on time sliding window | |
| CN115115125A (en) | Photovoltaic power interval probability prediction method based on deep learning fusion model | |
| CN116894504A (en) | Wind power cluster power ultra-short-term prediction model establishment method | |
| CN116345555A (en) | CNN-ISCA-LSTM model-based short-term photovoltaic power generation power prediction method | |
| CN113762591B (en) | Short-term electric quantity prediction method and system based on GRU and multi-core SVM countermeasure learning | |
| CN117132132A (en) | Photovoltaic power generation power prediction method based on meteorological data | |
| CN112836876A (en) | Power distribution network line load prediction method based on deep learning | |
| Peng et al. | Hmm-lstm for proactive traffic prediction in 6g wireless networks | |
| CN116418000A (en) | Bayesian state estimation method for unobservable distribution network based on embedded tide neural network | |
| CN113988395A (en) | Wind power ultra-short-term power prediction method based on SSD and dual attention mechanism BiGRU | |
| Shen et al. | An interval analysis scheme based on empirical error and MCMC to quantify uncertainty of wind speed | |
| Zhang et al. | Short-term power load forecasting based on wavelet transform and deep deterministic policy gradient | |
| Zhang et al. | Wind energy forecasting framework based on new perspective | |
| Zhang et al. | A method of load forecasting based on temporal convolutional network |
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 |