CN115034432A - Wind speed prediction method for wind generating set of wind power plant - Google Patents
Wind speed prediction method for wind generating set of wind power plant Download PDFInfo
- Publication number
- CN115034432A CN115034432A CN202210441115.9A CN202210441115A CN115034432A CN 115034432 A CN115034432 A CN 115034432A CN 202210441115 A CN202210441115 A CN 202210441115A CN 115034432 A CN115034432 A CN 115034432A
- Authority
- CN
- China
- Prior art keywords
- wind
- unit
- power plant
- wind speed
- wind power
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000013528 artificial neural network Methods 0.000 claims abstract description 30
- 230000008569 process Effects 0.000 claims abstract description 24
- 125000004122 cyclic group Chemical group 0.000 claims abstract description 16
- 238000007781 pre-processing Methods 0.000 claims abstract description 12
- 239000011159 matrix material Substances 0.000 claims description 44
- 238000004364 calculation method Methods 0.000 claims description 34
- 230000004927 fusion Effects 0.000 claims description 13
- 238000000605 extraction Methods 0.000 claims description 10
- 238000007499 fusion processing Methods 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 5
- 238000010248 power generation Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 9
- 230000006870 function Effects 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
Images
Classifications
-
- 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
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Wind Motors (AREA)
Abstract
The invention discloses a wind speed prediction method for a wind power plant unit. Aiming at the problems that the prediction precision is not high, the method is not suitable for large data processing and the modeling process is complex in the existing wind power plant unit wind speed prediction technology; the invention comprises the following steps: step 1, acquiring information of each unit of a wind power plant, and preprocessing data; step 2, extracting spatial features based on a graph network; step 3, extracting time characteristics based on a neural network of a gated cyclic unit; and 4, fusing space-time characteristics to realize the prediction of the wind speed of each fan in the wind power plant. The wind power plant unit wind speed prediction method has the advantages that the technical scheme is constructed by combining the spatial distribution characteristics of the wind power plant unit and the time characteristics of the wind speed data, the wind speed prediction method of the wind power plant unit is provided, and the effective prediction of the wind speed of the whole wind power plant unit is realized.
Description
Technical Field
The invention relates to the technical field of wind power, in particular to a wind speed prediction method for a wind generating set of a wind power plant.
Background
The wind speed is an important measurement index of wind energy, and the change of the wind speed has important influence on the control of a wind generating set, the generating capacity of a wind power plant and the like. Therefore, the wind speed of the wind power plant unit is predicted, the wind speed prediction precision of the wind power generator unit is improved, and the method has important significance for wind power generator unit control optimization, power grid dispatching and the like.
The existing wind speed prediction methods for wind power plant units can be mainly divided into the following categories: 1. wind speed prediction technologies based on mathematical statistics analysis, such as an autoregressive model, an ARIMA model and the like, mainly realize prediction of wind speed data according to relation analysis of historical time sequence data; 2. wind speed prediction technologies based on machine learning models, such as a support vector machine, a back propagation neural network, an Xgboost and the like, mainly realize prediction of wind speed data through data feature mining; 3. the technology is mainly used for realizing the prediction of the wind speed by modeling wind speed influence factors.
Although the method can realize the prediction of the wind speed of the wind power plant unit, certain limitation exists. For example, wind speed prediction techniques based on mathematical statistical analysis are susceptible to abnormal data in historical data and cannot be applied to big data processing; compared with a mathematical statistics technology, the wind speed prediction technology based on the machine learning model has stronger data feature analysis capability, but most of technologies still only consider the time feature influence and cannot mine deep features of data; the wind speed prediction technology based on the mechanism model needs to collect a large amount of data and has a complex calculation process.
For example, a chinese patent document discloses "a method for predicting wind speed at multiple points in a wind farm based on a convolutional recurrent neural network", which is disclosed in publication No. CN110889535A, published No. 2020-03-17, and includes the following steps: the method comprises the following steps: collecting operation data of a wind power plant, wherein the collected data comprises actually measured wind speeds and actually measured wind directions at the positions of a plurality of wind power units; step two: a convolution module of a multi-point wind speed prediction model in the wind power plant based on a convolution cyclic neural network is established according to the data collected in the step one; step three: establishing an LSTM module of a multi-point wind speed prediction model in the wind power plant based on the convolution cyclic neural network according to the first step; step four: connecting the outputs of the convolution module and the LSTM module, and a fifth step: the neural network model is trained with Mean Absolute Error (MAE) loss function indicators. According to the technical scheme, the LSTM module of the multi-point wind speed prediction model in the wind power plant based on the convolution circulation neural network is adopted, the LSTM module has disadvantages in parallel processing, and the problems of low prediction precision and time-consuming calculation can be caused when the processed data are excessive.
Disclosure of Invention
The invention provides a wind speed prediction method for a wind power plant unit, aiming at solving the problems of low prediction precision, inapplicability to big data processing and complex modeling process in the existing wind speed prediction technology for the wind power plant unit, and combining potential spatial characteristics and wind speed data time characteristics of the wind power plant unit from the data driving angle.
The technical problem of the invention is mainly solved by the following technical scheme:
the invention comprises the following steps:
and 4, fusing space-time characteristics to realize the prediction of the wind speed of each fan in the wind power plant: and performing feature fusion processing on the spatial features of each unit and the wind speed time features of each unit of the wind power plant, and realizing the prediction of the wind speed of each fan of the wind power plant through a multilayer full-connection layer network based on the space-time fusion features.
According to the technical scheme, the technical scheme is constructed by combining the two aspects of wind speed spatial distribution influence and wind speed variation trend of the wind power plant, a wind power plant unit network is constructed by adopting the arrangement position information of the wind power plant units, spatial correlation among the units is extracted, and wind speed spatial distribution characteristics of the wind power plant are obtained; extracting the time characteristics of the wind speed sequence of each unit based on a gated cyclic unit neural network by adopting wind speed information of the units of the wind power plant, and acquiring the wind speed variation trend characteristics of each unit of the wind power plant; and finally, effectively predicting the wind speed of each unit of the wind power plant by combining the spatial characteristics and the time characteristics.
Preferably, in step 1, geographical location information of each unit of the wind farm is acquired, and data is preprocessed: acquiring longitude and latitude of geographical position information of each unit of the wind power plant, and preprocessing the geographical position information;
setting the number of fans of a target wind power plant to be N, and setting the geographical position information set of each unit to be P ═ Lat i ,Lon i And the wind speed information of each unit is set as V ═ V i }. Wherein i is more than or equal to 1 and less than or equal to N, Lat i Indicating latitude, Lon, of the ith unit i Indicating longitude, V, of the ith unit i Representing the ith unit wind speed information set;
based on the setting, distance calculation processing is carried out on the geographical position information of the unit, and a distance set Dis ═ d is constructed ij },1≤i≤N,1≤j≤N,d ij The calculation formula is as follows:
ΔLat′ ij =Lat′ i -Lat′ j
ΔLon′ ij =Lon′ i -Lont′ j
in the formula, temp ij For intermediate calculation of variables, d ij The distance between the wind generating set i and the wind generating set j.
By adopting the scheme, the geographical position information of the unit can be calculated, and the data information is preprocessed, so that the data is more standard, and the calculation is convenient later.
Preferably, in step 1, the wind speed information of each unit of the wind farm is acquired, and the data is preprocessed: acquiring wind speed information of a unit of a wind power plant, and performing standardized processing and serialized processing on the wind speed data of the unit;
the normalized calculation formula is as follows:
in the formula, V i ={v it },v it Is wind speed data v 'at the moment t of the ith unit' it Wind speed data standardized at the moment t of the ith unit are obtained;
the results of the serialization process are as follows:
in the formula, n is the number of the wind speed data of each unit, T is the size of a historical time window, P is the size of a prediction window, X is a unit wind speed historical data set, and S is a unit predicted wind speed data set.
By adopting the scheme, the wind speed information of each unit of the wind power plant can be calculated, and the data information is subjected to standardized processing and serialized processing, so that the data is more standard.
Preferably, the specific content in step 2 is:
step 2-1, constructing a graph network based on geographical position information of the wind power plant unit:
taking a wind turbine generator of a wind power plant as a node V and a straight-line path of the wind turbine generator as an edge E, and constructing a wind turbine generator graph G (V, E, A), wherein A is an adjacent matrix of the graph G;
constructing a wind turbine generator adjacency matrix A according to a wind turbine generator distance calculation result Dis of a wind power plant, wherein the construction process is as follows:
wherein, sigma is the variance of Dis;
2-2, extracting the spatial feature information of the wind power plant unit based on the graph convolution network:
setting the graph convolution network input data as x, the graph convolution network calculation process is as follows:
g θ * G x=g θ (L)x=g θ (U∧U T )x=Ug θ (∧)U T x
L=D-A
in the formula, g θ Is a Laplace function with a parameter theta G For the operation of convolution of the drawing, L is a pullThe Prasiian matrix, A is the adjacency matrix, D is the degree matrix, U is the orthogonal matrix, U T Is the transpose of an orthogonal matrix, and Λ is a diagonal matrix composed of L eigenvalues.
The method and the device for acquiring the wind power plant unit space characteristic information are used for acquiring the wind power plant unit space characteristic information based on the graph network.
Preferably, when the graph network is large and the wind power plant unit data is large, the corresponding laplacian matrix eigenvalue solving time is long, so that the problem is solved by using the Chelbeccov polynomial, and the calculation process is as follows:
in the formula, T k (x)=2xT k-1 (x)-T k-2 (x),T 0 (x)=1,T 1 (x) X, K2 is the Chelby-Skoff polynomial order, θ k Is the coefficient preceding the kth polynomial in the Chelby-Schiff polynomial, L' is the normalized Laplace matrix, λ max Is the maximum eigenvalue of the laplace matrix L, I is the identity matrix.
The technical scheme is that the corresponding solving process is too complex when the graph network is large, namely the data of the wind power plant units are large, so that the solving process is simplified by adopting the Chelbe-Schefft polynomial.
Preferably, the specific content in step 3 is as follows:
3-1, constructing a gated cyclic unit neural network:
suppose gated cyclic unit neural network input is x t Then, the calculation process of the gated cyclic unit neural network is as follows:
z t =σ(W z x t +U z h t-1 )
r t =σ(W r x t +U r h t-1 )
wherein sigma is sigmoid activation function, o is dot product calculation, and z t To update the gate output result, r t To reset the output of the gate, h t For hidden layer output of results, W z 、W r 、W h X in the refresh gate, reset gate, cell unit, respectively t Weight matrix at input, U z 、U r 、U h Respectively a refresh gate, a reset gate, and h in a cell unit t-1 Inputting a weight matrix;
3-2, extracting time characteristics of wind speed data of the wind turbine generator based on a gated circulation unit neural network:
after the wind speed data of the wind turbine generator is serialized, the result X is divided into N groups according to the wind turbine generator, namely X ═ X 1 ,X 2 ,…,X N Respectively collecting X wind speed data sets of each unit in the wind power plant i And (3) performing time characteristic extraction, wherein the extraction process is as follows:
H i =GRU(X i )
H={H i }
in the formula, H i And (3) extracting time characteristics for the ith unit, wherein GRU represents a gated cyclic unit neural network, and H is a set of time characteristics extracted by the wind power plant unit.
The time characteristic extraction method is used for extracting the time characteristics of the wind speed information of the wind power plant units through the gate control circulation unit neural network based on the wind speed information of each wind power plant unit.
Preferably, the specific content in step 4 is:
step 4-1, feature fusion treatment:
carrying out characteristic fusion on the wind speed time characteristic and the spatial characteristic of the wind power plant unit, wherein the fusion process is as follows:
Z=H+g θ * G X
step 4-2, forecasting and outputting the wind speed of each unit of the wind power plant:
based on the fusion characteristics and the output prediction result of the multilayer full-connection layer, the calculation process is as follows:
Y=W y2 (W y1 Z+b y1 )+b y2
in the formula, W y1 、b y1 For the first layer fully connected layer weight and offset, W y2 、b y2 And the weight and the bias of the second layer full-connection layer, and Y is the output of the prediction result.
The method is used for performing characteristic fusion processing on the space characteristics of each unit and the wind speed time characteristics of each unit in the wind power plant, and realizing prediction of the wind speed of each fan in the wind power plant through a multi-layer full-connection layer network based on the space-time fusion characteristics.
The invention has the beneficial effects that: 1. the wind speed prediction method for the wind power plant unit is provided by combining the two aspects of the spatial distribution characteristics of the wind power plant unit and the time characteristics of the wind speed data and considering the construction technical scheme, so that the wind speed of the whole wind power plant unit is effectively predicted; 2. compared with a single method based on time characteristic statistical analysis, the method for mining and fusing the spatial characteristics can further improve the accuracy of wind speed prediction.
Drawings
FIG. 1 is a flow chart of the present invention.
FIG. 2 is a wind farm unit wind speed prediction model frame diagram of the invention.
Detailed Description
The technical scheme of the invention is further specifically described by the following embodiments and the accompanying drawings.
Example (b):
the wind speed prediction method for the wind power plant wind generating set in the embodiment is shown in fig. 1, and comprises the following steps:
step 1-1, acquiring geographical position information of each unit of a wind power plant, and preprocessing data: acquiring longitude and latitude of geographical position information of each unit of the wind power plant, and preprocessing the geographical position information;
setting the number of fans of a target wind power plant to be N, and setting the geographical position information set of each unit to be P ═ Lat i ,Lon i And the wind speed information of each unit is set as V ═ V i }. Wherein i is more than or equal to 1 and less than or equal to N, Lat i Indicating latitude, Lon, of the ith unit i Indicating longitude, V, of the ith unit i Representing the ith unit wind speed information set;
based on the setting, distance calculation processing is carried out on the geographical position information of the unit, and a distance set Dis ═ d is constructed ij },1≤i≤N,1≤j≤N,d ij The calculation formula is as follows:
ΔLat′ ij =Lat′ i -Lat′ j
ΔLon′ ij =Lon′ i -Lont′ j
in the formula, temp ij For intermediate calculation of variables, d ij For wind generating sets i and wind generatorsDistance between motor sets j;
step 1-2, acquiring wind speed information of each unit of the wind power plant, and preprocessing data: acquiring wind speed information of a unit of a wind power plant, and performing standardized processing and serialized processing on the wind speed data of the unit;
the normalized calculation formula is as follows:
in the formula, V i ={v it },v it Is the wind speed data v 'of the ith unit at the moment t' it Wind speed data after the t moment standardization of the ith unit is obtained;
the results of the serialization process are as follows:
in the formula, n is the number of the wind speed data of each unit, T is the size of a historical time window, P is the size of a prediction window, X is a unit wind speed historical data set, and S is a unit predicted wind speed data set.
step 2-1, constructing a graph network based on geographical position information of the wind power plant unit:
taking a wind turbine generator of a wind power plant as a node V and a straight-line path of the wind turbine generator as an edge E, and constructing a wind turbine generator graph G (V, E, A), wherein A is an adjacent matrix of the graph G;
constructing a wind turbine generator adjacency matrix A according to a wind turbine generator distance calculation result Dis of a wind power plant, wherein the construction process is as follows:
in the formula, sigma is the variance of Dis;
2-2, extracting the spatial feature information of the wind power plant unit based on the graph convolution network:
setting the graph convolution network input data as x, the graph convolution network calculation process is as follows:
g θ * G x=g θ (L)x=g θ (U∧U T )x=Ug θ (∧)U T x
L=D-A
in the formula, g θ Laplace function of parameter theta G For the graph convolution operation, L is the Laplace matrix, A is the adjacency matrix, D is the degree matrix, U is the orthogonal matrix, U T Is the transposition of an orthogonal matrix, and the lambada is a diagonal matrix formed by L characteristic values;
considering that the corresponding Laplace matrix eigenvalue solving time is long when the graph network is large and the wind power plant unit data is large, the problem is solved by adopting the Chelbeccov polynomial, and the calculation process is as follows:
in the formula, T k (x)=2xT k-1 (x)-T k-2 (x),T 0 (x)=1,T 1 (x) X, K2 is the Chelby-Skoff polynomial order, θ k Is the coefficient preceding the kth polynomial in the Chelbecco polynomial, L' is the normalized Laplace matrix, λ max Is the maximum eigenvalue of the laplace matrix L, I is the identity matrix.
3-1, constructing a gated cyclic unit neural network:
suppose gated cyclic unit neural network input is x t Then, the calculation process of the gated cyclic unit neural network is as follows:
z t =σ(W z x t +U z h t-1 )
r t =σ(W r x t +U r h t-1 )
wherein sigma is sigmoid activation function, o is dot product calculation, and z t To update the gate output result, r t To reset the output of the gate, h t For hidden layer output of results, W z 、W r 、W h X in the refresh gate, reset gate, cell unit, respectively t Weight matrix at input, Uz, U r 、U h Respectively a refresh gate, a reset gate, and h in a cell unit t-1 Inputting a weight matrix;
3-2, extracting time characteristics of wind speed data of the wind turbine generator based on a gated circulation unit neural network:
serializing wind speed data of wind turbine generatorThe latter result X is divided into N groups according to the unit, i.e. X ═ X 1 ,X 2 ,…,X N Respectively collecting X sets of wind speed data of each unit in the wind power plant i And (3) performing time characteristic extraction, wherein the extraction process is as follows:
H i =GRU(X i )
H={H i }
in the formula, H i And (3) extracting time characteristics for the ith unit, wherein GRU represents a gated cyclic unit neural network, and H is a set of time characteristics extracted by the wind power plant unit.
And 4, fusing space-time characteristics to realize the prediction of the wind speed of each fan of the wind power plant: performing feature fusion processing on the spatial features of each unit and the wind speed time features of each unit of the wind power plant, and realizing prediction of the wind speed of each fan of the wind power plant through a plurality of layers of full-connection layer networks based on the space-time fusion features; as shown in fig. 2, the specific content is:
step 4-1, feature fusion treatment:
carrying out characteristic fusion on the wind speed time characteristic and the spatial characteristic of the wind power plant unit, wherein the fusion process is as follows:
Z=H+g θ * G X
step 4-2, predicting and outputting the wind speed of each unit of the wind power plant:
based on the fusion characteristics and the output prediction result of the multilayer full-connection layer, the calculation process is as follows:
Y=W y2 (W y1 Z+b y1 )+b y2
in the formula, W y1 、b y1 For the first layer fully connected layer weight and offset, W y2 、b y2 And the weight and the bias of the second layer full-connection layer, and Y is the output of the prediction result.
In the embodiment, a wind turbine generator set graph network of a wind power plant is constructed by taking a fan as a point and based on longitude and latitude information of a wind turbine generator set; in the construction process of a wind power generation set diagram network of a wind power plant, an adjacency matrix calculation mode is provided to replace the traditional {0, 1} adjacency matrix representation mode, and the spatial relation of the wind power generation set is better represented; extracting wind speed data characteristics of a wind power plant wind generating set from two aspects of space and time, and extracting space characteristics from a wind power plant wind generating set graph network based on graph convolution operation; the time characteristics and the time-space characteristics are extracted from the wind speed data of the wind power plant wind generating set based on the gated circulation unit neural network, and the prediction precision of the wind speed of the whole wind power plant wind generating set is further improved.
It should be understood that the examples are for illustrative purposes only and are not intended to limit the scope of the present invention. Further, it should be understood that various changes or modifications of the present invention may be made by those skilled in the art after reading the teaching of the present invention, and such equivalents may fall within the scope of the present invention as defined in the appended claims.
Claims (7)
1. A wind speed prediction method for a wind generating set of a wind power plant is characterized by comprising the following steps:
step 1, obtaining information of each unit of a wind power plant, and preprocessing data: collecting geographical position information and wind speed information of each unit of the wind power plant, and performing data preprocessing on the geographical position information and the wind speed information;
step 2, extracting spatial features based on graph networks: constructing a graph network based on geographical position information of each unit of the wind power plant, and extracting spatial characteristic information of the units of the wind power plant through graph convolution operation;
step 3, time characteristic extraction based on the gated loop unit neural network: extracting time characteristics of wind speed information of the wind power plant units through a gate control circulation unit neural network based on the wind speed information of each wind power plant unit;
and 4, fusing space-time characteristics to realize the prediction of the wind speed of each fan in the wind power plant: and performing feature fusion processing on the spatial features of each unit and the wind speed time features of each unit of the wind power plant, and realizing the prediction of the wind speed of each fan of the wind power plant through a multilayer full-connection layer network based on the space-time fusion features.
2. The method for predicting the wind speed of the wind generating set of the wind farm according to claim 1, wherein the geographical position information of each set of the wind farm is obtained in the step 1, and the data is preprocessed: acquiring longitude and latitude of geographical position information of each unit of the wind power plant, and preprocessing the geographical position information;
setting the number of fans of a target wind power plant to be N, and setting the geographical position information set of each unit to be P ═ Lat i ,Lon i The wind speed information of each unit is set as V ═ V i }. Wherein i is more than or equal to 1 and less than or equal to N, Lat i Indicating latitude, Lon, of the ith unit i Indicating longitude, V, of the ith unit i Representing the ith unit wind speed information set;
based on the setting, distance calculation processing is carried out on the geographical position information of the unit, and a distance set Dis ═ d is constructed ij },1≤i≤N,1≤j≤N,d ij The calculation formula is as follows:
ΔLat′ ij =Lat′ i -Lat′ j
ΔLon′ ij =Lon′ i -Lont′ j
in the formula, temp ij For intermediate calculation of variables, d ij The distance between the wind generating set i and the wind generating set j.
3. The wind speed prediction method for the wind generating sets of the wind farm according to claim 2, characterized in that in the step 1, wind speed information of each set of the wind farm is obtained, and data are preprocessed: acquiring wind speed information of a unit of a wind power plant, and performing standardized processing and serialized processing on the wind speed data of the unit;
the normalized calculation formula is as follows:
in the formula, V i ={v it },v it Is the wind speed data v 'of the ith unit at the moment t' it Wind speed data after the t moment standardization of the ith unit is obtained;
the results of the serialization process are as follows:
in the formula, n is the number of wind speed data of each unit, T is the size of a historical time window, P is the size of a prediction window, X is a unit wind speed historical data set, and S is a unit predicted wind speed data set.
4. The wind speed prediction method for wind power generation sets in wind farm according to claim 1, characterized in that the specific content in step 2 is as follows:
step 2-1, constructing a graph network based on geographical position information of the wind power plant unit:
taking a wind turbine generator of a wind power plant as a node V and a linear path of the wind turbine generator as an edge E, and constructing a wind turbine generator graph G (V, E, A), wherein A is an adjacent matrix of the graph G;
constructing a wind turbine generator adjacency matrix A according to a wind turbine generator distance calculation result Dis of a wind power plant, wherein the construction process is as follows:
wherein, sigma is the variance of Dis;
2-2, extracting the spatial feature information of the wind power plant unit based on the graph convolution network:
setting the graph convolution network input data as x, the graph convolution network calculation process is as follows:
L=D-A
in the formula, g θ Is a Laplace function with a parameter theta G For graph convolution operations, L is the Laplace matrix, A is the adjacency matrix, D is the degree matrix, U is the orthogonal matrix, U T Is the transpose of an orthogonal matrix, and Λ is a diagonal matrix composed of L eigenvalues.
5. The wind speed prediction method for the wind turbine generator system of the wind farm according to claim 4, characterized in that when the graph network is large and the wind turbine generator system data is large, the corresponding Laplace matrix eigenvalue solution time is long, so that the problem is solved by using the Chelbeccov polynomial, and the calculation process is as follows:
in the formula, T k (x)=2xT k-1 (x)-T k-2 (x),T 0 (x)=1,T 1 (x) X, K2 is the Chelby-Skoff polynomial order, θ k Is the coefficient preceding the kth polynomial in the Chelby-Schiff polynomial, L' is the normalized Laplace matrix, λ max Is the maximum eigenvalue of the laplace matrix L, I is the identity matrix.
6. The wind speed prediction method for wind generating sets in wind farm according to claim 3, characterized in that the specific content in step 3 is as follows:
step 3-1, constructing a gated cyclic unit neural network:
suppose the gated cyclic unit neural network input is x t Then, the calculation process of the gated cyclic unit neural network is as follows:
z t =σ(W z x t +U z h t-1 )
r t =σ(W r x t +U r h t-1 )
wherein sigma is sigmoid activation function, o is dot product calculation, and z t To update the gate output result, r t To reset the gate output result, h t For the hidden layer to output the result, W z 、W r 、W h X in the refresh gate, reset gate, cell unit, respectively t Weight matrix at input, U z 、U r 、U h Respectively for updating the door and the weightIn the portal, cell unit h t-1 Inputting a weight matrix;
3-2, extracting time characteristics of wind speed data of the wind turbine generator based on a gated circulation unit neural network:
after the wind speed data of the wind turbine generator is serialized, the result X is divided into N groups according to the wind turbine generator, namely X ═ X 1 ,X 2 ,…,X N Respectively collecting X sets of wind speed data of each unit in the wind power plant i And (3) performing time characteristic extraction, wherein the extraction process is as follows:
H i =GRU(X i )
H={H i }
in the formula, H i And (4) extracting time characteristics for the ith unit, wherein GRU represents a gated cyclic unit neural network, and H is a set of time characteristics extracted by the wind power plant unit.
7. The wind speed prediction method for wind power generation sets in wind farm according to claim 1, characterized in that the specific content in step 4 is as follows:
step 4-1, feature fusion treatment:
carrying out characteristic fusion on the wind speed time characteristic and the spatial characteristic of the wind power plant unit, wherein the fusion process is as follows:
Z=H+g θ*G X
step 4-2, predicting and outputting the wind speed of each unit of the wind power plant:
based on the fusion characteristics and the output prediction result of the multilayer full-connection layer, the calculation process is as follows:
Y=W y2 (W y1 Z+b y1 )+b y2
in the formula, W y1 、b y1 For the first layer fully connected layer weight and offset, W y2 、b y2 And the weight and the bias of the second layer full-connection layer, and Y is the output of the prediction result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210441115.9A CN115034432A (en) | 2022-04-25 | 2022-04-25 | Wind speed prediction method for wind generating set of wind power plant |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210441115.9A CN115034432A (en) | 2022-04-25 | 2022-04-25 | Wind speed prediction method for wind generating set of wind power plant |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115034432A true CN115034432A (en) | 2022-09-09 |
Family
ID=83119494
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210441115.9A Pending CN115034432A (en) | 2022-04-25 | 2022-04-25 | Wind speed prediction method for wind generating set of wind power plant |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115034432A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117252448A (en) * | 2023-11-20 | 2023-12-19 | 华东交通大学 | Wind power prediction method based on space-time feature extraction and secondary decomposition polymerization |
-
2022
- 2022-04-25 CN CN202210441115.9A patent/CN115034432A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117252448A (en) * | 2023-11-20 | 2023-12-19 | 华东交通大学 | Wind power prediction method based on space-time feature extraction and secondary decomposition polymerization |
CN117252448B (en) * | 2023-11-20 | 2024-02-20 | 华东交通大学 | Wind power prediction method based on space-time feature extraction and secondary decomposition polymerization |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108448610B (en) | Short-term wind power prediction method based on deep learning | |
CN109242223B (en) | Quantum support vector machine evaluation and prediction method for urban public building fire risk | |
CN111860982A (en) | Wind power plant short-term wind power prediction method based on VMD-FCM-GRU | |
CN108985515B (en) | New energy output prediction method and system based on independent cyclic neural network | |
CN104463381B (en) | Building energy consumption Forecasting Methodology based on KPCA and WLSSVM | |
CN107169598A (en) | A kind of weather prediction method and system based on deep learning | |
CN111292525A (en) | Traffic flow prediction method based on neural network | |
CN111127246A (en) | Intelligent prediction method for transmission line engineering cost | |
CN112966853A (en) | Urban road network short-term traffic flow prediction method based on space-time residual error mixed model | |
CN112464996A (en) | Intelligent power grid intrusion detection method based on LSTM-XGboost | |
CN113420868A (en) | Traveling salesman problem solving method and system based on deep reinforcement learning | |
CN116187835A (en) | Data-driven-based method and system for estimating theoretical line loss interval of transformer area | |
CN114021483A (en) | Ultra-short-term wind power prediction method based on time domain characteristics and XGboost | |
CN116345555A (en) | CNN-ISCA-LSTM model-based short-term photovoltaic power generation power prediction method | |
CN112149896A (en) | Attention mechanism-based mechanical equipment multi-working-condition fault prediction method | |
CN117154690A (en) | Photovoltaic power generation power prediction method and system based on neural network | |
CN115034432A (en) | Wind speed prediction method for wind generating set of wind power plant | |
CN114004152B (en) | Multi-wind-field wind speed space-time prediction method based on graph convolution and recurrent neural network | |
CN112633591B (en) | Space searching method and device based on deep reinforcement learning | |
CN116722545B (en) | Photovoltaic power generation prediction method based on multi-source data and related equipment | |
CN116501444B (en) | Abnormal cloud edge collaborative monitoring and recovering system and method for virtual machine of intelligent network-connected automobile domain controller | |
CN117313795A (en) | Intelligent building energy consumption prediction method based on improved DBO-LSTM | |
CN115761654B (en) | Vehicle re-identification method | |
CN117197632A (en) | Transformer-based electron microscope pollen image target detection method | |
CN116070763A (en) | New wind power plant wind power prediction method and system based on gradient evolution |
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 |