CN115456259A - Wind power plant site selection optimization system and method based on mesoscale data - Google Patents
Wind power plant site selection optimization system and method based on mesoscale data Download PDFInfo
- Publication number
- CN115456259A CN115456259A CN202211036273.2A CN202211036273A CN115456259A CN 115456259 A CN115456259 A CN 115456259A CN 202211036273 A CN202211036273 A CN 202211036273A CN 115456259 A CN115456259 A CN 115456259A
- Authority
- CN
- China
- Prior art keywords
- wind
- vector
- classification
- data
- responsiveness
- 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
- 238000005457 optimization Methods 0.000 title claims abstract description 107
- 238000000034 method Methods 0.000 title claims abstract description 29
- 239000013598 vector Substances 0.000 claims description 224
- 230000004043 responsiveness Effects 0.000 claims description 84
- 238000012546 transfer Methods 0.000 claims description 59
- 239000011159 matrix material Substances 0.000 claims description 48
- 238000010248 power generation Methods 0.000 claims description 40
- 238000013527 convolutional neural network Methods 0.000 claims description 35
- 238000010586 diagram Methods 0.000 claims description 19
- 238000000605 extraction Methods 0.000 claims description 15
- 238000011156 evaluation Methods 0.000 claims description 14
- 230000007704 transition Effects 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 10
- 230000004913 activation Effects 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 4
- 238000011176 pooling Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 description 15
- 230000008859 change Effects 0.000 description 11
- 238000013528 artificial neural network Methods 0.000 description 6
- 238000013135 deep learning Methods 0.000 description 6
- 239000000284 extract Substances 0.000 description 6
- 230000008901 benefit Effects 0.000 description 5
- 230000004044 response Effects 0.000 description 4
- 241000668842 Lepidosaphes gloverii Species 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000003058 natural language processing Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000005215 recombination Methods 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- 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)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Mathematical Physics (AREA)
- General Business, Economics & Management (AREA)
- Educational Administration (AREA)
- Pure & Applied Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Computational Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Public Health (AREA)
- Computing Systems (AREA)
- Primary Health Care (AREA)
- Algebra (AREA)
- Water Supply & Treatment (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Wind Motors (AREA)
Abstract
The application relates to the field of intelligent site selection of wind power plants, and particularly discloses a wind power plant site selection optimization system and method based on mesoscale data.
Description
Technical Field
The invention relates to the field of intelligent site selection of wind power plants, in particular to a wind power plant site selection optimization system and method based on mesoscale data.
Background
At present, the sea area for an offshore wind power plant is limited, meanwhile, turbulence is low, wake flow recovery is slow, and optimal arrangement of wind turbines is one of effective means and key technologies for increasing benefits. The existing arrangement method is difficult to fully utilize a planning area of an offshore wind farm, and meanwhile, the problems that the offshore wind farm is developed in a centralized manner, the wind farms are often influenced mutually and the like are solved.
Therefore, an optimized wind farm site selection optimization scheme is expected.
At present, deep learning and neural networks have been widely applied in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks also exhibit a level close to or even exceeding that of humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.
In recent years, deep learning and development of a neural network provide solutions and schemes for site selection optimization of a wind power plant.
Disclosure of Invention
The present application is proposed to solve the above-mentioned technical problems. The embodiment of the application provides a wind power plant site selection optimization system based on mesoscale data and a method thereof, wherein the power generation performance characteristics of each offshore wind power plant are represented by the output electric power characteristics of each offshore wind power plant under a preset wind condition, then the site selection of the offshore wind power plant is carried out by combining the dynamic change characteristic information of wind data, furthermore, the transfer matrix among the offshore wind power plants is used for representing the implicit difference characteristics among the power generation performance of each offshore wind power plant, and then the high-dimensional implicit correlation characteristic extraction is carried out on the difference by using a convolutional neural network model, so that the optimization result of the site selection of the wind power plants is ensured.
According to one aspect of the application, a wind farm siting optimization system based on mesoscale data is provided, comprising:
the wind data acquisition module is used for acquiring wind data of a plurality of preset time points, wherein the wind data comprises wind speed and wind direction, and the wind direction is represented by cosine values or sine values of angles;
the wind data time sequence coding module is used for arranging the wind data of the plurality of preset time points into an input vector according to the time dimension and then obtaining a wind driving characteristic vector through a time sequence coder comprising a one-dimensional convolution layer;
the output power data acquisition module is used for acquiring output electric power values of a plurality of offshore wind farms at the plurality of preset time points;
the power data time sequence coding module is used for respectively arranging the output electric power values of each offshore wind farm at the plurality of preset time points into power input vectors and then obtaining a plurality of output power characteristic vectors by the time sequence coder containing the one-dimensional convolution layer;
the power generation performance evaluation module is used for respectively calculating the responsiveness estimation of each output power eigenvector in the plurality of output power eigenvectors relative to the wind-driven eigenvector to obtain a plurality of responsiveness matrixes;
the wind power performance difference evaluation module is used for calculating a transfer matrix between every two responsiveness matrixes in the responsiveness matrixes to obtain a plurality of transfer matrixes;
the difference correlation characteristic extraction module is used for arranging the plurality of transfer matrixes into a three-dimensional input tensor and obtaining a classification characteristic diagram through a convolutional neural network model serving as a characteristic extractor; and
and the site selection optimization result generation module is used for enabling the classification characteristic graph to pass through a classifier to obtain a classification result, and the classification result is used for indicating whether site selection schemes of the offshore wind farms need to be optimized or not.
In the wind farm site selection optimization system based on the mesoscale data, the wind data time sequence coding module comprises: the input vector construction unit is used for arranging the wind data of the plurality of preset time points into input vectors according to a time dimension; a full-concatenation coding unit, configured to perform full-concatenation coding on the input vector using a full-concatenation layer of the time sequence encoder according to the following formula to extract a high-dimensional implicit feature of a feature value at each position in the input vector, where the formula is:
wherein X is the input vector, Y is the output vector, W is the weight matrix, B is the offset vector,
represents a matrix multiplication; a one-dimensional convolution coding unit, configured to perform one-dimensional convolution coding on the input vector by using a one-dimensional convolution layer of the time-series encoder according to the following formula to extract a high-dimensional implicit correlation feature between feature values of each position in the input vector, where the formula is:
wherein, a is the width of the convolution kernel in the X direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, w is the size of the convolution kernel, and X represents the input vector.
In the wind farm site selection optimization system based on the mesoscale data, the power generation performance evaluation module is further configured to: respectively calculating the responsiveness estimation of each output power characteristic vector in the plurality of output power characteristic vectors relative to the wind-driven characteristic vector by the following formula to obtain a plurality of responsiveness matrixes;
wherein the formula is:
V i ＝M i *V c
wherein V i Representing each of said plurality of output power eigenvectors, V c Representing said wind-driven feature vector, M i Representing the plurality of responsiveness matrices.
In the wind farm site selection optimization system based on the mesoscale data, the wind power performance difference evaluation module is further configured to: calculating a transition matrix between every two responsiveness matrixes in the plurality of responsiveness matrixes to obtain a plurality of transition matrixes;
wherein the formula is:
M 1 ＝M k *M 2
wherein M is 1 And M 2 Respectively representing every two responsiveness matrices, M, of the plurality of responsiveness matrices k Representing the plurality of transition matrices.
In the wind farm site selection optimization system based on the mesoscale data, the difference correlation feature extraction module is further configured to: each layer of the convolutional neural network model as the feature extractor is respectively carried out in the forward transmission of the layer: performing convolution processing on input data to obtain a convolution characteristic diagram; performing mean pooling on the convolution feature map to obtain a pooled feature map; and performing nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the convolutional neural network model as the feature extractor is the classification feature map, and the input of the first layer of the convolutional neural network model as the feature extractor is the three-dimensional input tensor.
In the wind farm site selection optimization system based on the mesoscale data, the site selection optimization result generation module includes: the dimension reduction unit is used for calculating the global mean value of each feature matrix of the classification feature map along the channel dimension to obtain a reference feature vector; a classification feature vector generation unit configured to calculate, as a classification feature vector, a channel-recursive squeeze-excitation optimization vector of the reference feature vector, the channel-recursive squeeze-excitation optimization vector of the reference feature vector being related to a mean and a variance of the reference feature vector; and the classification unit is used for inputting the classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
In the wind farm site selection optimization system based on the mesoscale data, the classification feature vector generation unit is further configured to: calculating a recursive squeeze-excitation optimization vector of the channel of the reference feature vector as the classification feature vector according to the following formula;
wherein the formula is:
wherein v is i And v i ' represents the eigenvalue of the ith position of the reference eigenvector and the pass-recursive squeeze-excitation optimization vector, respectively, and μ and σ are the mean and variance of the eigenvalue set of the reference eigenvector, respectively, exp (·) represents the exponential operation of the eigenvalues, and the exponential operation raised by the eigenvalues represents the natural exponential function value raised by the eigenvalues.
In the wind farm site selection optimization system based on the mesoscale data, the classification unit is further configured to: processing the classification feature vector using the classifier to obtain the classification result with the following formula: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) I X }, wherein W 1 To W n As a weight matrix, B 1 To B n Is an offset vector, X isThe classification feature vector.
According to another aspect of the application, a wind farm site selection optimization method based on mesoscale data comprises the following steps:
acquiring wind data of a plurality of preset time points, wherein the wind data comprises wind speed and wind direction, and the wind direction is represented by cosine values or sine values of angles;
arranging the wind data of the plurality of preset time points into an input vector according to a time dimension, and then obtaining a wind driving characteristic vector through a time sequence encoder comprising a one-dimensional convolution layer;
acquiring output electric power values of a plurality of offshore wind farms at the plurality of preset time points;
arranging the output electric power values of each offshore wind farm at the plurality of preset time points into a power input vector respectively, and then obtaining a plurality of output power characteristic vectors by the time sequence encoder containing the one-dimensional convolution layer;
respectively calculating the responsiveness estimation of each output power characteristic vector in the plurality of output power characteristic vectors relative to the wind-driven characteristic vector to obtain a plurality of responsiveness matrixes;
calculating a transfer matrix between every two of the plurality of responsiveness matrixes to obtain a plurality of transfer matrixes;
arranging the plurality of transfer matrixes into a three-dimensional input tensor to obtain a classification characteristic diagram through a convolutional neural network model serving as a characteristic extractor; and
and passing the classification characteristic diagram through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the site selection schemes of the offshore wind farms need to be optimized.
In the method for optimizing the site selection of the wind farm based on the mesoscale data, after the wind data of the plurality of preset time points are arranged as input vectors according to the time dimension, the wind data are processed by a time sequence encoder comprising a one-dimensional convolutional layer to obtain a wind-driven characteristic vector, and the method comprises the following steps: arranging the wind data of the plurality of preset time points into an input vector according to a time dimension; using the full connection layer of the time sequence encoder to the input direction according to the following formulaPerforming full-concatenation coding to extract high-dimensional implicit features of feature values of each position in the input vector, wherein the formula is as follows:
wherein X is the input vector, Y is the output vector, W is the weight matrix, B is the offset vector,
represents a matrix multiplication; performing one-dimensional convolutional coding on the input vector by using a one-dimensional convolutional layer of the time sequence encoder according to the following formula to extract high-dimensional implicit correlation characteristics among characteristic values of all positions in the input vector, wherein the formula is as follows:
wherein, a is the width of the convolution kernel in the X direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, w is the size of the convolution kernel, and X represents the input vector.
In the above wind farm site selection optimization method based on mesoscale data, respectively calculating the responsiveness estimates of each output power feature vector in the plurality of output power feature vectors with respect to the wind-driven feature vector to obtain a plurality of responsiveness matrices, includes: respectively calculating the responsiveness estimation of each output power eigenvector in the plurality of output power eigenvectors relative to the wind-driven eigenvector by the following formula to obtain a plurality of responsiveness matrixes;
wherein the formula is:
V i ＝M i *V c
wherein V i Representing each of said plurality of output power eigenvectors, V c Representing said wind-driven feature vector, M i Representing the plurality of responsiveness matrices.
In the wind farm site selection optimization method based on the mesoscale data, calculating a transfer matrix between every two responsiveness matrixes in the plurality of responsiveness matrixes to obtain a plurality of transfer matrixes, including: calculating a transition matrix between every two responsiveness matrixes in the plurality of responsiveness matrixes to obtain a plurality of transition matrixes;
wherein the formula is:
M 1 ＝M k *M 2
wherein M is 1 And M 2 Respectively representing every two responsiveness matrices, M, of the plurality of responsiveness matrices k Representing the plurality of transition matrices.
In the wind farm site selection optimization method based on the mesoscale data, the step of arranging the plurality of transfer matrices into a three-dimensional input tensor to obtain a classification feature map through a convolutional neural network model serving as a feature extractor includes: each layer of the convolutional neural network model as the feature extractor is respectively carried out in the forward transmission of the layer: carrying out convolution processing on input data to obtain a convolution characteristic diagram; performing mean pooling on the convolution feature map to obtain a pooled feature map; and performing nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the convolutional neural network model as the feature extractor is the classification feature map, and the input of the first layer of the convolutional neural network model as the feature extractor is the three-dimensional input tensor.
In the wind farm site selection optimization method based on the mesoscale data, the classification feature map is passed through a classifier to obtain a classification result, and the classification result is used for indicating whether site selection schemes of the plurality of offshore wind farms need to be optimized or not, and the method comprises the following steps: calculating a global mean value of each feature matrix of the classification feature map along the channel dimension to obtain a reference feature vector; calculating a channel-recursive squeeze-excitation optimization vector of the reference feature vector as a classification feature vector, the channel-recursive squeeze-excitation optimization vector of the reference feature vector being related to a mean and a variance of the reference feature vector; and inputting the classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
In the wind farm site selection optimization method based on the mesoscale data, calculating a channel recursive squeeze-excitation optimization vector of the reference feature vector as a classification feature vector, and including: calculating a recursive squeeze-excitation optimization vector of the channel of the reference feature vector as the classification feature vector according to the following formula;
wherein the formula is:
wherein v is i And v i ' represents the eigenvalue of the ith position of the reference eigenvector and the channel-recursive press-excitation optimization vector, respectively, and μ and σ are the mean and variance of the eigenvalue set of the reference eigenvector, respectively, exp (-) represents the exponential operation of the eigenvalue, and the exponential operation raised to the eigenvalue represents the natural exponential function value raised to the eigenvalue.
In the wind farm site selection optimization method based on the mesoscale data, inputting the classification feature vector into a Softmax classification function of the classifier to obtain the classification result, including: processing the classification feature vector using the classifier to obtain the classification result with the following formula: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) I X }, wherein W 1 To W n As a weight matrix, B 1 To B n X is the classified feature vector.
Compared with the prior art, the wind farm site selection optimization system and method based on the mesoscale data have the advantages that the power generation performance characteristics of each offshore wind farm are represented by the output electric power characteristics of each offshore wind farm under the preset wind condition, then site selection of the offshore wind farm is carried out by combining with dynamic change characteristic information of wind data, furthermore, transfer matrixes among the offshore wind farms are used for representing implicit difference characteristics among the power generation performance of each offshore wind farm, and then a convolutional neural network model is used for carrying out high-dimensional implicit associated characteristic extraction on the differences, so that the optimization result of site selection of the wind farms is guaranteed.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally represent like parts or steps.
FIG. 1 is an application scenario diagram of a wind farm site selection optimization system based on mesoscale data according to an embodiment of the application.
FIG. 2 is a block diagram of a wind farm site selection optimization system based on mesoscale data according to an embodiment of the application.
FIG. 3 is a block diagram of a site selection optimization result generation module in the wind farm site selection optimization system based on mesoscale data according to the embodiment of the present application.
FIG. 4 is a flowchart of a wind farm site selection optimization method based on mesoscale data according to an embodiment of the application.
FIG. 5 is a schematic architecture diagram of a wind farm site selection optimization method based on mesoscale data according to an embodiment of the present application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only a few embodiments of the present application, and not all embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein.
Overview of a scene
As described above, at present, the offshore wind farm has a limited sea area, low turbulence and slow wake flow recovery, and the optimal arrangement of wind turbines is one of effective means and key technologies for increasing the yield. The existing arrangement method is difficult to fully utilize a planning area of an offshore wind farm, and meanwhile, the problems that the offshore wind farm is developed in a centralized manner, the wind farms are often influenced mutually and the like are solved.
Therefore, an optimized wind farm site selection optimization scheme is expected.
At present, deep learning and neural networks have been widely applied in the fields of computer vision, natural language processing, text signal processing, and the like. In addition, deep learning and neural networks also exhibit a level close to or even exceeding that of humans in the fields of image classification, object detection, semantic segmentation, text translation, and the like.
In recent years, deep learning and development of a neural network provide solutions and schemes for site selection optimization of a wind power plant.
Based on the above, the inventor considers that the site selection optimization of the wind power plant needs to be carried out based on the change characteristics of the environmental wind power and the output power generation performance characteristics of the offshore wind power plant. Therefore, in the technical scheme of the application, the power generation performance characteristics of each offshore wind farm are represented by the output electric power characteristics of each offshore wind farm under the preset wind condition, and then the site selection of the offshore wind farm is carried out by combining the dynamic change characteristic information of wind data. In addition, the inventor also considers that the power generation performance among the offshore wind farms is different, in order to pay attention to the implicit difference characteristic during site selection, the difference between the power generation performance among the offshore wind farms is further represented by a transfer matrix among the offshore wind farms, and a convolution neural network model is used for carrying out high-dimensional implicit correlation characteristic extraction on the difference, so that the optimization result of site selection of the wind farms is ensured.
Specifically, in the technical solution of the present application, first, wind data at a plurality of predetermined time points are collected by a wind speed station, where the wind data includes a wind speed and a wind direction, and the wind direction is expressed by a cosine value or a sine value of an angle. In particular, in the solution of the present application, the short, medium and long scales are assessed based on the number of a plurality of predetermined time points, whereas in the present application the site selection optimization of the offshore wind farm is performed based on medium scale data. It should be understood that, in order to fully utilize the implicit feature information of the dynamic change of the wind data at the plurality of predetermined time points, the wind data at the plurality of predetermined time points are further arranged into an input vector according to the time dimension and then encoded in a time-sequence encoder comprising a one-dimensional convolutional layer to obtain a wind-driven feature vector, considering that the wind data is dynamic in the time dimension. Accordingly, in one specific example, the time-series encoder is composed of fully-connected layers and one-dimensional convolutional layers which are alternately arranged, and extracts the correlation of the wind data in a time-series dimension through one-dimensional convolutional coding and extracts high-dimensional implicit features of the wind data through the fully-connected coding.
Then, the output electric power values of the plurality of offshore wind farms at the plurality of predetermined time points are also obtained through the power detector. It should be understood that, similarly, for the output electric power values of each offshore wind farm at the plurality of predetermined time points, considering that the offshore wind farm also has dynamic implicit association in the time sequence dimension, the output electric power values of each offshore wind farm at the plurality of predetermined time points are respectively arranged into a power input vector and then encoded in the time sequence encoder comprising the one-dimensional convolution layer, so as to extract the implicit association features of the dynamic changes of the output electric power values of the plurality of offshore wind farms at the plurality of predetermined time points, and thus obtain a plurality of output power feature vectors. In this way, the power generation performance characteristics of the respective offshore wind farms can be represented by the output electric power characteristics of the respective offshore wind farms under the predetermined wind conditions.
It should be understood that, since the dynamic variation characteristic of the output electric power of the offshore wind farm can be regarded as a responsiveness characteristic to the dynamic variation characteristic of the wind data in the high-dimensional characteristic space, in the technical solution of the present application, the responsiveness estimates of each output power characteristic vector in the plurality of output power characteristic vectors with respect to the wind-driven characteristic vector are further calculated respectively to obtain a plurality of responsiveness matrices.
In particular, in consideration of the difference between the power generation performances of the offshore wind farms, in order to highlight the difference characteristic, a transfer matrix between every two of the plurality of responsiveness matrices is further calculated to obtain a plurality of transfer matrices.
Further, after the plurality of transfer matrices are arranged into a three-dimensional input tensor so as to integrate the difference of the power generation performance of each offshore wind farm, the high-dimensional implicit association features of the difference between the power generation performance of the plurality of offshore wind farms are extracted by a convolutional neural network model serving as a feature extractor so as to obtain a classification feature map. In this way, the classification feature map can be used for obtaining a classification result for indicating whether the site selection schemes of the offshore wind farms need to be optimized or not through the classifier.
However, when the plurality of transfer matrices are arranged into a three-dimensional input tensor and a classification feature map is obtained through a convolutional neural network model as a feature extractor, since the channel dimension of the classification feature map represents the sample dimension of the plurality of transfer matrices, and the plurality of transfer matrices are represented as a transfer space between responsiveness estimates, there is a high possibility that sample inconsistency exists therebetween, and such sample inconsistency may cause expression inconsistency in the channel dimension of the classification feature map, which affects the classification performance of the classification feature map.
Therefore, in the technical solution of the present application, a global mean of each feature matrix along a channel of the classification feature map is first calculated to obtain a reference feature vector, and then a channel-recursive squeeze-excitation optimization vector of the reference feature vector is calculated, specifically:
wherein v is i And v i ' represents the eigenvalue of the ith position of the reference eigenvector and the channel-recursive squeeze-excitation optimization vector, respectively, and μ and σ are the mean and variance, respectively, of the eigenvalue set of the reference eigenvector, exp (-) represents the exponential operation of the eigenvalueThe exponential operation with the eigenvalue as the power represents a natural exponential function value with the eigenvalue as the power.
Here, the channel recursive squeeze-excitation optimization vector may activate channel recursive along-channel of feature distribution based on statistical characteristics of a feature set along the channel, so as to infer distribution of the features in each channel sampling dimension thereof, and a squeeze-excitation mechanism composed of a ReLU-Sigmoid function is employed to obtain a confidence value of attention enhancement along the channel dimension, so as to enhance expression consistency in the channel dimension, thereby improving classification performance of the classification feature map, and further improving accuracy of classification.
Based on this, the application provides a wind power plant site selection optimizing system based on mesoscale data, which includes: the wind data acquisition module is used for acquiring wind data of a plurality of preset time points, wherein the wind data comprises wind speed and wind direction, and the wind direction is expressed by cosine values or sine values of angles; the wind data time sequence coding module is used for arranging the wind data of the plurality of preset time points into an input vector according to the time dimension and then obtaining a wind driving characteristic vector through a time sequence coder comprising a one-dimensional convolution layer; the output power data acquisition module is used for acquiring output electric power values of a plurality of offshore wind farms at the plurality of preset time points; the power data time sequence coding module is used for respectively arranging the output electric power values of each offshore wind farm at the plurality of preset time points into power input vectors and then obtaining a plurality of output power characteristic vectors by the time sequence coder containing the one-dimensional convolution layer; the power generation performance evaluation module is used for respectively calculating the responsiveness estimation of each output power characteristic vector in the plurality of output power characteristic vectors relative to the wind-driven characteristic vector to obtain a plurality of responsiveness matrixes; the wind power performance difference evaluation module is used for calculating a transfer matrix between every two responsiveness matrixes in the responsiveness matrixes to obtain a plurality of transfer matrixes; the difference correlation characteristic extraction module is used for arranging the plurality of transfer matrixes into a three-dimensional input tensor and obtaining a classification characteristic diagram through a convolutional neural network model serving as a characteristic extractor; and the site selection optimization result generation module is used for enabling the classification characteristic graph to pass through a classifier to obtain a classification result, and the classification result is used for indicating whether site selection schemes of the offshore wind farms need to be optimized or not.
FIG. 1 illustrates an application scenario of a wind farm siting optimization system based on mesoscale data according to an embodiment of the present application. As shown in fig. 1, in this application scenario, first, wind data including wind speed and wind direction at a plurality of predetermined time points are collected by a wind speed station (e.g., W as illustrated in fig. 1), and output electric power values of a plurality of offshore wind farms (e.g., F as illustrated in fig. 1) at the plurality of predetermined time points are acquired by a power detector (e.g., P as illustrated in fig. 1). The obtained wind data and output electrical power values for the plurality of predetermined points in time are then input to a server (e.g. a cloud server S as illustrated in fig. 1) deployed with a wind farm siting optimization algorithm based on mesoscale data, wherein the server is capable of processing the wind data and output electrical power values for the plurality of predetermined points in time with a wind farm siting optimization algorithm based on mesoscale data to generate a classification result indicating whether or not the siting schemes of the plurality of offshore wind farms require optimization.
Having described the general principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.
Exemplary System
FIG. 2 illustrates a block diagram of a wind farm siting optimization system based on mesoscale data according to an embodiment of the application. As shown in fig. 2, a wind farm site selection optimization system 200 based on mesoscale data according to an embodiment of the present application includes: a wind data acquisition module 210, configured to acquire wind data at a plurality of predetermined time points, where the wind data includes a wind speed and a wind direction, and the wind direction is represented by a cosine value or a sine value of an angle; the wind data time sequence coding module 220 is configured to arrange the wind data at the plurality of predetermined time points into an input vector according to a time dimension, and then obtain a wind-driven feature vector through a time sequence coder including a one-dimensional convolutional layer; an output power data acquisition module 230, configured to acquire output electric power values of a plurality of offshore wind farms at the plurality of predetermined time points; a power data time sequence coding module 240, configured to arrange the output electric power values of each offshore wind farm at the multiple predetermined time points into power input vectors respectively, and then obtain multiple output power feature vectors by using the time sequence encoder including the one-dimensional convolution layer; a power generation performance evaluation module 250, configured to calculate responsiveness estimates of each output power eigenvector in the plurality of output power eigenvectors with respect to the wind-driven eigenvector to obtain a plurality of responsiveness matrices; the wind power performance difference evaluation module 260 is configured to calculate a transfer matrix between every two of the plurality of responsiveness matrices to obtain a plurality of transfer matrices; a difference correlation feature extraction module 270, configured to arrange the plurality of transfer matrices into a three-dimensional input tensor, and obtain a classification feature map through a convolutional neural network model serving as a feature extractor; and an optimization result generation module 280 for site selection, configured to pass the classification feature map through a classifier to obtain a classification result, where the classification result is used to indicate whether the site selection schemes of the offshore wind farms need to be optimized.
Specifically, in this embodiment of the application, the wind data acquisition module 210 and the wind data time sequence encoding module 220 are configured to acquire wind data at a plurality of predetermined time points, where the wind data includes a wind speed and a wind direction, the wind direction is represented by a cosine value or a sine value of an angle, and the wind data at the plurality of predetermined time points are arranged as an input vector according to a time dimension and then pass through a time sequence encoder including a one-dimensional convolution layer to obtain a wind-driven feature vector. As previously mentioned, it should be appreciated that site selection optimization in view of wind farms needs to be performed based on varying characteristics of the ambient wind and output power generation performance characteristics of offshore wind farms. Therefore, in the technical scheme of the application, the power generation performance characteristics of each offshore wind farm are represented by the output electric power characteristics of each offshore wind farm under the preset wind condition, and then the site selection of the offshore wind farm is carried out by combining the dynamic change characteristic information of wind data. In addition, the difference of the power generation performance among the offshore wind farms is also considered, in order to pay attention to the implicit difference characteristic during site selection, the difference of the power generation performance among the offshore wind farms is further represented by a transfer matrix among the offshore wind farms, and a convolutional neural network model is used for carrying out high-dimensional implicit associated characteristic extraction on the difference, so that the optimization result of site selection of the wind farms is ensured.
That is, specifically, in the technical solution of the present application, first, wind data at a plurality of predetermined time points is collected by a wind speed station, the wind data includes a wind speed and a wind direction, and the wind direction is expressed by a cosine value or a sine value of an angle. In particular, in the solution of the present application, the short, medium and long scales are evaluated based on the number of a plurality of predetermined time points, whereas in the present application the site selection optimization of the offshore wind farm is performed based on the medium scale data. Then, it should be understood that, in order to fully utilize the hidden feature information of the dynamic change of the wind data at the plurality of predetermined time points, the wind data at the plurality of predetermined time points are further arranged into an input vector according to the time dimension and then encoded in a time sequence encoder comprising a one-dimensional convolution layer to obtain a wind-driven feature vector, considering that the wind data is dynamic in the time dimension. Accordingly, in one specific example, the time-series encoder is composed of fully-connected layers and one-dimensional convolutional layers which are alternately arranged, and extracts the correlation of the wind data in a time-series dimension through one-dimensional convolutional coding and extracts high-dimensional implicit features of the wind data through the fully-connected coding.
More specifically, in this embodiment of the present application, the wind data time-series encoding module includes: the input vector construction unit is used for arranging the wind data of the plurality of preset time points into input vectors according to a time dimension; a full-concatenation coding unit, configured to perform full-concatenation coding on the input vector using a full-concatenation layer of the time sequence encoder according to the following formula to extract a high-dimensional implicit feature of a feature value at each position in the input vector, where the formula is:
wherein X is the input vector, Y is the output vector, W is the weight matrix, B is the offset vector,
represents a matrix multiplication; a one-dimensional convolution coding unit, configured to perform one-dimensional convolution coding on the input vector by using a one-dimensional convolution layer of the time-series encoder according to the following formula to extract a high-dimensional implicit correlation feature between feature values of each position in the input vector, where the formula is:
wherein, a is the width of the convolution kernel in the X direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, w is the size of the convolution kernel, and X represents the input vector.
Specifically, in this embodiment of the present application, the output power data collecting module 230 and the power data time sequence encoding module 240 are configured to obtain output electric power values of a plurality of offshore wind farms at the plurality of predetermined time points, and arrange the output electric power values of each offshore wind farm at the plurality of predetermined time points into power input vectors respectively, and then obtain a plurality of output power feature vectors by using the time sequence encoder including the one-dimensional convolution layer. That is, in the technical solution of the present application, further, the output electric power values of the plurality of offshore wind farms at the plurality of predetermined time points are obtained through a power detector. It should be understood that, similarly, for the output electric power values of each offshore wind farm at the plurality of predetermined time points, considering that the output electric power values also have dynamic implicit association in a time sequence dimension, after the output electric power values of each offshore wind farm at the plurality of predetermined time points are respectively arranged as a power input vector, the power input vector is encoded by the time sequence encoder comprising the one-dimensional convolutional layer, so as to extract the implicit association features of the dynamic changes of the output electric power values of the plurality of offshore wind farms at the plurality of predetermined time points, thereby obtaining a plurality of output power feature vectors. In this way, the power generation performance characteristics of the respective offshore wind farms can be represented by the output electric power characteristics of the respective offshore wind farms under the predetermined wind conditions.
Specifically, in this embodiment, the power generation performance evaluation module 250 and the wind power performance difference evaluation module 260 are configured to respectively calculate responsiveness estimates of each output power eigenvector of the plurality of output power eigenvectors with respect to the wind-driven eigenvector to obtain a plurality of responsiveness matrices, and calculate a transfer matrix between every two responsiveness matrices of the plurality of responsiveness matrices to obtain a plurality of transfer matrices. It should be understood that, since the dynamic variation characteristic of the output electric power of the offshore wind farm can be regarded as a responsiveness characteristic to the dynamic variation characteristic of the wind data in the high-dimensional characteristic space, in the technical solution of the present application, the responsiveness estimates of each output power characteristic vector in the plurality of output power characteristic vectors with respect to the wind-driven characteristic vector are further calculated respectively to obtain a plurality of responsiveness matrices. In particular, in consideration of the difference between the power generation performances of the offshore wind farms, in order to highlight the difference characteristic, the transfer matrix between every two response matrices in the plurality of response matrices is further calculated to obtain a plurality of transfer matrices.
More specifically, in an embodiment of the present application, the power generation performance evaluation module is further configured to: respectively calculating the responsiveness estimation of each output power characteristic vector in the plurality of output power characteristic vectors relative to the wind-driven characteristic vector by the following formula to obtain a plurality of responsiveness matrixes;
wherein the formula is:
V i ＝M i *V c
wherein V i Representing each of said plurality of output power eigenvectors, V c Representing said wind-driven feature vector, M i Representing the plurality of responsiveness matrices.
More specifically, in this embodiment of the application, the wind power performance difference evaluation module is further configured to: calculating a transition matrix between every two responsiveness matrixes in the plurality of responsiveness matrixes to obtain a plurality of transition matrixes;
wherein the formula is:
M 1 ＝M k *M 2
wherein M is 1 And M 2 Respectively represent every two responsiveness matrices, M, of the plurality of responsiveness matrices k Representing the plurality of transition matrices.
Specifically, in this embodiment of the present application, the difference correlation feature extraction module 270 is configured to arrange the plurality of transfer matrices into a three-dimensional input tensor through a convolutional neural network model serving as a feature extractor to obtain a classification feature map. That is, in the technical solution of the present application, further, after the plurality of transfer matrices are arranged as three-dimensional input tensors to integrate differences in power generation performance of each offshore wind farm, a convolutional neural network model serving as a feature extractor is used to extract high-dimensional implicit correlation features of differences in power generation performance among the offshore wind farms to obtain a classification feature map.
More specifically, in this embodiment of the application, the difference associated feature extraction module is further configured to: each layer of the convolutional neural network model as the feature extractor is respectively carried out in the forward transmission of the layer: performing convolution processing on input data to obtain a convolution characteristic diagram; performing mean pooling on the convolution feature map to obtain a pooled feature map; and performing nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the convolutional neural network model as the feature extractor is the classification feature map, and the input of the first layer of the convolutional neural network model as the feature extractor is the three-dimensional input tensor.
Specifically, in this embodiment of the present application, the site selection optimization result generating module 280 is configured to pass the classification feature map through a classifier to obtain a classification result, where the classification result is used to indicate whether optimization is needed for the site selection schemes of the offshore wind farms. That is, in the technical solution of the present application, the classification feature map is further processed by a classifier to obtain a classification result indicating whether the addressing scheme of the offshore wind farms needs to be optimized. However, it should be understood that when the plurality of transfer matrices are arranged into a three-dimensional input tensor to obtain the classification feature map through a convolutional neural network model as a feature extractor, since the channel dimension of the classification feature map represents the sample dimension of the plurality of transfer matrices, and the plurality of transfer matrices are represented as transfer spaces between responsiveness estimates, there is a high possibility that there is sample inconsistency therebetween, and such sample inconsistency may cause expression inconsistency in the channel dimension of the classification feature map, which affects the classification performance of the classification feature map. Therefore, in the technical solution of the present application, further, a global mean of each feature matrix along a channel of the classification feature map is first calculated to obtain a reference feature vector, and then a channel-recursive press-excitation optimization vector of the reference feature vector is calculated.
It should be understood that, here, the recursive squeeze-excitation optimization vector of the channel may activate the channel-wise recursion of feature distribution based on the statistical characteristics of the feature set along the channel, so as to infer the distribution of the features in each channel sampling dimension thereof, and a squeeze-excitation mechanism composed of a ReLU-Sigmoid function is adopted to obtain a confidence value of attention enhancement along the channel dimension, so as to enhance the expression consistency in the channel dimension, thereby improving the classification performance of the classification feature map, and further improving the classification accuracy.
More specifically, in this embodiment of the present application, the module for generating an address selection optimization result includes: first, a global mean of each feature matrix along a channel dimension of the classification feature map is calculated to obtain a reference feature vector. Then, a channel-recursive squeeze-fire optimization vector of the reference feature vector is calculated as a classification feature vector, the channel-recursive squeeze-fire optimization vector of the reference feature vector being related to the mean and variance of the reference feature vector. Accordingly, in one specific example, a recursive squeeze-excitation optimization vector of the channel of the reference feature vector is calculated as the classification feature vector in the following formula;
wherein the formula is:
wherein v is i And v i ' represents the eigenvalue of the ith position of the reference eigenvector and the channel-recursive press-excitation optimization vector, respectively, and μ and σ are the mean and variance of the eigenvalue set of the reference eigenvector, respectively, exp (-) represents the exponential operation of the eigenvalue, and the exponential operation raised to the eigenvalue represents the natural exponential function value raised to the eigenvalue. And finally, inputting the classification feature vector into a Softmax classification function of the classifier to obtain the classification result. Accordingly, in one specific example, the classification feature vector is processed using the classifier to obtain the classification result with the following formula: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) I X }, wherein W 1 To W n As a weight matrix, B 1 To B n Is a bias vector and X is the classification feature vector.
FIG. 3 illustrates a block diagram of a site selection optimization result generation module in a wind farm site selection optimization system based on mesoscale data according to an embodiment of the present application. As shown in fig. 3, the address selection optimization result generation module 280 includes: a dimension reduction unit 281 configured to calculate a global mean of each feature matrix of the classification feature map along a channel dimension to obtain a reference feature vector; a classification feature vector generation unit 282 configured to calculate, as a classification feature vector, a pass-recursive squeeze-excitation optimization vector of the reference feature vector, the pass-recursive squeeze-excitation optimization vector of the reference feature vector being related to a mean and a variance of the reference feature vector; a classifying unit 283, configured to input the classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
In summary, the wind farm site selection optimization system 200 based on the mesoscale data is illustrated, and the optimization result of the site selection of the wind farm is ensured by representing the power generation performance characteristics of each offshore wind farm by the output electric power characteristics of each offshore wind farm under the preset wind condition, then performing site selection of the offshore wind farm by combining the dynamic change characteristic information of wind data, further representing the implicit difference characteristics between the power generation performances of each offshore wind farm by the transfer matrix between each offshore wind farm, and performing high-dimensional implicit correlation characteristic extraction on the difference by using the convolutional neural network model.
As described above, the wind farm site selection optimization system 200 based on the mesoscale data according to the embodiment of the present application may be implemented in various terminal devices, such as a server of a wind farm site selection optimization algorithm based on the mesoscale data. In one example, the mesoscale data based wind farm site selection optimization system 200 according to embodiments of the present application may be integrated into the terminal device as one software module and/or hardware module. For example, the mesoscale data-based wind farm site selection optimization system 200 may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the wind farm site selection optimization system 200 based on the mesoscale data may also be one of many hardware modules of the terminal device.
Alternatively, in another example, the mesoscale data-based wind farm site selection optimization system 200 and the terminal device may also be separate devices, and the mesoscale data-based wind farm site selection optimization system 200 may be connected to the terminal device through a wired and/or wireless network and transmit the interaction information in an agreed data format.
Exemplary method
FIG. 4 illustrates a flow chart of a method for wind farm site selection optimization based on mesoscale data. As shown in fig. 4, the method for optimizing site selection of a wind farm based on mesoscale data according to the embodiment of the present application includes the steps of: s110, acquiring wind data of a plurality of preset time points, wherein the wind data comprises wind speed and wind direction, and the wind direction is expressed by cosine or sine of an angle; s120, arranging the wind data of the plurality of preset time points into an input vector according to a time dimension, and then obtaining a wind driving characteristic vector through a time sequence encoder comprising a one-dimensional convolutional layer; s130, acquiring output electric power values of a plurality of offshore wind farms at the plurality of preset time points; s140, after the output electric power values of the offshore wind farms at the preset time points are respectively arranged into power input vectors, the time sequence encoder comprising the one-dimensional convolutional layer obtains a plurality of output power characteristic vectors; s150, respectively calculating the responsiveness estimation of each output power eigenvector in the plurality of output power eigenvectors relative to the wind-driven eigenvector to obtain a plurality of responsiveness matrixes; s160, calculating a transfer matrix between every two responsiveness matrixes in the plurality of responsiveness matrixes to obtain a plurality of transfer matrixes; s170, arranging the plurality of transfer matrixes into a three-dimensional input tensor, and obtaining a classification feature map through a convolution neural network model serving as a feature extractor; and S180, enabling the classification characteristic graph to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the site selection schemes of the offshore wind farms need to be optimized or not.
FIG. 5 illustrates an architecture diagram of a wind farm site selection optimization method based on mesoscale data according to an embodiment of the application. As shown in fig. 5, in the network architecture of the wind farm site selection optimization method based on mesoscale data, firstly, arranging the obtained wind data (e.g., P1 as illustrated in fig. 5) of a plurality of predetermined time points into an input vector (e.g., V1 as illustrated in fig. 5) according to a time dimension, and then passing through a time-sequence encoder (e.g., E as illustrated in fig. 5) containing one-dimensional convolutional layers to obtain a wind-driven feature vector (e.g., VF1 as illustrated in fig. 5); then, arranging the obtained output electric power values (for example, P2 as illustrated in fig. 5) of each offshore wind farm at the plurality of predetermined time points respectively as a power input vector (for example, V2 as illustrated in fig. 5) and then obtaining a plurality of output power characteristic vectors (for example, VF2 as illustrated in fig. 5) by the time-series encoder comprising one-dimensional convolutional layers; then, respectively calculating responsiveness estimates of each of the plurality of output power eigenvectors with respect to the wind-driven eigenvector to obtain a plurality of responsiveness matrices (e.g., MF1 as illustrated in fig. 5); then, calculating a transfer matrix between every two of the plurality of responsiveness matrices to obtain a plurality of transfer matrices (e.g., MF2 as illustrated in fig. 5); then, arranging the plurality of transfer matrices as a three-dimensional input tensor (e.g., T as illustrated in fig. 5) through a convolutional neural network model (e.g., CNN as illustrated in fig. 5) as an feature extractor to obtain a classification feature map (e.g., FC as illustrated in fig. 5); and finally, passing the classification feature map through a classifier (e.g., a classifier as illustrated in fig. 5) to obtain a classification result, wherein the classification result is used for indicating whether the site selection schemes of the plurality of offshore wind farms need to be optimized.
More specifically, in steps S110 and S120, wind data of a plurality of predetermined time points is acquired, the wind data includes a wind speed and a wind direction, the wind direction is expressed by a cosine value or a sine value of an angle, and the wind data of the plurality of predetermined time points is arranged as an input vector according to a time dimension and then passes through a time sequence encoder comprising a one-dimensional convolution layer to obtain a wind-driven feature vector. It will be appreciated that site selection optimisation in view of wind farms needs to be performed based on varying characteristics of the ambient wind and output power generation performance characteristics of offshore wind farms. Therefore, in the technical scheme of the application, the output electric power characteristics of each offshore wind farm under the preset wind condition are selected to represent the power generation performance characteristics of each offshore wind farm, and then the dynamic change characteristic information of wind data is combined to select the site of the offshore wind farm. In addition, the difference of the power generation performance among the offshore wind farms is also considered, in order to pay attention to the implicit difference characteristic during site selection, the difference of the power generation performance among the offshore wind farms is further represented by a transfer matrix among the offshore wind farms, and a convolutional neural network model is used for carrying out high-dimensional implicit associated characteristic extraction on the difference, so that the optimization result of site selection of the wind farms is ensured.
That is, specifically, in the technical solution of the present application, first, wind data at a plurality of predetermined time points are collected by a wind speed station, the wind data includes a wind speed and a wind direction, and the wind direction is expressed by a cosine value or a sine value of an angle. In particular, in the solution of the present application, the short, medium and long scales are evaluated based on the number of a plurality of predetermined time points, whereas in the present application the site selection optimization of the offshore wind farm is performed based on the medium scale data. Then, it should be understood that, in order to fully utilize the hidden feature information of the dynamic change of the wind data at the plurality of predetermined time points, the wind data at the plurality of predetermined time points are further arranged into an input vector according to the time dimension and then encoded in a time sequence encoder comprising a one-dimensional convolution layer to obtain a wind-driven feature vector, considering that the wind data is dynamic in the time dimension. Accordingly, in one specific example, the time-series encoder is composed of fully-connected layers and one-dimensional convolutional layers which are alternately arranged, and extracts the correlation of the wind data in a time-series dimension through one-dimensional convolutional coding and extracts high-dimensional implicit features of the wind data through the fully-connected coding.
More specifically, in steps S130 and S140, the output electric power values of a plurality of offshore wind farms at the predetermined time points are obtained, and the output electric power values of each offshore wind farm at the predetermined time points are respectively arranged into a power input vector to obtain a plurality of output power characteristic vectors by the time-series encoder including the one-dimensional convolutional layer. That is, in the technical solution of the present application, further, the output electric power values of the plurality of offshore wind farms at the plurality of predetermined time points are obtained through a power detector. It should be understood that, similarly, for the output electric power values of each offshore wind farm at the plurality of predetermined time points, considering that the offshore wind farm also has dynamic implicit association in the time sequence dimension, the output electric power values of each offshore wind farm at the plurality of predetermined time points are respectively arranged into a power input vector and then encoded in the time sequence encoder comprising the one-dimensional convolution layer, so as to extract the implicit association features of the dynamic changes of the output electric power values of the plurality of offshore wind farms at the plurality of predetermined time points, and thus obtain a plurality of output power feature vectors. In this way, the power generation performance characteristics of the respective offshore wind farms can be represented by the output electric power characteristics of the respective offshore wind farms under the predetermined wind conditions.
More specifically, in step S150 and step S160, the responsiveness estimates of each of the plurality of output power eigenvectors with respect to the wind-driven eigenvector are respectively calculated to obtain a plurality of responsiveness matrices, and the transition matrix between each two of the plurality of responsiveness matrices is calculated to obtain a plurality of transition matrices. It should be understood that, since the dynamic variation characteristic of the output electric power of the offshore wind farm can be regarded as a responsiveness characteristic to the dynamic variation characteristic of the wind data in the high-dimensional characteristic space, in the technical solution of the present application, the responsiveness estimates of each output power characteristic vector in the plurality of output power characteristic vectors with respect to the wind-driven characteristic vector are further calculated respectively to obtain a plurality of responsiveness matrices. In particular, in consideration of the difference between the power generation performances of the offshore wind farms, in order to highlight the difference characteristic, the transfer matrix between every two response matrices in the plurality of response matrices is further calculated to obtain a plurality of transfer matrices.
More specifically, in step S170, the plurality of transfer matrices are arranged as a three-dimensional input tensor passing through a convolutional neural network model as an feature extractor to obtain a classification feature map. That is, in the technical solution of the present application, further, after the plurality of transfer matrices are arranged as three-dimensional input tensors to integrate differences in power generation performance of each offshore wind farm, a convolutional neural network model serving as a feature extractor is used to extract high-dimensional implicit correlation features of differences in power generation performance among the offshore wind farms to obtain a classification feature map.
More specifically, in step S180, the classification feature map is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the addressing schemes of the offshore wind farms need to be optimized. That is, in the technical solution of the present application, the classification result indicating whether the addressing schemes of the plurality of offshore wind farms need to be optimized can be obtained by further passing the classification feature map through a classifier. However, it should be understood that when the classification feature map is obtained by a convolutional neural network model as a feature extractor by arranging the plurality of transfer matrices into a three-dimensional input tensor, since the channel dimension of the classification feature map represents the sample dimension of the plurality of transfer matrices, and the plurality of transfer matrices are represented by a transfer space between responsivity estimates, there is a high possibility that there is sample inconsistency therebetween, and such sample inconsistency may cause expression inconsistency in the channel dimension of the classification feature map, which affects the classification performance of the classification feature map.
Therefore, in the technical solution of the present application, further, a global mean of each feature matrix along a channel of the classification feature map is first calculated to obtain a reference feature vector, and then a channel-recursive squeeze-excitation optimization vector of the reference feature vector is calculated. It should be understood that, here, the channel-recursive squeeze-excitation optimization vector may activate channel-recursive along-channel recursion of feature distribution based on statistical characteristics of a feature set along a channel, so as to infer distribution of features in each channel sampling dimension thereof, and a squeeze-excitation mechanism composed of a ReLU-Sigmoid function is adopted to obtain an attention-enhanced confidence value along a channel dimension, so as to enhance expression consistency in the channel dimension, thereby improving classification performance of the classification feature map, and further improving classification accuracy.
In summary, the method for optimizing the site selection of the wind farm based on the mesoscale data is illustrated, and the optimization result of the site selection of the wind farm is ensured by representing the power generation performance characteristics of each offshore wind farm by the output electric power characteristics of each offshore wind farm under the predetermined wind condition, then performing the site selection of the offshore wind farm by combining the dynamic change characteristic information of the wind data, further representing the implicit difference characteristics between the power generation performances of each offshore wind farm by the transfer matrix between each offshore wind farm, and performing high-dimensional implicit associated characteristic extraction on the difference by using the convolutional neural network model.
The basic principles of the present application have been described above with reference to specific embodiments, but it should be noted that advantages, effects, etc. mentioned in the present application are only examples and are not limiting, and the advantages, effects, etc. must not be considered to be possessed by various embodiments of the present application. Furthermore, the foregoing disclosure of specific details is for the purpose of illustration and description and is not intended to be limiting, since the foregoing disclosure is not intended to be exhaustive or to limit the disclosure to the precise details disclosed.
The block diagrams of devices, apparatuses, systems referred to in this application are only given as illustrative examples and are not intended to require or imply that the connections, arrangements, configurations, etc. must be made in the manner shown in the block diagrams. These devices, apparatuses, devices, systems may be connected, arranged, configured in any manner, as will be appreciated by one skilled in the art. Words such as "including," "comprising," "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably herein. The words "or" and "as used herein mean, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".
It should also be noted that in the devices, apparatuses, and methods of the present application, the components or steps may be decomposed and/or recombined. These decompositions and/or recombinations are to be considered as equivalents of the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit embodiments of the application to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.
Claims (10)
1. A wind power plant site selection optimization system based on mesoscale data is characterized by comprising the following components:
the wind data acquisition module is used for acquiring wind data of a plurality of preset time points, wherein the wind data comprises wind speed and wind direction, and the wind direction is expressed by cosine values or sine values of angles;
the wind data time sequence coding module is used for arranging the wind data of the plurality of preset time points into an input vector according to the time dimension and then obtaining a wind driving characteristic vector through a time sequence coder comprising a one-dimensional convolution layer;
the output power data acquisition module is used for acquiring output electric power values of a plurality of offshore wind farms at the plurality of preset time points;
the power data time sequence coding module is used for respectively arranging the output electric power values of each offshore wind farm at the plurality of preset time points into power input vectors and then obtaining a plurality of output power characteristic vectors by the time sequence coder containing the one-dimensional convolution layer;
the power generation performance evaluation module is used for respectively calculating the responsiveness estimation of each output power eigenvector in the plurality of output power eigenvectors relative to the wind-driven eigenvector to obtain a plurality of responsiveness matrixes;
the wind power performance difference evaluation module is used for calculating a transfer matrix between every two responsiveness matrixes in the responsiveness matrixes to obtain a plurality of transfer matrixes;
the difference correlation characteristic extraction module is used for arranging the plurality of transfer matrixes into a three-dimensional input tensor and obtaining a classification characteristic diagram through a convolutional neural network model serving as a characteristic extractor; and
and the site selection optimization result generation module is used for enabling the classification characteristic graph to pass through a classifier to obtain a classification result, and the classification result is used for indicating whether site selection schemes of the offshore wind farms need to be optimized or not.
2. A wind farm siting optimization system according to claim 1 based on mesoscale data, characterized in that said wind data time series encoding module comprises:
the input vector construction unit is used for arranging the wind data of the plurality of preset time points into input vectors according to a time dimension;
a full-connection coding unit, configured to perform full-connection coding on the input vector by using a full-connection layer of the time sequence encoder according to the following formula to extract a high-dimensional implicit feature of a feature value at each position in the input vector, where the formula is:
wherein X is the input vector, Y is the output vector, W is the weight matrix, B is the offset vector,
represents a matrix multiplication;
a one-dimensional convolution coding unit, configured to perform one-dimensional convolution coding on the input vector by using a one-dimensional convolution layer of the time sequence encoder according to the following formula to extract a high-dimensional implicit correlation feature between feature values of each position in the input vector, where the formula is:
wherein, a is the width of the convolution kernel in the X direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, w is the size of the convolution kernel, and X represents the input vector.
3. A mesoscale data based wind farm siting optimization system according to claim 2, wherein said power generation performance assessment module is further configured to: respectively calculating the responsiveness estimation of each output power eigenvector in the plurality of output power eigenvectors relative to the wind-driven eigenvector by the following formula to obtain a plurality of responsiveness matrixes;
wherein the formula is:
V i ＝M i *V c
wherein V i Representing each of said plurality of output power eigenvectors, V c Representing said wind-driven feature vector, M i Representing the plurality of responsiveness matrices.
4. The system of claim 3, wherein the wind power performance difference assessment module is further configured to: calculating a transition matrix between every two responsiveness matrixes in the plurality of responsiveness matrixes to obtain a plurality of transition matrixes;
wherein the formula is:
M 1 ＝M k *M 2
wherein M is 1 And M 2 Respectively representing every two responsiveness matrices, M, of the plurality of responsiveness matrices k Representing the plurality of transition matrices.
5. A wind farm siting optimization system according to claim 4 based on mesoscale data, characterized in that said difference correlation feature extraction module is further configured to: each layer of the convolutional neural network model as the feature extractor is respectively carried out in the forward transmission of the layer:
performing convolution processing on input data to obtain a convolution characteristic diagram;
performing mean pooling on the convolution feature map to obtain a pooled feature map; and
performing nonlinear activation on the pooled feature map to obtain an activated feature map;
wherein the output of the last layer of the convolutional neural network model as the feature extractor is the classification feature map, and the input of the first layer of the convolutional neural network model as the feature extractor is the three-dimensional input tensor.
6. A wind farm siting optimization system based on mesoscale data according to claim 5, characterized in that said siting optimization result generation module comprises:
the dimension reduction unit is used for calculating the global mean value of each feature matrix of the classification feature map along the channel dimension to obtain a reference feature vector;
a classification feature vector generation unit configured to calculate, as a classification feature vector, a channel-recursive squeeze-excitation optimization vector of the reference feature vector, the channel-recursive squeeze-excitation optimization vector of the reference feature vector being related to a mean and a variance of the reference feature vector;
and the classification unit is used for inputting the classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
7. A mesoscale data based wind farm siting optimization system according to claim 6, wherein said classification feature vector generation unit is further configured to: calculating a recursive squeeze-excitation optimization vector of the channel of the reference feature vector as the classification feature vector according to the following formula;
wherein the formula is:
wherein v is i And v i ' separately represent theAnd the characteristic value of the ith position of the reference characteristic vector and the channel recursive squeeze-excitation optimization vector, wherein mu and sigma are respectively the mean value and the variance of the characteristic value set of the reference characteristic vector, exp (·) represents the exponential operation of the characteristic value, and the exponential operation taking the characteristic value as power represents the natural exponential function value taking the characteristic value as power.
8. A mesoscale data based wind farm siting optimization system according to claim 7, characterized in that said classification unit is further configured to: processing the classification feature vector using the classifier to obtain the classification result with a formula: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) I X }, wherein W 1 To W n As a weight matrix, B 1 To B n Is a bias vector and X is the classification feature vector.
9. A wind power plant site selection optimization method based on mesoscale data is characterized by comprising the following steps:
acquiring wind data of a plurality of preset time points, wherein the wind data comprises wind speed and wind direction, and the wind direction is expressed by cosine or sine of an angle;
the wind data of the preset time points are arranged into an input vector according to the time dimension and then pass through a time sequence encoder comprising a one-dimensional convolution layer to obtain a wind driving characteristic vector;
acquiring output electric power values of a plurality of offshore wind farms at the plurality of preset time points;
arranging the output electric power values of each offshore wind farm at the plurality of preset time points into a power input vector respectively, and then obtaining a plurality of output power characteristic vectors by the time sequence encoder containing the one-dimensional convolution layer;
respectively calculating the responsiveness estimation of each output power characteristic vector in the plurality of output power characteristic vectors relative to the wind-driven characteristic vector to obtain a plurality of responsiveness matrixes;
calculating a transfer matrix between every two of the plurality of responsiveness matrixes to obtain a plurality of transfer matrixes;
arranging the plurality of transfer matrices into a three-dimensional input tensor, and obtaining a classification characteristic diagram through a convolutional neural network model serving as a characteristic extractor; and
and enabling the classification characteristic graph to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the site selection schemes of the offshore wind farms need to be optimized or not.
10. A wind farm siting optimization method according to claim 9 based on mesoscale data, wherein said separately calculating a responsiveness estimate of each of said plurality of output power eigenvectors with respect to said wind driven eigenvector to obtain a plurality of responsiveness matrices comprises:
respectively calculating the responsiveness estimation of each output power eigenvector in the plurality of output power eigenvectors relative to the wind-driven eigenvector by the following formula to obtain a plurality of responsiveness matrixes;
wherein the formula is:
V i ＝M i *V c
wherein V i Representing each of said plurality of output power eigenvectors, V c Representing said wind-driven feature vector, M i Representing the plurality of responsiveness matrices.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211036273.2A CN115456259A (en) | 2022-08-28 | 2022-08-28 | Wind power plant site selection optimization system and method based on mesoscale data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211036273.2A CN115456259A (en) | 2022-08-28 | 2022-08-28 | Wind power plant site selection optimization system and method based on mesoscale data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115456259A true CN115456259A (en) | 2022-12-09 |
Family
ID=84301015
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211036273.2A Pending CN115456259A (en) | 2022-08-28 | 2022-08-28 | Wind power plant site selection optimization system and method based on mesoscale data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115456259A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117094453A (en) * | 2023-10-20 | 2023-11-21 | 国网安徽省电力有限公司合肥供电公司 | Scheduling optimization system and method for virtual power plant |
-
2022
- 2022-08-28 CN CN202211036273.2A patent/CN115456259A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117094453A (en) * | 2023-10-20 | 2023-11-21 | 国网安徽省电力有限公司合肥供电公司 | Scheduling optimization system and method for virtual power plant |
| CN117094453B (en) * | 2023-10-20 | 2024-01-30 | 国网安徽省电力有限公司合肥供电公司 | Scheduling optimization system and method for virtual power plant |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhao et al. | Analog forecasting with dynamics-adapted kernels | |
| CN115013298B (en) | Real-time performance online monitoring system and monitoring method of sewage pump | |
| CN115099285A (en) | Intelligent detection method and system based on neural network model | |
| CN115834433B (en) | Data processing method and system based on Internet of things technology | |
| CN115577248A (en) | Fault diagnosis system and method for wind generating set | |
| CN115859437A (en) | Jacket underwater stress detection system based on distributed optical fiber sensing system | |
| CN117058622A (en) | Intelligent monitoring system and method for sewage treatment equipment | |
| CN115235612B (en) | Intelligent fault diagnosis system and method for servo motor | |
| CN115147655A (en) | Oil gas gathering and transportation monitoring system and method thereof | |
| CN115324843A (en) | Wind generating set fault diagnosis system and method based on monitoring data | |
| CN115456047A (en) | Offshore wind turbine structure state monitoring system and method based on artificial intelligence | |
| CN115359300A (en) | Intelligent processing and manufacturing system and method for wind driven generator component | |
| CN115357065B (en) | Remote intelligent dehumidification control system and method for offshore wind turbine | |
| CN115600140A (en) | Fan variable pitch system fault identification method and system based on multi-source data fusion | |
| CN115526202A (en) | Offshore wind turbine fault diagnosis system based on data driving and diagnosis method thereof | |
| CN115410069A (en) | Fault detection method and system based on multiple attention mechanism | |
| CN115456259A (en) | Wind power plant site selection optimization system and method based on mesoscale data | |
| CN114503131A (en) | Search device, search method, search program, and learning model search system | |
| CN115219845A (en) | Power grid fault diagnosis system and method | |
| CN115456053A (en) | Wind wheel heat dissipation system and method of offshore wind generating set | |
| CN116292130A (en) | Wind driven generator state monitoring system and method based on cooperative analysis | |
| CN117158923A (en) | Remote home-care monitoring method based on meta universe | |
| CN115788793A (en) | Global monitoring and early warning method and system thereof | |
| CN115564092A (en) | Short-time wind power prediction system and method for wind power plant | |
| CN114793086A (en) | Double-sided double-glass photovoltaic panel adjusting bracket and adjusting 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 |