CN115049169B - Regional power consumption prediction method, system and medium based on combination of frequency domain and spatial domain - Google Patents
Regional power consumption prediction method, system and medium based on combination of frequency domain and spatial domain Download PDFInfo
- Publication number
- CN115049169B CN115049169B CN202210978150.4A CN202210978150A CN115049169B CN 115049169 B CN115049169 B CN 115049169B CN 202210978150 A CN202210978150 A CN 202210978150A CN 115049169 B CN115049169 B CN 115049169B
- Authority
- CN
- China
- Prior art keywords
- frequency domain
- power grid
- power consumption
- layer
- data
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000000605 extraction Methods 0.000 claims abstract description 21
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 20
- 238000013528 artificial neural network Methods 0.000 claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims description 41
- 230000001932 seasonal effect Effects 0.000 claims description 25
- 230000008569 process Effects 0.000 claims description 14
- 230000005611 electricity Effects 0.000 claims description 13
- 238000003860 storage Methods 0.000 claims description 12
- 238000004458 analytical method Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 238000001228 spectrum Methods 0.000 claims description 6
- 102000002274 Matrix Metalloproteinases Human genes 0.000 claims description 4
- 108010000684 Matrix Metalloproteinases Proteins 0.000 claims description 4
- 230000007246 mechanism Effects 0.000 claims description 4
- 241001235534 Graphis <ascomycete fungus> Species 0.000 claims description 3
- 230000003213 activating effect Effects 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 239000000047 product Substances 0.000 claims 2
- 239000013589 supplement Substances 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 11
- 230000006870 function Effects 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 230000009471 action Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 230000007774 longterm Effects 0.000 description 3
- 230000010365 information processing Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- 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
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Development Economics (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Educational Administration (AREA)
- Health & Medical Sciences (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The application relates to a regional power consumption prediction method, a system and a medium based on combination of a frequency domain and a spatial domain, wherein the method comprises the following specific steps: according to the acquired historical data of the power grid, comprehensively analyzing historical power consumption data and position information in the historical data of the power grid, and performing season-trend item decomposition on the historical data of the power grid; performing encoder processing of a transform model aiming at input power grid historical data, wherein the encoder processing comprises frequency domain feature extraction based on Fourier transform and space domain feature extraction based on a graph convolution neural network; and performing decoding layer operation on the processed data by using a decoder of a transform model, and combining frequency domain characteristics, space domain characteristics and multi-head frequency domain sub-attention operation in a long-time sequence to realize accurate power consumption prediction of a power grid area. The method solves the problem that the long-time sequence is difficult to effectively model in the existing prediction method, and meanwhile, the frequency domain and space domain characteristics are extracted, and the characteristics are utilized to predict the power consumption.
Description
Technical Field
The application relates to the field of intelligent prediction of regional power consumption in the field of energy Internet, in particular to a regional power consumption prediction method, a regional power consumption prediction system and a regional power consumption prediction medium based on combination of a frequency domain and a spatial domain.
Background
With the continuous increase of the scale of the power grid, the information processing bottleneck problem of the traditional power grid can be effectively solved by the application of the artificial intelligence technology in the power grid. The power consumption of the user is important reference data of the intelligent power grid, and the power consumption data of the user is analyzed and predicted to further know the rule of the power consumption data of the user, so that planning and scheduling decisions of power grid power can be assisted. The existing power grid user power consumption prediction method mainly uses a model based on an autoregressive or recurrent neural network. The two models have poor processing capability on long-time sequences, and are difficult to acquire long-term dependence information of the sequences, so that the long-term dependence information cannot be effectively predicted. The Transformer model is proposed by Google corporation in 2017, is a sequence information processing model based on an encoder-decoder and a self-attention mechanism, and can effectively acquire long-term dependence characteristics in a sequence so as to improve the sequence prediction effect. The power consumption data of the users in the power grid has the characteristics of long sequence and much data, and the problem of overlarge calculated amount is faced by directly applying the Transformer model.
The power consumption of the user has obvious seasonality and periodicity, and in addition, the power consumption is greatly related to the distribution of the spatial position of the user, and the power consumption has certain frequency domain characteristics as time sequence data. Such features are difficult to obtain by conventional Transformer models.
Disclosure of Invention
The embodiment of the application aims to provide a regional power consumption prediction method, a regional power consumption prediction system and a regional power consumption prediction medium based on combination of a frequency domain and a spatial domain, a fast Fourier transform method and a graph convolution method are fused in a transform, a user power consumption data prediction model based on characteristics of the frequency domain and the spatial domain is constructed, and the feature extraction capability of the transform model is improved.
In order to achieve the above purpose, the present application provides the following technical solutions:
in a first aspect, an embodiment of the present application provides a method for predicting area power consumption based on a combination of a frequency domain and a spatial domain, including the following specific steps:
according to the acquired historical data of the power grid, comprehensively analyzing historical power consumption data and position information in the historical data of the power grid, and performing season-trend item decomposition on the historical data of the power grid;
performing encoder processing of a transform model aiming at input power grid historical data, wherein the encoder processing comprises frequency domain feature extraction based on Fourier transform and space domain feature extraction based on a graph convolution neural network;
and performing decoding layer operation on the processed data by using a decoder of a transform model, and realizing accurate power consumption prediction of the power grid region by combining frequency domain characteristics, space domain characteristics and multi-head frequency domain self-attention operation in a long-time sequence.
The historical power consumption data and the position information in the historical data of the comprehensive analysis power grid are specifically that the historical power consumption data comprises a user power consumption time sequenceThe user position information passes through a directed graphIs shown in whichIs a collection of user locations, in commonA user then hasThe elements are selected from the group consisting of,is the edge of the directed graph,the method is characterized in that an adjacency matrix for storing the distance between users is formed by splicing the time sequence and the position information of each power consumption data of the users into a matrix;
The method for decomposing the historical data of the power grid by the season-trend item specifically comprises the step of carrying out time series on the power consumption of a user by a transform modelPerforming sequence decomposition into seasonal itemsAnd trend itemThe specific calculation method is shown in the following formula:
wherein,in order to average out the pooling operation,operations are complemented by 0 at both ends of the sequence to keep the length of the sequence unchanged after the pooling operation.
The frequency domain feature extraction based on the fourier transform is specifically,
time series of electricity consumption of userIn the step (1), the first step,is the length of the sequence, and is,for the characteristic quantity, willUser electricity consumption time series of dimensionSending the data into a frequency domain characteristic layer for Fourier transform to extract frequency domain characteristics, wherein the specific operation of the frequency domain characteristic layer is to firstly carry out Fourier transform on the dataLinear mapping is performed to map it intoDimensional query matrixIn whichIs a preset number with proper size, and then is pairedIs subjected to fast Fourier transform to obtainDimensional spectrum matrixFor reducing the amount of calculation and preventing overfitting, the method carries out sparse operationRandom sampling is carried out to obtainDimension matrixWhereinIs far less thanSecond using a layer of randomly initialized convolutional neural network andis convolved to obtainMatrix ofFinally to is alignedPerforming zero-filling operation to make its dimension becomeAnd performing inverse fast Fourier transform to obtain an output;
The spatial domain feature extraction based on the graph convolution neural network specifically comprises the following steps of extracting a matrixSending into graph volume network layer, defining in-graphThe formula for the convolution of the above graph is:
whereinAre the parameters of the convolution kernel and,is a matrix of the unit, and is,is shown as a drawingDegree matrix, matrixObtaining the corresponding dimension of the user to be predicted after the graph convolution layer and carrying out zero filling to obtain the output of the spatial domain characteristic layer。
The structure of the encoder in the encoder processing of the transformer model aiming at the input power grid historical data is determined byThe same coding layer is formed, and the coder processes the output of the frequency domain characteristic layerAnd spatial feature layerAdding the data, decomposing the sequence to obtain seasonal terms, sending the seasonal terms into a feedforward neural network, decomposing the sequence again to obtain the output of the coding layer, the input of the first layer in the coder is the original data, the output of the previous layer in each coding layer in the back is the input of the layer, and finally the coder outputs the coding result。
The decoder using the transform model performs a decoding layer operation on the processed data specifically,
first, the season itemAfter the frequency domain characteristic layer and the space domain characteristic layer are input, the output of the characteristic layer is summedAdding and carrying out sequence decomposition to obtain seasonal componentsAnd trend component;
The second step, the seasonal component obtained in the previous stepAnd the encoder outputs the encoding resultPerforming frequency domain self-attention operation;
thirdly, outputting the frequency domain self attention and the seasonal component obtained in the first stepAdding the seasonal components and then performing sequence decomposition to obtain seasonal componentsAnd trend componentWill beAfter passing through a feedforward neural network,,After addition, the output of the decoder is obtained。
The frequency domain self-attention operation process is as follows: first of all, multi is introduced through the linear layerHead mechanism, willQuery matrix mapped to more channels by linear layersIn the same way, willMapping into value matrix V and key matrix through linear layerThen to,V, carrying out fast Fourier transform and random sampling to obtain a sampled frequency spectrum matrixSecond, the matrix is dividedPerforming inner product operation, activating function and then performing AND through softmaxPerforming inner product operation, and finally performing zero filling and fast Fourier inverse transformation on the obtained result to obtain frequency domain self-attention outputThe formula is expressed as follows,
In a second aspect, an embodiment of the present application provides a regional power consumption prediction system based on a combination of a frequency domain and a spatial domain, the system including,
the power grid historical data acquisition and analysis module is used for acquiring power grid historical data and comprehensively analyzing historical power consumption data and position information in the power grid historical data;
the season-trend item decomposition module is used for performing season-trend item decomposition on the historical data of the power grid;
the characteristic extraction module is used for extracting the frequency domain characteristic of the power grid historical data based on Fourier transform and the spatial domain characteristic of the power grid historical data based on a graph convolution neural network;
and the power consumption prediction module of the power grid area is used for combining frequency domain characteristics, airspace characteristics and multi-head frequency domain sub-attention operation in a long-time sequence to realize accurate power consumption prediction of the power grid area.
In a third aspect, an embodiment of the present application provides a computer-readable storage medium, which stores program codes, and when the program codes are executed by a processor, the method for predicting regional power consumption based on combination of a frequency domain and a spatial domain as described above is implemented.
Compared with the prior art, the invention has the beneficial effects that: a fast Fourier transform method and a graph convolution method are fused in a transform, a user power consumption data prediction model based on frequency domain and spatial domain characteristics is constructed, and the characteristic extraction capability of the transform model is improved. Meanwhile, the calculated amount is reduced by combining sparse operation, and finally the power consumption of the power grid user is predicted.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments of the present application will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and that those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.
FIG. 1 is a flowchart of a regional power consumption prediction method based on a combination of a frequency domain and a spatial domain according to an embodiment of the present application;
FIG. 2 is a flow chart of an encoder process of a transform model according to an embodiment of the present application;
FIG. 3 is a flowchart illustrating the operation of a decoding layer according to an embodiment of the present application;
FIG. 4 is a block diagram of a model for predicting power consumption of a user according to an embodiment of the present disclosure;
FIG. 5 is a schematic block diagram of a regional power consumption prediction system based on a combination of a frequency domain and a spatial domain according to an embodiment of the present disclosure;
FIG. 6 is a graph showing the actual amount of electricity used and the predicted amount of electricity used in 100 hours according to the embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application. It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
The terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising a," "8230," "8230," or "comprising" does not exclude the presence of additional like elements in a process, method, article, or apparatus that comprises the element.
The terms "first," "second," and the like, are used solely to distinguish one entity or action from another entity or action without necessarily being construed as indicating or implying any actual such relationship or order between such entities or actions.
Referring to fig. 1, an embodiment of the present application provides a method for predicting area power consumption based on a combination of a frequency domain and a spatial domain, including the following specific steps:
s1, according to the acquired historical data of the power grid, comprehensively analyzing historical power consumption data and position information in the historical data of the power grid, and performing season-trend item decomposition on the historical data of the power grid;
s2, performing encoder processing of a transform model aiming at input power grid historical data, wherein the encoder processing comprises frequency domain feature extraction based on Fourier transform and space domain feature extraction based on a graph convolution neural network;
and S3, decoding layer operation is carried out on the processed data by using a decoder of a transform model, and accurate power consumption prediction of the power grid region is realized by combining frequency domain characteristics, space domain characteristics and multi-head frequency domain sub-attention operation in a long-time sequence.
The historical power consumption data and the position information in the historical data of the comprehensive analysis power grid are specifically that the historical power consumption data comprises a user power consumption time sequenceThe user position information passes through a directed graphIs shown in whichIs a collection of user locations, in commonA user then hasThe elements are selected from the group consisting of,is the edge of the directed graph,the method is characterized in that an adjacency matrix for storing the distance between users is formed, and the time sequence and the position information of each power consumption data of the users are spliced into the matrix;
The method for decomposing the historical data of the power grid by the season-trend item specifically comprises the step of carrying out time series on the power consumption of a user by a transform modelPerforming sequence decomposition into seasonal itemsAnd trend itemThe specific calculation method is shown by the following formula:
wherein,in order to average out the pooling operation,operations are padded with 0's at both ends of the sequence to keep the length of the sequence unchanged after the pooling operation.
As shown in figure 2 of the drawings, in which,
s21, the frequency domain feature extraction based on the Fourier transform specifically comprises the steps of,
time series of electricity consumption of userIn (1),in order to be the length of the sequence,for the characteristic quantity, willUser power consumption time series of dimensionSending the frequency domain characteristic layer to perform Fourier transform to extract frequency domain characteristics, wherein the specific operation of the frequency domain characteristic layer is to firstly carry out Fourier transform on the frequency domain characteristic layerLinear mapping is performed to map it intoDimensional query matrixWhereinIs a preset number with proper size, and then is pairedIs subjected to fast Fourier transform to obtainDimensional spectrum matrixIn order to reduce the amount of calculation and prevent overfitting, sparse operation is carried out, and the specific method is toRandom sampling is carried out to obtainDimension matrixWhereinIs far less thanSecond using a layer of randomly initialized convolutional neural network andis convolved to obtainMatrix arrayFinally to make a pairPerforming zero-filling operation to make its dimension becomeAnd performing inverse fast Fourier transform to obtain an output;
The spatial domain feature extraction based on the graph convolution neural network specifically comprises the following steps of extracting a matrixSending into graph volume network layer, defining in-graphThe above graph convolution formula is:
whereinAre the parameters of the convolution kernel and,is a matrix of the unit, and is,is shown as a drawingDegree matrix, matrixObtaining the corresponding dimension of the user to be predicted after the graph convolution layer and carrying out zero filling to obtain the output of the spatial domain characteristic layer。
S23, the structure of an encoder in the encoder processing of the transformer model aiming at the input power grid historical data is determined byThe same coding layer is formed, and the coder processes the output of the frequency domain characteristic layerAnd spatial feature layerAdding the data, decomposing the sequence to obtain seasonal terms, sending the seasonal terms into a feedforward neural network, decomposing the sequence again to obtain the output of the coding layer, the input of the first layer in the coder is the original data, the output of the previous layer in each coding layer in the back is the input of the layer, and finally the coder outputs the coding result。
As shown in fig. 3, the decoding layer operation performed on the processed data by the decoder using the transform model is specifically that the decoder data input operation first extracts half of the seasonal item obtained by the decomposition in step S1, and the half sequence length is equal toAfter that, 0 is added to make the total length equal toAnd then input into a decoder for processing.
S31, selecting the season itemAfter the frequency domain characteristic layer and the spatial domain characteristic layer are input, the output of the characteristic layer is compared withAdding and carrying out sequence decomposition to obtain seasonal componentsAnd trend component;
S32, the seasonal component obtained in the last stepAnd the encoder outputs the encoding resultPerforming frequency domain self-attention operation;
s33, outputting the frequency domain from attention and the seasonal component obtained in the first stepAdding the obtained seasonal components and performing sequence decomposition to obtain seasonal componentsSum trend componentWill beAfter passing through feedforward neural network and,,adding to obtain the output of the decoder。
The frequency domain self-attention operation process is as follows: first, a multi-head mechanism is introduced through the linear layerQuery matrix mapped to more channels by linear layersIn the same way, willMapping into value matrix V and key matrix through linear layerThen is aligned with,V, carrying out fast Fourier transform and random sampling to obtain a sampled frequency spectrum matrixSecondly, the matrix is divided intoPerforming inner product operation, activating a function through softmax, and performing the operationPerforming inner product operation, and finally performing zero filling and fast Fourier inverse transformation on the obtained result to obtain frequency domain self-attention outputThe formula is expressed as follows,
The method for predicting the user power consumption based on the transform of the frequency domain and the spatial domain features solves the problem that the long-time sequence is difficult to effectively model in the conventional prediction method, extracts the frequency domain and the spatial domain features at the same time, and predicts the power consumption by using the features.
As shown in fig. 4, a structure diagram of a user power consumption prediction model based on a transform model of frequency domain and spatial domain features according to an embodiment of the present application is provided.
As shown in fig. 5, an embodiment of the present application provides a regional power consumption prediction system based on a combination of a frequency domain and a spatial domain, the system includes,
the power grid historical data acquisition and analysis module 1 is used for acquiring power grid historical data and comprehensively analyzing historical power consumption data and position information in the power grid historical data;
the season-trend item decomposition module 2 is used for performing season-trend item decomposition on historical data of the power grid;
the characteristic extraction module 3 is used for performing frequency domain characteristic extraction based on Fourier transform and space domain characteristic extraction based on a graph convolution neural network on the historical data of the power grid;
and the power consumption prediction module 4 for the power grid region is used for realizing accurate power consumption prediction of the power grid region by combining frequency domain characteristics, spatial domain characteristics and multi-head frequency domain sub-attention operation in a long time sequence.
An embodiment of the present application further provides a computer-readable storage medium, where program codes are stored, and when the program codes are executed by a processor, the method for intelligently predicting regional power consumption as described above is implemented.
As shown in fig. 6, by using the method for predicting electricity consumption in an area based on combination of a frequency domain and an airspace according to the embodiment of the present application, the data of the electricity consumption of users in a certain cell in wuhan is analyzed to obtain a predicted electricity consumption curve in 100 hours, and the actual electricity consumption curve in 100 hours is contrastingly analyzed, so that it can be found that the similarity between the predicted electricity consumption curve obtained by the prediction method of the present application and the actual electricity consumption curve is high, and thus the effectiveness of the method of the present application can be determined.
The method for predicting the user power consumption of the transform based on the frequency domain and spatial domain features solves the problem that the long-time sequence is difficult to effectively model in the existing prediction method, extracts the frequency domain and spatial domain features, predicts the power consumption by using the features, and is high in prediction accuracy.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). The memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising a," "8230," "8230," or "comprising" does not exclude the presence of additional identical elements in the process, method, article, or apparatus comprising the element.
The above description is only an example of the present application and is not intended to limit the scope of the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims (3)
1. A regional power consumption prediction method based on combination of a frequency domain and a spatial domain is characterized by comprising the following specific steps:
according to the acquired historical data of the power grid, comprehensively analyzing historical power consumption data and position information in the historical data of the power grid, and performing season-trend item decomposition on the historical data of the power grid;
performing encoder processing of a transform model aiming at input power grid historical data, wherein the encoder processing comprises frequency domain feature extraction based on Fourier transform and space domain feature extraction based on a graph convolution neural network;
performing decoding layer operation on the processed data by using a decoder of a transform model, and realizing accurate power consumption prediction of a power grid area by combining frequency domain characteristics, space domain characteristics and multi-head frequency domain sub-attention operation in a long-time sequence;
the historical power consumption data and the position information in the historical data of the comprehensive analysis power grid are specifically that the historical power consumption data comprises a user power consumption time sequenceThe user position information passes through a directed graphIs shown in whichIs a collection of user locations, in commonA user is then provided withThe elements are selected from the group consisting of,is the edge of the directed graph,the method is characterized in that an adjacency matrix for storing the distance between users is formed by splicing the time sequence and the position information of each power consumption data of the users into a matrix;
The method for decomposing the historical data of the power grid by the season-trend item specifically comprises the step of carrying out time series on the power consumption of a user by a transform modelPerforming sequence decomposition into seasonal itemsAnd trend itemThe specific calculation method is shown in the following formula:
wherein,in order to average out the pooling operation,operations supplement 0 at both ends of the sequence to keep the length of the sequence unchanged after pooling operations;
the frequency domain feature extraction based on the fourier transform is specifically,
time series of electricity consumption of userIn the step (1), the first step,in order to be the length of the sequence,for the characteristic quantity, willUser electricity consumption time series of dimensionSending the data into a frequency domain characteristic layer for Fourier transform to extract frequency domain characteristics, wherein the specific operation of the frequency domain characteristic layer is to firstly carry out Fourier transform on the dataLinear mapping is performed to map it intoDimensional query matrixWhereinIs a preset number with proper size, and then is pairedIs subjected to fast Fourier transform to obtainDimensional spectrum matrixIn order to reduce the amount of calculation and prevent overfitting, sparse operation is carried out, and the specific method is toRandom sampling is carried out to obtainDimension matrixIn whichIs far less thanSecond using a layer of randomly initialized convolutional neural network andby convolution to obtainMatrix ofFinally to is alignedPerforming zero-filling operation to make its dimension becomeAnd performing inverse fast Fourier transform to obtain an output;
The spatial domain feature extraction based on the graph convolution neural network specifically comprises the following steps of extracting a matrixSending into graph volume network layer, defining in-graphThe above graph convolution formula is:
whereinAre the parameters of the convolution kernel and,is a matrix of the unit, and is,is shown as a drawingDegree matrix, matrixObtaining the corresponding dimension of the user to be predicted after the graph convolution layer and carrying out zero filling to obtain the output of the spatial domain characteristic layer;
The structure of an encoder in the encoder processing of the transformer model aiming at the input power grid historical data is determined byThe same coding layer is formed, and the encoder processes the output of the frequency domain characteristic layerAnd spatial feature layerAdding the data, decomposing the sequence to obtain seasonal terms, sending the seasonal terms into a feedforward neural network, decomposing the sequence again to obtain the output of the coding layer, the input of the first layer in the coder is the original data, the output of the previous layer in each coding layer in the back is the input of the layer, and finally the coder outputs the coding result;
The decoder using the transform model performs a decoding layer operation on the processed data specifically,
first, the season itemAfter the frequency domain characteristic layer and the space domain characteristic layer are input, the output of the characteristic layer is summedAdding and carrying out sequence decomposition to obtain seasonal componentsAnd trend component;
The second step, the seasonal component obtained in the previous stepAnd the encoder outputs the encoding resultCarrying out frequency domain self-attention operation;
thirdly, outputting the frequency domain self attention and the seasonal component obtained in the first stepAdding the obtained seasonal components and performing sequence decomposition to obtain seasonal componentsAnd trend componentWill beAfter passing through a feedforward neural network,,Adding to obtain the output of the decoder;
The frequency domain self-attention operation process is as follows: first, a multi-head mechanism is introduced through the linear layer, andquery matrix mapped to more channels by linear layersIn the same way, willMapping into value matrix V and key matrix through linear layerThen to,V, carrying out fast Fourier transform and random sampling to obtain a sampled frequency spectrum matrixSecondly, the matrix is divided intoPerforming inner product operation, activating function and then performing AND through softmaxPerforming inner product operation, and finally performing zero filling and inverse fast Fourier transform on the obtained result to obtain frequency domain self-attention outputThe formula is expressed as follows,
2. A regional power consumption prediction system based on a combination of frequency domain and spatial domain, wherein the system uses the method of claim 1 to predict, the system comprising,
the power grid historical data acquisition and analysis module is used for acquiring power grid historical data and comprehensively analyzing historical power consumption data and position information in the power grid historical data;
the season-trend item decomposition module is used for performing season-trend item decomposition on historical data of the power grid;
the characteristic extraction module is used for extracting the frequency domain characteristic of the power grid historical data based on Fourier transform and the spatial domain characteristic of the power grid historical data based on a graph convolution neural network;
and the power consumption prediction module of the power grid area is used for combining the frequency domain characteristics, the spatial domain characteristics and the multi-head frequency domain sub-attention operation in the long-time sequence to realize accurate power consumption prediction of the power grid area.
3. A computer-readable storage medium, characterized in that the computer-readable storage medium stores program code which, when executed by a processor, implements the steps of the method for regional power usage prediction based on a combination of frequency and spatial domains as claimed in claim 1.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210978150.4A CN115049169B (en) | 2022-08-16 | 2022-08-16 | Regional power consumption prediction method, system and medium based on combination of frequency domain and spatial domain |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210978150.4A CN115049169B (en) | 2022-08-16 | 2022-08-16 | Regional power consumption prediction method, system and medium based on combination of frequency domain and spatial domain |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115049169A CN115049169A (en) | 2022-09-13 |
CN115049169B true CN115049169B (en) | 2022-10-28 |
Family
ID=83168205
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210978150.4A Expired - Fee Related CN115049169B (en) | 2022-08-16 | 2022-08-16 | Regional power consumption prediction method, system and medium based on combination of frequency domain and spatial domain |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115049169B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115545361B (en) * | 2022-12-02 | 2023-05-09 | 国网湖北省电力有限公司信息通信公司 | Method, system and medium for predicting climate environment of power grid transmission line |
CN116109016A (en) * | 2023-04-11 | 2023-05-12 | 昆仑数智科技有限责任公司 | Energy consumption prediction model training and energy consumption prediction method and device |
CN116415744B (en) * | 2023-06-12 | 2023-09-19 | 深圳大学 | Power prediction method and device based on deep learning and storage medium |
CN118233035B (en) * | 2024-05-27 | 2024-08-06 | 烟台大学 | Multiband spectrum prediction method and system based on graph convolution inversion transform |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2016063660A1 (en) * | 2014-10-23 | 2016-04-28 | 株式会社日立製作所 | Power system monitoring device and power system monitoring method |
CN111784043A (en) * | 2020-06-29 | 2020-10-16 | 南京工程学院 | Accurate prediction method for power selling amount of power distribution station area based on modal GRU learning network |
CN114154700A (en) * | 2021-11-25 | 2022-03-08 | 浙江工业大学 | User power consumption prediction method based on transformer model |
CN114519471A (en) * | 2022-03-28 | 2022-05-20 | 杭州电子科技大学 | Electric load prediction method based on time sequence data periodicity |
CN114723583A (en) * | 2022-03-24 | 2022-07-08 | 天津三源电力信息技术股份有限公司 | Unstructured electric power big data analysis method based on deep learning |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113935534A (en) * | 2021-10-20 | 2022-01-14 | 华能大理风力发电有限公司洱源分公司 | Photovoltaic day-ahead prediction method, system and equipment based on convolution transform architecture |
CN114819388A (en) * | 2022-05-18 | 2022-07-29 | 江苏方天电力技术有限公司 | Condenser vacuum degree prediction method and device based on frequency domain information guidance |
-
2022
- 2022-08-16 CN CN202210978150.4A patent/CN115049169B/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2016063660A1 (en) * | 2014-10-23 | 2016-04-28 | 株式会社日立製作所 | Power system monitoring device and power system monitoring method |
CN111784043A (en) * | 2020-06-29 | 2020-10-16 | 南京工程学院 | Accurate prediction method for power selling amount of power distribution station area based on modal GRU learning network |
CN114154700A (en) * | 2021-11-25 | 2022-03-08 | 浙江工业大学 | User power consumption prediction method based on transformer model |
CN114723583A (en) * | 2022-03-24 | 2022-07-08 | 天津三源电力信息技术股份有限公司 | Unstructured electric power big data analysis method based on deep learning |
CN114519471A (en) * | 2022-03-28 | 2022-05-20 | 杭州电子科技大学 | Electric load prediction method based on time sequence data periodicity |
Non-Patent Citations (1)
Title |
---|
基于提升小波的时间序列分析法的电力负荷预测;张帆等;《电气自动化》;20170530(第03期);第76-80页 * |
Also Published As
Publication number | Publication date |
---|---|
CN115049169A (en) | 2022-09-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115049169B (en) | Regional power consumption prediction method, system and medium based on combination of frequency domain and spatial domain | |
CN116361256B (en) | Data synchronization method and system based on log analysis | |
Shankaranarayana et al. | Attention augmented convolutional transformer for tabular time-series | |
Xing et al. | Towards accurate binary neural networks via modeling contextual dependencies | |
Bertola et al. | Darboux transformations and random point processes | |
Tack et al. | Online adaptation of language models with a memory of amortized contexts | |
Xu et al. | Predicting Multidimensional Environmental Factor Trends in Greenhouse Microclimates Using a Hybrid Ensemble Approach | |
CN115545361B (en) | Method, system and medium for predicting climate environment of power grid transmission line | |
Li et al. | Daily PM 10 concentration forecasting based on multiscale fusion support vector regression | |
Fatema et al. | Analysing and forecasting electricity demand and price using deep learning model during the COVID-19 pandemic | |
CN116777066A (en) | Photovoltaic power probability prediction method and device based on foundation cloud image guided depth decomposition | |
Mao et al. | Time-varying graph signals reconstruction | |
CN108599773B (en) | Vibration signal data compression acquisition method based on deterministic measurement matrix | |
CN114819425A (en) | Intelligent prediction method and system for regional power consumption and storage medium | |
Fan et al. | CPI big data prediction based on wavelet twin support vector machine | |
Ling et al. | Diffusion model-based probabilistic downscaling for 180-year East Asian climate reconstruction | |
CN115203073A (en) | Memory prefetching method, device and medium | |
Yang et al. | Sparse bayesian variable selection with correlation prior for forecasting macroeconomic variable using highly correlated predictors | |
Mordvinov et al. | Long-term changes in total solar irradiance and their predictions | |
Reinbold et al. | Parameterized splitting of summed volume tables | |
CN118171783B (en) | Energy consumption prediction model training method, prediction method, device, medium and equipment | |
CN111736845B (en) | Coding method and device | |
Long et al. | A CNN channel pruning low-bit framework using weight quantization with sparse group lasso regularization | |
CN117290669B (en) | Optical fiber temperature sensing signal noise reduction method, device and medium based on deep learning | |
Kloska et al. | Towards Symbolic Time Series Representation Improved by Kernel Density Estimators |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20221028 |
|
CF01 | Termination of patent right due to non-payment of annual fee |