CN118657408B - A regional electric power prediction method, system, device and storage medium - Google Patents
A regional electric power prediction method, system, device and storage medium Download PDFInfo
- Publication number
- CN118657408B CN118657408B CN202411139259.4A CN202411139259A CN118657408B CN 118657408 B CN118657408 B CN 118657408B CN 202411139259 A CN202411139259 A CN 202411139259A CN 118657408 B CN118657408 B CN 118657408B
- Authority
- CN
- China
- Prior art keywords
- power
- nodes
- node
- distance
- electric
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for AC mains or AC distribution networks
- H02J3/004—Generation forecast, e.g. methods or systems for forecasting future energy generation
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for AC mains or AC distribution networks
- H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
- H02J3/381—Dispersed generators
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/10—Power transmission or distribution systems management focussing at grid-level, e.g. load flow analysis, node profile computation, meshed network optimisation, active network management or spinning reserve management
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2300/00—Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
- H02J2300/20—The dispersed energy generation being of renewable origin
- H02J2300/22—The renewable source being solar energy
- H02J2300/24—The renewable source being solar energy of photovoltaic origin
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2300/00—Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
- H02J2300/20—The dispersed energy generation being of renewable origin
- H02J2300/28—The renewable source being wind energy
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- General Business, Economics & Management (AREA)
- Educational Administration (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Power Engineering (AREA)
- Tourism & Hospitality (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Development Economics (AREA)
- Primary Health Care (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a regional electric power prediction method, a regional electric power prediction system, regional electric power prediction equipment and a regional electric power prediction storage medium, and relates to the technical field of energy, wherein the regional electric power prediction method comprises the following steps: performing spatial gridding on the target area to obtain a plurality of grids only comprising single power nodes; synchronizing the time of each grid node; calculating the electrical distance between the nodes with electrical correlation, and dividing the nodes with electrical correlation onto a power path according to the electrical distance; calculating the proximity centrality of each node in each power path based on the electrical distance; the power nodes are distributed with attention weights according to the electrical distance and the proximity centrality; the matrix obtained after the processing is arranged according to the time stamp sequence, and the power of a target area at the future moment is predicted by using a three-dimensional convolution network with different time granularity; according to the method, a three-dimensional convolutional neural network combined with an attention mechanism is adopted to conduct power prediction, space-time characteristics of power data are extracted under low model complexity, and then power of a target area is predicted.
Description
Technical Field
The invention relates to the technical field of energy, in particular to a regional electric power prediction method, a regional electric power prediction system, regional electric power prediction equipment and a regional electric power storage medium.
Background
In the age of energy diversification, regional electric power prediction has exceeded the category of single energy types, and it covers comprehensive prediction of power consumption from the conventional energy and renewable energy to terminals such as household electricity. This comprehensive predictive capability is critical to accommodate the challenges currently faced by power systems. Nowadays, the large-scale integration of renewable energy sources brings volatility and uncertainty to a power grid, and the diversification and individuation demands of terminal energy consumption modes such as household electricity consumption and the like also put higher demands on the supply and demand balance of the power grid. The comprehensive power prediction not only can help a power grid operator to better manage the stable operation of the power grid and ensure the effective distribution of energy sources, but also can meet the personalized requirements of end users on energy source management and improve the efficiency and satisfaction of energy source use.
The current regional electric power prediction research is mainly independently carried out aiming at the fields of photovoltaic power generation power, wind power generation power, household energy requirements and the like. In performing these predictive tasks, models are typically selected in advance based on data characteristics, and the respective characteristics are taken into account in the selection and adjustment of the models. This approach relies on human prior feature analysis to guide the prediction process. However, this approach limits the universality of the model in the global power prediction task, and when other aspects of data need to be predicted, the model needs to be retrained, even the model needs to be replaced, so that various powers of the target area cannot be effectively predicted;
The main methods used in the power prediction field at present are divided into two main types, one type is prediction based on a traditional statistical method, and the method is simple and easy to realize, has strong interpretation, but cannot effectively capture complex nonlinear relations; the other type is the combination of multiple deep learning models, such as the combination of CNN and LSTM for prediction, and the method can capture complex nonlinear relations, but the difficulty of model adjustment and optimization is correspondingly increased due to the increase of the number of super parameters caused by model cascading.
Disclosure of Invention
Aiming at the defects that the prior art cannot effectively capture complex nonlinear relations and the number of super parameters is increased due to model cascading, and the difficulty of model adjustment and optimization is correspondingly improved, the invention provides a regional electric power prediction method, a regional electric power prediction system, regional electric power prediction equipment and a regional electric power storage medium.
A regional electric power prediction method comprising the steps of:
collecting all power nodes in a target area containing absorption power or transmission power equipment;
extracting the space-time dynamic characteristics of each power node;
predicting the power of the target area at the future moment according to the space-time dynamic characteristics of each power node;
The method for extracting the space-time dynamic characteristics of each power node specifically comprises the following steps:
performing spatial gridding on the target area to obtain a plurality of grids only comprising single power nodes;
obtaining the electrical distance between the power nodes with electrical correlation by adopting an equivalent impedance method, and dividing the power nodes with electrical correlation onto a power path according to the electrical distance;
Calculating the proximity centrality of each power node in each power path based on the electrical distance;
The power nodes are distributed with attention weights according to the electrical distance and the proximity centrality;
and inputting the power node data with the attention weight assigned to a three-dimensional convolution network with different time granularities to obtain the space-time dynamic characteristics of the power nodes.
Further, the step of performing spatial meshing on the target area to obtain a plurality of meshes including only a single power node specifically includes the following steps:
Grid division with different sizes is carried out according to the electric energy density of different positions in the target area; selecting a grid with the largest size to divide the areas with sparse power node distribution and concentrated electric energy density; taking the grid with the smallest size as a unit grid, and dividing other grids according to the unit grid size;
Distributing the power nodes into each grid to obtain a two-dimensional grid containing the power nodes; wherein each grid contains only a single power node; if a power node spans multiple grids, dividing the power node into one grid without other power nodes according to a minimum center distance standard.
Further, the method for obtaining the electrical distance between the power nodes with electrical correlation by adopting the equivalent impedance method specifically comprises the following steps:
For power nodes in a single power supply network in a target area, using a modulus value of equivalent impedance between the power nodes as an electrical distance between the power nodes;
for power nodes in a multi-power network in a target area, splitting the multi-power network into a plurality of single-power networks through a downstream tracking method for processing to obtain impedance among the power nodes under the action of any power source; and after the equivalent impedance between the corresponding nodes under the respective action of the power supplies is connected in parallel, taking the modulus value as the electrical distance between the power nodes.
Further, the calculating the proximity centrality of each power node in each power path based on the electrical distance is expressed as a calculation formula of the proximity centrality:
Wherein, Representing the average value of the electrical distances of node i to points in power transfer relationship therewith, N representing the total number of nodes on the power path with node i,Representing the electrical distance between node i and node j,Indicating the proximity centrality of node i.
Further, the power node is assigned attention weight according to the electrical distance and the proximity centrality, and specifically comprises the following steps:
Normalizing the electrical distance and the proximity centrality, and recording the electrical distance between two power nodes on a certain power path after normalization as ; If there is a node electrically associated with the plurality of nodes, an average of the plurality of electrical distances is employed;
the value of the proximity centrality in the power path is noted as ；
Weight distribution is carried out on the electrical distance and the proximity centrality; wherein if the electrical distance weight isThen the approximate centrality weight isAnd (2) andThe attention weight of the power node is obtained as。
The invention also includes a regional electric power prediction system comprising:
the acquisition module is used for acquiring all power nodes in a target area containing power absorption or transmission equipment;
the extraction module is used for extracting the space-time dynamic characteristics of each power node;
the prediction module is used for predicting the power of the target area at the future moment according to the space-time dynamic characteristics of each power node;
wherein, the extraction module includes:
the dividing unit is used for carrying out space gridding on the target area to obtain a plurality of grids only comprising single power nodes;
The electric distance calculation unit is used for obtaining the electric distance between the power nodes with electric correlation by adopting an equivalent impedance method, and dividing the power nodes with electric correlation onto a power path according to the electric distance;
a proximity centrality calculating unit for calculating proximity centrality of each power node in each power path based on the electrical distance;
an allocation unit for allocating attention weights to the power nodes according to the electrical distance and the proximity centrality;
and the extraction unit is used for inputting the power node data after the attention weight is distributed into the three-dimensional convolution network with different time granularity, so as to obtain the space-time dynamic characteristic of the power node.
The invention also includes a regional electric power prediction computer device comprising: a memory, a processor and a computer program stored in the memory, the processor implementing the steps of the regional electric power prediction method when executing the computer program.
Further, a readable storage medium stores a computer program comprising program instructions for performing the steps of the regional electric power prediction method when executed by a processor.
The invention provides a regional electric power prediction method, a regional electric power prediction system, regional electric power prediction equipment and a storage medium, which have the following beneficial effects:
According to the invention, the grid division of all power nodes in the target area is performed to ensure the effective meshing of the power nodes in all directions in the target area; the connection tightness degree between the power nodes is measured by adopting an electric distance, the power nodes with electric correlation are found out according to the electric distance and are divided into a power path, the internal connection and evolution rule of power dynamics on a specific power path are revealed by intensively analyzing the power variation of the nodes on the same power path, the variation situation of the power of the nodes is predicted better, the power nodes are distributed with attention weights according to the electric distance and the approaching centrality, the characteristic learning capacity of the three-dimensional convolutional neural network on the time scale and the space scale is utilized, the time-space dynamic characteristic of the power node data with the attention weights distributed is extracted under the low model complexity, and the power of the future moment of the area is predicted; according to the method, various powers in the target area are comprehensively considered, the difficulty of model adjustment and optimization is reduced while the complex nonlinear relation is effectively captured, and further comprehensive prediction of the power of the target area is achieved.
Drawings
FIG. 1 is a schematic diagram of effective meshing of a target area in an embodiment of the present invention; FIG. 1a is a schematic diagram showing the distribution of power nodes in an area where power prediction is required; FIG. 1b is a diagram showing a grid distribution of different sizes according to the density of power nodes in different portions of the target area; fig. 1 c shows a standardized distribution diagram of a grid using a minimum grid as a unit grid; d of fig. 1 represents a schematic diagram of the nearby allocation of power nodes into a grid; FIG. 1 e shows a schematic diagram of the distribution with rows and columns without power nodes removed; f of fig. 1 shows a distribution diagram of the target area after effective meshing;
FIG. 2 is a schematic diagram of attention weight distribution in an embodiment of the present invention;
FIG. 3 is a schematic diagram of a three-dimensional convolutional neural network in an embodiment of the present invention;
fig. 4 is a flowchart of a method for predicting regional electric power according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments.
The invention provides a regional electric power prediction method, which extracts the space-time characteristics of power data under low model complexity so as to predict the power of a target region, as shown in fig. 4; the method specifically comprises the following steps:
S1, performing space meshing operation on a target area (the area can be any area containing power absorption or transmission equipment, such as a certain photovoltaic power plant, a certain town, a certain cell and the like, as shown in a of fig. 1), and performing meshing with different sizes according to the power node densities at different positions in the target area so as to ensure effective meshing of power nodes in all directions in the target area.
For areas with sparse power node distribution and small electric energy density, larger grid division can be used; for areas where the power nodes are densely distributed and the power density is high, smaller meshing may be used, as shown in fig. 1 b.
In this data allocation pattern, each grid is designed to contain only a single power node, and there is no case where two or more power nodes are present in one grid; there may be cases where a power node spans multiple grids, which may then be partitioned into one of the grids without other power nodes by a minimum center distance criterion. Gridding meeting the above criteria is referred to as effective gridding.
After the above processing, an initial two-dimensional grid table is obtained, and the grid table is not regular, as shown in b of fig. 1.
S2, taking the grid with the smallest size as a unit grid, and dividing other grids according to the unit grid size. If the rest part is less than one unit grid, the rest part is complemented according to the unit grid; as shown in fig. 1c, the mesh complement operation is performed during the conversion from fig. 1 b to fig. 1 c.
And S3, distributing the power nodes into nearby grids to form a two-dimensional grid containing the power nodes.
And (3) for the situation that the power node is completely in a certain grid, directly distributing the power node into the grid to form a two-dimensional grid containing the power node, and then placing the sampled data into the grid.
And for the case that the power node crosses the grid, taking the minimum center distance as a judging basis for grid attribution. If the grid selected according to the minimum center distance criterion contains other power nodes, the grid with the next smallest center distance is selected, and so on. If the center distances of the power nodes to a plurality of grids are the same, one of the grids is selected randomly.
For any one power node, its data will be mapped uniquely into one grid, but this does not guarantee that all grids will be assigned data, i.e. there may be some grids that are not filled with data, but all existing power node data will have a corresponding grid. Meshing and power node allocation as shown in d of fig. 1, the dark mesh region represents the nearby mesh to which the power node belongs.
S4, removing rows and columns without power nodes in the two-dimensional data grid to form a new two-dimensional data grid, wherein e in fig. 1 to f in fig. 1 represent the removal process.
In view of the fact that the interdependencies between power nodes in a power system do not significantly depend on their geographical proximity, the rows and columns without power nodes are deleted in order to reduce the complexity of data processing and the data dimension in subsequent analysis, reducing the dimension of the original data while preserving part of the geographical location information.
S5, regarding each grid area containing the power nodes as an Internet of things node, and distributing certain communication, calculation and storage capacities for the Internet of things node.
S6, microsecond time synchronization and millisecond data transmission of all nodes are achieved by utilizing the internet of things technology.
And S7, selecting a corresponding sampling frequency according to the expected prediction precision, filling the acquired power value into a corresponding grid, acquiring power data for a period of time, and adding a time stamp for the power data.
When recording data, for the node of the transmitting power, the related power value is recorded as a positive value; for nodes that absorb power, the relevant power data is recorded as negative values.
The sampling frequency should be at least the same order of magnitude as the prediction accuracy to ensure that the desired prediction accuracy is achieved. For the present study, power prediction at a time granularity of millisecond and above can be achieved due to limitations of millisecond time transfer techniques.
S8, calculating the electrical distance between nodes with electrical correlation, namely power (electric energy) circulation relation by using an equivalent impedance method; the electrical distance is an indicator for measuring the tightness of the connection between nodes in an electrical network. The smaller the electrical distance between the nodes, the tighter the connection between them. And calculating the electrical distance between the nodes, and taking the electrical distance as one of the basis for attention weight distribution of the follow-up convolutional neural network.
For a single power network, the modulus of the equivalent impedance between nodes is directly used as the electrical distance between the nodes. For a multi-power network, the multi-power network is split into a plurality of single-power networks through downstream tracking (namely, current path identification is carried out according to the actual flow path of current in an electrical system), and the electrical distance is calculated according to the processing method of the single-power networks; then according to the trend tracking (namely according to the identified current path, calculating and analyzing the voltage, current and power flow of each node in the power system), the impedance between each node under the action of any power supply can be obtained; and for the synthetic power flow, the equivalent impedance between the corresponding nodes when the power supplies act respectively is connected in parallel, and the power supply is taken as the electrical distance between the nodes after the modulus value is taken.
S9, dividing the nodes with electrical association into one power path.
From the electrical distance, a power node with an electrical association can be found. Dividing the power change of the node on the same power path into one power path, and analyzing the power change of the node on the same power path in a centralized way is helpful for revealing the internal connection and evolution rule of the power dynamic on a specific power path, so that the change condition of the node power is predicted better.
S10, calculating the proximity centrality of each node in each path.
Proximity centrality is an indicator that measures the importance of nodes in a network. It is defined as the reciprocal of the average distance of one node from the other, the shorter the distance, the higher the proximity centrality of the node. Nodes with high proximity centrality are generally considered critical nodes in the network, with greater control capability and impact on other nodes. And calculating the proximity centrality of each node by using the electrical distance, and taking the proximity centrality as one of the basis for attention weight distribution of the follow-up convolutional neural network.
The approximate centrality is calculated as follows:
In the above-mentioned method, the step of, Representing the average value of the electrical distances of node i to points in power transfer relationship therewith, N representing the total number of nodes on the power path with node i,Representing the electrical distance between node i and node j,Indicating the proximity centrality of node i.
And S11, distributing attention weight to the power nodes according to the electric distance and the proximity centrality.
Different attention patterns are generated according to different power paths. The size of the attention map is the same as the input matrix size, and the attention map is given a corresponding positional attention weight along the power path. The attention weight of each power node on the power path may be determined in accordance with the following manner.
Firstly, normalizing the electrical distance and the index approaching to the centrality, and recording the electrical distance between two nodes on a certain power path after normalization asFor nodes electrically associated with a plurality of nodes, an average of a plurality of electrical distances is used; the value of proximity centrality in the power path is. Then, the two indexes of the electric distance and the approaching centrality are assigned with weights, if the electric distance weight isThen the approximate centrality weight isWherein. Finally, the attention weight of the node is as follows。
Given a concentration weight distribution as shown in fig. 2, the lines of different thickness in fig. 2 represent different power paths; the thickness of the line indicates the magnitude of the electrical distance between the nodes, and the thicker the line indicates the closer the electrical distance between the nodes.
S12, the matrix obtained after the processing is arranged according to the time stamp sequence, and the power of the target area at the future moment is predicted by using three-dimensional convolution networks with different time granularities.
And constructing a three-dimensional convolutional neural network, and capturing the space-time dynamic characteristics of the power data by utilizing the characteristic learning capability of the three-dimensional convolutional neural network on a time scale and a space scale, so that the power at the future moment is predicted according to the space-time characteristics of the data.
A convolutional network is given as shown in fig. 3: assume that the input matrix 1 has dimensions 4 x4, which represents a sequence of power samples with three time steps, where each time step contains a4 x4 matrix of values.
First, attention mechanisms are introduced, and the importance or weight of different areas in the input matrix 1 is adjusted by adding attention layers with dimensions of 4×4, so as to obtain a matrix 2. Then, the time and space information in the matrix 2 is extracted through a convolution kernel 1 of 2 multiplied by 2, and a matrix 3 is obtained. A2 x 2 pooling layer is used to reduce the dimensions of the matrix 3 while retaining the most important features, resulting in a matrix 4. And then the convolution kernel 2 with the size of 1 multiplied by 3 is used for further extracting time information to obtain a matrix 5, then nonlinearity is introduced through an activation layer, the nonlinearity is mapped to a target dimension through a full connection layer, and finally residual calculation and weight updating are carried out.
In addition, by constructing three-dimensional convolution networks with different time granularities, the influence of accumulated prediction errors on the final prediction result can be effectively reduced. The invention also provides a feasible scheme for respectively constructing the three-dimensional convolutional neural network with time granularity of 1s, 5s and 10 s. The network with the time granularity of 10s can reduce the accumulated prediction error of the network with the time granularity of 1s and 5s to a certain extent; a network with a time granularity of 5s can reduce the accumulated prediction error of a network with a time granularity of 1s to some extent.
Based on the same inventive concept, the invention also provides a regional electric power prediction system, comprising:
and the acquisition module is used for acquiring all power nodes in the target area containing the power absorption or transmission equipment.
And the extraction module is used for extracting the space-time dynamic characteristics of each power node.
And the prediction module is used for predicting the power of the target area at the future moment according to the space-time dynamic characteristics of each power node.
Wherein, the extraction module includes:
And the dividing unit is used for carrying out space gridding on the target area to obtain a plurality of grids only comprising a single power node.
And the electrical distance calculation unit is used for acquiring the electrical distance between the power nodes with electrical correlation by adopting an equivalent impedance method and dividing the power nodes with electrical correlation into a power path according to the electrical distance.
And the proximity centrality calculating unit is used for calculating the proximity centrality of each power node in each power path based on the electrical distance.
And the distribution unit is used for distributing attention weight values to the power nodes according to the electrical distance and the proximity centrality.
And the extraction unit is used for inputting the power node data after the attention weight is distributed into the three-dimensional convolution network with different time granularity, so as to obtain the space-time dynamic characteristic of the power node.
Based on the same inventive concept, the invention also provides a regional electric power prediction computer device, comprising: a memory, a processor and a computer program stored in the memory, the processor implementing the steps of the regional electric power prediction method when the computer program is executed.
Based on the same inventive concept, the invention also proposes a readable storage medium storing a computer program comprising program instructions for executing the steps of the regional electric power prediction method when the program instructions are executed by a processor.
The foregoing is only a preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art, who is within the scope of the present invention, should make equivalent substitutions or modifications according to the technical scheme of the present invention and the inventive concept thereof, and should be covered by the scope of the present invention.
Claims (8)
1. A regional electric power prediction method, characterized by comprising the steps of:
collecting all power nodes in a target area containing absorption power or transmission power equipment;
extracting the space-time dynamic characteristics of each power node;
predicting the power of the target area at the future moment according to the space-time dynamic characteristics of each power node;
The method for extracting the space-time dynamic characteristics of each power node specifically comprises the following steps:
Performing spatial gridding on the target area to obtain a plurality of grids only comprising one power node;
obtaining the electrical distance between the power nodes with electrical correlation by adopting an equivalent impedance method, and dividing the power nodes with electrical correlation onto a power path according to the electrical distance;
Calculating the proximity centrality of each power node in each power path based on the electrical distance;
The power nodes are distributed with attention weights according to the electrical distance and the proximity centrality;
Inputting the power node data with the attention weight distributed into a three-dimensional convolution network with different time granularities to obtain the space-time dynamic characteristics of the power nodes; the space-time dynamic characteristic is that a dynamic rule of power of a power node under the comprehensive influence of time, electric distance and approaching centrality is obtained through three-dimensional convolution operation.
2. The method for predicting regional electric power according to claim 1, wherein the spatial gridding of the target region is performed to obtain a plurality of grids including only one power node, and specifically comprises the steps of:
Grid division with different sizes is carried out according to the electric energy density of different positions in the target area; selecting a grid with the largest size to divide the areas with sparse power node distribution and concentrated electric energy density; taking the grid with the smallest size as a unit grid, and dividing other grids according to the unit grid size;
Distributing the power nodes into each grid to obtain a two-dimensional grid containing the power nodes; wherein each grid contains only one power node; if a power node spans multiple grids, dividing the power node into one grid without other power nodes according to a minimum center distance standard.
3. The method for predicting regional electric power according to claim 1, wherein the step of obtaining the electric distance between the power nodes having the electric association by using the equivalent impedance method comprises the steps of:
For power nodes in a single power supply network in a target area, using a modulus value of equivalent impedance between the power nodes as an electrical distance between the power nodes;
for power nodes in a multi-power network in a target area, splitting the multi-power network into a plurality of single-power networks through a downstream tracking method for processing to obtain impedance among the power nodes under the action of any power source; and after the equivalent impedance between the corresponding nodes under the respective action of the power supplies is connected in parallel, taking the modulus value as the electrical distance between the power nodes.
4. The method of claim 1, wherein the calculating the proximity centrality of each power node in each power path based on the electrical distance is expressed by a formula:
where d i represents the average value of the electrical distances from node i to points in power transfer relationship therewith, N represents the total number of nodes on the power path with node i, d ij represents the electrical distance between node i and node j, and CC i represents the proximity centrality of node i.
5. The method for predicting regional electric power according to claim 4, wherein the assigning attention weights to power nodes according to electric distance and proximity centrality comprises the steps of:
normalizing the electrical distance and the proximity centrality, and recording the electrical distance between two power nodes on a certain power path after normalization as a i; if there is a node electrically associated with the plurality of nodes, an average of the plurality of electrical distances is employed;
the value normalized by the approximate centrality in the power path is denoted b i;
Weight distribution is carried out on the electrical distance and the proximity centrality; if the electrical distance weight is λ 1, the proximity centrality weight is 1- λ 1, and 1 > λ 1 > 0, and the attention weight of the power node is a i·λ1+bi·(1-λ1).
6. A regional electric power prediction system, comprising:
the acquisition module is used for acquiring all power nodes in a target area containing power absorption or transmission equipment;
the extraction module is used for extracting the space-time dynamic characteristics of each power node;
the prediction module is used for predicting the power of the target area at the future moment according to the space-time dynamic characteristics of each power node;
wherein, the extraction module includes:
the dividing unit is used for carrying out space gridding on the target area to obtain a plurality of grids which only comprise one power node;
The electric distance calculation unit is used for obtaining the electric distance between the power nodes with electric correlation by adopting an equivalent impedance method, and dividing the power nodes with electric correlation onto a power path according to the electric distance;
a proximity centrality calculating unit for calculating proximity centrality of each power node in each power path based on the electrical distance;
an allocation unit for allocating attention weights to the power nodes according to the electrical distance and the proximity centrality;
The extraction unit is used for inputting the power node data with the attention weight distributed into a three-dimensional convolution network with different time granularity to obtain the space-time dynamic characteristics of the power nodes; the space-time dynamic characteristic is that a dynamic rule of power of a power node under the comprehensive influence of time, electric distance and approaching centrality is obtained through three-dimensional convolution operation.
7. An area electric power prediction computer device, comprising: memory, a processor and a computer program stored in the memory, which processor, when executing the computer program, implements the steps of the regional electric power prediction method of any one of claims 1-5.
8. A readable storage medium, characterized in that the readable storage medium stores a computer program comprising program instructions for performing the steps of the regional electric power prediction method of any one of claims 1-5 when executed by a processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411139259.4A CN118657408B (en) | 2024-08-20 | 2024-08-20 | A regional electric power prediction method, system, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411139259.4A CN118657408B (en) | 2024-08-20 | 2024-08-20 | A regional electric power prediction method, system, device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118657408A CN118657408A (en) | 2024-09-17 |
| CN118657408B true CN118657408B (en) | 2024-11-26 |
Family
ID=92698114
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202411139259.4A Active CN118657408B (en) | 2024-08-20 | 2024-08-20 | A regional electric power prediction method, system, device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118657408B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116960952A (en) * | 2023-07-12 | 2023-10-27 | 山东大学 | A method and system for predicting photovoltaic power generation |
| CN117559415A (en) * | 2023-11-14 | 2024-02-13 | 重庆邮电大学 | Mobile virtual power plant dispatching system and method considering slope angle grid division |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2707498A1 (en) * | 2007-12-12 | 2009-06-18 | Alan Mcdonnell | Electric power distribution methods and apparatus |
| CN112487658B (en) * | 2020-12-14 | 2022-09-16 | 重庆邮电大学 | Method, device and system for identifying key nodes of power grid |
| CN117151261A (en) * | 2022-05-20 | 2023-12-01 | 华北电力大学（保定） | Ultra-short-term grid prediction method for distributed photovoltaic power plant cluster power |
| CN116245249A (en) * | 2023-03-15 | 2023-06-09 | 东南大学 | A deep learning-based online spatio-temporal forecasting method for power grids |
-
2024
- 2024-08-20 CN CN202411139259.4A patent/CN118657408B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116960952A (en) * | 2023-07-12 | 2023-10-27 | 山东大学 | A method and system for predicting photovoltaic power generation |
| CN117559415A (en) * | 2023-11-14 | 2024-02-13 | 重庆邮电大学 | Mobile virtual power plant dispatching system and method considering slope angle grid division |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118657408A (en) | 2024-09-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106529731A (en) | Regional power grid photovoltaic power station cluster division method | |
| CN112329997A (en) | Power demand load prediction method and system, electronic device, and storage medium | |
| Zhao et al. | Improved dynamic programming for reservoir flood control operation | |
| CN117713238B (en) | Random optimization operation strategy combining photovoltaic power generation and energy storage micro-grid | |
| CN118199061B (en) | Short-term power prediction method and system for renewable energy sources | |
| CN117744916A (en) | Energy storage capacity prediction methods, devices, computer equipment and readable storage media | |
| CN118863121A (en) | A distributed regional power generation prediction method based on stacked integrated model | |
| CN117216572A (en) | Wind-solar power generation load prediction method for multi-scene application | |
| CN120222412A (en) | A method for distinguishing inertia and inertia response characteristics in power system frequency dynamics | |
| CN114118839B (en) | A method and device for evaluating power grid resilience, electronic equipment, and storage medium | |
| CN118657408B (en) | A regional electric power prediction method, system, device and storage medium | |
| Rosato et al. | A Combined Deep Learning Approach for Time Series Prediction in Energy Environments | |
| CN118630747A (en) | Power load forecasting method, device, equipment and medium | |
| CN117350447A (en) | A multi-source heterogeneous power data fusion algorithm suitable for power grids | |
| CN117394308A (en) | A multi-time scale wind power power prediction method, system and electronic equipment | |
| CN110264010A (en) | Novel rural area electric power saturation load forecasting method | |
| CN115905360A (en) | Abnormal data measurement identification method and device based on random construction matrix | |
| CN117541019B (en) | A medium- and long-term robust optimal dispatching method, system and storage medium for cascade reservoirs | |
| CN120151690B (en) | Electric energy meter and concentrator capable of enhancing communication efficiency | |
| CN119294769B (en) | Deployment method, device and computer equipment of all-vanadium liquid flow battery energy storage system | |
| CN119227769B (en) | Model pruning method, device and storage medium of heterogeneous computing cluster | |
| CN120150112A (en) | Distributed energy distribution network prediction and coordinated allocation method and system | |
| CN119448238A (en) | Substation power supply optimization system, method, terminal equipment and storage medium based on multivariate data fusion analysis | |
| CN120513450A (en) | Photovoltaic power prediction method, device, electronic device and storage medium | |
| Luis | Framework for Scenario Generation and Reduction in Photovoltaic-Integrated Generation Commitment |
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 |