CN114971053A - Training method and device for online prediction model of network line loss rate of low-voltage transformer area - Google Patents

Training method and device for online prediction model of network line loss rate of low-voltage transformer area Download PDF

Info

Publication number
CN114971053A
CN114971053A CN202210635589.7A CN202210635589A CN114971053A CN 114971053 A CN114971053 A CN 114971053A CN 202210635589 A CN202210635589 A CN 202210635589A CN 114971053 A CN114971053 A CN 114971053A
Authority
CN
China
Prior art keywords
area
loss rate
line loss
space
node
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210635589.7A
Other languages
Chinese (zh)
Inventor
张洋瑞
陶鹏
申洪涛
韩桂楠
刘晓瑜
贾永良
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Corp of China SGCC
Marketing Service Center of State Grid Hebei Electric Power Co Ltd
Original Assignee
State Grid Corp of China SGCC
Marketing Service Center of State Grid Hebei Electric Power Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by State Grid Corp of China SGCC, Marketing Service Center of State Grid Hebei Electric Power Co Ltd filed Critical State Grid Corp of China SGCC
Priority to CN202210635589.7A priority Critical patent/CN114971053A/en
Publication of CN114971053A publication Critical patent/CN114971053A/en
Pending legal-status Critical Current

Links

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
    • 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
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • 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

Abstract

The invention provides a training method and a device of a low-voltage transformer area network line loss rate online prediction model, wherein the method comprises the following steps: screening target electrical characteristic quantity data which are strongly related to the line loss rate in each distribution area according to the collected historical line loss rate data and historical electrical characteristic quantity data which correspond to each distribution area in the low-voltage distribution area network; carrying out graph signal processing according to the target electrical characteristic quantity data of each transformer area and the acquired geographic information to obtain a space-time characteristic matrix of each transformer area; and updating the weight of the preset neural network model of each distribution area according to the output result of each space-time characteristic matrix after the preset neural network model of the corresponding distribution area is input and the historical line loss rate data of the corresponding distribution area, and obtaining the trained line loss rate online prediction model of each distribution area when the training stopping condition is met. The model training method provided by the invention can obtain a comprehensive and accurate online prediction model of the line loss rate of the transformer area, thereby realizing the accurate online prediction of the line loss rate of each transformer area in the low-voltage transformer area network.

Description

Training method and device for online prediction model of network line loss rate of low-voltage transformer area
Technical Field
The invention relates to the technical field of line loss rate prediction, in particular to a training method and a device of a low-voltage transformer area network line loss rate online prediction model.
Background
In recent years, economic construction of China is rapidly developed, urban construction is fast, power consumption requirements are gradually increased, and pressure is brought to power construction due to the increase of living electric appliances and working electric appliances of people. The line loss rate is an important index for measuring the technical economy of a power grid, plays an important role in evaluating the economic operation of a novel power market system, is also one of key management contents of a power enterprise, and comprehensively reflects the planning design, production operation and operation management level of the power system.
In recent years, the coverage rate of intelligent acquisition equipment in each provincial and urban area is greatly improved, the availability of power measurement data is improved, massive data provides support for operation and maintenance tasks such as line loss analysis and prediction, and the possibility of online prediction of the line loss rate of a power distribution network in a whole low-voltage distribution area is improved.
Disclosure of Invention
The embodiment of the invention provides a training method and a training device for a low-voltage transformer area network line loss rate online prediction model, which aim to solve the technical problems that in the prior art, the prediction of the transformer area line loss rate is limited to the prediction of the line loss rate of a certain transformer area, the prediction precision is low, the speed is low, and the accurate prediction and monitoring management of the line loss rate of each transformer area in the whole low-voltage transformer area network cannot be realized.
In a first aspect, an embodiment of the present invention provides a method for training an online prediction model of a network line loss rate of a low-voltage transformer area, including:
acquiring historical line loss rate data, historical electrical characteristic quantity data and geographic information corresponding to each distribution area in a low-voltage distribution area network;
screening out target electrical characteristic quantity data which are strongly related to the line loss rate in each transformer area according to the historical line loss rate data and the historical electrical characteristic quantity data which correspond to each transformer area;
carrying out graph signal processing according to the target electrical characteristic quantity data and the geographic information of each transformer area to obtain a space-time characteristic matrix of each transformer area in the low-voltage transformer area network;
and updating the weight of the preset neural network model of each distribution area according to the output result of each space-time characteristic matrix after the preset neural network model of the corresponding distribution area is input and the historical line loss rate data of the corresponding distribution area until the model training stopping condition is met, and obtaining the trained line loss rate online prediction model of each distribution area.
In a possible implementation manner, the performing graph signal processing according to the target electrical characteristic quantity data and the geographic information of each station area to obtain a space-time characteristic matrix of each station area in the low-voltage station area network includes:
constructing a station area node undirected graph of the low-voltage station area network according to the target electrical characteristic quantity data and the geographic information of each station area;
and carrying out direct space-time characteristic extraction and indirect space-time characteristic extraction on the basis of the graph signals of all the station area nodes in the station area node undirected graph to obtain a space-time characteristic matrix of each station area in the low-voltage station area network.
In a possible implementation manner, the obtaining a space-time feature matrix of each cell in the low-voltage cell network based on the direct space-time feature extraction and the indirect space-time feature extraction performed on the basis of the graph signal of each cell node in the cell node undirected graph includes:
for each station area node in the station area node undirected graph, marking the station area node directly connected with the station area node as a connected node, and performing direct space-time feature extraction on the corresponding connected node based on a graph signal of each connected node of the station area node to obtain an indirect space-time feature matrix of the station area node relative to each connected node;
and performing direct space-time feature extraction on the distribution area node based on the graph signal of the distribution area node and each indirect space-time feature matrix to obtain the space-time feature matrix of the distribution area node.
In a possible implementation manner, the marking a station area node directly connected to the station area node as a connected node, and performing direct space-time feature extraction on the corresponding connected node based on a graph signal of each connected node of the station area node to obtain an indirect space-time feature matrix of the station area node with respect to each connected node includes:
according to
Figure BDA0003680178710000031
Performing direct space-time feature extraction on corresponding connected nodes to obtain an indirect space-time feature matrix of the distribution room node relative to each connected node;
wherein H (m+1) An indirect space-time feature matrix representing the station area nodes relative to each connected node,
Figure BDA0003680178710000032
representing an adjacency matrix A and an identity matrix I N And, D represents
Figure BDA0003680178710000033
The degree matrix of (c) is,
Figure BDA0003680178710000034
pair of representations
Figure BDA0003680178710000035
Carrying out normalization treatment, W (m) Representing weight vectors of m station nodes directly connected to each connected node, m being a positive integer, σ representing a Relu activation function, H (m) Representing the graph signals of the connected nodes before the spatio-temporal feature extraction.
In a possible implementation manner, the performing, on the basis of the graph signal of the station area node and each indirect space-time feature matrix, direct space-time feature extraction on the station area node to obtain a space-time feature matrix of the station area node includes:
according to
Figure BDA0003680178710000036
Performing direct space-time feature extraction on the platform area node, and recording the direct space-time feature matrix as a space-time feature matrix of the platform area node;
wherein F represents the space-time characteristic matrix of the station area node,
Figure BDA0003680178710000037
presentation pair
Figure BDA0003680178710000038
Carrying out a normalization process H (m+1) Representing an indirect space-time characteristic matrix, σ, of the station area nodes with respect to each of the connected nodes 1 Denotes the Softmax activation function, W (n) And the weight vector represents n connected nodes directly connected with the station area node, and n is a positive integer.
In a possible implementation manner, the updating the weight of the preset neural network model of each distribution room according to the output result of each space-time feature matrix after the preset neural network model of the corresponding distribution room is input and the historical line loss rate data of the corresponding distribution room includes:
according to
Figure BDA0003680178710000039
Updating the weight of the preset neural network model of each transformer area;
wherein Q is t+1 Weight Q representing real-time update of the preset neural network model of each distribution area after the current time t t Representing the weight of the preset neural network model of each distribution area for current prediction, alpha representing the learning rate, m t Representing the weight variation, J (theta) is a loss function which is determined by the output result of each space-time characteristic matrix after being input into the preset neural network model of the corresponding transformer area and the historical line loss rate data of the corresponding transformer area,
Figure BDA0003680178710000041
in a possible implementation manner, after obtaining the trained line loss rate online prediction model for each cell, the method further includes:
s1: carrying out graph signal processing on the acquired real-time target electrical characteristic quantity data and the geographic information corresponding to each distribution area in the low-voltage distribution area network to obtain a time-space characteristic matrix of each distribution area in the low-voltage distribution area network, and inputting the time-space characteristic matrix of each distribution area into an online line loss rate prediction model of the corresponding distribution area to obtain an online line loss rate prediction result of each distribution area in the low-voltage distribution area network;
s2: and updating the weight of the online line loss rate prediction model of the corresponding distribution area based on the online line loss rate prediction result of each distribution area and the actual line loss rate data corresponding to each distribution area, and continuing to jump to S1 for execution according to the updated online line loss rate prediction model of each distribution area.
In a second aspect, an embodiment of the present invention provides a training apparatus for an online prediction model of a network line loss rate of a low-voltage transformer area, including:
the data acquisition module is used for acquiring historical line loss rate data, historical electrical characteristic quantity data and geographic information corresponding to each distribution area in the low-voltage distribution area network;
the characteristic screening module is used for screening target electrical characteristic quantity data which are strongly related to the line loss rate in each transformer area according to the historical line loss rate data and the historical electrical characteristic quantity data which correspond to each transformer area;
the characteristic extraction module is used for carrying out graph signal processing according to the target electrical characteristic quantity data and the geographic information of each distribution area to obtain a space-time characteristic matrix of each distribution area in the low-voltage distribution area network;
and the model training module is used for updating the weight of the preset neural network model of each station area according to the output result of each time-space characteristic matrix after the preset neural network model of the corresponding station area is input and the historical line loss rate data of the corresponding station area until the model training stopping condition is met, and obtaining the trained line loss rate online prediction model of each station area.
In a third aspect, an embodiment of the present invention provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements the steps of the method according to the first aspect or any possible implementation manner of the first aspect when executing the computer program.
In a fourth aspect, the present invention provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the computer program implements the steps of the method according to the first aspect or any one of the possible implementation manners of the first aspect.
The embodiment of the invention provides a method and a device for training a low-voltage transformer area network line loss rate online prediction model, which are characterized in that historical line loss rate data, historical electrical characteristic quantity data and geographic information corresponding to each transformer area in a low-voltage transformer area network are collected; screening target electrical characteristic quantity data which are strongly related to the line loss rate in each transformer area according to the historical line loss rate data and the historical electrical characteristic quantity data which correspond to each transformer area; then, carrying out graph signal processing according to the target electrical characteristic quantity data and the geographic information of each transformer area to obtain a space-time characteristic matrix of each transformer area in the low-voltage transformer area network; and finally, updating the weight of the preset neural network model of each distribution area according to the output result of each space-time characteristic matrix after the preset neural network model of the corresponding distribution area is input and the historical line loss rate data of the corresponding distribution area until the model training stopping condition is met, and obtaining the trained line loss rate online prediction model of each distribution area. The method for training the online prediction model of the line loss rate of the low-voltage transformer area network provided by the embodiment of the invention can obtain a comprehensive and accurate online prediction model of the line loss rate of each transformer area in the low-voltage transformer area network, further realize the accurate online prediction of the line loss rate of each transformer area node in the whole low-voltage transformer area network, realize the monitoring and management of the line loss rate of the whole network transformer area, be beneficial to enhancing the management strength of the line loss work by a power department, improve the inspection speed of the line loss rate of the low-voltage transformer area network and ensure the smooth and stable operation of a power grid.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the embodiments or the prior art descriptions will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise.
Fig. 1 is a flowchart illustrating an implementation of a training method for a low-voltage transformer area network line loss rate online prediction model according to an embodiment of the present invention;
fig. 2 is an undirected graph illustration of a distribution room node of a low-voltage distribution room network according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a training apparatus of an online low-voltage transformer area network line loss rate prediction model according to an embodiment of the present invention;
fig. 4 is a schematic diagram of an electronic device provided in an embodiment of the present invention.
Detailed Description
In the following description, for purposes of explanation and not limitation, specific details are set forth, such as particular system structures, techniques, etc. in order to provide a thorough understanding of the embodiments of the invention. It will be apparent, however, to one skilled in the art that the present invention may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.
In order to make the objects, technical solutions and advantages of the present invention more apparent, the following description is made by way of specific embodiments with reference to the accompanying drawings.
Referring to fig. 1, fig. 1 is a flowchart illustrating an implementation of a method for training a low-voltage network line loss rate online prediction model according to an embodiment of the present invention, where the method includes:
s101: and acquiring historical line loss rate data, historical electrical characteristic quantity data and geographic information corresponding to each distribution area in the low-voltage distribution area network.
In this embodiment, historical line loss rate data, historical electrical characteristic quantity data, and geographic information corresponding to each distribution area in the low-voltage distribution area network can be collected; illustratively, the collected historical electrical characteristic quantity data of each station zone may include: active power, reactive power, transformer loss, power supply radius, total line length, three-phase load unbalance loss, load factor and the like, which are not limited in the application.
S102: and screening out target electrical characteristic quantity data which are strongly related to the line loss rate in each transformer area according to the historical line loss rate data and the historical electrical characteristic quantity data which correspond to each transformer area.
In this embodiment, can be according to
Figure BDA0003680178710000071
And calculating the correlation between the historical electrical characteristic quantity data and the historical line loss rate data corresponding to each station area. For example, a correlation threshold may be set to 0.7, and target electrical characteristic quantity data with line loss rate correlation higher than the set threshold corresponding to each distribution area is screened out, where R is the correlation between the historical electrical characteristic quantity data and the historical line loss rate data, N is the length of the sequence, and R is the length of the sequence k And S k Respectively a sequence of historical electrical characteristic quantity data and a sequence of historical line loss rate data,
Figure BDA0003680178710000072
are each R k Sequence and S k Average of the sequences.
Further, in this embodiment, before the target electrical characteristic quantity data, which is strongly related to the line loss rate, in each cell is screened out according to the historical line loss rate data and the historical electrical characteristic quantity data corresponding to each cell, the historical electrical characteristic quantity data corresponding to each cell may be preprocessed.
The preprocessing of the historical electrical characteristic quantity data corresponding to each distribution area comprises the following steps: missing data in historical electrical characteristic quantity data is filled, a softlmpute module special for filling missing values in fancyimpute can be used for filling the data, and the softlmpute module is used for filling the data through iterative soft threshold processing of SVD (singular value decomposition), so that the data are adaptive to a model and are matched with the requirements of the model; other methods may be used to fill in missing data in the historical electrical characteristic quantity data, which is not limited in this application.
In this embodiment, missing data in the historical electrical characteristic quantity data of each distribution room is filled before the target electrical characteristic quantity data is screened, so that relatively comprehensive and accurate historical electrical characteristic quantity data can be obtained, and comprehensive and accurate historical electrical characteristic quantity data is provided for the subsequent training of the line loss rate prediction model of each distribution room.
S103: and carrying out graph signal processing according to the target electrical characteristic quantity data and the geographic information of each transformer area to obtain a space-time characteristic matrix of each transformer area in the low-voltage transformer area network.
In this embodiment, the map signal processing may be performed on the target electrical characteristic quantity data and the geographic information of each station area to obtain a space-time characteristic matrix of each station area in the low-voltage station area network.
Optionally, as a specific implementation manner of the training method for the online low-voltage transformer area network line loss rate prediction model provided in the embodiment of the present invention, the method performs graph signal processing according to the target electrical characteristic quantity data and the geographic information of each transformer area to obtain a time-space characteristic matrix of each transformer area in the low-voltage transformer area network, and includes:
and constructing a station area node undirected graph of the low-voltage station area network according to the target electrical characteristic quantity data and the geographic information of each station area.
And carrying out direct space-time characteristic extraction and indirect space-time characteristic extraction on the basis of the graph signals of all the station area nodes in the station area node undirected graph to obtain a space-time characteristic matrix of each station area in the low-voltage station area network.
In this embodiment, a topology structure of a station area in a low-voltage station area network is described by constructing a station area node undirected graph G ═ V, E, W of the low-voltage station area network according to target electrical characteristic quantity data and geographic information of each station area, in the station area node undirected graph, V is a set of station area nodes in the low-voltage station area network, the set includes N station area nodes, each node includes target electrical characteristic quantity data of a corresponding station area, E is a set of edges in the station area node undirected graph and is used for describing an adjacency relation between the nodes, and W is a trained attention weight matrix and is used for describing a spatial dependency relation between the nodes. And (3) carrying out direct space-time feature extraction and indirect space-time feature extraction on the basis of the graph signals of all the station nodes in the station node undirected graph G (V, E, W), and obtaining a space-time feature matrix of each station in the low-voltage station network. Referring to fig. 2, fig. 2 is a schematic diagram of a distribution area node undirected graph of a low-voltage distribution area network according to an embodiment of the present invention, which is only an example and is not intended to limit the present invention.
Optionally, as a specific implementation manner of the training method for the online prediction model of the network line loss rate of the low-voltage distribution room provided in the embodiment of the present invention, the method for obtaining the space-time feature matrix of each distribution room in the low-voltage distribution room network based on the graph signals of each distribution room node in the undirected graph of the distribution room nodes performs direct space-time feature extraction and indirect space-time feature extraction, and includes:
and aiming at each station area node in the station area node undirected graph, marking the station area nodes directly connected with the station area node as connected nodes, and performing direct space-time feature extraction on the corresponding connected nodes based on the graph signal of each connected node of the station area node to obtain an indirect space-time feature matrix of the station area node relative to each connected node.
And carrying out direct space-time feature extraction on the station area node based on the graph signal of the station area node and each indirect space-time feature matrix to obtain the space-time feature matrix of the station area node.
In the embodiment, the space-time feature extraction operation is performed twice on each station area node in the low-voltage station area network, so that the feature relationship among the station area nodes in the low-voltage station area network can be fully excavated, the space-time feature of each station area node can be fully extracted, and data support is provided for training an accurate online low-voltage station area network line loss rate prediction model.
Optionally, as a specific implementation manner of the training method for the low-voltage transformer area network line loss rate online prediction model provided in the embodiment of the present invention, the transformer area nodes directly connected to the transformer area nodes are marked as connected nodes, and direct spatio-temporal feature extraction is performed on the corresponding connected nodes based on a graph signal of each connected node of the transformer area nodes, so as to obtain an indirect spatio-temporal feature matrix of the transformer area nodes relative to each connected node, where the method includes:
according to
Figure BDA0003680178710000091
And (3) performing direct space-time feature extraction on the corresponding connected nodes to obtain an indirect space-time feature matrix of the distribution area node relative to each connected node.
Wherein H (m+1) An indirect space-time feature matrix representing the station area nodes relative to each connected node,
Figure BDA0003680178710000092
representing an adjacency matrix A and an identity matrix I N And, D represents
Figure BDA0003680178710000093
The degree matrix of (c) is,
Figure BDA0003680178710000094
presentation pair
Figure BDA0003680178710000095
Carrying out normalization treatment, W (m) Representing m zones directly connected to each connected nodeWeight vector of node, m is positive integer, σ represents Relu activation function, H (m) Representing the graph signal of each connected node before spatio-temporal feature extraction.
In this embodiment, the adjacency matrix a is an N × N matrix constructed from the relationship between N station nodes in the station node undirected graph G ═ V, E, W. H (m) And a graph signal representing each connected node of the station area node before the time-space feature extraction is not carried out, wherein the graph signal comprises target electrical feature quantity data which are strongly related to the line loss rate in the screened station area. The normalized adjacency matrix and the matrix H are combined (m) As input to a first direct spatio-temporal feature extraction operation, based on
Figure BDA0003680178710000096
And calculating a direct space-time characteristic matrix of the connected node layer of the station area node, and marking the direct space-time characteristic matrix as an indirect space-time characteristic matrix of the station area node relative to each connected node.
Optionally, as a specific implementation manner of the training method for the online prediction model of the network line loss rate of the low-voltage transformer area provided by the embodiment of the present invention, the method for performing direct spatio-temporal feature extraction on the transformer area node based on the graph signal of the transformer area node and each indirect spatio-temporal feature matrix to obtain the spatio-temporal feature matrix of the transformer area node includes:
according to
Figure BDA0003680178710000101
And performing direct space-time feature extraction on the station area node, and recording the direct space-time feature matrix as a space-time feature matrix of the station area node.
Wherein F represents the space-time characteristic matrix of the station area node,
Figure BDA0003680178710000102
presentation pair
Figure BDA0003680178710000103
Carrying out a normalization process H (m+1) Indicating that the station node is connected with respect to each connectionIndirect space-time characteristic matrix of nodes, σ 1 Denotes the Softmax activation function, W (n) And the weight vector represents n connected nodes directly connected with the station area node, and n is a positive integer.
In this embodiment, the normalized adjacency matrix and the indirect space-time characteristic matrix H of the station area node with respect to each connected node are used (m+1) As input for a second direct spatio-temporal feature extraction operation, based on
Figure BDA0003680178710000104
And calculating a direct space-time characteristic matrix of the station area node at the moment, and recording the direct space-time characteristic matrix as the space-time characteristic matrix of the station area node.
In this embodiment, after the spatio-temporal features of the connected nodes are updated, the direct spatio-temporal feature extraction operation is performed on the station area node based on the updated spatio-temporal features of the connected nodes by performing the direct spatio-temporal feature extraction operation on the connected nodes of the station area node. The method has the advantages that the space-time feature extraction operation is performed twice for each station area node in the low-voltage station area network, the feature relation among the station area nodes in the low-voltage station area network can be fully excavated, the space-time feature of each station area node in the whole low-voltage station area network is fully extracted, and data support is provided for training an accurate online prediction model of the line loss rate of the low-voltage station area network.
S104: and updating the weight of the preset neural network model of each distribution area according to the output result of each space-time characteristic matrix after the preset neural network model of the corresponding distribution area is input and the historical line loss rate data of the corresponding distribution area until the model training stopping condition is met, and obtaining the trained line loss rate online prediction model of each distribution area.
Optionally, as a specific implementation manner of the training method for the online low-voltage transformer area network line loss rate prediction model provided in the embodiment of the present invention, updating the weight of the preset neural network model of each transformer area according to the output result of each spatio-temporal feature matrix after the preset neural network model of the corresponding transformer area is input and the historical line loss rate data of the corresponding transformer area, where the method includes:
according to
Figure BDA0003680178710000111
And updating the weight of the preset neural network model of each station area.
Wherein Q is t+1 Weight Q representing real-time update of the preset neural network model of each distribution area after the current time t t Representing the weight of the preset neural network model of each distribution area for current prediction, alpha representing the learning rate, m t Representing the weight variation, J (theta) is a loss function which is determined by the output result of each space-time characteristic matrix after being input into the preset neural network model of the corresponding transformer area and the historical line loss rate data of the corresponding transformer area,
Figure BDA0003680178710000112
in this embodiment, the preset neural network model may be a convolutional neural network model, a Long Short Term Memory (LSTM) model, a cyclic neural network model, and the like, which is not limited in this application.
In the present example, by introducing m t And updating the gradient of the preset neural network model corresponding to each distribution area in the low-voltage distribution area network on line, thereby realizing the on-line updating of the weight of the preset neural network model corresponding to each distribution area.
Optionally, as a specific implementation manner of the training method for the online prediction model of the network line loss rate of the low-voltage transformer area provided in the embodiment of the present invention, after obtaining the trained online prediction model of the line loss rate of each transformer area, the method further includes:
s1: and carrying out graph signal processing on the acquired real-time target electrical characteristic quantity data and geographic information corresponding to each distribution area in the low-voltage distribution area network to obtain a space-time characteristic matrix of each distribution area in the low-voltage distribution area network, and inputting the space-time characteristic matrix of each distribution area into the online line loss rate prediction model of the corresponding distribution area to obtain an online line loss rate prediction result of each distribution area in the low-voltage distribution area network.
S2: and updating the weight of the online line loss rate prediction model of the corresponding distribution area based on the online line loss rate prediction result of each distribution area and the actual line loss rate data corresponding to each distribution area, and continuing to jump to S1 for execution according to the updated online line loss rate prediction model of each distribution area.
In this embodiment, based on a trained online prediction model of the line loss rate of each cell in the low-voltage cell network, real-time target electrical characteristic quantity data and geographic information corresponding to each cell in the low-voltage cell network are acquired, graph signals are processed on the data to obtain a space-time characteristic matrix of each cell, and the space-time characteristic matrix of each cell is input into the online prediction model of the line loss rate of the corresponding cell to obtain an online prediction result of the line loss rate of each cell; and updating the weight of the corresponding online line loss rate prediction model of the transformer area according to the obtained online line loss rate prediction result of each transformer area and the actual line loss rate data corresponding to each transformer area. And after the online line loss rate prediction model of each distribution area is updated, the online line loss rate prediction of each distribution area in the low-voltage distribution area network is carried out next time.
In this embodiment, each time the line loss rate result of each cell in the low-voltage cell network is predicted, the weight of the line loss rate online prediction model of the corresponding cell is updated according to the predicted line loss rate result of each cell and the actual line loss rate corresponding to each cell. By continuously updating the weight of the online line loss rate prediction model of each distribution area in the low-voltage distribution area network, a more accurate online line loss rate prediction model of the distribution area is obtained, and then accurate online line loss rate prediction of each distribution area node in the whole low-voltage distribution area network is realized.
The embodiment of the invention provides a method and a device for training a low-voltage transformer area network line loss rate online prediction model, which are characterized in that historical line loss rate data, historical electrical characteristic quantity data and geographic information corresponding to each transformer area in a low-voltage transformer area network are collected; screening target electrical characteristic quantity data which are strongly related to the line loss rate in each transformer area according to the historical line loss rate data and the historical electrical characteristic quantity data which correspond to each transformer area; then, carrying out graph signal processing according to the target electrical characteristic quantity data and the geographic information of each transformer area to obtain a space-time characteristic matrix of each transformer area in the low-voltage transformer area network; and finally, updating the weight of the preset neural network model of each distribution area according to the output result of each space-time characteristic matrix after the preset neural network model of the corresponding distribution area is input and the historical line loss rate data of the corresponding distribution area until the model training stopping condition is met, and obtaining the trained line loss rate online prediction model of each distribution area. The method for training the online prediction model of the line loss rate of the low-voltage transformer area network provided by the embodiment of the invention can obtain a comprehensive and accurate online prediction model of the line loss rate of each transformer area in the low-voltage transformer area network, further realize the accurate online prediction of the line loss rate of each transformer area node in the whole low-voltage transformer area network, realize the monitoring and management of the line loss rate of the whole network transformer area, be beneficial to enhancing the management strength of the line loss work by a power department, improve the inspection speed of the line loss rate of the low-voltage transformer area network and ensure the smooth and stable operation of a power grid.
It should be understood that, the sequence numbers of the steps in the foregoing embodiments do not imply an execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.
The following are embodiments of the apparatus of the invention, reference being made to the corresponding method embodiments described above for details which are not described in detail therein.
Fig. 3 is a schematic structural diagram of a training apparatus for a low-voltage station area network line loss rate online prediction model according to an embodiment of the present invention, which, for convenience of description, only shows parts related to the embodiment of the present invention, and the details are as follows:
as shown in fig. 3, the training apparatus for the online prediction model of the network line loss rate of the low-voltage transformer area provided in the embodiment of the present invention includes:
the data acquisition module 301 is configured to acquire historical line loss rate data, historical electrical characteristic quantity data, and geographic information corresponding to each distribution area in the low-voltage distribution area network.
The feature screening module 302 is configured to screen out target electrical feature data, which is strongly related to the line loss rate, in each distribution area according to the historical line loss rate data and the historical electrical feature data corresponding to each distribution area.
And the feature extraction module 303 is configured to perform graph signal processing according to the target electrical feature data and the geographic information of each distribution area to obtain a space-time feature matrix of each distribution area in the low-voltage distribution area network.
And the model training module 304 is configured to update the weight of the preset neural network model of each distribution room according to the output result of each space-time feature matrix after the preset neural network model of the corresponding distribution room is input and the historical line loss rate data of the corresponding distribution room, until a model training stop condition is met, to obtain a trained line loss rate online prediction model of each distribution room.
The embodiment of the invention collects the historical line loss rate data, the historical electrical characteristic quantity data and the geographic information corresponding to each distribution area in the low-voltage distribution area network; screening target electrical characteristic quantity data which are strongly related to the line loss rate in each transformer area according to the historical line loss rate data and the historical electrical characteristic quantity data which correspond to each transformer area; then, carrying out graph signal processing according to the target electrical characteristic quantity data and the geographic information of each transformer area to obtain a space-time characteristic matrix of each transformer area in the low-voltage transformer area network; and finally, updating the weight of the preset neural network model of each distribution area according to the output result of each space-time characteristic matrix after the preset neural network model of the corresponding distribution area is input and the historical line loss rate data of the corresponding distribution area until the model training stopping condition is met, and obtaining the trained line loss rate online prediction model of each distribution area. The training device of the online prediction model of the line loss rate of the low-voltage transformer area network provided by the embodiment of the invention can obtain a comprehensive and accurate online prediction model of the line loss rate of each transformer area in the low-voltage transformer area network, further realize the accurate online prediction of the line loss rate of each transformer area node in the whole low-voltage transformer area network, realize the monitoring and management of the line loss rate of the whole network transformer area, be beneficial to enhancing the management strength of the line loss work by the power department, improve the inspection speed of the line loss rate of the low-voltage transformer area network and ensure the smooth and stable operation of a power grid.
Optionally, as a specific implementation manner of the training device for the online low-voltage transformer area network line loss rate prediction model provided in the embodiment of the present invention, the device may further include: and the preprocessing module is used for preprocessing the historical electrical characteristic quantity data corresponding to each distribution area.
The preprocessing of the historical electrical characteristic quantity data corresponding to each distribution area comprises the following steps: missing data in historical electrical characteristic quantity data is filled, a softlmpute module special for filling missing values in fancyimpute can be used for filling the data, and the softlmpute module is used for filling the data through iterative soft threshold processing of SVD (singular value decomposition), so that the data are adaptive to a model and are matched with the requirements of the model; other methods may be used to fill in missing data in the historical electrical characteristic quantity data, which is not limited in this application.
Optionally, as a specific implementation manner of the training device for the online low-voltage transformer area network line loss rate prediction model provided in the embodiment of the present invention, the feature extraction module 303 performs graph signal processing according to the target electrical feature data and the geographic information of each transformer area to obtain a space-time feature matrix of each transformer area in the low-voltage transformer area network, and is specifically configured to:
and constructing a station area node undirected graph of the low-voltage station area network according to the target electrical characteristic quantity data and the geographic information of each station area.
And carrying out direct space-time characteristic extraction and indirect space-time characteristic extraction on the basis of the graph signals of all the station area nodes in the station area node undirected graph to obtain a space-time characteristic matrix of each station area in the low-voltage station area network.
Optionally, as a specific implementation manner of the training device for the online low-voltage transformer area network line loss rate prediction model provided in the embodiment of the present invention, the feature extraction module 303 performs direct space-time feature extraction and indirect space-time feature extraction based on a graph signal of each transformer area node in a transformer area node undirected graph, to obtain a space-time feature matrix of each transformer area in the low-voltage transformer area network, and is specifically configured to:
and aiming at each station area node in the station area node undirected graph, marking the station area nodes directly connected with the station area node as connected nodes, and performing direct space-time feature extraction on the corresponding connected nodes based on the graph signal of each connected node of the station area node to obtain an indirect space-time feature matrix of the station area node relative to each connected node.
And carrying out direct space-time feature extraction on the station area node based on the graph signal of the station area node and each indirect space-time feature matrix to obtain the space-time feature matrix of the station area node.
Optionally, as a specific implementation manner of the training apparatus for the low-voltage transformer area network line loss rate online prediction model provided in the embodiment of the present invention, the feature extraction module 303 marks a transformer area node directly connected to the transformer area node as a connected node, and performs direct space-time feature extraction on the corresponding connected node based on a graph signal of each connected node of the transformer area node to obtain an indirect space-time feature matrix of the transformer area node relative to each connected node, and is specifically configured to:
according to
Figure BDA0003680178710000151
And (4) carrying out direct space-time feature extraction on corresponding connected nodes to obtain an indirect space-time feature matrix of the distribution room node relative to each connected node.
Wherein H (m+1) An indirect space-time feature matrix representing the station area nodes relative to each connected node,
Figure BDA0003680178710000152
representing an adjacency matrix A and an identity matrix I N And, D table
Figure BDA0003680178710000153
The degree matrix is shown as a function of,
Figure BDA0003680178710000154
presentation pair
Figure BDA0003680178710000155
Carrying out normalization treatment, W (m) Representing weight vectors of m station nodes directly connected to each connected node, m being a positive integer, σ representing a Relu activation function, H (m) Representing the graph signals of the connected nodes before the spatio-temporal feature extraction.
Optionally, as a specific implementation manner of the training device for the low-voltage transformer area network line loss rate online prediction model provided in the embodiment of the present invention, the feature extraction module 303 performs direct space-time feature extraction on the transformer area node based on the graph signal of the transformer area node and each indirect space-time feature matrix to obtain the space-time feature matrix of the transformer area node, and is specifically configured to:
according to
Figure BDA0003680178710000156
And performing direct space-time feature extraction on the station area node, and recording the direct space-time feature matrix as a space-time feature matrix of the station area node.
Wherein F represents the space-time characteristic matrix of the station area node,
Figure BDA0003680178710000157
presentation pair
Figure BDA0003680178710000158
Carrying out a normalization process H (m+1) Representing an indirect space-time characteristic matrix, σ, of the station area nodes with respect to each of the connected nodes 1 Denotes the Softmax activation function, W (n) And the weight vector represents n connected nodes directly connected with the station area node, and n is a positive integer.
Optionally, as a specific implementation manner of the training apparatus for the low-voltage transformer area network line loss rate online prediction model provided in the embodiment of the present invention, the model training module 304 updates the weight of the preset neural network model of each transformer area according to the output result of each spatio-temporal feature matrix after the preset neural network model of the corresponding transformer area is input and the historical line loss rate data of the corresponding transformer area, and is specifically configured to:
according to
Figure BDA0003680178710000161
And updating the weight of the preset neural network model of each station area.
Wherein Q is t+1 Weight Q representing real-time update of the preset neural network model of each distribution area after the current time t t Representing individual zonesPresetting the weight of the current prediction of the neural network model, wherein alpha represents the learning rate and m t Representing the weight variation, J (theta) is a loss function which is determined by the output result of each space-time characteristic matrix after being input into the preset neural network model of the corresponding transformer area and the historical line loss rate data of the corresponding transformer area,
Figure BDA0003680178710000162
optionally, as a specific implementation manner of the training device for the low-voltage station area network line loss rate online prediction model provided in the embodiment of the present invention, after obtaining the trained line loss rate online prediction model of each station area, the model training module 304 is further specifically configured to:
s1: and carrying out graph signal processing on the acquired real-time target electrical characteristic quantity data and geographic information corresponding to each distribution area in the low-voltage distribution area network to obtain a space-time characteristic matrix of each distribution area in the low-voltage distribution area network, and inputting the space-time characteristic matrix of each distribution area into the online line loss rate prediction model of the corresponding distribution area to obtain an online line loss rate prediction result of each distribution area in the low-voltage distribution area network.
S2: and updating the weight of the online line loss rate prediction model of the corresponding distribution area based on the online line loss rate prediction result of each distribution area and the actual line loss rate data corresponding to each distribution area, and continuing to jump to S1 for execution according to the updated online line loss rate prediction model of each distribution area.
Fig. 4 is a schematic diagram of an electronic device provided in an embodiment of the present invention. As shown in fig. 4, the electronic apparatus 4 of this embodiment includes: a processor 40, a memory 41 and a computer program 42 stored in said memory 41 and executable on said processor 40. The processor 40 executes the computer program 42 to implement the steps in the above-mentioned embodiments of the training method for the network line loss rate online prediction model of the low-voltage transformer area, such as the steps 101 to 104 shown in fig. 1. Alternatively, the processor 40, when executing the computer program 42, implements the functions of the modules/units in the above-mentioned device embodiments, such as the functions of the modules 301 to 304 shown in fig. 3.
Illustratively, the computer program 42 may be partitioned into one or more modules/units that are stored in the memory 41 and executed by the processor 40 to implement the present invention. The one or more modules/units may be a series of computer program instruction segments capable of performing specific functions, which are used to describe the execution of the computer program 42 in the electronic device 4. For example, the computer program 42 may be divided into the modules 301 to 304 shown in fig. 3.
The electronic device 4 may be a desktop computer, a notebook, a palm computer, a cloud server, or other computing devices. The electronic device 4 may include, but is not limited to, a processor 40, a memory 41. Those skilled in the art will appreciate that fig. 4 is merely an example of an electronic device 4 and does not constitute a limitation of the electronic device 4 and may include more or fewer components than shown, or some components may be combined, or different components, e.g., the electronic device may also include input-output devices, network access devices, buses, etc.
The Processor 40 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory 41 may be an internal storage unit of the electronic device 4, such as a hard disk or a memory of the electronic device 4. The memory 41 may also be an external storage device of the electronic device 4, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like, provided on the electronic device 4. Further, the memory 41 may also include both an internal storage unit and an external storage device of the electronic device 4. The memory 41 is used for storing the computer program and other programs and data required by the electronic device. The memory 41 may also be used to temporarily store data that has been output or is to be output.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-mentioned functions. Each functional unit and module in the embodiments may be integrated in one processing unit, or each unit may exist alone physically, or two or more units are integrated in one unit, and the integrated unit may be implemented in a form of hardware, or in a form of software functional unit. In addition, specific names of the functional units and modules are only used for distinguishing one functional unit from another, and are not used for limiting the protection scope of the present application. The specific working processes of the units and modules in the system may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and reference may be made to the related descriptions of other embodiments for parts that are not described or illustrated in a certain embodiment.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
In the embodiments provided in the present invention, it should be understood that the disclosed apparatus/electronic device and method may be implemented in other ways. For example, the above-described apparatus/electronic device embodiments are merely illustrative, and for example, the division of the modules or units is only one logical division, and there may be other divisions when actually implemented, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit may be implemented in the form of hardware, or may also be implemented in the form of a software functional unit.
The integrated modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the method according to the embodiments of the present invention may also be implemented by a computer program, which may be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the embodiments of the method for training the network line loss rate online prediction model of the low-voltage distribution area may be implemented. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain suitable additions or subtractions depending on the requirements of legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable media may not include electrical carrier signals or telecommunication signals in accordance with legislation and patent practice.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present invention, and are intended to be included within the scope of the present invention.

Claims (10)

1. A training method of a low-voltage transformer area network line loss rate online prediction model is characterized by comprising the following steps:
acquiring historical line loss rate data, historical electrical characteristic quantity data and geographic information corresponding to each distribution area in a low-voltage distribution area network;
screening out target electrical characteristic quantity data which are strongly related to the line loss rate in each transformer area according to the historical line loss rate data and the historical electrical characteristic quantity data which correspond to each transformer area;
carrying out graph signal processing according to the target electrical characteristic quantity data and the geographic information of each transformer area to obtain a space-time characteristic matrix of each transformer area in the low-voltage transformer area network;
and updating the weight of the preset neural network model of each distribution area according to the output result of each space-time characteristic matrix after the preset neural network model of the corresponding distribution area is input and the historical line loss rate data of the corresponding distribution area until the model training stopping condition is met, and obtaining the trained line loss rate online prediction model of each distribution area.
2. The method for training the on-line prediction model of the low-voltage transformer area network line loss rate according to claim 1, wherein the step of performing graph signal processing according to the target electrical characteristic quantity data and the geographic information of each transformer area to obtain the space-time characteristic matrix of each transformer area in the low-voltage transformer area network comprises the following steps:
constructing a station area node undirected graph of the low-voltage station area network according to the target electrical characteristic quantity data and the geographic information of each station area;
and performing direct space-time characteristic extraction and indirect space-time characteristic extraction on the basis of the graph signals of all the station area nodes in the station area node undirected graph to obtain a space-time characteristic matrix of each station area in the low-voltage station area network.
3. The method for training the low-voltage transformer area network line loss rate online prediction model according to claim 2, wherein the obtaining of the space-time feature matrix of each transformer area in the low-voltage transformer area network by performing direct space-time feature extraction and indirect space-time feature extraction based on the graph signal of each transformer area node in the undirected graph of transformer area nodes comprises:
for each station area node in the station area node undirected graph, marking the station area node directly connected with the station area node as a connected node, and performing direct space-time feature extraction on the corresponding connected node based on a graph signal of each connected node of the station area node to obtain an indirect space-time feature matrix of the station area node relative to each connected node;
and carrying out direct space-time feature extraction on the station area node based on the graph signal of the station area node and each indirect space-time feature matrix to obtain the space-time feature matrix of the station area node.
4. The method for training the low-voltage transformer area network line loss rate online prediction model as claimed in claim 3, wherein the step of marking the transformer area nodes directly connected with the transformer area nodes as connected nodes and performing direct space-time feature extraction on the corresponding connected nodes based on the graph signal of each connected node of the transformer area nodes to obtain the indirect space-time feature matrix of the transformer area nodes relative to each connected node comprises:
according to
Figure FDA0003680178700000021
Performing direct space-time feature extraction on corresponding connected nodes to obtain an indirect space-time feature matrix of the distribution room node relative to each connected node;
wherein H (m+1) An indirect space-time feature matrix representing the station area nodes relative to each connected node,
Figure FDA0003680178700000022
Figure FDA0003680178700000023
representing an adjacency matrix A and an identity matrix I N And, D represents
Figure FDA0003680178700000024
The degree matrix of (c) is,
Figure FDA0003680178700000025
presentation pair
Figure FDA0003680178700000026
Carrying out normalization treatment, W (m) Representing weight vectors of m station nodes directly connected to each connected node, m being a positive integer, σ representing a Relu activation function, H (m) Representing the graph signal of each connected node before spatio-temporal feature extraction.
5. The method for training the low-voltage transformer area network line loss rate online prediction model as claimed in claim 4, wherein the step of performing direct space-time feature extraction on the transformer area node based on the graph signal of the transformer area node and each indirect space-time feature matrix to obtain the space-time feature matrix of the transformer area node comprises:
according to
Figure FDA0003680178700000027
Performing direct space-time feature extraction on the platform area node, and recording the direct space-time feature matrix as a space-time feature matrix of the platform area node;
wherein F represents the space-time characteristic matrix of the station area node,
Figure FDA0003680178700000028
presentation pair
Figure FDA0003680178700000029
Carrying out a normalization process H (m+1) Representing an indirect space-time characteristic matrix, σ, of the station area nodes with respect to each of the connected nodes 1 Denotes the Softmax activation function, W (n) And the weight vector represents n connected nodes directly connected with the platform area node, and n is a positive integer.
6. The method for training the low-voltage transformer area network line loss rate online prediction model according to any one of claims 1 to 5, wherein the updating the weight of the preset neural network model of each transformer area according to the output result of each spatio-temporal feature matrix after the preset neural network model of the corresponding transformer area is input and the historical line loss rate data of the corresponding transformer area comprises:
according to
Figure FDA0003680178700000031
Updating the weight of the preset neural network model of each transformer area;
wherein Q is t+1 Weight Q representing real-time update of the preset neural network model of each distribution area after the current time t t Representing the weight of the preset neural network model of each distribution area for current prediction, alpha representing the learning rate, m t The amount of change in the weight is represented,j (theta) is a loss function which is determined by the output result of each space-time characteristic matrix after being input into the preset neural network model of the corresponding distribution room and the historical line loss rate data of the corresponding distribution room,
Figure FDA0003680178700000032
7. the method for training the low-voltage network line loss rate online prediction model according to any one of claims 1-5, further comprising, after obtaining the trained line loss rate online prediction model for each cell:
s1: carrying out graph signal processing on the acquired real-time target electrical characteristic quantity data and the geographic information corresponding to each distribution area in the low-voltage distribution area network to obtain a time-space characteristic matrix of each distribution area in the low-voltage distribution area network, and inputting the time-space characteristic matrix of each distribution area into an online line loss rate prediction model of the corresponding distribution area to obtain an online line loss rate prediction result of each distribution area in the low-voltage distribution area network;
s2: and updating the weight of the line loss rate online prediction model of the corresponding distribution area based on the line loss rate online prediction result of each distribution area and the actual line loss rate data corresponding to each distribution area, and continuing to jump to S1 for execution according to the updated line loss rate online prediction model of each distribution area.
8. A training device of a low-voltage transformer area network line loss rate online prediction model is characterized by comprising:
the data acquisition module is used for acquiring historical line loss rate data, historical electrical characteristic quantity data and geographic information corresponding to each distribution area in the low-voltage distribution area network;
the characteristic screening module is used for screening target electrical characteristic quantity data which are strongly related to the line loss rate in each transformer area according to the historical line loss rate data and the historical electrical characteristic quantity data which correspond to each transformer area;
the characteristic extraction module is used for carrying out graph signal processing according to the target electrical characteristic quantity data and the geographic information of each distribution area to obtain a space-time characteristic matrix of each distribution area in the low-voltage distribution area network;
and the model training module is used for updating the weight of the preset neural network model of each station area according to the output result of each time-space characteristic matrix after the preset neural network model of the corresponding station area is input and the historical line loss rate data of the corresponding station area until the model training stopping condition is met, and obtaining the trained line loss rate online prediction model of each station area.
9. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the steps of the method according to any of the preceding claims 1 to 7 are implemented when the computer program is executed by the processor.
10. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 7.
CN202210635589.7A 2022-06-06 2022-06-06 Training method and device for online prediction model of network line loss rate of low-voltage transformer area Pending CN114971053A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210635589.7A CN114971053A (en) 2022-06-06 2022-06-06 Training method and device for online prediction model of network line loss rate of low-voltage transformer area

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210635589.7A CN114971053A (en) 2022-06-06 2022-06-06 Training method and device for online prediction model of network line loss rate of low-voltage transformer area

Publications (1)

Publication Number Publication Date
CN114971053A true CN114971053A (en) 2022-08-30

Family

ID=82960013

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210635589.7A Pending CN114971053A (en) 2022-06-06 2022-06-06 Training method and device for online prediction model of network line loss rate of low-voltage transformer area

Country Status (1)

Country Link
CN (1) CN114971053A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116882589A (en) * 2023-09-04 2023-10-13 国网天津市电力公司营销服务中心 Online line loss rate prediction method based on Bayesian optimization deep neural network

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116882589A (en) * 2023-09-04 2023-10-13 国网天津市电力公司营销服务中心 Online line loss rate prediction method based on Bayesian optimization deep neural network

Similar Documents

Publication Publication Date Title
EP3822880A1 (en) Load prediction method and apparatus based on neural network
CN111680841B (en) Short-term load prediction method, system and terminal equipment based on principal component analysis
CN115796932B (en) Project cost prediction method and device, electronic equipment and storage medium
CN110689162A (en) Bus load prediction method, device and system based on user side classification
CN112308335A (en) Short-term electricity price prediction method and device based on xgboost algorithm
CN115660233A (en) Photovoltaic power prediction method and device, electronic equipment and storage medium
CN114021425B (en) Power system operation data modeling and feature selection method and device, electronic equipment and storage medium
CN115759365A (en) Photovoltaic power generation power prediction method and related equipment
CN114971053A (en) Training method and device for online prediction model of network line loss rate of low-voltage transformer area
CN111371088A (en) Method and system for correcting SVG control strategy based on BP neural network
CN114461618A (en) Metering missing data completion method and device and terminal equipment
CN113592192A (en) Short-term power load prediction method and device and terminal equipment
CN115796338A (en) Photovoltaic power generation power prediction model construction and photovoltaic power generation power prediction method
CN115619449A (en) Deep learning model-based fertilizer price prediction method, device and storage medium
CN113516275A (en) Power distribution network ultra-short term load prediction method and device and terminal equipment
CN115185804A (en) Server performance prediction method, system, terminal and storage medium
CN115187134A (en) Grid-based power distribution network planning method and device and terminal equipment
CN111339664B (en) Layer-by-layer load modeling method and system based on edge calculation
CN114154415A (en) Equipment life prediction method and device
CN113705957A (en) User cluster baseline load estimation method and device and terminal equipment
CN112766609A (en) Power consumption prediction method based on cloud computing
CN111652420A (en) Real-time load prediction system
CN112632846B (en) Power transmission section limit probability assessment method of power system and electronic equipment
CN117196123B (en) Data control method, device and equipment for digital production of power distribution equipment
CN114638171A (en) Power grid project investment prediction method and device, storage medium and equipment

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