Urban population migration prediction method and device based on power consumption data, and medium

Abstract

The invention relates to an urban population migration prediction method and device based on power consumption data and a medium, and belongs to the technical field of data processing, and the method comprises the following steps: constructing an urban structure chart and an edge weight matrix, and obtaining the power consumption data of each node to construct a power consumption characteristic matrix; and building an urban population migration prediction model, wherein the urban population migration prediction model comprises a spatial feature extraction module, a directed edge feature generation module and a migration prediction module. Inputting the edge weight matrix and the power utilization characteristic matrix into a spatial characteristic extraction module to obtain a spatial characteristic vector of each node; and inputting the spatial feature vectors and the edge weight matrix into a directed edge feature generation module, and outputting directed edge feature vectors among the nodes. And obtaining population flow among the nodes, and constructing a directed edge population flow feature matrix. And inputting the directed edge population flow characteristic matrix into a migration prediction module, and outputting a population migration prediction matrix containing each directed edge population migration prediction amount. The directed edge feature generation module effectively solves the directionality problem of population flow.

Description

Technical Field

[0001] This invention relates to a method, device, and medium for predicting urban population migration based on electricity consumption data, belonging to the field of data processing technology. Background Technology

[0002] Urban population migration forecasting is a fundamental technical issue in urban planning, resource allocation, traffic management, and public service provision. Accurately predicting inter-city population flow trends has significant practical implications for businesses in developing commercial networks and for transportation departments in optimizing transport capacity.

[0003] Traditional population migration forecasting primarily relies on statistical survey data, including census data, household registration data, and sampling survey data. Collecting this data requires significant human and material resources, and the data update cycle is long, typically on an annual basis, making it difficult to meet the needs of monitoring real-time dynamic population flow trends. While statistical survey data is highly authoritative, its inherent lag means that forecast results cannot promptly reflect the impact of short-term emergencies or seasonal changes on population flow. Furthermore, traditional statistical methods often employ mathematical models such as regression analysis, gravity models, and intervention opportunity models. These models have relatively fixed structures and limited ability to characterize complex nonlinear spatiotemporal dependencies, resulting in forecast accuracy that is insufficient for the requirements of refined urban management. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention proposes a method, device, and medium for predicting urban population migration based on electricity consumption data.

[0005] The technical solution of the present invention is as follows: On the one hand, this invention provides a method for predicting urban population migration based on electricity consumption data, comprising the following steps: Construct the city structure map and the edge weight matrix of the city structure map; Obtain electricity consumption data for each node in the urban structure map, and construct an electricity consumption feature matrix for each node based on the electricity consumption data; A model for predicting urban population migration is constructed, including a spatial feature extraction module, a directed edge feature generation module, and a migration prediction module. Input the edge weight matrix and the electricity consumption feature matrix into the spatial feature extraction module, and output the spatial feature vector of each node; The spatial feature vector and edge weight matrix are input into the directed edge feature generation module, which outputs the directed edge feature vector between each node. Obtain the population flow between each node, and construct a directed edge population flow feature matrix based on the population flow and the directed edge feature vector; The directed edge population flow feature matrix is ​​input into the migration prediction module, and the population migration prediction matrix is ​​output, including the population migration prediction amount for each directed edge.

[0006] Preferably, a city structure map is constructed using cities as nodes; Calculate the distance between each node in the city structure graph, use the reciprocal of the distance as the edge weight between the corresponding nodes, and construct the edge weight matrix of the city structure graph based on the edge weight.

[0007] Preferably, the electricity consumption data includes weather-corrected residential electricity consumption, industrial electricity consumption, commercial electricity consumption, other electricity consumption, temperature data, whether it is a holiday, and major events and activities; The meteorologically corrected residential electricity consumption is obtained based on residential electricity consumption and temperature data; The electricity consumption data is Z-score standardized, and the electricity consumption feature matrix of each node is constructed based on the Z-score standardized electricity consumption data.

[0008] Preferably, the weather-corrected residential electricity consumption is obtained based on residential electricity consumption and temperature data, and the specific steps are as follows: Historical residential electricity consumption was decomposed using STL to obtain historical trend electricity consumption and historical periodic electricity consumption. Calculate the historical residual electricity consumption based on the historical residential electricity consumption, historical trend electricity consumption, and historical periodic electricity consumption. Construct a temperature fitting function to fit the linear relationship between historical temperature data and historical residual electricity consumption; The residual electricity consumption of residents is obtained based on temperature data and temperature fitting functions; The meteorological-corrected residential electricity consumption is calculated based on the residual electricity consumption and the residential electricity consumption.

[0009] Preferably, the spatial feature extraction module is constructed using a two-layer graph convolutional network, as expressed by the formula: ; In the formula, Indicates the first Nodes Spatial feature vector at time step, Indicates the first Nodes Electricity consumption characteristic matrix at time points. Represents the edge weight matrix. This represents the Sigmoid activation function. Represents the ReLU activation function. Represents a symmetric normalized adjacency matrix. This represents the weight matrix of the first layer of the graph convolutional network. This represents the weight matrix of the second-layer graph convolutional network; The symmetric normalized adjacency matrix is ​​expressed by the formula: ; In the formula, This represents the self-connection weight matrix. Represents the self-connection weight matrix The degree matrix; The self-connection weight matrix is ​​expressed by the formula: ; In the formula, Represents the identity matrix.

[0010] Preferably, the spatial feature vector and edge weight matrix are input into the directed edge feature generation module, as expressed by the formula: ; In the formula, Indicates the first The node to the first Nodes The directed edge feature vector at time step 1. This represents a multilayer perceptron (MLP). This represents the concatenation operation function. Indicates the first Nodes Spatial feature vector at time step, Indicates the first Nodes Electricity consumption characteristic matrix at time points. Represents the edge weight matrix. The node and the first The edge weights of each node.

[0011] Preferably, a directed edge population flow feature matrix is ​​constructed based on the population flow volume and the directed edge feature vector, expressed by the formula: ; In the formula, Indicates the first The node to the first Nodes The directed edge population flow feature vector at time step. Indicates the first The node to the first Nodes Population movement at any given time; Construct a directed edge population flow feature matrix based on the directed edge population flow feature vector.

[0012] Preferably, the migration prediction module is constructed using a GRU model; The directed edge population flow feature matrix is ​​used as input to the GRU model, and the output is the hidden state matrix, expressed by the formula: ; ; ; ; In the formula, Indicates update gate Output at any moment This indicates updating the gate weight matrix. express The directed edge population flow characteristic matrix at time t. This indicates updating the gate bias vector. express The hidden state matrix at time step 1. Indicates door reset Output at any moment This represents the reset gate weight matrix. This indicates the reset gate bias vector. Indicates the candidate hidden state Output at any moment It represents the Hadamah accumulation. This represents the hyperbolic tangent activation function. This represents the candidate hidden state weight matrix. This represents the candidate hidden state bias vector. express The hidden state matrix at time step; The population migration prediction matrix is ​​obtained based on the hidden state matrix, expressed by the formula: ; In the formula, express Population migration prediction matrix at time point, This represents the weight matrix of the fully connected layer. This represents the bias vector of the fully connected layer.

[0013] In another aspect, the present invention also provides an electronic device having a computer program stored thereon, which, when executed by a processor, implements the urban population migration prediction method based on electricity consumption data as described in any embodiment of the present invention.

[0014] In another aspect, the present invention also provides a computer-readable storage medium for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the urban population migration prediction method based on electricity consumption data as described in any embodiment of the present invention.

[0015] The present invention has the following beneficial effects: 1. This invention constructs a graph structure using cities as nodes and the reciprocal of the distance between cities as edge weights, which aligns with the gravity model theory, i.e., geographical proximity promotes population interaction. Graph convolutional networks, by aggregating feature information from neighboring nodes, can capture higher-order spatial dependencies, ensuring that each city's representation not only includes its own electricity consumption characteristics but also incorporates relevant information from surrounding cities. This design allows the model to understand the push-pull effects between cities; for example, the economic prosperity of surrounding cities may attract population outflow from the city, thus more accurately depicting the complex patterns of regional population migration and improving the ability to model spatial heterogeneity.

[0016] 2. The directed edge feature generation module designed in this invention effectively solves the problem of the directionality of population flow. Traditional undirected graph models cannot distinguish between the different mechanisms of population movement from city A to city B and from city B to city A, while actual population migration has clear directionality and asymmetry. By concatenating the spatial feature vectors of the starting and ending points and combining them with edge weight information, the directed edge representation is learned using a multilayer perceptron. This allows for the separate modeling of the push factors at the origin and the pull factors at the destination. The edge weights, as a quantification of physical distance cost, further enrich the semantic information of the directed edges, significantly improving the spatial resolution and accuracy of the prediction.

[0017] 3. This invention employs a time-series model to capture the dynamic evolution of population flow, achieving effective prediction of migration trends. Population flow exhibits significant time dependence; changes in flow over a past period influence future trends. After extracting the spatiotemporal features of directed edges, this invention uses a temporal recursive network to model this dynamic nature, automatically learning the persistence, periodicity, and trend of flow patterns. The model uses a gating mechanism to control the retention and forgetting of historical information, dynamically adjusting the weight of historical data at different times on the current prediction. This end-to-end architecture avoids the complex manual feature engineering of traditional methods, allowing the model to automatically discover hidden temporal patterns from the data. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the implementation of the method in an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] It should be understood that the step numbers used in the text are for ease of description only and are not intended to limit the order in which the steps are performed.

[0021] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0022] The terms “comprising” and “including” indicate the presence of the described feature, whole, step, operation, element and / or component, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0023] The term “and / or” refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes these combinations.

[0024] Example 1: See Figure 1 This embodiment provides a method for predicting urban population migration based on electricity consumption data, including the following steps: Construct the city structure map and the edge weight matrix of the city structure map; Obtain electricity consumption data for each node in the urban structure map, and construct an electricity consumption feature matrix for each node based on the electricity consumption data; A model for predicting urban population migration is constructed, including a spatial feature extraction module, a directed edge feature generation module, and a migration prediction module. Input the edge weight matrix and the electricity consumption feature matrix into the spatial feature extraction module, and output the spatial feature vector of each node; The spatial feature vector and edge weight matrix are input into the directed edge feature generation module, which outputs the directed edge feature vector between each node. Obtain the population flow between each node, and construct a directed edge population flow feature matrix based on the population flow and the directed edge feature vector; The directed edge population flow feature matrix is ​​input into the migration prediction module, and the population migration prediction matrix is ​​output, including the population migration prediction of each directed edge, that is, the population migration prediction between cities.

[0025] Preferably, a city structure map is constructed using cities as nodes, expressed by the formula: ; ; ; ; In the formula, A diagram representing the city's structure. Represents a set of nodes. Denotes the set of edges. Indicates the first 1 node Indicates the first The node and the first The edges corresponding to each node. Indicates the number of cities; Calculate the distance between each node in the city structure graph, take the reciprocal of the distance as the edge weight between the corresponding nodes, and construct the edge weight matrix of the city structure graph based on the edge weight. Specifically, the distance refers to the straight-line distance between the two corresponding cities, for the first... The node cannot communicate with the first node. Since each node has an edge, the edge weights on the main diagonal of the edge weight matrix are all 0.

[0026] Using the reciprocal of the straight-line distance between two cities as the edge weight draws inspiration from the gravitational model in physics, which states that the closer the distance, the stronger the interaction.

[0027] Preferably, the electricity consumption data includes weather-corrected residential electricity consumption, industrial electricity consumption, commercial electricity consumption, other electricity consumption, temperature data, whether it is a holiday, and major events and activities (such as sports events, concerts, and other electricity-related activities). The meteorologically corrected residential electricity consumption is obtained based on residential electricity consumption and temperature data; The electricity consumption data is Z-score standardized, and the electricity consumption feature matrix of each node is constructed based on the Z-score standardized electricity consumption data.

[0028] Industrial electricity consumption, commercial electricity consumption, and other electricity consumption represent the level of production activity, commercial prosperity, and the intensity of public service operation. Their fluctuations can effectively reflect the potential for population mobility due to work, business activities, and other purposes.

[0029] Other electricity consumption includes lighting for schools and welfare institutions, as well as public lighting such as streetlights in residential areas.

[0030] Residential electricity consumption is a highly sensitive indicator reflecting the size of the resident population and their daily activity levels within a region. However, residential electricity consumption is significantly affected by meteorological factors such as temperature (e.g., air conditioning in summer and heating in winter).

[0031] Preferably, the weather-corrected residential electricity consumption is obtained based on residential electricity consumption and temperature data, and the specific steps are as follows: Historical residential electricity consumption was decomposed using STL to obtain historical trend electricity consumption and historical periodic electricity consumption. Based on the aforementioned historical residential electricity consumption, historical trend electricity consumption, and historical periodic electricity consumption, the historical residual electricity consumption is calculated, expressed by the formula: ; In the formula, Indicates the first Historical residual electricity consumption of each node Indicates the first Historical residential electricity consumption at each node Indicates the first Historical electricity consumption trends at each node Indicates the first Historical periodic electricity consumption of each node; A temperature fitting function is constructed to fit the linear relationship between historical temperature data and historical residual electricity consumption, expressed as the formula: ; In the formula, Indicates the first Historical temperature data for each node, This represents the temperature fitting function; During the high-temperature period of summer, historical temperature data and historical residual electricity consumption show a moderate positive correlation; during the low-temperature period of winter, a moderate negative correlation is observed. This strongly supports the hypothesis that residents' temperature control equipment load is only significantly activated when the ambient temperature exceeds the human comfort zone, and that its power consumption is approximately linearly related to the temperature difference exceeding the comfort zone. Therefore, a piecewise linear regression model is used as the temperature fitting function to describe the response relationship between historical temperature data and historical residual electricity consumption.

[0032] The temperature fitting function is expressed by the following formula: ; In the formula, Represents the intercept term. Indicates the temperature slope. Indicates the first The slope of each linear segment. Indicates the total number of segments. Indicates the first A temperature threshold, This indicates an indicator function, if the condition is... Output 1 if true, otherwise output 0. Represents the random error term; For summer cooling and winter heating scenarios, the total number of segments In this embodiment, the value is set to 2, and the temperature threshold of the temperature fitting function is determined using the differential evolution algorithm. Then, the least squares method is used to determine each temperature threshold. The temperature fitting function. The differential evolution algorithm is expressed by the formula: ; ; ; In the formula, This represents a piecewise linear regression model. This represents the predicted value of historical residual electricity consumption. Describes the minimum value function. Represents the maximum value function; Taking the residential electricity consumption of a certain city as an example, the temperature threshold obtained using the differential evolution algorithm... Given 15.8606 and 25.3206, the temperature fitting function can be expanded as follows: ; The residual electricity consumption for residential use, obtained based on temperature data and a temperature fitting function, is expressed by the following formula: ; In the formula, Indicates the first Nodes The residual electricity consumption of residents at any given time. Indicates the first Nodes Temperature data at any given time; Based on the residual electricity consumption and residential electricity consumption, the meteorologically corrected residential electricity consumption is calculated and expressed by the formula: ; In the formula, Indicates the first Nodes Residential electricity consumption after weather correction. Indicates the first Nodes Residential electricity consumption at any given time.

[0033] Preferably, the spatial feature extraction module is constructed using a two-layer graph convolutional network, as expressed by the formula: ; In the formula, Indicates the first Nodes Spatial feature vector at time step, Indicates the first Nodes Electricity consumption characteristic matrix at time points. Represents the edge weight matrix. This represents the Sigmoid activation function. Represents the ReLU activation function. Represents a symmetric normalized adjacency matrix. This represents the weight matrix of the first layer of the graph convolutional network. This represents the weight matrix of the second-layer graph convolutional network; The symmetric normalized adjacency matrix is ​​expressed by the formula: ; In the formula, This represents the self-connection weight matrix. Represents the self-connection weight matrix The degree matrix; The self-connection weight matrix is ​​expressed by the formula: ; In the formula, Represents the identity matrix; The degree matrix is ​​expressed by the formula: ; In the formula, The degree matrix of the representation One diagonal element, Represents the self-connection weight matrix No. Line number The elements of the column.

[0034] The design of a two-layer graph convolutional network effectively expands the receptive field of nodes to capture higher-order spatial dependencies by stacking convolutional layers.

[0035] Preferably, the spatial feature vector and edge weight matrix are input into the directed edge feature generation module, as expressed by the formula: ; In the formula, Indicates the first The node to the first Nodes The directed edge feature vector at time step 1. This represents a multilayer perceptron (MLP). This represents the concatenation operation function. Indicates the first Nodes Spatial feature vector at time step, Indicates the first Nodes Electricity consumption characteristic matrix at time points. Represents the edge weight matrix. The node and the first The edge weights of each node.

[0036] Spatial feature vectors Represents the starting node Thrust information, spatial feature vector Represents the target node The tension information, edge weights This represents the physical cost or ease of this flow. (The starting node...) Spatial feature vectors Target Node Spatial feature vectors and the corresponding edge weights The concatenated information is then fed into a multilayer perceptron (MLP) to learn the complex interactions between these three parts, resulting in directed edge feature vectors. .

[0037] The directed edge feature generation module can model the asymmetric influence between nodes, significantly enhancing the model's ability to represent directional information in the graph.

[0038] Preferably, a directed edge population flow feature matrix is ​​constructed based on the population flow volume and the directed edge feature vector, expressed by the formula: ; In the formula, Indicates the first The node to the first Nodes The directed edge population flow feature vector at time step. Indicates the first The node to the first Nodes Population movement at any given time; Construct a directed edge population flow feature matrix based on the directed edge population flow feature vector.

[0039] Incorporating population mobility as an explicit autoregressive information into the directed edge feature vector allows the subsequent GRU model to learn not only the external variables (implied in the directed edge feature vector) but also... Changes driven by population movement can also directly capture the endogenous time dependence of population movement itself, thus providing a more comprehensive and profound understanding of the spatiotemporal dynamics of population movement.

[0040] Preferably, the migration prediction module is constructed using a GRU model; The directed edge population flow feature matrix is ​​input into the GRU model, and the output is the hidden state matrix, expressed by the formula: ; ; ; ; ; In the formula, Indicates update gate Output at any moment This indicates updating the gate weight matrix. express The directed edge population flow characteristic matrix at time t. This indicates updating the gate bias vector. express The hidden state matrix at time step 1. Indicates door reset Output at any moment This represents the reset gate weight matrix. This indicates the reset gate bias vector. Indicates the candidate hidden state Output at any moment It represents the Hadamah accumulation. This represents the hyperbolic tangent activation function. This represents the candidate hidden state weight matrix. This represents the candidate hidden state bias vector. express The hidden state matrix at time step 1. Indicates the training period; GRU model after After rounds of training, a population migration prediction matrix is ​​obtained based on the hidden state matrix, expressed by the formula: ; In the formula, express Population migration prediction matrix at time point, This represents the weight matrix of the fully connected layer. This represents the bias vector of the fully connected layer.

[0041] The highly abstract hidden state matrix output by GRU is decoded and projected linearly or non-linearly into the output space of the target prediction task, flexibly completing the necessary transformation from the feature dimension to the final prediction dimension.

[0042] Example 2: This embodiment provides an electronic device that stores a computer program. When the computer program is executed by a processor, it implements the urban population migration prediction method based on electricity consumption data as described in any embodiment of the present invention.

[0043] Example 3: This embodiment provides a computer-readable storage medium for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the urban population migration prediction method based on electricity consumption data as described in any embodiment of the present invention.

[0044] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, A and B simultaneously, or B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, at least one of a, b, and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0045] Those skilled in the art will recognize that the units and algorithm steps described in the embodiments disclosed herein can be implemented using electronic hardware, computer software, or a combination of electronic hardware and software. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0046] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0047] In the several embodiments provided in this application, any function, if implemented as a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0048] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for predicting urban population migration based on electricity consumption data, characterized in that, Includes the following steps: Construct a city structure graph and an edge weight matrix for the city structure graph, with cities as nodes; Obtain electricity consumption data for each node in the urban structure map, and construct an electricity consumption feature matrix for each node based on the electricity consumption data; A model for predicting urban population migration is constructed, including a spatial feature extraction module, a directed edge feature generation module, and a migration prediction module. Input the edge weight matrix and the electricity consumption feature matrix into the spatial feature extraction module, and output the spatial feature vector of each node; The spatial feature vector and edge weight matrix are input into the directed edge feature generation module, which outputs the directed edge feature vector between each node. Obtain the population flow between each node, and construct a directed edge population flow feature matrix based on the population flow and the directed edge feature vector; The directed edge population flow feature matrix is ​​input into the migration prediction module, and the population migration prediction matrix is ​​output, including the population migration prediction amount for each directed edge.

2. The urban population migration prediction method based on electricity consumption data according to claim 1, characterized in that, Calculate the distance between each node in the city structure graph, use the reciprocal of the distance as the edge weight between the corresponding nodes, and construct the edge weight matrix of the city structure graph based on the edge weight.

3. The urban population migration prediction method based on electricity consumption data according to claim 2, characterized in that, The electricity consumption data includes weather-corrected residential electricity consumption, industrial electricity consumption, commercial electricity consumption, other electricity consumption, temperature data, whether there are holidays, and major events and activities; The meteorologically corrected residential electricity consumption is obtained based on residential electricity consumption and temperature data; The electricity consumption data is Z-score standardized, and the electricity consumption feature matrix of each node is constructed based on the Z-score standardized electricity consumption data.

4. The urban population migration prediction method based on electricity consumption data according to claim 3, characterized in that, The weather-corrected residential electricity consumption is obtained based on residential electricity consumption and temperature data. The specific steps are as follows: Historical residential electricity consumption was decomposed using STL to obtain historical trend electricity consumption and historical periodic electricity consumption. Calculate the historical residual electricity consumption based on the historical residential electricity consumption, historical trend electricity consumption, and historical periodic electricity consumption. Construct a temperature fitting function to fit the linear relationship between historical temperature data and historical residual electricity consumption; The residual electricity consumption of residents is obtained based on temperature data and temperature fitting functions; The meteorological-corrected residential electricity consumption is calculated based on the residual electricity consumption and the residential electricity consumption.

5. The urban population migration prediction method based on electricity consumption data according to claim 3, characterized in that, The spatial feature extraction module is constructed using a two-layer graph convolutional network, expressed by the formula: ； In the formula, Indicates the first Nodes Spatial feature vector at time step, Indicates the first Nodes Electricity consumption characteristic matrix at time points. Represents the edge weight matrix. This represents the Sigmoid activation function. Represents the ReLU activation function. Represents a symmetric normalized adjacency matrix. This represents the weight matrix of the first layer of the graph convolutional network. This represents the weight matrix of the second-layer graph convolutional network; The symmetric normalized adjacency matrix is ​​expressed by the formula: ； In the formula, This represents the self-connection weight matrix. Represents the self-connection weight matrix The degree matrix; The self-connection weight matrix is ​​expressed by the formula: ； In the formula, Represents the identity matrix.

6. The urban population migration prediction method based on electricity consumption data according to claim 5, characterized in that, The spatial feature vector and edge weight matrix are input into the directed edge feature generation module, expressed as a formula: ； In the formula, Indicates the first The node to the first Nodes The directed edge feature vector at time step 1. This represents a multilayer perceptron (MLP). This represents the concatenation operation function. Indicates the first Nodes Spatial feature vector at time step, Indicates the first Nodes Electricity consumption characteristic matrix at time points. Represents the edge weight matrix. The node and the first The edge weights of each node.

7. The urban population migration prediction method based on electricity consumption data according to claim 6, characterized in that, Based on the population flow volume and the directed edge feature vector, a directed edge population flow feature matrix is ​​constructed, expressed by the formula: ； In the formula, Indicates the first The node to the first Nodes The directed edge population flow feature vector at time step. Indicates the first The node to the first Nodes Population movement at any given time; Construct a directed edge population flow feature matrix based on the directed edge population flow feature vector.

8. The urban population migration prediction method based on electricity consumption data according to claim 7, characterized in that, The migration prediction module is constructed using a GRU model. The directed edge population flow feature matrix is ​​used as input to the GRU model, and the output is the hidden state matrix, expressed by the formula: ； ； ； ； In the formula, Indicates update gate Output at any moment This indicates updating the gate weight matrix. express The directed edge population flow characteristic matrix at time t. This indicates updating the gate bias vector. express The hidden state matrix at time step 1. Indicates door reset Output at any moment This represents the reset gate weight matrix. This indicates the reset gate bias vector. Indicates the candidate hidden state Output at any moment It represents the Hadamah accumulation. This represents the hyperbolic tangent activation function. This represents the candidate hidden state weight matrix. This represents the candidate hidden state bias vector. express The hidden state matrix at time step; The population migration prediction matrix is ​​obtained based on the hidden state matrix, expressed by the formula: ； In the formula, express Population migration prediction matrix at time point, This represents the weight matrix of the fully connected layer. This represents the bias vector of the fully connected layer.

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, When the processor executes the program, it implements the urban population migration prediction method based on electricity consumption data as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the urban population migration prediction method based on electricity consumption data as described in any one of claims 1 to 8.

Images