Multi-temporal-spatial-state main multi-source collaborative modeling method and device

Abstract

The invention discloses a multi-temporal-spatial-state main multi-source collaborative modeling method and device. The method comprises the following steps of S1, performing hierarchical processing on multi-temporal-spatial-state data; s2, performing multi-time-space state feature extraction and correlation analysis on the processed data to form a multi-time-space correlation matrix; s3, performing main multi-source collaborative modeling based on the multiple time-space incidence matrixes and outputting a result; and S4, predicting and generating a scheduling scheme by using an output result. During work, on the basis of multi-time-space state data, through combination of multi-time-space correlation analysis and a main model-collaborative sub-model collaborative modeling method, global and local features are integrated, and through dynamic weight adjustment, accurate prediction and optimization of the operation state of the power distribution network are realized, and finally, comprehensive support is provided for scheduling, optimization and emergency response.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent distribution network scheduling, and particularly to a multi-temporal and multi-source collaborative modeling method and device. Background Art

[0002] With the continuous expansion of the scale of the distribution network and the increase in operation complexity, traditional distribution network state modeling methods are facing great challenges. Most existing methods rely on static models or simplified calculations based on linear assumptions, and cannot accurately handle various dynamic situations that occur during the spatio-temporal changes of the distribution network. In addition, the data sources involved in the operation process of the distribution network are complex, and how to efficiently integrate, analyze and make decisions remains an urgent problem to be solved.

[0003] Traditional distribution network modeling is usually based on a single model, and the calculation time is long, making it difficult to achieve real-time scheduling. In the case of complex spatio-temporal changes such as load fluctuations and equipment failures, the response speed and accuracy of traditional models cannot meet the actual needs. Therefore, how to combine multi-source data and design a multi-temporal and multi-state modeling method that can dynamically and adaptively adjust has become an urgent problem in this field. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art.

[0005] The technical solution of the present invention is: a multi-temporal and multi-source collaborative modeling method, including the following steps:

[0006] Step S1, hierarchical processing of multi-temporal data;

[0007] Step S2, extracting and correlating multi-temporal features from the processed data to form a multi-temporal correlation matrix;

[0008] Step S3, performing multi-source collaborative modeling based on the multi-temporal correlation matrix and outputting the result;

[0009] Step S4, using the output result for prediction and generating a scheduling plan.

[0010] In step S1, it further includes hierarchical acquisition and preprocessing of multi-temporal data;

[0011] Among them, the multi-temporal data obtained hierarchically includes: short-term data, medium-term data, and long-term data,

[0012] Preprocessing the hierarchically obtained data includes: missing value filling, outlier removal, and time alignment.

[0013] The short-term data includes real-time operation data related to load fluctuations, voltage changes, and power flow, with a time scale from seconds to minutes;

[0014] The medium-term temporal data includes equipment status and power dispatching records, with a time scale from hours to days;

[0015] The long-term temporal data includes historical operation records and power grid planning parameters, with a time scale from months to years.

[0016] In step S2, it further includes time feature extraction and spatial feature extraction;

[0017] Among them, the time feature extraction includes:

[0018] Based on the time series analysis method, the following time features are extracted from short-term, medium-term, and long-term temporal data:

[0019] Short-term temporal features: load fluctuation frequency, voltage change trend;

[0020] Medium-term temporal features: equipment utilization rate change, fault record frequency;

[0021] Long-term temporal features: historical load growth rate, dispatching efficiency change;

[0022] The spatial feature extraction includes:

[0023] Combined with the power grid topology information and regional operation data, the following spatial features are extracted: node load distribution, line congestion situation, and power flow characteristics between regions.

[0024] The multi-temporal and spatial correlation matrix A is:

[0025] A = {a i,j,k}, i ∈ V, j ∈ V, k ∈ τ

[0026] In the formula, a i,j,k : The correlation value between nodes i and j at time k; V: The set of all nodes in the distribution network; τ: The time range; i and j are natural numbers.

[0027] In step S3, it further includes designing a main model and collaborative sub-models based on the multi-temporal and spatial correlation matrix, and the fusion of the main model and collaborative sub-models;

[0028] Among them, the main model is used to capture global features; the collaborative sub-models are used to capture local area features. The design of the main model includes:

[0029] Input representation: The multi-temporal and spatial correlation matrix A;

[0030] Global spatial feature extraction, extracting the whole network spatial features through a graph convolutional network:

[0031]

[0032] Wherein, The feature of node v in the (l + 1)-th layer in the figure; σ: Sigmoid activation function; N(v): The set of neighbor nodes of node v; c vu : The connection weight between node v and node u in the adjacency matrix; w (l) : The weight matrix of graph convolution in the l-th layer; h u (l) : The feature of node u in the l-th layer; b (l) : The bias vector in the l-th layer;

[0033] Global time feature extraction, using LSTM to model the time dimension:

[0034] h t = LSTM({h t-Δ , …, h t})

[0035] Wherein, h t : The hidden state of the global time t, representing the feature representation at the current moment; h t-Δ : The hidden state of the global time t - Δ;

[0036] Global feature fusion, fusing spatial and time features:

[0037]

[0038] Wherein, H global : Global feature vector; W g , b g : Parameters for global feature fusion; The feature of node v in the l-th layer in the figure;

[0039] Output global prediction, predicting the target according to the global feature:

[0040]

[0041] Wherein, Global prediction value.

[0042] The design of the collaborative sub-model includes:

[0043] Input sub-matrix, extracting the sub-matrix A of the local area from the multi-spatiotemporal correlation matrix A sub :

[0044] A sub = A[i:j, i:j, 1:K]

[0045] Wherein,

[0046] i:j: Node range, representing the subset of nodes in the local area;

[0047] 1: K: Time range, representing the selected time interval;

[0048] Local spatial feature extraction, extracting the whole network spatial features through a graph convolutional network:

[0049]

[0050] Where,

[0051] c vu : The connection weight between node v and node u in the adjacency matrix;

[0052] N sub (v): The set of neighbors of node v in the sub-region;

[0053] Local weights and biases;

[0054] Local time feature extraction, using LSTM to model the time dimension:

[0055] h t，sub = LSTM({h t-Δ，sub , ···, h t，sub})

[0056] Where, h t，sub : The hidden state at local time t; h t-Δ，sub : The hidden state at local time t - Δ;

[0057] Local feature fusion, fusing spatial and time features:

[0058]

[0059] Where,

[0060] H local : Local feature vector; W l : The weight matrix for local feature fusion, b l : The bias term for local feature fusion;

[0061] Output local prediction, predicting the target based on the global features:

[0062]

[0063] Where,

[0064] Local prediction value

[0065] The fusion of the main model and the collaborative sub-model includes:

[0066] Dynamically adjusting the outputs of the main model and the collaborative sub-model according to the weights of the spatio-temporal correlation matrix :

[0067]

[0068] Wherein,

[0069] w global : The weight of the main model;

[0070] w local : The weight of each collaborative sub-model.

[0071] Main model weight:

[0072]

[0073] Collaborative sub-model weight:

[0074]

[0075] Wherein,

[0076] γ: Adjustment parameter;

[0077] sub: Indicates under local features;

[0078] Var(A): The variance of the global matrix A;

[0079] Var(A sub ): Sub-matrix A sub 's variance.

[0080] A main multi-source collaborative modeling device for multi-temporal and spatial states, comprising:

[0081] A layering module for layering processing of multi-temporal and spatial state data;

[0082] An association module for extracting and associating multi-temporal and spatial state features from the processed data to form a multi-temporal and spatial association matrix;

[0083] A modeling module for performing main multi-source collaborative modeling based on the multi-temporal and spatial association matrix and outputting results;

[0084] An output module for using the output results to make predictions and generate a scheduling plan.

[0085] In the working process of the present invention, based on multi-temporal and spatial state data, by combining multi-temporal and spatial association analysis with the main model - collaborative sub-model collaborative modeling method, integrating global and local features, precise prediction and optimization of the operation state of the distribution network are achieved through dynamic weight adjustment, and finally comprehensive support is provided for scheduling, optimization, and emergency response. Brief Description of the Drawings

[0086] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the accompanying drawings required for the description of the specific embodiments or the prior art. In the accompanying drawings, the parts are not necessarily drawn to actual scale.

[0087] Figure 1 is the method flowchart of the present invention. Specific embodiments

[0088] As shown in the present invention Figure 1 a multi-temporal and multi-spatial main multi-source collaborative modeling method is provided, including the following steps:

[0089] Step S1, hierarchical processing of multi-temporal and multi-spatial data;

[0090] Step S2, extracting and correlating multi-temporal and multi-spatial features from the processed data to form a multi-temporal and multi-spatial correlation matrix;

[0091] Step S3, performing main multi-source collaborative modeling based on the multi-temporal and multi-spatial correlation matrix and outputting the result;

[0092] Step S4, using the output result to make predictions and generate a scheduling plan.

[0093] The present invention adopts hierarchical data processing and standardized preprocessing technologies to ensure data consistency and integrity. By performing multi-temporal and multi-spatial correlation analysis to construct a multi-temporal and multi-spatial feature matrix, it dynamically captures the temporal and spatial changes in the power grid operation.

[0094] Through feature fusion and dynamic weight optimization of the main model and collaborative sub-models, it integrates multi-temporal and multi-spatial data and improves the modeling efficiency.

[0095] At the same time, the main model and collaborative sub-models can jointly process global and local dynamic features and adapt to complex changes in different scenarios.

[0096] The present invention can predict power flow changes, load trends, and fault probabilities in real time, providing rapid support for scheduling optimization.

[0097] In step S1, it further includes hierarchical acquisition and preprocessing of multi-temporal and multi-spatial data;

[0098] From the operation data of the distribution network, the data is hierarchically acquired based on the time scale and spatial attributes of the data;

[0099] Among them, the hierarchically acquired multi-temporal and multi-spatial data includes: short-term data, medium-term data, and long-term data,

[0100] Preprocessing the hierarchically acquired data includes: filling missing values, removing outliers, and time alignment.

[0101] Missing value filling: Use interpolation or historical data to complete missing values;

[0102] Outlier removal: Set thresholds to remove abnormal data that does not conform to reality;

[0103] Time alignment: Unify the time scales of different data to ensure the synchronization of hierarchical data.

[0104] Short-term temporal data includes real-time operation data related to load fluctuations, voltage changes, and power flow, with a time scale from seconds to minutes;

[0105] Medium-term temporal data includes equipment status and power dispatch records, with a time scale from hours to days;

[0106] Long-term temporal data includes historical operation records and grid planning parameters, with a time scale from months to years.

[0107] Step S1 realizes the systematic management of multi-source data, ensuring the integrity and consistency of the model input data; hierarchical processing of data with different time scales provides a basis for subsequent spatio-temporal feature extraction.

[0108] In step S2, it further includes time feature extraction and space feature extraction;

[0109] Among them, the time feature extraction includes:

[0110] Based on time series analysis methods, extract the following time features from short-term, medium-term, and long-term temporal data:

[0111] Short-term temporal features: Load fluctuation frequency, voltage change trend;

[0112] Medium-term temporal features: Equipment utilization rate change, fault record frequency;

[0113] Long-term temporal features: Historical load growth rate, dispatch efficiency change;

[0114] The space feature extraction includes:

[0115] Combined with grid topology information and regional operation data, extract the following space features: Node load distribution; Line congestion situation; Power flow characteristics between regions.

[0116] Among them, the time series analysis method includes the following steps:

[0117] (1) Construct a spatio-temporal graph

[0118] Space: Construct a graph structure of the distribution network δ=(v, ε), where v represents substations and equipment, and ε represents line connection relationships;

[0119] Time: For each time step t, generate a snapshot graph δt 。

[0120] (2) Graph neural network model, graph convolution is used to extract spatial features:

[0121]

[0122] (3) Time series modeling, combined with the output of the graph neural network, uses LSTM to capture dynamic changes in the time dimension:

[0123] h t = LSTM({h t-Δ , ···, h t )

[0124] (4) Multi-task loss function, integrating the errors of spatial prediction and time prediction:

[0125]

[0126] (5) Prediction and optimization, output the future operating state according to spatio-temporal features

[0127] Where:

[0128] h t : Hidden state at time t, representing the feature representation at the current moment;

[0129] σ: Sigmoid activation function;

[0130] Feature of node v in the l-th layer of the graph;

[0131] N(v): Set of neighbor nodes of node v;

[0132] c vu : Connection weight between node v and node u in the adjacency matrix;

[0133] w (l) : Weight matrix of the l-th layer graph convolution;

[0134] b (l) : Bias vector of the l-th layer;

[0135] Δ: Time step;

[0136] Represent the errors of spatial feature prediction and time feature prediction respectively;

[0137] λ1, λ2: Weight parameters in the loss function;

[0138] Predicted value output by the model.

[0139] Multi - spatio - temporal correlation analysis includes:

[0140] By analyzing the correlation between data at different time scales and spatial dimensions, a multi - spatio - temporal correlation matrix is constructed to describe the operating characteristics of the distribution network under different spatio - temporal conditions.

[0141] Unify the processing of hierarchical data (short - term, medium - term, long - term) and spatial topology data, and represent it as the input feature set X:

[0142] X = {X T , X S}

[0143] X T : Temporal dimension features, including short - term (such as real - time load fluctuations), medium - term (such as changes in equipment utilization rate), and long - term (such as historical load growth rate).

[0144] X S : Spatial dimension features, including grid node attributes (such as load) and edge attributes (such as power flow).

[0145] Temporal feature extraction; The changes in the temporal dimension are modeled through time - series analysis to extract the following features:

[0146] F T，short : Short - term features, such as load fluctuation frequency;

[0147] F T，mid : Medium - term features, such as scheduling changes in equipment status;

[0148] F T，long : Long - term features, such as load growth trend;

[0149] These features are represented as the temporal feature vector F T :

[0150] F T = [F T，short , F T，mid , F T，long

[0151] Spatial feature extraction; The changes in the spatial dimension are extracted through the analysis of the power grid topology map:

[0152] Node features F S，node : Node load, voltage level, etc.;

[0153] Edge features F S，edge : Power flow, power grid line impedance, etc.;

[0154] The spatial feature vector F S is represented as:

[0155] F S = [F​S，node , F S，edge

[0156] The construction of spatio-temporal correlation features calculates the spatio-temporal correlation feature F through time series analysis and spatial feature fusion T,S :

[0157]

[0158] Indicates the cross-calculation of spatio-temporal features (such as inner product or convolution).

[0159] Construct a comprehensive multi-spatio-temporal correlation matrix to describe the spatio-temporal relationship of the distribution network:

[0160] A = {a i,j,k}, i ∈ V, j ∈ V, k ∈ τ

[0161] In the formula,

[0162] A: The multi-spatio-temporal correlation matrix, describing the spatio-temporal correlation relationship between nodes;

[0163] a i,j,k : The correlation value between nodes i and j at time k, usually expressed as correlation or weight;

[0164] V: The set of all nodes in the distribution network;

[0165] τ: The time range (such as from short-term second-level to long-term annual-level); i and j are natural numbers.

[0166] The calculation formula of the correlation value; the spatio-temporal correlation value a i,j,k Based on the fusion calculation of node features and time features:

[0167]

[0168] In the formula,

[0169] The correlation between nodes i and j in the time dimension;

[0170] The similarity between nodes i and j in the spatial dimension.

[0171] Step S2 extracts the key spatio-temporal features of the power grid operation to comprehensively describe the dynamic behavior of the power grid;

[0172] Establish spatio-temporal correlation to provide input features for collaborative modeling and optimization.

[0173] In step S3, it further includes designing the main model and collaborative sub-models based on the multi-spatio-temporal correlation matrix, and the fusion of the main model and collaborative sub-models;

[0174] ​Among them, the main model: captures global features, such as power flow distribution, voltage stability, etc.;

[0175] The collaborative sub-model: captures local area features, such as the load status of specific nodes and line congestion conditions. The design of the main model includes:

[0176] (1) Input representation: multi-temporal and spatial correlation matrix A;

[0177] (2) Global spatial feature extraction, extracting the whole network spatial features through a graph convolutional network:

[0178]

[0179] In the formula, The feature of node v in the (l + 1)-th layer of the graph; σ: Sigmoid activation function; N(v): the set of neighbor nodes of node v; c vu : the connection weight between node v and node u in the adjacency matrix; w (l) : the weight matrix of the l-th layer of graph convolution; h u (l) : the feature of node u in the l-th layer; b (l) : the bias vector of the l-th layer;

[0180] (3) Global temporal feature extraction, using LSTM to model the time dimension:

[0181] h t = LSTM({h t-Δ , ···, h t})

[0182] In the formula, h t : the hidden state at time t, representing the feature representation at the current moment; h t-Δ : the hidden state at global time t - Δ;

[0183] (4) Global feature fusion, fusing spatial and temporal features:

[0184]

[0185] In the formula, H global : global feature vector; W g , b g : parameters for global feature fusion; The feature of node v in the l-th layer of the graph;

[0186] (5) Output global prediction, predicting the target according to the global features:

[0187]

[0188] In the formula, Global prediction value.

[0189] The collaborative sub - model design includes:

[0190] (1) Input sub - matrix, extract the sub - matrix \(A_{}\) of the local area from the multi - spatio - temporal correlation matrix \(A\): sub :

[0191] \(A_{}\) sub = \(A[i:j,i:j,1:K]\)

[0192] In the formula,

[0193] \(i:j\): Node range, representing the subset of nodes in the local area;

[0194] \(1:K\): Time range, representing the selected time interval;

[0195] (2) Local spatial feature extraction, extract the global spatial features through the graph convolutional network:

[0196]

[0197] In the formula,

[0198] \(c_{}\) vu : The connection weight between node \(v\) and node \(u\) in the adjacency matrix;

[0199] \(N\) sub \((v)\): The neighbor set of node \(v\) in the sub - area;

[0200] Local weights and biases;

[0201] (3) Local temporal feature extraction, use LSTM to model the time dimension:

[0202] \(h_{}\) t，sub = LSTM(\(\{h_{}\), ···, \(h_{}\)\}) t-Δ，sub , ···, \(h_{}\) t，sub}

[0203] In the formula, \(h_{}\) t，sub : The hidden state at local time \(t\); \(h_{}\) t-Δ，sub : The hidden state at local time \(t - \Delta\);

[0204] (4) Local feature fusion, fuse spatial and temporal features:

[0205]

[0206] In the formula,

[0207] \(H_{}\) local : Local feature vector; \(W_{}\) l : The weight matrix for local feature fusion, \(b_{}\) l: Bias term for local feature fusion;

[0208] (5) Output local prediction, and predict the target based on the global feature:

[0209]

[0210] In the formula,

[0211] Local prediction value

[0212] The fusion of the main model and the collaborative sub - models includes:

[0213] Dynamically adjust the outputs of the main model and the collaborative sub - models according to the weights of the spatio - temporal correlation matrix

[0214]

[0215] In the formula,

[0216] w global : Weight of the main model, reflecting the importance of the global feature;

[0217] w local : Weight of each collaborative sub - model, reflecting the importance of the local feature.

[0218] Main model weight:

[0219]

[0220] Collaborative sub - model weight:

[0221]

[0222] In the formula,

[0223] γ: Regulation parameter, used to control the influence degree of volatility on the weight. When γ > 0, the weight is more inclined to smooth features with less fluctuations.

[0224] γ controls the sensitivity of volatility to the weight.

[0225] Large value (γ ≥ 1): The influence of volatility is more significant, and the weight allocation is sensitive to fluctuations.

[0226] Small value (γ ≤ 1): The weight allocation is smoother, ignoring small - scale fluctuations.

[0227] sub: Represents under local features;

[0228] Var(A): Variance of the global matrix A

[0229] Var(A sub ): Sub - matrix A subVariance

[0230] Var(A) and Var(A sub ): Volatility

[0231] By calculating the variance of the matrix, the stability or volatility of global or local spatio-temporal features is measured.

[0232] A larger variance indicates that the feature has strong dynamic changes or uncertainties.

[0233] A smaller variance indicates that the feature is relatively stable and can provide a more reliable basis for modeling.

[0234] exp(-γ·Var(A)): Suppression of volatility;

[0235] The negative exponential function is used to suppress volatility. The greater the volatility, the smaller the impact of the weight.

[0236] When Var is very small, exp(-γ·Var(A))≈1, indicating that the feature is more reliable and a higher weight is assigned.

[0237] Dynamic balance between the main model and the collaborative sub-model:

[0238] When the global feature (A) has small fluctuations and high stability, the weight w of the main model global will be higher;

[0239] When the local feature (A sub ) has smaller fluctuations and higher stability in certain regions, the corresponding sub-model weight w local will be higher.

[0240] Step S3 realizes the linkage optimization between the main model and the collaborative sub-model, improving the adaptability of modeling; ensuring that the model can capture both global and local features simultaneously.

[0241] The result of collaborative modeling (i.e., the comprehensive output of the main model and the collaborative sub-model, including both the global analysis of the operation state of the distribution network and the refined modeling of local areas, finally forming a comprehensive analysis result of global + local, static + dynamic, description + prediction) is used as the input of dynamic response to support the operation optimization and decision-making of the power grid.

[0242] Step S4 includes: real-time state prediction, using the output result of collaborative modeling to predict the future operation state of the power grid, including: power flow changes; load distribution trends; fault probabilities.

[0243] According to the prediction results, a scheduling plan is dynamically generated: optimizing line power distribution; starting standby power supplies; reducing the priority of non-critical loads.

[0244] Step S4 provides real-time power grid state prediction to support accurate dispatching decisions; quickly generates emergency response plans in case of emergencies to ensure power supply safety.

[0245] The present invention also provides a multi-temporal and multi-spatial main multi-source collaborative modeling device, including:

[0246] A layering module for multi-temporal and multi-spatial data layering processing;

[0247] An association module for extracting multi-temporal and multi-spatial features from the processed data and performing association analysis to form a multi-temporal and multi-spatial association matrix;

[0248] A modeling module for performing main multi-source collaborative modeling based on the multi-temporal and multi-spatial association matrix and outputting results;

[0249] An output module for using the output results to make predictions and generate dispatching plans.

[0250] Based on multi-temporal and multi-spatial data, the present invention combines multi-temporal and multi-spatial association analysis with the main model - collaborative sub-model collaborative modeling method, and realizes accurate prediction and optimization of the operation state of the distribution network through dynamic weight adjustment, and finally provides comprehensive support for dispatching, optimization and emergency response.

[0251] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A main multi-source collaborative modeling method for multiple spatio-temporal states, characterized in that, It includes the following steps: Step S1: Hierarchical processing of multi-temporal and multi-spatial data; Step S2: Extract multi-temporal and multi-spatial features from the processed data and conduct correlation analysis to form a multi-temporal and multi-spatial correlation matrix; Step S3: Based on the multi-temporal and multi-spatial correlation matrix, conduct primary multi-source collaborative modeling and output the results; Step S4: Utilize the output results to make predictions and generate a scheduling plan.

2. A primary multi-source collaborative modeling method for multi-temporal and multi-spatial states according to claim 1, wherein in Step S1, it further includes hierarchical acquisition and preprocessing of multi-temporal and multi-spatial data; wherein, the hierarchically acquired multi-temporal and multi-spatial data includes: short-term temporal data, medium-term temporal data, and long-term temporal data, and preprocessing the hierarchically acquired data includes: filling missing values, removing outliers, and time alignment.

3. A primary multi-source collaborative modeling method for multi-temporal and multi-spatial states according to claim 2, wherein the short-term temporal data includes real-time operation data related to load fluctuations, voltage changes, and power flow, with a time scale from seconds to minutes; the medium-term temporal data includes equipment status and power dispatch records, with a time scale from hours to days; the long-term temporal data includes historical operation records and grid planning parameters, with a time scale from months to years.

4. A primary multi-source collaborative modeling method for multi-temporal and multi-spatial states according to claim 1, wherein in Step S2, it further includes time feature extraction and spatial feature extraction; wherein, the time feature extraction includes: Based on time series analysis methods, extract the following time features from short-term, medium-term, and long-term temporal data: Short-term temporal features: load fluctuation frequency, voltage change trend; Medium-term temporal features: equipment utilization rate change, fault record frequency; Long-term temporal features: historical load growth rate, dispatch efficiency change; The spatial feature extraction includes: Combining grid topology information and regional operation data, extract the following spatial features: node load distribution, line congestion situation, and power flow characteristics between regions.

5. A primary multi-source collaborative modeling method for multi-temporal and multi-spatial states according to claim 4, wherein the multi-temporal and multi-spatial correlation matrix A is: A = {a i,j,k}, i ∈ V, j ∈ V, k ∈ τ where a i,j,k : the association value between nodes i and j at time k; V: the set of all nodes in the distribution network; τ: the time range; i and j are natural numbers.

6. A primary multi-source collaborative modeling method for multi-temporal and multi-spatial states according to claim 1, wherein in Step S3, it further includes designing a primary model and collaborative sub-models based on the multi-temporal and multi-spatial correlation matrix, and the fusion of the primary model and collaborative sub-models; wherein, the primary model is used to capture global features; the collaborative sub-models are used to capture local regional features.

7. A primary multi-source collaborative modeling method for multi-temporal and multi-spatial states according to claim 6, wherein the design of the primary model includes: Input representation: multi-temporal and multi-spatial correlation matrix A; Global spatial feature extraction, extracting the whole network spatial features through a graph convolutional network: In the formula, The feature of node v in the (l + 1)-th layer in the figure; σ: Sigmoid activation function; N(v): The set of neighbor nodes of node v; c vu : The connection weight between node v and node u in the adjacency matrix; w (l) : The weight matrix of graph convolution in the l-th layer; The feature of node u in the l-th layer; b (l) : The bias vector in the l-th layer; Global time feature extraction, using LSTM to model the time dimension: h t = LSTM({h t-Δ , ···, h t ) where h t : the hidden state at the global time t, representing the feature representation at the current moment; h t-Δ : the hidden state at the global time t - Δ; Global feature fusion, fusing spatial and time features: Where, H global : Global feature vector; W g , b g : Parameters for global feature fusion; The feature of node v in the l-th layer in the figure; Output global prediction, predicting the target according to the global features: In the formula, Global prediction value.

8. A primary multi-source collaborative modeling method for multi-temporal and multi-spatial states according to claim 7, wherein the design of the collaborative sub-models includes: Input sub-matrix, extracting the sub-matrix A of the local area from the multi-temporal and spatial correlation matrix A sub : A sub = A[i:j, i:j, 1:K] In the formula, i:j: Node range, representing a subset of nodes in the local area; 1: K: The time range, representing the selected time interval; Local spatial feature extraction, extracting the whole network spatial features through a graph convolutional network: where c vu : the connection weight between node v and node u in the adjacency matrix; N sub (v): the set of neighbors of node v in the sub-region; Local weights and biases; Local temporal feature extraction, using LSTM to model the temporal dimension: h t，sub = LSTM({h t-Δ，sub , ···, h t，sub}) where h t，sub is the hidden state at local time t; h t-Δ，sub is the hidden state at local time t - Δ. Local feature fusion, fusing spatial and temporal features: Wherein, H local : Local feature vector; W l : Weight matrix for local feature fusion, b l : Bias term for local feature fusion; Output local prediction, predicting the target according to the global features: Wherein, Local prediction value.

9. A multi-spatiotemporal main multi-source collaborative modeling method according to claim 8, characterized in that The fusion of the main model and the collaborative sub-model includes: Dynamically adjust the outputs of the main model and the collaborative sub-models according to the weights of the spatio-temporal correlation matrix where, w global : the weight of the main model; w local : the weight of each collaborative sub-model; Weights of the main model: Weights of the collaborative sub-model: Wherein, γ: Adjustment parameter; sub: Represents under local features; Var(A): Variance of the global matrix A; Var(A sub ): The variance of sub-matrix A sub .

10. A main multi-source collaborative modeling device with multiple space-time states, characterized in that, Includes: A hierarchical module for hierarchical processing of multi-spatiotemporal data; An association module for extracting and associating multi-spatiotemporal features from the processed data to form a multi-spatiotemporal association matrix; A modeling module for performing main multi-source collaborative modeling based on the multi-spatiotemporal association matrix and outputting results; An output module for using the output results to make predictions and generate a scheduling plan.

Images