A distributed energy storage collaborative scheduling method and system for energy exchange fusion

Abstract

The application discloses a kind of distributed energy storage collaborative scheduling methods and systems for energy exchange fusion, it is related to energy storage scheduling related technical field, including: collecting traffic energy feature data, power distribution network operating state data, constructs traffic-power distribution network coupling space-time data graph;Carry out regional collaborative control, formulate initial collaborative scheduling strategy;Build regional traffic digital twin to simulate execution, generate simulation scheduling result and carry out reverse balance optimization, generate distributed energy storage collaborative control instruction;Distributed energy storage collaborative control instruction is associated with multiple traffic energy nodes, and generates energy exchange fusion collaborative scheduling scheme.The technical problems that the prior art exists that the energy exchange data is difficult to deeply fuse and space-time representation, collaborative control strategy lacks dynamic closed-loop verification and adaptive optimization are solved, the technical effects that the collaborative utilization efficiency of distributed energy storage resources in traffic-power distribution network coupling scene is improved and the robust adaptive capacity of scheduling strategy to real-time working condition is achieved.

Description

Technical Field

[0001] This invention relates to the field of energy storage scheduling technology, specifically to a distributed energy storage collaborative scheduling method and system oriented towards energy exchange integration. Background Technology

[0002] Traffic flow and energy flow no longer operate independently, but are deeply intertwined through charging behavior, load fluctuations, and energy storage charging and discharging. This "traffic-energy fusion" model poses a severe challenge to the voltage stability, peak-valley balance, and equipment utilization of the distribution network. Currently, energy storage scheduling for traffic scenarios has significant limitations. Using charging stations or battery swapping stations as units and managing energy storage charging and discharging based on local traffic flow or historical charging curves ignores the real-time carrying capacity of the upper-level distribution network. This easily leads to peak electricity consumption overlapping during peak traffic periods, forming "peak on peak," which exacerbates the risk of heavy overload in the distribution area. With the goal of optimizing the distribution network flow, traffic energy consumption nodes are treated as rigid loads or adjustable resource pools, and a unified command issuance mode is adopted. However, traffic energy consumption has strong regional heterogeneity and event randomness, such as highway traffic during holidays and urban commuting tidal phenomena. Centralized management cannot accurately capture the dynamic energy consumption intentions and spatial constraints of each node, resulting in a disconnect between dispatch commands and actual traffic behavior and large execution deviations. In addition, most existing traffic and energy integration systems simply overlay traffic situation data with distribution network electrical quantities or empirically correlate them, lacking a unified graph data modeling framework. They cannot effectively characterize the non-electrical coupling relationship between traffic network topology and distribution network feeder topology. For example, if the charging load of a highway service area surges, its impact will be transmitted to the upstream substation along the distribution network feeder, but it will also be affected by the spatial game of the flow behavior of adjacent service areas.

[0003] Therefore, current technologies face challenges such as the difficulty in deeply integrating and spatiotemporally representing energy exchange data, and the lack of dynamic closed-loop verification and adaptive optimization for collaborative control strategies. Summary of the Invention

[0004] This application provides a distributed energy storage collaborative scheduling method and system for energy exchange integration, which solves the technical problems in the prior art, such as the difficulty in deep integration and spatiotemporal representation of energy exchange data, and the lack of dynamic closed-loop verification and adaptive optimization of collaborative control strategies. It achieves the technical effect of improving the collaborative utilization efficiency of distributed energy storage resources in the transportation-distribution network coupling scenario and the robust adaptability of scheduling strategies to real-time operating conditions.

[0005] This application provides a distributed energy storage collaborative scheduling method for energy integration, the method comprising: collecting traffic energy consumption characteristic data and distribution network operation status data to construct a traffic-distribution network coupled spatiotemporal data map; performing regional collaborative control based on the traffic-distribution network coupled spatiotemporal data map to formulate an initial collaborative scheduling strategy; constructing a regional traffic digital twin to simulate the execution of the initial collaborative scheduling strategy, generating simulated scheduling results to perform balanced optimization of the initial collaborative scheduling strategy, and generating distributed energy storage collaborative control instructions; associating the distributed energy storage collaborative control instructions with multiple traffic energy consumption nodes to generate an energy integration collaborative scheduling scheme.

[0006] This application also provides a distributed energy storage collaborative scheduling system for energy integration, the system comprising: a data acquisition module for collecting traffic energy consumption characteristic data and distribution network operation status data, and constructing a traffic-distribution network coupled spatiotemporal data map; a scheduling strategy formulation module for performing regional collaborative control based on the traffic-distribution network coupled spatiotemporal data map, and formulating an initial collaborative scheduling strategy; a control command generation module for constructing a regional traffic digital twin to simulate the execution of the initial collaborative scheduling strategy, generating simulated scheduling results to perform balanced optimization of the initial collaborative scheduling strategy, and generating distributed energy storage collaborative control commands; and a scheduling scheme generation module for associating the distributed energy storage collaborative control commands with multiple traffic energy consumption nodes to generate an energy integration collaborative scheduling scheme.

[0007] This application proposes a distributed energy storage collaborative scheduling method and system for integrated transportation and energy use. The method involves collecting traffic energy consumption characteristic data and distribution network operation status data to construct a spatiotemporal data map of traffic-distribution network coupling. It then performs regional collaborative control, formulates an initial collaborative scheduling strategy, constructs a regional traffic digital twin for simulation execution, generates simulation scheduling results, performs reverse equilibrium optimization, and generates distributed energy storage collaborative control commands. Finally, it associates these commands with multiple traffic energy consumption nodes to generate a traffic-energy integration collaborative scheduling scheme. This approach addresses the technical challenges of deep integration and spatiotemporal representation of traffic energy data in existing technologies, as well as the lack of dynamic closed-loop verification and adaptive optimization of collaborative control strategies. It achieves the technical effect of improving the collaborative utilization efficiency of distributed energy storage resources in traffic-distribution network coupling scenarios and enhancing the robustness of scheduling strategies to real-time operating conditions. Attached Figure Description

[0008] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments of this disclosure will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0009] Figure 1 This is a schematic diagram of a distributed energy storage collaborative scheduling method for energy exchange integration provided in an embodiment of this application.

[0010] Figure 2 This is a schematic diagram illustrating the process of constructing a transportation-distribution network coupled spatiotemporal data graph in a distributed energy storage collaborative scheduling method for energy integration provided in an embodiment of this application.

[0011] Figure 3 This is a schematic diagram of a distributed energy storage collaborative scheduling system structure for energy exchange integration, provided as an embodiment of this application.

[0012] Explanation of reference numerals in the attached diagram: Data acquisition module 10, scheduling strategy formulation module 20, control command generation module 30, and scheduling scheme generation module 40. Detailed Implementation

[0013] To further illustrate the technical means and effects adopted by the present invention in order to achieve the intended purpose, the following detailed description is provided in conjunction with the accompanying drawings and preferred embodiments, based on the specific implementation methods, structures, features and effects of the present invention.

[0014] This application provides a distributed energy storage collaborative scheduling method for energy convergence, such as... Figure 1 As shown, the method includes: Step S100: Collect traffic energy consumption characteristic data and power distribution network operation status data to construct a traffic-power distribution network coupled spatiotemporal data map.

[0015] Step S100 further includes: collecting traffic energy consumption characteristic data through roadside sensing devices and collecting power distribution network operation status data through power distribution automation terminals; performing multi-dimensional state identification on the traffic energy consumption characteristic data to construct a traffic flow state energy matrix; extracting electrical features from the power distribution network operation status data to construct a power grid operation status feature matrix; performing node energy supply and demand mapping based on the traffic flow state energy matrix and the power grid operation status feature matrix to construct an energy supply and demand relationship; performing spatiotemporal alignment and fusion of the traffic flow state energy matrix and the power grid operation status feature matrix according to the energy supply and demand relationship to construct an initial coupled spatiotemporal data graph; and optimizing the graph structure of the initial coupled spatiotemporal data graph to construct a traffic-power distribution network coupled spatiotemporal data graph.

[0016] Preferably, traffic energy consumption characteristic data are collected through roadside sensing devices such as highway gantry systems, traffic radar, checkpoint cameras, geomagnetic coils, and vehicle-road cooperative roadside units. This includes, but is not limited to, instantaneous vehicle speed, headway, lane occupancy rate, vehicle type classification, and vehicle destination / route planning information. These data represent a set of traffic behavior indicators that characterize the future charging load demand of the node, such as traffic flow per unit time, average dwell time, estimated distribution of remaining charge (SOC), and expected charging probability. Distribution network operation status data are collected through distribution automation terminals such as feeder terminal units (FTU), substation terminal units (DTU), distribution transformer monitoring terminals (TTU), and smart meters / PMUs. This data includes node voltage amplitude and phase angle, line active / reactive power, current amplitude, transformer tap status, and distributed power generation (photovoltaic / wind power) output values, characterizing the real-time electrical operation profile of each node in the distribution network.

[0017] Preferably, the raw traffic data collected by roadside sensing devices is organized by spatial nodes (e.g., each service area / charging station) and time windows (e.g., every 5 minutes). Then, for each node within each time window, multiple traffic state dimensions are calculated, including traffic flow rate, average vehicle speed, time occupancy, queue length, and predicted charging demand intensity. These are combined into a two-dimensional matrix to determine the traffic flow dynamic energy matrix. Here, the row index is the spatial node number, the column index is the time section number, and the matrix elements are multi-dimensional feature vectors. This matrix is ​​continuously updated as the time window slides. Data collected by the distribution automation terminal... The electrical quantities are organized according to the same set of spatial nodes (distribution network node numbers) and time windows. After filtering, denoising and state estimation of the original electrical quantities, the key electrical indicators of each node are calculated, including voltage per unit value, node injected active power, node injected reactive power, line load rate, node short-circuit capacity margin, etc., and combined into a matrix to determine the power grid operating state characteristic matrix. In this matrix, the row index is the distribution network node number, the column index is the time section number, and the matrix elements are multi-dimensional electrical feature vectors, which characterize the real-time electrical safety margin and power carrying capacity of each node in the distribution network at each time section.

[0018] Preferably, traffic nodes and distribution network nodes are not in a one-to-one correspondence in physical topology. A single distribution network feeder may supply power to multiple service areas, and a large service area may be supplied by multiple feeders working together. Therefore, node energy supply and demand mapping is performed based on the traffic flow state energy matrix and the power grid operation state characteristic matrix. Specifically, taking the distribution network node as a reference, the energy demand in the traffic flow state energy matrix is ​​powered and superimposed along the power supply topology path and mapped to the distribution network node supplying it. For example, highway service area A is supplied by distribution network node N. 10 If the power supply has an expected charging demand of P_A, then P_A is labeled as node N. 10 The "traffic-derived load" attribute of the distribution network node, and the remaining capacity of the distribution network node (rated capacity minus current load) constitutes the power supply capacity of the node; after the mapping is completed, each distribution network node obtains two attributes at the same time: the total expected energy demand of traffic nodes in the power supply area of ​​the node and the power adjustment margin of the node itself, thus forming the energy supply and demand relationship at the node level.

[0019] Preferably, the time granularity of traffic data differs from that of power grid data. Time alignment operations unify the two to the same time baseline. For example, using a coarser granularity as a benchmark, the power grid data is averaged or sampled cross-sections according to time windows to ensure that the time indices of the two matrices are completely consistent. The coordinates of traffic nodes are mapped to the geographic-topological coordinate space of distribution network nodes, ensuring that the load demand of each traffic node is associated with the correct power supply distribution network node. The traffic situation feature vector in the traffic flow potential energy matrix and the electrical feature vector in the power grid operation status feature matrix are spliced ​​or aggregated according to the same index to form an enhanced feature vector containing information from both sides. This is then organized in a graph structure, where the nodes of the graph are distribution network feeder nodes, i.e., each node carries the fused feature vector. The edges of the graph are the power supply topology connections of the distribution network, i.e., the electrical connection relationships between the bus, feeder, transformer, and load. Each edge carries impedance parameters and current transmission power to determine the initial coupled spatiotemporal data graph.

[0020] Preferably, the initial coupled spatiotemporal data graph directly adopts the original distribution network topology. However, this may result in an excessive number of nodes, some nodes being electrically close but having a high topology hop count, or redundant intermediate nodes affecting graph computation efficiency. Therefore, graph structure optimization is performed, such as... Figure 2 As shown, multiple nodes within the same power supply area that are electrically coupled and have consistent control behaviors are aggregated into a supernode to reduce the graph size; the intensity of traffic-power grid interaction is introduced as an auxiliary edge weight; strong coupling virtual edges are added to nodes that are electrically close but topologically non-adjacent, so that the graph structure more accurately reflects the real physical-information coupling relationship; nodes with no traffic load but pure residential load are marked as pure power grid nodes and given different processing strategies in graph calculation; finally, a spatiotemporal data graph of traffic-distribution network coupling is obtained, in which the feature vector of each node already contains information from both traffic and power grid.

[0021] Step S200: Based on the transportation-distribution network coupled spatiotemporal data map, perform regional collaborative control and formulate an initial collaborative scheduling strategy.

[0022] Step S200 further includes: performing spatiotemporal feature learning based on the transportation-distribution network coupled spatiotemporal data map to generate a local transportation-distribution network interaction dynamic vector; performing hierarchical aggregation of the nodes of the transportation-distribution network coupled spatiotemporal data map according to the local transportation-distribution network interaction dynamic vector to obtain a multi-layer aggregation result; performing regulation partitioning according to the multi-layer aggregation result to construct a multi-granularity collaborative control unit; performing multi-agent collaborative decision allocation based on the multi-granularity collaborative control unit to determine multiple decision agents to interact and coordinate, and generating an initial power adjustment command; performing consistency verification on the initial power adjustment command, and formulating the initial collaborative scheduling strategy based on the verification result.

[0023] Preferably, the spatial and temporal dimensions of the traffic-power distribution network coupled spatiotemporal data graph are extracted simultaneously. For each target node in the graph, the feature information of adjacent nodes is aggregated along its adjacent edges. The directionality and weight of the edges are considered during aggregation, so that the feature update of each node includes the state influence of its electrical neighbors and traffic-related neighbors. The graph features of the current time segment are used to perform temporal correlation modeling with the graph features of the historical time segment to extract the trend and periodicity of the state evolution of each node over time. The local traffic-power grid interaction dynamic vector is output, which represents the degree and direction of the influence of traffic situation changes in the local area where the node is located on the power grid state of the node under the current time segment, as well as the reverse constraint ability of the power grid operation state of the node on the traffic nodes in its power supply area. The specific numerical values ​​characterize the intensity, response delay and coupling direction of the interaction influence between the two systems. The local dimension reflects the differences in the energy interaction characteristics of different spatial locations.

[0024] Preferably, the topology of the input local traffic-grid interaction dynamic vector and the original data graph is used to group nodes with similar local interaction dynamic vectors into the same group to form a supernode at the next higher level. Specifically, with the original nodes as the finest granularity layer, a similarity threshold is set based on the similarity of the local interaction dynamic vectors, such as cosine similarity or Euclidean distance. Adjacent or topologically similar nodes with vector similarity higher than the threshold are merged to form supernodes. The feature of each supernode is a weighted average of the features of the original nodes it contains, with the weight being the node capacity or importance coefficient. Taking the supernode as the object, its corresponding upper-level aggregation vector is compared and merged again to form another supernode. This process continues until a preset number of aggregation layers is reached (such as 3 or 4 layers), or until the number of nodes in the current layer drops below a preset threshold. The multi-layer aggregation result is output, which includes the node set and its membership relationship for each layer from the finest to the coarsest granularity. Each layer corresponds to a different spatial scale, with the bottom layer being a single energy storage / load node, the middle layer being a power supply area, and the top layer being the entire control area.

[0025] Preferably, based on the aggregation grouping of a specific layer in the hierarchical aggregation result, the spatial range covered by each supernode of that layer is defined as a control partition. Each control partition contains all the original nodes and sub-supernodes of its next layer, forming a spatially continuous region with similar interactive influence characteristics. The number of partitions is determined by the number of supernodes in that layer. Then, the control authority levels assigned to each control partition are combined. That is, for each control partition, the control granularity level adopted by the partition is determined according to its layer number in the hierarchical aggregation result and the degree of heterogeneity of the nodes within the partition. Among them, the control granularity of the partition in the finer aggregation layer is high, and instructions are issued independently to each energy storage unit in the partition; the control granularity of the partition in the coarser aggregation layer is low, and the total power instruction is issued to the partition as a whole, with internal allocation autonomously by the local controller. Different partitions obtain different control granularities, forming a multi-granularity collaborative control unit.

[0026] Preferably, each multi-granularity collaborative control unit is mapped as a decision-making agent, capable of observing the traffic situation and power grid status within its own zone, outputting power regulation actions within the allowable range of its own control granularity, interacting with other entities within its own zone and decision-making agents in adjacent zones, and then performing multi-agent collaborative decision allocation. That is, the global optimization objectives such as minimizing network loss, minimizing voltage over-limit risk, and maximizing energy storage utilization are decomposed into sub-objectives for each decision-making agent according to the adjustable capacity ratio and response speed differences of each zone. While pursuing its own sub-objective, each agent needs to coordinate and constrain each other through communication interaction between agents. For example, there is electrical coupling between adjacent zones, and the power adjustment of one zone affects the voltage and power flow of adjacent zones. Each agent reaches a locally feasible solution through multiple rounds of interaction, and the summaries form an initial power regulation instruction set, with each energy storage unit or each control unit corresponding to a power regulation value.

[0027] Preferably, the initial power regulation commands may exhibit two types of inconsistencies at the global level: power balance inconsistency (the sum of discharge commands and charging commands from all energy storage units does not match, leading to a total system power imbalance) and safety constraint inconsistency (after independent regulation by each zone, the voltage amplitude or power flow direction calculated at the zone boundary nodes is discontinuous, violating Kirchhoff's laws). In such cases, the initial power regulation commands are aggregated to the global coordinator, substituted into the distribution network power flow model for global Newton-Raphson power flow calculation, and then the system's active / reactive power balance, whether the voltage at each node is within safe operating limits, and whether each line / transformer is overloaded are checked. If all checks pass, the command set is directly output as the initial coordinated scheduling strategy. If the checks fail, nodes exceeding limits or power imbalances are identified and fed back as correction values ​​to the corresponding decision-making agent, requiring it to readjust local commands and re-aggregate and check until global consistency is met. The final initial coordinated scheduling strategy includes the active / reactive power setpoints for each energy storage unit or each multi-granularity control unit in the next control cycle, as well as the response time windows for each unit.

[0028] Furthermore, step S200 also includes: introducing a graph attention network to perform multi-scale message passing along multiple types of edges of the traffic-power grid coupled spatiotemporal data graph, extracting a multi-scale dataset, wherein the multi-scale dataset contains traffic situation change data of multiple nodes and power grid state fluctuation impact data of multiple nodes; introducing a temporal convolutional network to perform node evolution analysis along the time axis to obtain node state change trend parameters of multiple historical time steps; performing spatial processing on the traffic situation change data of multiple nodes and the power grid state fluctuation impact data of multiple nodes to construct spatial neighborhood information, performing temporal compression on the node state change trend parameters of multiple historical time steps to construct temporal context information; and performing hybrid learning on the spatial neighborhood information and the temporal context information to generate the local traffic-power grid interaction impact dynamic vector.

[0029] Preferably, a multi-type edge input graph attention network is used to input the spatiotemporal data graph of the traffic-distribution network coupling. For each target node in the graph, when aggregating the features of its neighboring nodes, an attention weight is learned for each neighboring node, with a value ranging from 0 to 1. The sum of the weights of all neighbors of the same node is normalized to 1. The magnitude of this weight depends on the correlation between the features of the target node and the features of its neighboring nodes; the stronger the correlation, the higher the weight, and the greater the contribution of the neighbor's features to the update of the target node's features. Calculations are performed along different types of edges at different graph diffusion scales (e.g., 1-hop neighbors, 2-hop neighbors, 3-hop neighbors). Each distribution network node obtains electrical quantity information from adjacent distribution network nodes, and the attention weights are... Depending on the tightness of electrical coupling, each traffic node obtains traffic situation information from neighboring traffic nodes, and the attention weight depends on the correlation of traffic flow. Distribution network nodes obtain energy demand information from the traffic nodes they supply, and traffic nodes obtain power margin information from the distribution network nodes that supply them, and the attention weight depends on the power supply capacity ratio. Among them, 1-hop aggregation obtains direct neighborhood influence, 2-hop aggregation obtains indirect influence, and 3-hop aggregation obtains influence over a wider range. The output is a multi-scale dataset, which contains the updated feature vector of each node after aggregation of different types of side information at different spatial scales, including traffic situation change data of multiple nodes and power grid state fluctuation influence data of multiple nodes.

[0030] Preferably, each node is input with its feature sequences across multiple historical time steps, which are then subjected to temporal convolution. This involves using causal convolution and dilated convolution in the time dimension to ensure that the output at time t depends only on the input at time t and earlier. Fixed-interval holes are inserted between the elements of the convolution kernel to allow the kernel to cover a longer time range without increasing the number of parameters. Through multi-layer stacking, with each layer employing a different dilation rate, the network can capture the evolutionary patterns at different time scales, from short-term mutations to long-term trends. The output parameters represent the node's state change trend across multiple historical time steps, including but not limited to the first-order difference of the feature sequences, the second-order difference of the feature sequences, the autocorrelation coefficient of the feature sequences, and the extreme values ​​and magnitudes of the feature sequences. These parameters constitute a quantitative description of the node's state change trend within the historical time window, meaning each node obtains a trend parameter vector.

[0031] Preferably, the traffic situation change data and power grid state fluctuation impact data of multiple nodes are spatially processed. For each target node, the feature data of all nodes in its spatial neighborhood are summarized. The feature vectors of each node in the neighborhood are hierarchically encoded according to the neighborhood distance. The features of the 1-hop neighboring node enter the first encoding channel, the features of the 2-hop neighboring node enter the second encoding channel, and the features of the 3-hop neighboring node enter the third encoding channel. Within each channel, the features of the neighboring nodes are subjected to max pooling or average pooling with the sorting unchanged to obtain the spatial neighborhood information vector of that distance layer, which represents the sum of the spatial impact of the state of all nodes within different spatial radii around the node on the node.

[0032] Preferably, the trend parameters of each node at multiple historical time steps are arranged into a sequence in chronological order and input into a temporal compression module, such as the encoder part of a bidirectional long short-term memory network, to compress it into a fixed-length vector. The compression process preserves key temporal features in the sequence, including long-term dependencies, abrupt change moments, and periodic patterns, to determine the temporal context information vector of the node. Then, spatial neighborhood information and temporal context information are fused. This fusion process is optimized through end-to-end backpropagation during the training phase to learn the optimal interaction encoding method between spatial and temporal features. The fused vector undergoes nonlinear transformation and dimensionality reduction through multiple fully connected layers, outputting a fixed-length vector. Each dimension of the vector represents a certain aspect of the traffic-power grid interaction, and finally outputs a local traffic-power grid interaction dynamic vector. This vector represents a compact numerical encoding of the traffic-power grid interaction state of the local area where the node is located, considering the node's own state, the traffic and power grid states of all nodes in its spatial neighborhood, and the evolution of the node within a historical time window at the current time segment.

[0033] Furthermore, step S200 also includes: traversing the multi-layer aggregation results to extract multiple regional aggregation nodes, the multiple regional aggregation nodes containing regional aggregation dynamic vectors; performing similarity analysis on the regional aggregation dynamic vectors according to the multiple regional aggregation nodes to calculate dynamic vector similarity; automatically dividing the multiple regional aggregation nodes based on the dynamic vector similarity to obtain multiple control partitions, the multiple control partitions having coupling characteristics; and performing multi-layer control granularity matching between the multiple control partitions and the multiple regional aggregation nodes according to the coupling characteristics to construct the multi-granularity collaborative control unit.

[0034] Preferably, a specific layer is selected from the input multi-layer aggregation results. Each supernode in the layer is traversed and extracted as a regional aggregation node. The number of regional aggregation nodes is equal to the total number of supernodes in the layer. Each time a supernode is aggregated, the comprehensive feature vector of the supernode is calculated. This vector is obtained by normalizing the weighted average of the feature vectors of all nodes in its lower layer. The regional aggregation dynamic vector of the node is determined, which represents the comprehensive numerical characterization of the traffic-grid interaction characteristics of all original nodes covered by the regional aggregation node. This includes the weighted fusion of information such as the total traffic load demand of all nodes in the region, the total adjustable capacity of the power grid, and the intensity of the overall traffic-grid interaction of the region.

[0035] Preferably, cosine similarity or Pearson correlation coefficient is used to pairwise combine all regional aggregation nodes and calculate the dynamic vector similarity between nodes. Two regional aggregation nodes with high similarity indicate that the dominant mode of traffic-grid interaction is consistent. For example, if the vectors of two nodes have high grid constraint dimension values ​​and low traffic demand dimension values, it indicates that they belong to the "grid-constrained" region; or if the vectors of two nodes show similar periodic peak-valley patterns in the traffic temporal fluctuation dimension. The regional aggregation nodes are used as graph nodes, and similarity is used as the connection weight between nodes to construct a similarity graph. Then, Laplace matrix eigenvalue decomposition is performed on the graph to map the nodes to a low-dimensional feature space. K-means clustering is performed in the low-dimensional space to finally divide the nodes into several groups, obtaining multiple control zones. Multiple control zones have coupling characteristics. For example, urban central business districts may exhibit coupling characteristics of "high traffic load, low grid margin, and daytime bi-peak type", while suburban highway service areas may exhibit coupling characteristics of "drastic traffic load fluctuation, sufficient grid margin, and holiday impact type". Finally, for each control partition, the regional aggregation nodes contained in the partition are identified, and the appropriate level of control granularity is determined based on the coupling characteristics of the partition. Among them, nodes with high heterogeneity and high response speed requirements in the coupling characteristics are matched with fine-grained control, while nodes with high consistency and low response speed requirements in the coupling characteristics are matched with coarse-grained control. Finally, the spatial boundary, coupling characteristic type, and matched control level of each control partition are encapsulated into a control unit to obtain a multi-granularity collaborative control unit.

[0036] Step S300: Construct a regional traffic digital twin to simulate the execution of the initial collaborative scheduling strategy, generate simulated scheduling results to perform balanced optimization on the initial collaborative scheduling strategy, and generate distributed energy storage collaborative control instructions.

[0037] Step S300 further includes: synchronizing the initial collaborative scheduling strategy to the regional traffic digital twin for parallel simulation execution at multiple time scales to generate simulation scheduling results; performing cooperative game theory based on the simulation scheduling results to calculate the cooperative game contribution; performing equilibrium optimization on the initial collaborative scheduling strategy according to the cooperative game contribution, generating equilibrium optimization results to modify the initial collaborative scheduling strategy, and generating a strategy modification parameter set; and performing energy storage control integration based on the strategy modification parameter set to generate distributed energy storage collaborative control instructions.

[0038] Furthermore, step S300 also includes activating the parallel simulation engine of the digital twin, the parallel simulation engine comprising a traffic flow simulation module, a charging behavior simulation module, and a power grid simulation module; and performing joint simulation synchronously and iteratively on the traffic flow simulation module, the charging behavior simulation module, and the power grid simulation module: S1: driving the traffic flow simulation module and the charging behavior simulation module to perform minute-level updates, generating traffic flow simulation update data and charging behavior simulation update data; S2: driving the power grid simulation module to perform second-level updates, generating power grid simulation data; and performing closed-loop deduction based on the traffic flow simulation update data, the charging behavior simulation update data, and the power grid simulation data to generate the simulation scheduling result.

[0039] Preferably, the regional traffic digital twin is a virtual simulation component constructed in digital space that completely corresponds to the real physical area. It includes three core simulation modules: a traffic flow simulation module for simulating vehicle driving, route selection, and entry / exit from service areas; a charging behavior simulation module for simulating charging decisions, charging power selection, and SOC changes of electric vehicles after arriving at a service area; and a power grid simulation module for simulating power flow distribution, node voltage changes, and line power transmission. The power command values ​​in the initial coordinated scheduling strategy are used as external control inputs for each energy storage unit in the digital twin, and written into the energy storage control interface of the digital twin to execute the strategy. The physical processes of the three simulation modules have different time constants and rates of change. The time scale of traffic flow changes is on the order of minutes, such as vehicle... The arrival rate is statistically measured in 5-15 minute windows, the power change scale of charging behavior is on the minute level, the power adjustment cycle of charging piles is usually on the 1-5 minute level, and the time scale of the electromagnetic transients and electromechanical transients of the power grid is on the second level or even the millisecond level. Specifically, under the premise of synchronous progress in physical time, the three modules each use a time step that matches their own dynamic characteristics for progress calculation. The traffic flow module and the charging behavior module update the status on the minute level, and the power grid module updates the status on the second level. Finally, the simulation scheduling results are output, including but not limited to the voltage amplitude of each node at each time section, the transmission power and load rate of each line at each time section, the actual charging and discharging power and SOC change curve of each energy storage unit at each time section, the charging pile utilization rate and queuing waiting time of each traffic node, and the records of network loss and voltage over-limit of the entire system.

[0040] Preferably, each energy storage unit or each multi-granularity control unit is regarded as a player in a cooperative game, and the contribution of the cooperative game is quantified by Shapley value. Specifically, for any energy storage unit, it is "removed" from the current set of players, and the difference in global payoff before and after removal is calculated. The larger the absolute value of the difference, the greater the contribution of the energy storage unit to the global payoff. Each energy storage unit is removed one by one, and the system operation after the removal of the unit is re-simulated. The marginal decrease in global payoff is calculated, and the contribution of the cooperative game is output. This value is positive. The larger the value, the greater the role of the unit in achieving the global scheduling goal. The contribution is positively correlated with the capacity, response speed, installation location of the energy storage unit, and its actual usage in the simulation process.

[0041] Furthermore, step S300 also includes: sorting the cooperative game contribution values ​​to construct a contribution value sequence; performing extreme value matching between the contribution value sequence and the distributed energy storage units to determine the first energy storage unit and the second energy storage unit, wherein the first energy storage unit has the largest contribution value and the second energy storage unit has the smallest contribution value; adding an adjustment task to the second energy storage unit, calculating the release adjustment space, allocating the release adjustment space to the first energy storage unit for optimization, and generating the equilibrium optimization result.

[0042] Preferably, the system performs equilibrium optimization by inputting the cooperative game contribution of each energy storage unit and the initial collaborative scheduling strategy. The goal is to transfer the scheduling load from units with low contribution to units with high contribution. Specifically, the contributions of each unit are sorted, and the units with the highest and lowest contributions are identified. For low-contribution units, the power regulation command value allocated to them in the initial strategy is reduced, releasing part of their adjustable capacity. The released regulation capacity is then redistributed to high-contribution units. This process is iterated repeatedly, and the new contribution distribution after each adjustment is checked until the contribution gap between units is reduced to within a preset threshold range, such as the maximum contribution. If the ratio to the minimum contribution is less than 1.2, after iterative adjustment, the power command values ​​of all energy storage units will shift relative to the initial strategy, i.e., some units will increase and some units will decrease, forming a strategy correction parameter set. Each energy storage unit corresponds to a correction amount, indicating the magnitude by which the power command of that unit needs to be adjusted up or down from the initial value. Then, each correction amount in the strategy correction parameter set is superimposed on the power command value of the corresponding unit in the initial collaborative scheduling strategy to obtain the corrected new command value. Finally, it is converted into a distributed energy storage collaborative control command output, which at least includes the command batch number and generation time, command validity period, and control parameters of each energy storage unit.

[0043] Step S400: Associate the distributed energy storage collaborative control command with multiple transportation energy consumption nodes to generate a transportation energy fusion collaborative scheduling scheme.

[0044] Step S400 further includes: introducing multiple traffic energy consumption nodes, creating profiles for the multiple traffic energy consumption nodes to obtain multi-dimensional attribute profile parameters; performing correlation analysis between the multiple traffic energy consumption nodes and distributed energy storage units based on the multi-dimensional attribute profile parameters to construct correlation relationships; decomposing the distributed energy storage collaborative control instructions layer by layer to the multiple traffic energy consumption nodes according to the correlation relationships to construct a node-level scheduling scheme; constructing distribution network operation constraints, coordinating and verifying the node-level scheduling scheme according to the distribution network operation constraints, and generating the traffic energy fusion collaborative scheduling scheme.

[0045] Preferably, multiple traffic energy consumption nodes are acquired, namely all physical terminals or stations related to traffic energy consumption, including but not limited to highway service area charging stations, urban public fast charging stations, bus-dedicated charging stations, battery swapping stations, and parking facilities with V2G functionality. Each node occupies a specific geographical location, has an electrical connection with a specific power supply node in the distribution network, and its load characteristics are determined by the charging behavior of passing / parking electric vehicles. For each traffic energy consumption node, its static and dynamic attributes are collected and calculated from multiple dimensions to form a comprehensive description of the node's energy consumption behavior characteristics. Specifically, this includes static dimensions such as node type, number and type of charging piles, transformer capacity, and rated capacity of energy storage devices; quasi-static dimensions such as average daily service vehicle trips, average daily charging volume, and average daily charging duration; and dynamic dimensions such as the current real-time number of vehicles in queue and estimated queuing time, the current SOC distribution of vehicles charging and remaining charging time, and the current node's total charging power and adjustable potential. Finally, multi-dimensional attribute profile parameters are output to fully characterize all the key features of the node as a controlled unit in the traffic energy integration system.

[0046] Preferably, a correlation analysis is performed on multiple traffic energy consumption nodes and distributed energy storage units based on multi-dimensional attribute profile parameters. The electrical distance between the traffic energy consumption nodes and the energy storage units is the primary criterion, supplemented by a comprehensive evaluation based on supply and demand matching. Specifically, the impedance value or topology hop count along the power supply topology path from the distribution network node where the energy storage unit is connected to the distribution network node where the traffic energy consumption node is connected is considered. The smaller the electrical distance, the more direct the support effect of the energy storage unit on the load change of the traffic node and the shorter the response delay. An electrical distance threshold is set, such as an impedance modulus not exceeding 0.05 pu or a topology hop count not exceeding 0.05 pu. After three hops, energy storage units with electrical distances below a threshold are added to the candidate set for association of the traffic node. Then, the matching degree between the energy storage unit's capacity and the traffic node's predicted load demand is further evaluated. The matching degree is calculated as the energy storage unit's maximum discharge power divided by the traffic node's predicted peak load. A matching degree closer to 1 indicates a higher matching degree, meaning the energy storage can just cover the node's load fluctuations. A matching degree much less than 1 indicates insufficient support capacity; a matching degree much greater than 1 indicates severe overcapacity and poor economic efficiency. A reasonable matching degree range is set, such as 0.8 to 1.5, and energy storage units with matching degrees outside this range are removed. Finally, each traffic energy consumption node is associated with one or more energy storage units, and the association relationship is output.

[0047] Preferably, the power command of the energy storage unit is traced back to the traffic energy consumption node it serves for energy storage power decomposition. Specifically, for each energy storage unit, the set of all associated traffic energy consumption nodes is determined, and the current load demand forecast and voltage sensitivity factor of each node are obtained, representing the degree of impact of the node's unit load change on the distribution network voltage. The allocation weight of each node is calculated, and the node with larger load demand and higher voltage sensitivity receives a larger share of energy storage power allocation. Then, the commanded power of the energy storage unit is allocated to each associated node according to the weight. For each traffic energy consumption node, the allocated power obtained from all associated energy storage units is summed to obtain the total energy storage support power of the node. Then, the total energy storage support power of all traffic energy consumption nodes at each time segment is arranged in chronological order. Combining the node's own load forecast value and energy storage support power, the net load curve of the node after energy storage support is calculated, where net load = original forecast load - energy storage support power. The node-level scheduling scheme is output, including the original load forecast value, allocated energy storage support power value, and net load value of each node at each time segment.

[0048] Preferably, the safety boundary conditions that the distribution network must meet during operation are determined, including but not limited to node voltage constraints, branch power flow constraints, node power balance constraints, transformer capacity constraints, and energy storage operation constraints. Then, the node-level scheduling scheme is substituted into the distribution network power flow calculation model to calculate the voltage amplitude of each node, the power of each line, and the load rate of the entire network at each time segment. The power flow calculation results are compared with the distribution network operation constraints item by item. If all constraints are met, the verification is passed. If the voltage of a certain node is lower than the lower limit or higher than the upper limit at a certain time segment, or the load rate of a certain line exceeds the rated value, a coordinated correction is triggered. Finally, the verified node-level scheduling scheme is output as a traffic energy fusion collaborative scheduling scheme, which includes the original load forecast value of each traffic energy consumption node at each time segment, the total energy storage support power obtained by each traffic energy consumption node at each time segment, the net load of each traffic energy consumption node at each time segment, the final output power command value of each distributed energy storage unit at each time segment, and the power flow calculation results of the entire network at each time segment, as well as the safety verification pass indicator.

[0049] In the above text, refer to Figure 1 A method for coordinated scheduling of distributed energy storage oriented towards energy exchange integration according to an embodiment of the present invention is described in detail. Next, reference will be made to... Figure 3 This invention describes a distributed energy storage collaborative scheduling system oriented towards energy exchange integration.

[0050] According to an embodiment of the present invention, a distributed energy storage collaborative scheduling system oriented towards energy exchange integration is used to solve the technical problems existing in the prior art, such as the difficulty in deep integration and spatiotemporal representation of energy exchange data, and the lack of dynamic closed-loop verification and adaptive optimization of collaborative control strategies. This achieves the technical effect of improving the collaborative utilization efficiency of distributed energy storage resources in transportation-distribution network coupled scenarios and the robust adaptability of scheduling strategies to real-time operating conditions. Figure 3 As shown, a distributed energy storage collaborative scheduling system for energy exchange integration includes: a data acquisition module 10, a scheduling strategy formulation module 20, a control command generation module 30, and a scheduling scheme generation module 40.

[0051] The data acquisition module 10 is used to collect traffic energy consumption characteristic data and distribution network operation status data to construct a traffic-distribution network coupled spatiotemporal data map; the scheduling strategy formulation module 20 is used to perform regional collaborative control based on the traffic-distribution network coupled spatiotemporal data map and formulate an initial collaborative scheduling strategy; the control command generation module 30 is used to construct a regional traffic digital twin to simulate the execution of the initial collaborative scheduling strategy, generate simulated scheduling results to perform balanced optimization of the initial collaborative scheduling strategy, and generate distributed energy storage collaborative control commands; the scheduling scheme generation module 40 is used to associate the distributed energy storage collaborative control commands with multiple traffic energy consumption nodes to generate a traffic-energy fusion collaborative scheduling scheme.

[0052] The specific configuration of the data acquisition module 10 will be described in detail below. The data acquisition module 10 further includes: collecting traffic energy consumption characteristic data through roadside sensing devices and collecting power distribution network operation status data through power distribution automation terminals; performing multi-dimensional state identification on the traffic energy consumption characteristic data to construct a traffic flow potential energy matrix; extracting electrical features from the power distribution network operation status data to construct a power grid operation status feature matrix; performing node energy supply and demand mapping based on the traffic flow potential energy matrix and the power grid operation status feature matrix to construct an energy supply and demand relationship; performing spatiotemporal alignment and fusion of the traffic flow potential energy matrix and the power grid operation status feature matrix according to the energy supply and demand relationship to construct an initial coupled spatiotemporal data graph; and optimizing the graph structure of the initial coupled spatiotemporal data graph to construct a traffic-power distribution network coupled spatiotemporal data graph.

[0053] The specific configuration of the scheduling strategy formulation module 20 will be described in detail below. The scheduling strategy formulation module 20 further includes: performing spatiotemporal feature learning based on the traffic-distribution network coupled spatiotemporal data graph to generate a local traffic-distribution network interaction dynamic vector; performing hierarchical aggregation of the nodes in the traffic-distribution network coupled spatiotemporal data graph according to the local traffic-distribution network interaction dynamic vector to obtain multi-level aggregation results; performing control partitioning according to the multi-level aggregation results to construct a multi-granularity collaborative control unit; performing multi-agent collaborative decision allocation based on the multi-granularity collaborative control unit to determine multiple decision-making agents for interactive coordination and generate an initial power adjustment command; performing consistency verification on the initial power adjustment command; and formulating the initial collaborative scheduling strategy based on the verification results.

[0054] The specific configuration of the scheduling strategy formulation module 20 will be described in detail below. The scheduling strategy formulation module 20 further includes: introducing a graph attention network to perform multi-scale message passing along multiple types of edges of the traffic-distribution network coupled spatiotemporal data graph, extracting a multi-scale dataset containing traffic situation change data and power grid state fluctuation impact data for multiple nodes; introducing a temporal convolutional network to perform node evolution analysis along the time axis, obtaining node state change trend parameters for multiple historical time steps; performing spatial processing on the traffic situation change data and power grid state fluctuation impact data for multiple nodes to construct spatial neighborhood information; performing temporal compression on the node state change trend parameters for multiple historical time steps to construct temporal context information; and performing hybrid learning on the spatial neighborhood information and the temporal context information to generate the local traffic-power grid interaction dynamic vector.

[0055] The specific configuration of the scheduling strategy formulation module 20 will be described in detail below. The scheduling strategy formulation module 20 further includes: traversing the multi-layer aggregation results to extract multiple regional aggregation nodes, each containing a regional aggregation dynamic vector; performing similarity analysis on the regional aggregation dynamic vectors according to the multiple regional aggregation nodes to calculate the dynamic vector similarity; automatically dividing the multiple regional aggregation nodes based on the dynamic vector similarity to obtain multiple control partitions, each control partition having coupling characteristics; and performing multi-layer control granularity matching between the multiple control partitions and the multiple regional aggregation nodes according to the coupling characteristics to construct the multi-granularity collaborative control unit.

[0056] The specific configuration of the control command generation module 30 will be described in detail below. The control command generation module 30 further includes: synchronizing the initial collaborative scheduling strategy to the regional traffic digital twin for multi-timescale parallel simulation execution, generating simulated scheduling results; performing cooperative game theory based on the simulated scheduling results, calculating the cooperative game contribution; performing equilibrium optimization on the initial collaborative scheduling strategy according to the cooperative game contribution, generating equilibrium optimization results to modify the initial collaborative scheduling strategy, generating a strategy modification parameter set; and performing energy storage control integration based on the strategy modification parameter set to generate distributed energy storage collaborative control commands.

[0057] The specific configuration of the control command generation module 30 will be described in detail below. The control command generation module 30 further includes: activating the parallel simulation engine of the digital twin, which includes a traffic flow simulation module, a charging behavior simulation module, and a power grid simulation module; performing joint simulation and synchronous iterative operation of the traffic flow simulation module, the charging behavior simulation module, and the power grid simulation module: S1: driving the traffic flow simulation module and the charging behavior simulation module to perform minute-level updates, generating traffic flow simulation update data and charging behavior simulation update data; S2: driving the power grid simulation module to perform second-level updates, generating power grid simulation data; performing closed-loop deduction based on the traffic flow simulation update data, the charging behavior simulation update data, and the power grid simulation data to generate the simulation scheduling result.

[0058] The specific configuration of the control command generation module 30 will be described in detail below. The control command generation module 30 further includes: sorting the cooperative game contribution values ​​to construct a contribution value sequence; performing extreme value matching between the contribution value sequence and the distributed energy storage units to determine a first energy storage unit and a second energy storage unit, wherein the first energy storage unit has the largest contribution value and the second energy storage unit has the smallest contribution value; adding an adjustment task to the second energy storage unit, calculating the release of adjustment space, allocating the released adjustment space to the first energy storage unit for optimization, and generating the equilibrium optimization result.

[0059] The specific configuration of the scheduling scheme generation module 40 will be described in detail below. The scheduling scheme generation module 40 further includes: introducing multiple traffic energy-consuming nodes, creating profiles of these nodes to obtain multi-dimensional attribute profile parameters; performing correlation analysis between the multiple traffic energy-consuming nodes and distributed energy storage units based on the multi-dimensional attribute profile parameters to construct correlation relationships; decomposing the distributed energy storage collaborative control instructions layer by layer to the multiple traffic energy-consuming nodes according to the correlation relationships to construct a node-level scheduling scheme; constructing distribution network operation constraints, coordinating and verifying the node-level scheduling scheme according to the distribution network operation constraints, and generating the traffic energy fusion collaborative scheduling scheme.

[0060] The distributed energy storage collaborative scheduling system for energy exchange integration provided in this embodiment of the invention can execute the distributed energy storage collaborative scheduling method for energy exchange integration provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0061] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A distributed energy storage collaborative scheduling method for energy exchange integration, characterized in that, The method includes: Collect traffic energy consumption characteristic data and power distribution network operation status data to construct a spatiotemporal data map of traffic-power distribution network coupling. include: Traffic energy consumption characteristic data are collected through roadside sensing devices, and power distribution network operation status data are collected through power distribution automation terminals; Multi-dimensional state identification is performed on the traffic energy consumption characteristic data to construct a traffic flow state energy matrix; electrical features are extracted from the power distribution network operation status data to construct a power grid operation status feature matrix. Based on the traffic flow state energy matrix and the power grid operation state characteristic matrix, node energy supply and demand mapping is performed to construct energy supply and demand relationships. Based on the energy supply and demand relationship, the traffic flow state energy matrix and the power grid operation state characteristic matrix are spatiotemporally aligned and fused to construct an initial coupled spatiotemporal data map. The initial coupled spatiotemporal data graph is optimized to construct a transportation-distribution network coupled spatiotemporal data graph. Regional collaborative control is performed based on the aforementioned transportation-distribution network coupled spatiotemporal data map, and an initial collaborative scheduling strategy is formulated. include: Based on the spatiotemporal data graph of the transportation-power grid coupling, spatiotemporal feature learning is performed to generate a dynamic vector of local transportation-power grid interaction influence. Based on the dynamic vector of local traffic-power grid interaction, the nodes of the traffic-distribution network coupled spatiotemporal data graph are hierarchically aggregated to obtain multi-layer aggregation results; According to the multi-layer aggregation results, the control partition is performed, a multi-granularity collaborative control unit is constructed, and multi-agent collaborative decision allocation is performed based on the multi-granularity collaborative control unit. Multiple decision agents are determined to interact and coordinate to generate an initial power adjustment command. The initial power adjustment command is subjected to consistency verification, and the initial cooperative scheduling strategy is formulated based on the verification result. A regional traffic digital twin is constructed to simulate the execution of the initial collaborative scheduling strategy. The simulated scheduling results are then used to perform balanced optimization on the initial collaborative scheduling strategy, and distributed energy storage collaborative control commands are generated. include: The initial collaborative scheduling strategy is synchronized to the regional traffic digital twin for parallel simulation execution at multiple time scales to generate simulation scheduling results; Based on the simulated scheduling results, a cooperative game is conducted, and the contribution of the cooperative game is calculated. The initial collaborative scheduling strategy is optimized based on the contribution of the cooperative game, and the optimization result is used to correct the initial collaborative scheduling strategy, thereby generating a set of strategy correction parameters. Based on the strategy, the parameter set is modified to generate distributed energy storage collaborative control commands through energy storage control integration. The distributed energy storage collaborative control command is associated with multiple transportation energy consumption nodes to generate a transportation energy fusion collaborative scheduling scheme.

2. The distributed energy storage collaborative scheduling method for energy convergence as described in claim 1, characterized in that, Based on the aforementioned traffic-power grid coupled spatiotemporal data graph, spatiotemporal feature learning is performed to generate a local dynamic vector of traffic-power grid interaction influence. The method includes: A graph attention network is introduced to perform multi-scale message passing along the multi-type edges of the traffic-distribution network coupled spatiotemporal data graph, and a multi-scale dataset is extracted. The multi-scale dataset contains traffic situation change data of multiple nodes and power grid state fluctuation impact data of multiple nodes. A temporal convolutional network is introduced to perform node evolution analysis along the time axis, and the node state change trend parameters at multiple historical time steps are obtained. Spatial processing is performed on the traffic situation change data of the multiple nodes and the power grid state fluctuation impact data to construct spatial neighborhood information. The node state change trend parameters of multiple historical time steps are temporally compressed to construct temporal context information. The spatial neighborhood information and the temporal context information are mixed and learned to generate the local traffic-power grid interaction dynamic vector.

3. The distributed energy storage collaborative scheduling method for energy exchange integration as described in claim 1, characterized in that, Based on the aforementioned multi-layer aggregation results, a multi-granularity collaborative control unit is constructed by performing regulatory partitioning. The method includes: The multi-layer aggregation results are traversed to extract multiple regional aggregation nodes, each of which contains a regional aggregation dynamic vector. Similarity analysis is performed on the dynamic vectors of the region aggregation based on the multiple region aggregation nodes, and the dynamic vector similarity is calculated. Based on the dynamic vector similarity, the multiple region aggregation nodes are automatically divided to obtain multiple control partitions, and the multiple control partitions have coupling characteristics. The multiple control partitions are matched with the multiple regional aggregation nodes according to their coupling characteristics to construct the multi-granularity collaborative control unit.

4. The distributed energy storage collaborative scheduling method for energy exchange integration as described in claim 1, characterized in that, The method involves synchronizing the initial collaborative scheduling strategy to the regional traffic digital twin for parallel simulation execution across multiple time scales, generating simulation scheduling results, and includes: The parallel simulation engine of the digital twin is activated, which includes a traffic flow simulation module, a charging behavior simulation module, and a power grid simulation module. The traffic flow simulation module, the charging behavior simulation module, and the power grid simulation module are jointly simulated and run synchronously and iteratively. S1: Drive the traffic flow simulation module and the charging behavior simulation module to perform minute-level updates, generating traffic flow simulation update data and charging behavior simulation update data; S2: Drive the power grid simulation module to perform second-level updates and generate power grid simulation data; The simulation scheduling result is generated by performing closed-loop deduction based on the traffic flow simulation update data, the charging behavior simulation update data, and the power grid simulation data.

5. A distributed energy storage collaborative scheduling method for energy exchange integration as described in claim 1, characterized in that, The initial cooperative scheduling strategy is optimized based on the cooperative game contribution to generate an equilibrium optimization result. The method includes: The contribution values ​​of the cooperative game are sorted to construct a contribution value sequence; Based on the extreme value matching between the contribution sequence and the distributed energy storage unit, the first energy storage unit and the second energy storage unit are determined, wherein the first energy storage unit has the largest contribution value and the second energy storage unit has the smallest contribution value; An adjustment task is added to the second energy storage unit to calculate the release of adjustment space, and the release of adjustment space is allocated to the first energy storage unit for optimization to generate the balance optimization result.

6. The distributed energy storage collaborative scheduling method for energy convergence as described in claim 1, characterized in that, Associating the distributed energy storage collaborative control commands with multiple transportation energy consumption nodes to generate a transportation energy fusion collaborative scheduling scheme, the method includes: Multiple traffic energy consumption nodes are introduced, and the multiple traffic energy consumption nodes are profiled to obtain multi-dimensional attribute profile parameters. Based on the multi-dimensional attribute profile parameters, a correlation analysis is performed on the multiple traffic energy consumption nodes and distributed energy storage units to construct the correlation relationship; Based on the aforementioned relationship, the distributed energy storage collaborative control command is decomposed layer by layer to multiple traffic energy consumption nodes to construct a node-level scheduling scheme. Construct distribution network operation constraints, coordinate and verify the node-level scheduling scheme according to the distribution network operation constraints, and generate the energy-integrated collaborative scheduling scheme.

7. A distributed energy storage collaborative scheduling system for energy exchange integration, characterized in that, The system is used to implement the distributed energy storage collaborative scheduling method for energy exchange integration as described in any one of claims 1 to 6, and the system comprises: The data acquisition module is used to collect traffic energy consumption characteristic data and power distribution network operation status data, and to construct a traffic-power distribution network coupled spatiotemporal data map; The scheduling strategy formulation module is used to perform regional collaborative control based on the transportation-distribution network coupled spatiotemporal data map and formulate an initial collaborative scheduling strategy. The control command generation module is used to construct a regional traffic digital twin to simulate the execution of the initial collaborative scheduling strategy, generate simulated scheduling results to perform balanced optimization on the initial collaborative scheduling strategy, and generate distributed energy storage collaborative control commands. The scheduling scheme generation module is used to associate the distributed energy storage collaborative control command with multiple transportation energy consumption nodes to generate a transportation energy fusion collaborative scheduling scheme.

Images