Urban rail transit flexible direct current traction power supply method based on energy router

Abstract

The invention relates to an urban rail transit flexible direct current traction power supply method based on an energy router. The method comprises the following steps: constructing each structured energy intention corresponding to a train control terminal according to train running state data corresponding to the train control terminal; respectively analyzing space-time information in each structured energy intention to obtain an intention perception map corresponding to the train control terminal; performing energy efficiency analysis on each lattice point in the intention perception map to obtain a lattice point efficiency analysis structure data set; analyzing the power supply behavior of each train corresponding to the train control terminal according to the lattice point efficiency analysis structure data set to obtain power supply behavior execution data corresponding to each train; the power supply behavior execution data is used for carrying out flexible direct-current traction power supply on the train. By adopting the method, the sensing precision and the regulation and control capability of the power supply behavior of the train can be effectively improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent power supply technology, and in particular to a flexible DC traction power supply method for urban rail transit based on an energy router. Background Art

[0002] Traditionally, subway trains have generally used a DC traction power supply system, primarily via a third rail or overhead catenary. The supply voltage is typically DC 750V, DC 1500V, or DC 3000V. Traction substations convert the high-voltage AC power into the required DC voltage, which is then delivered to the train via power lines. The train's traction motors are controlled by traction inverters or choppers for speed regulation and drive. While DC traction power supply systems for subway trains are widely used, real-time monitoring data for these systems is limited, making it difficult to accurately and timely capture dynamic changes in traction voltage, current, and load status. This results in insufficient precision in sensing and controlling the train's power supply behavior. Summary of the Invention

[0003] Based on this, it is necessary to provide an urban rail transit flexible DC traction power supply method based on an energy router that can effectively improve the perception accuracy and control capability of the train's power supply behavior to address the above technical problems.

[0004] In a first aspect, the present application provides an urban rail transit flexible DC traction power supply method based on an energy router, comprising: constructing structured energy intentions corresponding to the train control terminals according to train operation status data corresponding to the train control terminals; Respectively analyzing the spatiotemporal information in each structured energy intention to obtain an intention perception map corresponding to the train control terminal; Performing energy efficiency analysis on each lattice point in the intention perception map to obtain a lattice point efficiency analysis structure data set; According to the lattice point efficiency analysis structure data set, the power supply behavior of each train corresponding to the train control terminal is analyzed to obtain the power supply behavior execution data corresponding to each train; the power supply behavior execution data is used to provide flexible DC traction power supply to the train.

[0005] In a second aspect, the present application also provides an urban rail transit flexible DC traction power supply device based on an energy router, comprising: An energy intention construction module is used to construct each structured energy intention corresponding to the train control terminal according to the train operation status data corresponding to the train control terminal; A perception map obtaining module is used to respectively analyze the spatiotemporal information in each structured energy intention to obtain an intention perception map corresponding to the train control terminal; an efficiency analysis execution module, configured to perform energy efficiency analysis on each lattice point in the intention perception map to obtain a lattice point efficiency analysis structure data set; The power supply data acquisition module is used to analyze the power supply behavior of each train corresponding to the train control terminal according to the lattice point efficiency analysis structure data set, and obtain the power supply behavior execution data corresponding to each train; the power supply behavior execution data is used to provide flexible DC traction power supply to the train.

[0006] The above-mentioned urban rail transit flexible DC traction power supply method and device based on an energy router obtains the train operation status data corresponding to the train control terminal, constructs the corresponding structured energy intent, and then extracts and analyzes the spatiotemporal information in each energy intent to generate an intention perception map reflecting the train's energy demand and operation dynamics. It also performs fine-grained energy efficiency analysis on each lattice point in the map to form a complete lattice point efficiency analysis structure data set. Further, the power supply behavior of the train in different operation stages is deeply modeled and analyzed to obtain the power supply behavior execution data of each train at a specific spatiotemporal position, providing a dynamic and accurate energy scheduling basis for the subsequent flexible DC traction power supply system. It can not only effectively improve the perception accuracy and control capability of the train's power supply behavior, but also significantly enhance the adaptability and energy allocation efficiency of the train traction power supply system to complex operation scenarios, thereby effectively reducing energy consumption, reducing power supply pressure, and improving the intelligence level and operation safety of the entire system. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0008] Figure 1 This is a diagram of an application environment of an urban rail transit flexible DC traction power supply method based on an energy router in one embodiment; Figure 2 This is a flow chart of a flexible DC traction power supply method for urban rail transit based on an energy router in one embodiment; Figure 3 FIG1 is a flow chart of a method for obtaining a structure data set for analyzing the first lattice point efficiency in one embodiment; Figure 4 FIG1 is a flow chart of a method for obtaining a structure data set for analyzing the second lattice point efficiency in one embodiment; Figure 5Schematic diagram of a flow chart of a first method for obtaining dynamic response collaborative analysis data in one embodiment; Figure 6 Schematic diagram of the flow of two methods for obtaining dynamic response collaborative analysis data in one embodiment; Figure 7 Schematic diagram of a process for obtaining a node state evolution tensor in one embodiment; Figure 8 A flowchart of a method for obtaining data from a first power supply behavior in one embodiment; Figure 9 Schematic diagram of a flow chart of a method for obtaining data from a second power supply behavior in one embodiment; Figure 10 A flowchart of a method for obtaining an intention perception map in one embodiment. DETAILED DESCRIPTION

[0009] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0010] The embodiment of the present application provides a flexible DC traction power supply method for urban rail transit based on an energy router, which can be applied to Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store data that server 104 needs to process. The data storage system can be integrated with server 104 or placed on a cloud or other network server. Server 104 can be implemented as a standalone server or a server cluster consisting of multiple servers.

[0011] In an exemplary embodiment, Figure 2 As shown in the figure, a flexible DC traction power supply method for urban rail transit based on energy router is provided. Figure 1 The server in FIG. 1 is used as an example to illustrate the method, including the following steps 202 to 206. Among them: Step 202: constructing structured energy intentions corresponding to the train control terminal according to the train operation status data corresponding to the train control terminal.

[0012] Step 204 , respectively analyze the spatiotemporal information in each structured energy intention to obtain the intention perception map corresponding to the train control terminal.

[0013] Step 206 , performing energy efficiency analysis on each lattice point in the intention perception map to obtain a lattice point efficiency analysis structure data set.

[0014] Step 208 : Analyze the power supply behavior of each train corresponding to the train control terminal according to the lattice point efficiency analysis structure data set to obtain power supply behavior execution data corresponding to each train.

[0015] Among them, the train control terminal can be a computer control system deployed in the urban rail transit dispatching center or section control center, which is used to receive real-time operation status data, structured energy intent, and location information uploaded by all trains in the jurisdiction, and uniformly process, analyze, and identify their behavior, thereby realizing centralized coordination and intelligent decision-making of train operation scheduling and flexible power supply behavior. This terminal usually integrates functional modules such as train operation status management, energy supply intent modeling, graph behavior reconstruction, path scheduling generation, and energy control instruction issuance. It is not only the global perception node of the flexible DC traction power supply system, but also the decision-making center for multi-train behavior prediction, conflict mediation, and optimal allocation of energy supply resources. Its output directly affects the power supply control layer's joint response to each energy router, power supply node, and energy source.

[0016] Among them, the train operation status data can be a collection of multi-dimensional operation information collected and uploaded by the train's sensors in real time, including but not limited to the train's current position (spatial coordinates or track number), current speed and acceleration, current operation stage (such as starting, cruising, deceleration, entering the station), braking and traction mode, remaining power demand, load status and train schedule deviation, etc.

[0017] Among them, structured energy intent can be a multi-field behavior object generated by semantic modeling and behavior recognition of train operation status data, which is used to describe the specific needs and behavioral characteristics of the train for energy supply during future operation. Its typical fields include injection type (such as traction acceleration, regenerative braking), target track segment or position, injection time window, priority level, behavior triggering conditions, energy intensity level, etc. It is parsable, combinable and schedulable, and is the original behavior expression unit of flexible power supply scheduling.

[0018] Among them, the intention perception map can be a graph structure model constructed based on the results of structured energy intention analysis by modeling the spatiotemporal coordinates, behavioral relationships and causal logic of the injection targets. Its nodes represent the track segments or positions pointed to by specific injection behaviors, and the edges represent the temporal relationships, resource dependencies or logical triggering relationships between behaviors. Each node in the map also contains attributes such as behavior type, injection time window, behavior intensity and priority weight, forming an energy supply demand map for multiple trains and multiple behaviors in the spatiotemporal dimension.

[0019] Among them, the lattice point can be the smallest analysis unit formed by discretely dividing the operating space and scheduling time axis of the urban rail network. It usually represents a candidate point where energy injection behavior occurs at a certain track position within a specific time period. The lattice point has both spatiotemporal identification (such as track number + time window) and carries specific behavioral intentions (such as energy injection or energy transfer). It is the basic granularity unit for energy efficiency analysis, behavioral path planning and flexible power supply scheduling execution.

[0020] Among them, energy efficiency analysis can be a process of multi-dimensional quantitative evaluation of the accessibility, execution stability and scheduling economy of a certain lattice point or path behavior in the power supply system, taking into account multiple factors such as the transmission loss, response delay, node status trend, and behavior coordination of the energy supply path. Its goal is to dynamically screen out the energy injection path and strategy combination with the highest execution value and energy utilization efficiency in complex, multi-path, and multi-source energy supply scenarios.

[0021] Among them, the lattice point efficiency analysis structure data set can be a set of structured evaluation results generated for all target lattice points after the energy efficiency analysis is completed. Each record in the data set corresponds to the efficiency evaluation indicators of a lattice point and its optional energy supply path, such as behavioral response success rate, path injection tolerance, node stability score, collaborative capability mark and scheduling risk level, etc., which are used to support the planning and generation of subsequent train power supply behaviors and the overall flexible scheduling optimization of the system.

[0022] Among them, the power supply behavior can be the actual energy injection behavior planned to be executed by the train at a specific time and space position based on the structured energy intention. It is specifically manifested as the whole process of energy transmission from the power supply source node to the target track segment or train node through the energy router and path structure. The power supply behavior not only includes the physical energy flow, but also includes the coordination of control behavior, timing execution judgment and path selection strategy. It is the specific implementation action of the system's execution intention at the behavioral graph level.

[0023] Among them, the power supply behavior execution data can be a structured instruction set or behavior contract data generated for the control system after analyzing the lattice point efficiency structure data and the train intention map. It usually includes fields such as the power supply path number, injection start and end time, list of participating power supply nodes, injection mode (such as synchronization / segmentation), scheduling priority, allowable tolerance range and behavior credit weight, which are used for the actual scheduling execution of the flexible DC power supply system and support behavior feedback and coordination mechanism.

[0024] Among them, flexible DC traction power supply can be an intelligent power supply method based on energy routers and graphical behavior perception mechanisms. Its characteristics are that on the premise of ensuring the continuity and safety of power supply, according to the dynamic operation needs of the train, the status of the power supply system and the path coordination capability, the power supply path, injection method, injection sequence and other execution parameters are adjusted in real time to achieve a flexible combination of multi-path, multi-node and multi-source injection, and with the ability of conflict mediation, behavior prediction and task reconfiguration. It is a new traction power supply method that is different from the traditional centralized rigid power supply method.

[0025] Specifically, based on train operating status data collected by the train control terminal, including but not limited to the current train's position coordinates, speed profile, acceleration, power demand, braking status, load level, train timetable plan, and future target travel sections, the embedded intent recognition model performs semantic modeling and behavior recognition on this status information. Based on pre-set intent type templates (such as traction injection, braking feedback, regenerative braking coordination, and static charging), the operating status is converted into multi-field, semantically clear structured energy intents. These structured energy intents include not only the basic type identifier of the injection behavior, but also the spatial identifier of the target track section or power supply interface (such as track number or train section number), the expected time range or time window for initiating injection (such as 3.20s to 3.24s), the state trigger conditions on which the injection behavior depends (such as control logic events such as "entry completed" and "door closed"), behavior intensity parameters (such as required power level), and behavior priority weights (used as a reference for behavior sorting during conflict resolution).

[0026] After obtaining each structured energy intention, the system first uniformly analyzes the spatial target information (such as track number, train location, and power supply node identifier) and temporal demand information (such as planned injection time window, trigger time, and tolerance range) contained within it. Each injection intention is spatially anchored to a specific track segment, station interval, or power supply access point through the topological mapping of the rail transit network. Temporally, it is converted into a unified time window representation (e.g., relative time intervals based on the train map reference time scale). The multiple parsed energy intentions are then associated and modeled in a unified spatiotemporal coordinate system, constructing a directed graph structure with track segments as nodes and temporal logic or causal dependencies between injection intentions as edges. This is known as the initial intention perception graph. In this initial intention perception graph, each node carries not only spatial location attributes but also the corresponding injection behavior type, target time window, behavior execution priority, and credibility label. Edges convey information such as injection order, behavior linkage conditions, and resource contention probability or sharing opportunities. Then, the graph structure of the initial intention perception map is further optimized and semantically enhanced (such as aggregating repeated targets, identifying time conflicts, and extracting behavioral synergy potential), and finally an intention perception map of the train with a complete topological structure, spatiotemporal distribution semantics, and behavioral causal logic is output.

[0027] Using the spatiotemporal coordinates corresponding to key behavior nodes defined in the intent-aware graph as analysis lattice points, each lattice point is considered an independent candidate region for triggering energy injection behavior. Based on the energy injection demand type, execution time window, and spatial positioning information of that lattice point, the set of candidate energy supply nodes and their connection paths that could potentially constitute effective energy supply are identified. Based on this, a multi-source energy supply transmission structure model is constructed. For each energy supply path, its corresponding path behavior trajectory (i.e., the sequence and timing relationship of energy routing nodes required to travel from the energy supply node to the lattice point) is extracted. Combined with the time trend data on the recent response capability of the energy supply node in the node state evolution tensor, its effectiveness, response stability, and behavior deviation risk within the current scheduling cycle are analyzed. The path behavior trajectories are further reconstructed into behavioral chains to analyze whether there are conflict windows in the timing of each path and whether there is energy injection synergy potential in control, and the energy injection synchronization tolerance between paths is calculated; finally, by fusing and analyzing the multi-path response characteristics with the energy supply node status trends, a multidimensional structured data set is formed, including lattice point positions, energy supply possibilities, path synergy capabilities, energy injection stability scores and behavioral risk indicators, which is called the lattice point efficiency analysis structured data set.

[0028] Based on the injection behavior node corresponding to each train control terminal in the intention perception graph, a comprehensive behavioral planning analysis is performed on the associated lattice point injection demand and energy supply possibilities. This results in a selection of power supply path combinations that meet timing consistency, high path coordination capability, and excellent energy supply stability from the lattice point efficiency analysis structure dataset. Using the feasible injection time window as the scheduling boundary, combined with the prediction results of the response trends of related energy supply nodes in the node state evolution tensor, a set of multi-path coordinated injection candidate solutions is constructed. Each combined path in the solution set is then scored for feasibility and ranked based on multi-dimensional indicators such as maximizing intention completion rate, behavior execution consistency, energy supply conflict probability, and scheduling flexibility to identify the optimal path behavior combination. Based on the train's current operating state and the control characteristics of the required injection behavior, fine-tuning is performed within the feasible injection time window (e.g., advancing or delaying the injection time, switching the injection mode to synchronous or segmented injection). Finally, power supply behavior execution data is generated, including the target path link, energy source node number, injection start and end times, injection mode type, control behavior priority, execution tolerance range, and feedback strategy flag. The power supply behavior execution data not only has highly structured features that can be directly identified and executed by the flexible controller or energy router, but also has behavioral contract attributes, supporting subsequent contract-driven flexible energy injection operations for flexible DC traction power supply to the train.

[0029] In the aforementioned energy router-based flexible DC traction power supply method for urban rail transit, train operating status data corresponding to the train control terminal is acquired to construct a corresponding structured energy intent. The spatiotemporal information within each energy intent is then extracted and analyzed to generate an intent-aware graph reflecting the train's energy demand and operating dynamics. Fine-grained energy efficiency analysis is then performed on each lattice point in the graph, forming a complete lattice point efficiency analysis structured data set. Further, in-depth modeling and analysis of the train's power supply behavior during different operating phases is performed to obtain execution data for each train's power supply behavior at a specific spatiotemporal location, providing a dynamic and accurate basis for energy scheduling for the subsequent flexible DC traction power supply system. This not only effectively improves the perception accuracy and control capabilities of the train's power supply behavior, but also significantly enhances the train's traction power supply system's adaptability to complex operating scenarios and energy allocation efficiency, thereby effectively reducing energy consumption, alleviating power supply pressure, and improving the overall system's intelligence and operational safety.

[0030] In an exemplary embodiment, Figure 3 As shown, energy efficiency analysis is performed on each lattice point in the intention perception map to obtain a lattice point efficiency analysis structure data set, including steps 302 to 306. Step 302 : for any lattice point, according to the energy supply demand of the lattice point, identify the candidate energy supply nodes corresponding to the lattice point from the intention perception map.

[0031] Step 304 : constructing a multi-source energy supply transmission structure model based on the energy transmission paths between the lattice points and each candidate energy supply node.

[0032] Step 306 , solving the energy supply parameters and state parameters in the multi-source energy supply transmission structure model to obtain a lattice point efficiency analysis structure data set.

[0033] Among them, the energy supply demand can be the energy usage request expressed by the train control terminal based on its operating status, behavioral intention and traction conditions in the flexible DC traction power supply system. It is usually manifested as the expected behavior of a train to inject energy into the external power supply system within a specific time and space position range and within a specific time window. Its demand characteristics usually include injection type (such as traction, braking feedback), target power level, injection duration, scheduling priority and behavior triggering conditions.

[0034] Among them, candidate energy supply nodes can be a set of energy nodes that have controllable power supply capabilities, physical path connectivity, and dispatch response potential within the target time window, selected from the entire energy network map after the energy supply demand at a specific lattice point is identified. Based on topological accessibility, path feasibility, and node responsiveness analysis, the system selects these nodes based on topological accessibility, path feasibility, and node responsiveness. This set can include fixed feeder nodes (such as substation outlets), energy routers (supporting multi-path switching), on-board energy storage systems (with feedback capabilities), or other dynamic energy injection nodes, and serves as the starting point for power supply path construction and efficiency evaluation.

[0035] An energy transmission path can be a combination of physical and control links starting from a candidate energy supply node, passing through multiple energy control devices, switching nodes, routers, and relay devices in the power supply network, and ultimately leading to the target lattice point. This path not only includes static physical connection relationships but also includes the switching status of nodes along the path, control response logic, energy transmission mode (constant voltage, current limiting, PWM modulation, etc.), and path scheduling window information. In flexible power supply systems, paths are typically dynamically reconfigurable, and multiple paths can exist in parallel or cross-link to provide coordinated energy.

[0036] The multi-source energy transmission structure model can be a comprehensive network behavior model constructed based on the energy supply requirements of a specific lattice point. It is used to simultaneously describe the overall organizational relationship of all energy supply paths, path structures, control mechanisms, and behavioral characteristics from different supply sources acceptable to that lattice point. The model uses the lattice point as the sink and the candidate energy supply nodes as the multi-source starting points. All available paths are constructed as a directed graph structure. Each node in the path contains energy supply capacity parameters, state behavior characteristics, and control coordination labels.

[0037] Energy supply parameters are a set of key indicators used to quantify the ability of an energy transmission path or energy supply node to transmit electrical energy to a target lattice point within a specific time window. These parameters primarily include adjustable output power range, unit transmission efficiency, path impedance, maximum supported current, voltage stability level, and energy response rate. Determining these parameters requires dynamic estimation based on node state evolution trends, historical behavior records, and path topology characteristics. These parameters are essential data for evaluating path feasibility and energy supply stability.

[0038] Among them, the state parameters can be a set of information that describes the behavioral response characteristics, control consistency and operational uncertainty of the energy supply path and its internal nodes within a specific scheduling cycle, usually including the node response delay prediction value, behavioral jitter amplitude, behavior trigger probability, control synchronization tolerance, node activity, risk factors and historical statistical ratio of path execution failure.

[0039] Specifically, for any lattice point to be analyzed, the system first identifies its energy injection target type (e.g., traction power supply, brake feedback reception, regenerative energy replenishment, etc.), target power level, expected injection time window, path constraints, and response priority based on the behavior annotation information of the lattice point in the intention-aware graph. This allows the system to construct a set of energy supply requirement parameters for that lattice point. Spatial adjacency search and temporal association calculations are then performed across the entire intention-aware graph to identify all energy nodes that have a reachable path relationship with the lattice point in the track topology and are likely to provide power within the injection time window. These include, but are not limited to, fixed power access points (e.g., substation feeder nodes), regional energy routers (relay nodes that support dynamic switching), regenerative train nodes with reverse energy feeding capabilities, energy storage devices, and supercapacitor arrays. After the initial selection of candidate nodes, the historical response capability trend of the power supply nodes recorded in the node state evolution tensor, the current adjustable power range and the control logic characteristics (such as whether collaborative energy injection is supported, whether a master-slave switching mechanism is available, etc.) are further combined to eliminate nodes that are unstable or have execution risks within the target time window, and finally form candidate energy supply nodes that meet path accessibility, behavioral responsiveness and scheduling window synchronization.

[0040] After identifying the candidate energy supply nodes corresponding to a lattice point, the lattice point is used as the target energy supply terminal. Based on the traction power supply network topology (including energy router layout, feeder structure, switchable channels, and control logic), a graph search algorithm (such as multi-source feasible path traversal under time window constraints) is used to construct a set of available energy supply paths from each candidate supply node to the target lattice point. These paths include key elements such as energy control nodes, switchgear, converter groups, and power routing devices along the path. Each intermediate node is annotated with its connection relationship, control mode (e.g., switching control, voltage control, current limiting control), energy loss estimate, and maximum power limit. Furthermore, behavioral dependencies between paths are established, including logical dependencies such as whether the paths can operate in parallel, whether there are shared relay nodes, and whether synchronous startup or master-slave switching is required. Furthermore, time characteristics are incorporated into the path structure, recording the response delay, control execution time, and scheduling tolerance required for each path from initiation at the supply node to arrival at the lattice point. This results in a multi-source, multi-path multi-source energy supply transmission structure model. The multi-source energy transmission structure model ultimately forms a space-time network diagram with energy supply nodes as source points, lattice points as sink points, and energy supply paths as directed edges in the diagram. Each edge carries energy transmission parameters, control behavior characteristics, and synchronization constraint labels.

[0041] Based on each multi-source energy transmission structure model, for each valid energy supply path from a candidate energy supply node to a target lattice point in the multi-source energy transmission structure model, the behavioral capabilities and state evolution trends of key nodes within the path are quantitatively solved, combined with the previously acquired node state evolution tensor. The energy supply parameters and state parameters for the current scheduling period are extracted. Energy supply parameters include, but are not limited to, the path's current maximum available power, unit energy transmission efficiency (taking into account power loss and control response loss), total transmission impedance, load dynamic adaptability, and transmission stability indicators. State parameters include the predicted response delay of each control node in the path, behavior trigger sensitivity, time window in which control intervention may occur, historical scheduling success rate, synchronization coordination tolerance bounds, and behavioral uncertainty measures. Using tensor modeling, these parameters are mapped into multidimensional structured data items indexed by lattice point, path ID, and time window. All available paths are then normalized and evaluated using a unified behavior scoring system to identify the path and energy supply node combinations with the highest behavioral execution value for each lattice point within the current period. Finally, the above analysis results are aggregated into a lattice point efficiency analysis structure dataset. Each record in the lattice point efficiency analysis structure dataset corresponds to a specific lattice point, including a set of optional energy supply paths, a comprehensive energy supply efficiency score for each path, a behavioral risk level, a scheduling feasibility label, and dynamic energy supply trend prediction information.

[0042] In this embodiment, by dynamically identifying candidate energy supply nodes from the intention perception map under the drive of energy supply demand for any lattice point, and further constructing a multi-source energy supply transmission structure model based on the path topology and control connectivity relationship between it and the lattice point, and then combining the energy supply parameters and state parameters carried by each path for quantitative solution and behavior evaluation, it is possible to achieve a systematic and multi-dimensional analysis of the energy injection feasibility, transmission efficiency, power supply risk and scheduling flexibility of the target lattice point, which not only improves the precise adaptation capability of the energy injection behavior, but also significantly enhances the intelligence and flexibility of the power supply path selection, effectively breaking through the limitations of the traditional static path selection and single-source energy supply model in complex urban rail transit scenarios, with strong scheduling rigidity and delayed behavior response, and has higher energy utilization efficiency and system regulation stability.

[0043] In an exemplary embodiment, Figure 4 As shown, solving the energy supply parameters and state parameters in the multi-source energy supply transmission structure model to obtain the lattice point efficiency analysis structure data set includes steps 402 to 408. Among them: Step 402: construct an energy behavior trajectory corresponding to each energy supply path in any multi-source energy supply transmission structure model.

[0044] Step 404 : Perform behavior chain reconstruction analysis on each energy behavior trajectory to obtain dynamic response collaborative analysis data.

[0045] In step 406 , energy supply capability trend analysis and scheduling expectation deviation analysis are performed on each candidate energy supply node in each energy behavior trajectory to obtain a node state evolution tensor.

[0046] Step 408 : Fusing the dynamic response collaborative analysis data and the node state evolution tensor to obtain a lattice point efficiency analysis structure data set.

[0047] Among them, the energy supply path can be an energy transmission link in the flexible DC power supply system, starting from a candidate energy supply node, passing through multiple energy control components (such as energy routers, switching devices, DC bus segments, etc.) and finally reaching the target lattice point. It not only reflects the flow route of electric energy in space, but also includes the control characteristics, connection relationships, scheduling response mechanisms and time synchronization constraints of each node in the path.

[0048] The energy behavior trajectory is the result of analyzing the sequential behavior of the energy supply path at the execution level. It describes the dynamic process of energy scheduling, path switching, and behavioral response of each control node along the path as power is transmitted from the energy supply node to the lattice point. The trajectory not only records each node's role (e.g., initiator, relay, terminal), control mode (e.g., active, passive), and power flow direction, but also includes response timing, execution priority, and behavioral logic relationships, depicting the complete execution chain of the path behavior.

[0049] Behavior chain reconstruction analysis can be a process of parsing and reorganizing the timing dependencies, control trigger mechanisms, and coordinated execution conditions of each behavior node in the energy behavior trajectory. The goal is to identify key behavior segments, bottleneck nodes, timing conflict windows, and resource competition relationships in the energy supply path. By constructing an event sequence chain diagram or behavior dependency diagram, the system can reconstruct the execution logic structure of the behavior path and analyze its compatibility and control risks within the scheduling window, providing a foundation for the coordinated scheduling and behavior synchronization design of multi-path energy supply solutions.

[0050] Dynamic response coordination analysis data can be structured analysis results generated for each energy supply path based on behavioral chain reconstruction analysis, describing its overall response characteristics, behavioral stability, and cross-path coordination capabilities. This data typically includes behavioral chain execution latency, path control conflict probability, key node influence weights, synchronization and coordination tolerance windows, and behavioral consistency scores. These data are key indicators for measuring whether a path can stably, collaboratively, and efficiently complete energy injection behaviors in actual scheduling.

[0051] Energy supply capacity trend analysis involves time series modeling and trend forecasting of candidate energy supply nodes' actual power supply behavior over historical scheduling cycles, identifying the trajectory of their power supply capacity changes within future scheduling windows. Analysis results typically include power output capacity curves, behavioral stability trends, and response success rate fluctuations, aiming to capture whether a node is experiencing capacity growth, stability, or degradation.

[0052] Scheduling expectation deviation analysis can model and quantify the response delays, behavioral deviations, and intention deviations exhibited by energy supply nodes in actual scheduling behaviors, analyzing the execution errors and behavioral mismatches that may occur under different scheduling contexts (such as high load, resource conflicts, and concurrent energy injection). The results of this analysis, expressed as a deviation tensor or error model, are important dimensions for measuring node execution predictability and scheduling risk levels, and can help establish protection boundaries and tolerance mechanisms in the development of behavioral contracts.

[0053] Among them, the node state evolution tensor can be a behavior prediction model constructed based on multi-dimensional indicators after integrating the results of energy supply capacity trend analysis and scheduling expectation deviation analysis. The dimensions of the tensor usually include node identification, time window, behavior scenario, etc., which are used to represent the energy supply capacity distribution, behavior deviation probability, response risk level and scheduling adaptability of the node under different temporal and spatial scheduling conditions.

[0054] Specifically, for any multi-source energy transmission structure model, any energy supply path from a candidate energy supply node to a target lattice point is considered a dynamic behavior propagation chain. According to the physical sequence of the energy supply process and the control trigger logic, the various functional nodes involved in the path (such as energy routers, DC circuit breakers, energy storage modules, and controllable converters) are arranged in sequence. Each node is assigned a behavior label, including key parameters such as node role (such as source node, relay node, and end node), behavior type (such as active injection, path switching, and energy convergence), control trigger mechanism (such as timed start, state trigger, and synchronous trigger), response delay prediction, and energy transmission directionality. The system also records information such as the overall path startup delay, signal propagation chain, path impedance accumulation, control scheduling dependencies, and behavior coordination methods, forming a high-dimensional trajectory structure that describes the temporal, spatial, and logical evolution of behavioral elements in the energy supply chain, known as the energy behavior trajectory.

[0055] After constructing the energy behavior trajectories corresponding to each energy supply path, the behavior chain modeling framework is used to extract the triggering logic (e.g., start conditions, control signal dependencies, feedback triggering mechanisms), behavior response times (e.g., earliest startable time, average response latency, execution duration), and execution priorities of each behavior node in each energy behavior trajectory. Based on this, a temporal dependency diagram is constructed between the behavior nodes, forming an event sequence chain diagram. This event sequence diagram is then structured and parsed to identify critical paths (i.e., the sequence of nodes that has the greatest impact on the overall path energy injection timeliness), bottleneck behavior segments (e.g., the slowest responding or most resource-constrained control nodes), potential timing conflict windows (e.g., multiple control behaviors competing for the same control resource), and resource preemption regions (e.g., multiple paths accessing shared control resources during the same time period). The potential for collaborative energy injection between the behavior chains is further analyzed, and their synchronization tolerances and compatibility indicators within the scheduling window are evaluated, resulting in a collaborative analysis of multiple paths and behavior chains. Finally, the above structures and data are integrated and encoded into dynamic response collaborative analysis data to characterize the coordination, response efficiency and potential conflict risks of the energy transmission control chain during the dynamic execution of the energy supply path.

[0056] For each candidate energy supply node involved in an energy behavior trajectory, a supply capacity trend analysis is conducted based on the node's historical scheduling records, operating status logs, and behavioral response data. Specifically, a time-variable energy supply response capacity time series is constructed to reflect the node's actual power supply performance within different historical time windows, including changes in power output capacity, fluctuations in response success rate, control behavior lag, and behavioral degradation patterns. Trend modeling algorithms (such as exponentially weighted moving average, sliding window statistics, or deep prediction models) are used to predict the node's energy supply behavior within the future target scheduling cycle, generating supply capacity change curves for different time periods. Subsequently, combining multiple rounds of scheduling history and behavioral trajectory data, the behavior deviations of each node under different scheduling contexts are analyzed. This identifies the node's response deviation, startup delay, and intention deviation when faced with varying scheduling load intensities, concurrent path pressures, or special control scenarios (such as simultaneous injection and multi-path confluence). This generates scheduling expectation deviation data based on temporal, spatial, and contextual dimensions. The above energy supply capacity trend data and scheduling expectation deviation data are integrated to construct a multidimensional structured tensor, namely the node state evolution tensor. This tensor uses node identification, time window, and behavioral context as index dimensions to record the energy supply capacity distribution, behavioral uncertainty, risk level, and scheduling adaptability that the node may exhibit in each future spatiotemporal scheduling unit.

[0057] After obtaining dynamic response collaborative analysis data and node state evolution tensors, the system uses lattice points as index units and, based on the path mapping relationship between energy behavior trajectories, pairs the dynamic response characteristics of each energy supply path (including critical path delay, timing conflict probability, behavior chain consistency index, synchronization tolerance range, etc.) with the state evolution information of its corresponding energy supply node within the current time window (including energy supply capacity prediction value, behavior stability level, expected deviation probability, and node risk coefficient) to construct a "path-node" cascade behavior matrix. This matrix is then subjected to multi-dimensional scoring and ranking analysis. Efficiency aggregation calculations and behavioral adaptation judgments are performed on all available path combinations, considering behavior execution feasibility, path combination synergy, response stability, and risk tolerance. This generates a comprehensive performance evaluation result with multi-path decision-making basis for each lattice point. These fused calculation results are then packaged in a structured form as a lattice point efficiency analysis structured dataset.

[0058] In this embodiment, by constructing the energy behavior trajectory corresponding to each energy supply path in the multi-source energy supply transmission structure model, and obtaining the path-level dynamic response coordination characteristics based on the behavior chain reconstruction analysis, and then combining the energy supply capacity trend analysis and scheduling expectation deviation modeling of the candidate energy supply nodes in the path, a node state evolution tensor is generated, and then these two types of dynamic behavior data are fused and processed to form a lattice point efficiency analysis structure data set, which can realize comprehensive modeling and prediction of the energy supply path execution efficiency, behavior consistency, response stability and power supply node reliability at the system scheduling layer; and perform multi-dimensional fusion analysis at the timing dimension, behavior logic layer and node state evolution layer, which significantly improves the intelligence, precision and risk perception capabilities of power supply scheduling, and provides more adaptive scheduling decision support for flexible traction power supply in scenarios of multiple trains concurrently, path intersection and resource competition in urban rail transit.

[0059] In an exemplary embodiment, Figure 5 As shown, each energy behavior trajectory is reconstructed and analyzed in a behavior chain to obtain dynamic response collaborative analysis data, including steps 502 to 508. Step 502 : Perform response time series analysis on each behavior node in each energy behavior trajectory to obtain an event sequence chain diagram.

[0060] Step 504 , identifying the critical path bottleneck nodes, timing conflict windows, and resource preemption areas of each behavior chain, and building a behavior conflict detection model.

[0061] Step 506 : Perform synchronization tolerance analysis on each behavior chain to obtain a multi-source coordinated energy injection timing matching diagram.

[0062] Step 508 : Apply the event sequence chain diagram and the multi-source collaborative energy injection timing matching diagram to the behavior conflict detection model to obtain dynamic response collaborative analysis data.

[0063] Among them, response time series analysis can be aimed at the execution process of each control behavior node in the energy behavior trajectory. The system extracts its key time attributes such as control trigger time, response delay, execution duration, earliest start time, latest completion time and behavior tolerance interval in the time dimension, constructs a node-level time series data set in the order of events, and performs statistical modeling and time-dependent analysis on the sequence to reveal the timing laws, bottleneck delay characteristics and potential execution uncertainties of the entire energy supply path at the control execution level.

[0064] The event sequence chain diagram is a behavioral event-level graph structure model constructed based on the results of response time series analysis. Each node in the diagram represents a control behavior event in the energy behavior trajectory, and the edges represent the temporal dependencies, triggering logic, or execution control flow between events. This diagram fully expresses the execution process and behavioral chain structure of a specific energy supply path in the time dimension, supporting critical path identification, event conflict analysis, and scheduling logic verification. It serves as the foundation for behavioral-level scheduling modeling and behavioral conflict detection.

[0065] Among them, the behavior chain can be a set of ordered control behaviors with clear timing logic extracted from a certain energy supply path or energy behavior trajectory. It is usually manifested as a control execution sequence from the energy supply node to the lattice point, including control actions such as starting the switch, path switching, route scheduling, and energy storage transfer. Each behavior chain has its own execution dependency structure, control delay structure and resource access logic.

[0066] A critical path bottleneck node can be one or more control nodes in a behavior chain or event sequence diagram that have the greatest impact on the overall execution efficiency of the behavior. These nodes often become critical delay control points in the path due to their long control response time, high execution lag, or severe resource contention. Identifying bottleneck nodes helps optimize path response speed, adjust injection timing, or reallocate control resources. It is a core step in achieving refined scheduling and timely response in flexible power supply systems.

[0067] Timing conflict windows can be high-risk areas where multiple action chains overlap in time, and the actions within these intervals face resource exclusion, control dependency conflicts, or scheduling logic inconsistencies, preventing concurrent execution within the same timeframe. These conflict windows can lead to control failures, power supply delays, or action interruptions, necessitating avoidance or reconciliation through scheduling adjustments, path reconstruction, or time window reallocation.

[0068] Among them, the resource preemption area can be an area in which a key power supply resource (such as energy routers, energy storage modules, control nodes) is shared and the scheduling time overlaps in multiple behavior chains or energy behavior trajectories. There is a clear resource competition relationship between the behaviors in these areas. If there is no priority mechanism or coordination strategy, it is easy to cause power supply path failure or scheduling conflict.

[0069] Among them, the behavior conflict detection model can be a composite analysis framework that integrates event sequence chain diagrams, synchronous matching structures, resource usage diagrams, and scheduling rule libraries. It is used to identify potential behavior conflict risks when multiple energy supply paths are simultaneously scheduled or selected, including timing conflicts, resource preemption, control logic contradictions, and path logic mutual exclusion.

[0070] Synchronization tolerance analysis targets key control nodes between two or more behavior chains, analyzing the alignment potential and maximum permissible offset between their execution time windows, assessing whether they can execute collaboratively without conflict within the same scheduling cycle. This analysis, typically output as a synchronized time window, tolerance threshold, and alignment confidence, is a key prerequisite for determining whether multiple paths can be concurrently powered, synchronously injected, or master-slave scheduled.

[0071] Among them, the multi-source collaborative energy injection timing matching graph can be a graph structure model generated by integrating the time docking capabilities, synchronization tolerance relationships and control behavior compatibility between multiple energy supply paths. The nodes represent energy supply paths or behavior chains, the edges represent the possibility of collaborative energy injection between paths, and the edge weights are quantified as the width of the synchronous execution window, the alignment probability or the behavior compatibility coefficient; the graph is used to identify which paths can form a coordinated execution group, and support multi-path fusion scheduling, behavior contract formulation and system-level energy routing optimization.

[0072] Specifically, for each energy behavior trajectory, the analysis unit is based on each behavior node within the trajectory. Key time parameters during the energy supply path execution are extracted for each node, including trigger start time, control response time, execution duration, earliest start time, latest deadline time, mean response delay, and maximum allowable deviation. This constitutes a node-level response time series. Based on this time series data and combined with the control logic and physical transmission relationships between nodes in the trajectory, an event-driven timing dependency graph is constructed. Each event node represents a control behavior operation (e.g., energy router activation, circuit breaker closing, energy storage module activation), and the direction of the edges in the graph indicates the triggering logic or sequential dependency between preceding and following events. The event-driven timing dependency graph is further structured into multiple subchains based on the behavior logic. Each subchain is a behavior chain, representing a complete response path for a set of ordered control behaviors from the energy supply node to the lattice point. The resulting event sequence chain diagram not only reflects the absolute and relative temporal order of each behavior event, but also captures the scheduling dependencies, coordination windows, and control interlocking structures between behaviors.

[0073] For each behavior chain in the event sequence diagram, we first identify the critical path bottleneck node—the control behavior node that has the dominant influence on the chain's overall response time. These nodes typically exhibit the longest response time, the greatest transmission delay, or the most constrained associated resources. This identification method combines the calculation of the longest behavior path, node response delay threshold comparison, and behavior impact weight assessment. Next, based on the temporal intersection between multiple behavior chains, we perform an alignment analysis of the time overlap of nodes in different behavior chains within the scheduling window to determine whether there is a timing conflict window. This is when multiple behaviors overlap in execution but the underlying control resources or control logic are incompatible, potentially leading to concurrent execution failures. We further evaluate the resource hierarchy's usage status to detect resource preemption zones, where nodes in multiple behavior chains compete for the same control resources (e.g., the same power supply node, the same energy router channel, a shared switch, etc.) without an effective isolation strategy or priority mechanism. Based on the identification results of the bottleneck nodes, conflict windows and preemptive areas mentioned above, a behavioral conflict detection model is constructed. The model uses the event sequence chain diagram as the underlying structure, integrates the timing overlap matrix, resource conflict table and priority rule set, and has the capabilities of conflict prediction, execution risk scoring and control behavior coordination judgment.

[0074] Key event nodes of each behavior chain are extracted, including path starting control points, energy injection points, and important relay control nodes. For each key node, an adjustable start interval and behavior execution time window are generated to construct a behavior chain time range model. Through cross-path time window intersection analysis and tolerance boundary calculation, the existence of synchronized execution sections between chains is identified. Specifically, whether control events on multiple paths within the same scheduling cycle can be coordinated within a certain time tolerance range is determined. Simultaneously, the behavior chains' delay adaptability, scheduling fault tolerance, and synchronization accuracy requirements during the alignment process are evaluated, and an inter-path coordination index is established. This cross-path synchronization relationship is modeled as a multi-source coordinated energy injection timing matching graph, where nodes represent behavior chains, edges represent logical connections between chains with synchronization injection potential, and edge weights are encoded as synchronized time window widths, time alignment credibility, or coordinated execution weights.

[0075] The event sequence chain graph and the multi-source collaborative energy injection timing matching graph are jointly input and applied to a pre-set behavior conflict detection model. The behavior conflict detection model first calibrates the execution timing distribution of each behavior chain within the target scheduling window and its resource usage range based on the node response order, dependency relationships, and bottleneck node markings in the event sequence chain graph. Then, combined with the synchronization relationship edge weights in the timing matching graph, all chain pairs with potential timing coordination relationships are compared one by one, calculating their feasibility, risk level, and reconciliation cost for achieving synchronous energy injection in actual scheduling scenarios. Simultaneously, the resource layer is mapped and expanded in the conflict detection model, uniformly modeling the access requests of shared energy interfaces, control nodes, and relay devices in the path as a resource scheduling tensor. This model then integrates time windows, control priorities, and conflict probability parameters for a joint solution. The dynamic response collaborative analysis data output by the final behavioral conflict detection model includes: behavioral chain response capability indicators for each path (such as execution success probability and response delay distribution), coordination potential scores between paths (such as coordination degree and synchronization probability), conflict sensitivity labels (such as high-risk path segments), and behavioral fusion recommendation strategies (such as step-by-step execution, delayed scheduling, and master-slave energy injection relationships).

[0076] In this embodiment, by performing response time series analysis on each behavior node in each energy behavior trajectory, an event sequence chain diagram is constructed, and the critical path bottleneck nodes, timing conflict windows, and resource preemption areas in each behavior chain are further identified to form a behavior conflict detection model. At the same time, synchronization tolerance analysis is performed on each behavior chain to generate a multi-source collaborative energy injection timing matching diagram. The above-mentioned graph structure is then applied to the behavior conflict detection model to output dynamic response collaborative analysis data. This method can realize global conflict detection, synchronous adaptation, and risk prediction of multi-path and multi-node power supply behaviors at the timing, resource, and control logic levels. It can accurately model and perceive the timing conflicts, resource competition, and execution risks in the complex urban rail transit power supply environment at the dynamic behavior chain and multi-path collaborative levels, effectively improving the concurrent safety of multi-train flexible power supply, the collaborative efficiency of the energy injection path, and the dynamic adaptability of the power supply behavior.

[0077] In an exemplary embodiment, Figure 6 As shown, the event sequence chain diagram and the multi-source collaborative energy injection timing matching diagram are applied to the behavior conflict detection model to obtain dynamic response collaborative analysis data, including steps 602 to 608. Step 602: Map the node response relationship of each behavior chain into a behavior propagation path set.

[0078] Step 604 : Cross-analyze the path coordination tolerance information and the behavior propagation path set in the multi-source coordinated energy injection timing matching graph to construct an inter-path behavior synchronization conflict table.

[0079] Step 606 : Input the inter-path behavior synchronization conflict table into the behavior conflict detection model to derive the conflict prediction tensor of each lattice point.

[0080] Step 608: Perform tensor coordination analysis on each conflict prediction tensor to generate dynamic response coordination analysis data.

[0081] Among them, node response relationship mapping can be a process of structurally modeling the execution sequence and response delay relationship in the energy behavior chain based on the time dependency, trigger mechanism and control logic between control behavior nodes.

[0082] The behavior propagation path set can be a set of sequential control paths constructed from multiple behavior chains through response relationship mapping. Each path represents a complete behavior execution sequence from the starting behavior node to the target injection node. This set records the directionality, control rhythm, timing structure, and dependency nodes of behavior propagation. It is comparable and combinable, and is a key analysis target for conflict detection, path coordination evaluation, and dynamic scheduling optimization. It can be used to reveal the propagation logic and scheduling coupling characteristics between multiple paths in the system at the execution level.

[0083] Among them, the path coordination tolerance information can be the time coordination data obtained after the alignment evaluation between any two or more energy supply paths during the synchronization tolerance analysis process, including indicators such as the maximum allowable time offset of each path at the key control node, adjustable window width, synchronization trigger threshold and collaborative execution credibility.

[0084] The inter-path behavior synchronization conflict table is a conflict record structure constructed by cross-analyzing the behavior propagation path set and path coordination tolerance information. It is used to express risk factors such as time conflicts, resource preemption, and behavioral incompatibility that may arise between multiple paths during scheduling. This table uses path pairs as the basic index unit and includes fields such as conflict type, conflict level, conflict node pair, compatibility score, and recommended reconciliation strategy.

[0085] Among them, the conflict prediction tensor can be a multidimensional structured data body constructed with lattice points, path combinations, time windows and conflict types as index dimensions, which records the behavioral conflict types, occurrence probabilities, conflict intensity levels and reconcilability estimates that may be caused by different power supply path combinations at specific time and space locations.

[0086] Tensor coordination analysis can be a multi-dimensional coordinated assessment of the resolvability of conflict states, the alternative nature of behavioral paths, and scheduling flexibility based on the conflict prediction tensor. This analysis identifies the conflict resolution potential of each path combination in the conflict tensor. Through reconciliation cost estimation, resource reconstruction feasibility assessment, and behavioral rescheduling strategy simulation, it generates path reconstruction suggestions, behavioral priority adjustment plans, and system scheduling fault tolerance recommendations, providing a tensor-level intelligent decision-making basis for optimizing energy injection paths and minimizing scheduling risks.

[0087] Specifically, the response logic of each behavior chain is expanded respectively. According to the response time, trigger sequence logic, resource access dependency and behavior execution mode of each node in the behavior chain, the continuous behavior chain between nodes is identified, and the causal relationship, control signal propagation path, response time deviation and key event sequence in these chains are encoded and processed to form a path structure of "starting behavior node→relay control node→target injection node". Finally, each path is defined as a complete behavior propagation unit and integrated as a behavior propagation path set.

[0088] Taking the path coordination tolerance information and behavior propagation path set in the multi-source coordinated energy injection timing matching graph as input, within a unified timing parsing framework, we first extract the time alignment information between any path pairs from the multi-source coordinated energy injection timing matching graph, including the synchronization execution tolerance range, maximum allowable time offset, coordination trigger window length, and path behavior compatibility score. Simultaneously, we extract the actual time distribution of key control events, response sequence, inter-node delay relationship, and control rhythm pattern of each path from the behavior propagation path set. By matching the key behavior node groups between path pairs, we analyze whether their time overlap intervals within the scheduling window are within the coordination tolerance range. We then identify cooperable path pairs, potential timing conflict path pairs, and completely incompatible path pairs, assigning conflict intensity levels to each pair. The results of this analysis are then used to construct a table of inter-path behavior synchronization conflicts.

[0089] The inter-path behavioral synchronization conflict table serves as the core input payload and is fed into the behavioral conflict detection model for multi-dimensional conflict prediction reasoning. The model first performs a scheduling mapping for each pair of potentially conflicting path combinations in the conflict table. Based on the associated target lattice point number, the time window in which the paths participate, the conflict event level, and resource occupancy, it establishes a "path pair-time segment-resource type-lattice point" correspondence. The conflict detection model then uses a built-in conflict identification rule set and resource scheduling tensor. By integrating the path conflict level, the probability of inter-event synchronization failure, the length of the conflict window, and the overlap of key resources, it calculates the conflict intensity, conflict type probability distribution, and behavioral reconciliation probability for each lattice point within the current scheduling cycle. The model then outputs a structured, high-dimensional tensor, called the conflict prediction tensor. The index dimensions of the conflict prediction tensor include lattice point ID, path combination ID, time period, conflict level, and scheduling impact weight. Each tensor cell records the probability and severity of a control conflict caused by a path combination at a specific spatiotemporal location.

[0090] Each conflict unit in each conflict prediction tensor is clustered and classified to identify high-frequency conflict patterns, concentrated conflict segments, and systemic conflict trends. Then, based on the node state evolution tensor and the scheduling policy library, a set of preset path reconciliation mechanisms (such as time window fine-tuning, behavior chain splitting, path delay activation, and master-slave switching injection) are invoked to simulate the reconciliation potential and calculate the adaptability of each conflict unit. Multi-dimensional indicators are introduced into the analysis process, such as conflict mitigation cost, reconciliation path reprioritization cost, resource reallocation feasibility, synchronization injection tolerance width, and scheduling behavior consistency score, to form a reconciliation score tensor. This is then jointly reconstructed with the original conflict prediction tensor to obtain dynamic response collaborative analysis data.

[0091] In this embodiment, the node response relationships in each behavior chain are mapped into a structured set of behavior propagation paths. Cross-analysis is performed in conjunction with the path coordination tolerance information in the multi-source collaborative injection timing matching diagram to construct an inter-path behavior synchronization conflict table. This conflict table is then input into the behavior conflict detection model to derive the conflict prediction tensor for each lattice point. Finally, coordination analysis is performed on the conflict tensor to generate dynamic response coordination analysis data. This approach not only enables the early perception and quantification of potential scheduling conflicts, but also outputs reconciliation strategies and scheduling recommendations based on behavioral-level tensor analysis, achieving dynamic risk mitigation and behavior optimization at the power supply path level. This significantly improves the coordination, safety, and system compatibility of flexible DC traction power supply in urban rail transit systems, including multi-train injection.

[0092] In an exemplary embodiment, Figure 7As shown, the energy supply capability trend analysis and scheduling expectation deviation analysis are performed on each candidate energy supply node in each energy behavior trajectory to obtain the node state evolution tensor, including steps 702 to 708. Step 702: Obtain historical response behavior data, energy supply intention record data, and node response trajectory of each candidate energy supply node.

[0093] Step 704 : Perform time series trend modeling on the historical response behavior data to construct a time scale energy supply capability trend function.

[0094] Step 706 : extract the energy supply intention record data and the offset error and prediction deviation of the node response trajectory in different scheduling contexts, and construct a scheduling expectation deviation mapping tensor.

[0095] In step 708, the time-scale energy supply capability trend function is applied to the scheduling expected deviation mapping tensor to obtain the node state evolution tensor.

[0096] Among them, historical response behavior data can be a set of control response information generated by the energy supply node when executing scheduling instructions, participating in energy supply path control or completing energy injection tasks, mainly including the time when the node receives the control signal, actual startup delay, power output value and change rate, energy injection duration, behavior interruption frequency, execution success rate, etc.

[0097] Among them, the energy supply intention record data can be the power supply expectation information issued by the power supply system or the dispatching center to a specific energy supply node in a certain period of time, including the target lattice point where the node is designated for power supply, the planned energy injection time window, the expected output power, the behavior execution conditions (such as synchronous triggering, prior path dependency), the task priority, and the path identity, etc.

[0098] A node response trajectory can be a time series of the actual operation behavior and performance indicator changes of a power supply node during continuous scheduling cycles or the execution of multiple behavioral paths. It typically includes information such as control response moments, injection behavior occurrence times, power output dynamics, control state transitions, and failure records within a continuous time period.

[0099] Time series trend modeling involves analyzing the historical response behavior data of nodes over time, identifying trends, behavioral patterns, periodic patterns, and fluctuation boundaries through statistical methods or machine learning algorithms, and constructing model functions that can be used to predict future behavior. Common methods include moving average, weighted sliding window, ARIMA, and LSTM. The goal is to extract the behavioral evolution characteristics of power supply nodes, enabling continuous prediction of power supply capacity or response performance at a certain point in the future in the form of functions.

[0100] The time-scale energy supply capability trend function is a behavior prediction function generated based on the results of time-series trend modeling. It describes the time-varying trend of a particular energy supply node's power supply capability (e.g., power level, response speed, and stability) along a continuous time axis. This function uses time as an independent variable and outputs a set of multi-dimensional capability predictions. This function can be used to infer the node's availability, scheduling adaptability, and behavioral confidence within any time window. It is the core modeling result for dynamic resource evaluation and path scheduling calculations.

[0101] Among them, the scheduling context can refer to the collection of environmental states and system configuration conditions that affect the execution effect of power supply behavior scheduling, usually including path topology, resource load intensity, the number of trains scheduled at the same time, the current control strategy (such as priority strategy or tolerance setting), the type of execution behavior (synchronous / asynchronous), etc.

[0102] The scheduling expectation deviation mapping tensor can be a multi-dimensional tensor data structured by comparing the power supply intention record data with the node's actual response trajectory, classifying it according to the scheduling context, and extracting indicators such as response offset, behavioral error, and performance deviation. The dimensions of this tensor typically include node number, time window, scheduling context, and deviation indicators (such as response delay, power deviation, and behavioral consistency deviation). It is used to describe and predict the behavioral uncertainty and execution risk of power supply nodes under different scheduling scenarios.

[0103] Specifically, for each candidate energy supply node involved in an energy behavior trajectory, the complete control and execution data recorded during the historical scheduling cycle are retrieved, including historical response behavior data (such as startup delay, execution success rate, and control signal trigger timing), the node's energy supply intention record under each task (such as target lattice point, expected energy supply time window, and behavior priority), and the actual response trajectory (including actual startup time, power supply duration, power output curve, etc.).

[0104] Based on the historical response behavior data of any candidate energy supply node, the node's energy supply behavior sequence over multiple scheduling cycles is time-rearranged and normalized. Subsequently, key behavioral indicators within each time slice (such as response start-up delay, actual energy supply duration, output power mean and fluctuation, and behavior success rate) are mapped into time series feature vectors based on continuous time slices. Time series modeling techniques (including but not limited to ARIMA models, exponentially weighted moving average algorithms, and LSTM prediction models based on recurrent neural networks) are used to perform trend fitting and behavioral evolution prediction on the node's historical behavior trajectory. The modeling process not only focuses on the changing trends of individual indicators but also comprehensively considers the co-evolutionary relationships between multiple behavioral characteristics. This results in a continuous function with time as the independent variable and a behavioral capability vector as the output, constructing a time-scale energy supply capability trend function. This function outputs the node's predicted available power level, response credibility level, and behavioral stability distribution within any target time window.

[0105] Since the energy supply intention record data includes the expected start time, target energy supply range, expected output power, behavior priority, and control dependency conditions, while the node response trajectory includes the node's actual response time, power output curve, behavior interruption frequency, and response anomaly flags, by comparing the energy supply intention record data and node response trajectory, the node's deviation error indicators are calculated for each scheduling context (such as different train numbers, resource load intensity, control strategy, or path topology conditions), including start advance / delay time, power output deviation, behavior execution failure probability, and behavior consistency deviation rate, and these error values are normalized. The deviation data is organized into a tensor based on "node identity × time window × scheduling context characteristics × deviation dimension" to construct a scheduling expectation deviation mapping tensor.

[0106] Based on the time-scale energy supply capability trend function, the static energy supply capability level of the node is predicted within each future target time window, including the output power range, response delay lower limit, and behavior success probability. Subsequently, within the same time window and corresponding scheduling context, the corresponding offset error data and behavior deviation distribution information from the scheduling expectation deviation mapping tensor are extracted to correct and perturb the output of the time-scale energy supply capability trend function. This process introduces a behavior confidence adjustment factor and a deviation interference coefficient to dynamically adjust the output credibility range and behavior fluctuation boundary of the time-scale energy supply capability trend function within a local time period. The trend prediction values and deviation adjustment results within all time windows are uniformly expressed in tensors to form a node state evolution tensor, which is indexed by "node ID × time window × behavior capability dimension × uncertainty level × scheduling background conditions."

[0107] In this embodiment, by obtaining the historical response behavior data, energy supply intention record data and node response trajectory of the candidate energy supply nodes, and combining the temporal trend modeling of the historical response data, a time-scale energy supply capacity trend function is constructed. At the same time, based on the error information between the energy supply intention and the actual response behavior in different scheduling contexts, a scheduling expectation deviation mapping tensor is constructed, and finally the tensor is applied to the trend function to obtain a structured node state evolution tensor, which can achieve high-resolution predictive expression in the time dimension and the scheduling environment dimension, significantly improving the intelligence level and forward-looking control capability of the flexible DC traction power supply system in node availability assessment, path dynamic scheduling and multi-node concurrent coordination, and effectively enhancing the system's perception, avoidance and adaptation capabilities to behavioral uncertainty and scheduling conflicts in complex operating environments.

[0108] In an exemplary embodiment, Figure 8 As shown, based on the lattice point efficiency analysis structure data set, the power supply behavior of each train corresponding to the train control terminal is analyzed to obtain the power supply behavior execution data corresponding to each train, including steps 802 to 806. Step 802 : Identify a set of regulated power supply paths and a feasible energy injection time window from a lattice point efficiency analysis structure data set.

[0109] Step 804 : Under the condition of the feasible injection time window as a constraint, a collaborative injection path combination set is constructed according to the regulated power supply path set and the node state evolution tensor.

[0110] Step 806 , based on the intention perception map and the feasible injection time window, the spatiotemporal coordination of injection of each train is planned to obtain the power supply behavior execution data corresponding to each train.

[0111] The set of regulated power supply paths can be the set of all power supply paths available for scheduling within the current scheduling cycle, selected based on the lattice point efficiency analysis structure dataset in the flexible DC power supply system. Each path in this set meets the basic physical connectivity, node status availability, behavior chain coordination, and scheduling strategy execution requirements, and its control logic and power supply capability have been confirmed to be regulated through the efficiency evaluation mechanism.

[0112] The feasible injection time window can be one or more time intervals within the system scheduling timeframe where injection is permitted, determined by combining the train operation plan, lattice point injection demand timing, path resource availability, and behavioral tolerance analysis. This time window must not only meet the target time period requirements of the train injection intent but also consider path node state evolution trends, control trigger accessibility, and conflict avoidance strategies. It is a key boundary condition for synchronizing injection planning, behavioral path constraints, and time period scheduling.

[0113] A collaborative injection path combination set can be a set of two or more energy supply paths, capable of parallel injection or coordinated execution, while meeting the same injection mission objective (e.g., injecting energy on the same train or a group of trains at similar times and within the same section). Each combination features concurrent power supply in terms of physical connection and behavioral mechanisms for synchronous startup, resource sharing, or master-slave switching in terms of control logic. Optimized for injection efficiency, behavioral consistency, and system scheduling coordination, it serves as a high-level structural unit for achieving multi-source path convergence scheduling and flexible energy supply configuration.

[0114] The spatiotemporal coordination of energy injection can be described as a system-level energy injection scheduling relationship. During the scheduling process, the system comprehensively assesses the energy injection intentions and system resource status of multiple trains, looking at whether there are conflicts, overlaps, or opportunities for coordination in time and space. Based on this, the system then executes behavior planning, resource reallocation, and path optimization. This assessment includes the alignment of time windows between train energy injection behaviors, the degree of path overlap, control synchronization tolerance, behavior priority matching, and system resource availability.

[0115] Specifically, the lattice point efficiency analysis structure data set is traversed to extract the efficient energy supply path information related to the target lattice point corresponding to each train, including path number, transmission efficiency score, synchronization capability label and risk level, and the path set that is in an executable state within the current scheduling cycle is screened out as the control power supply path set; at the same time, according to the time attributes recorded in the lattice point efficiency analysis structure data set and the system scheduling tolerance, the feasible energy injection time window that meets the behavior execution constraints is identified, that is, the time segment that allows the lattice point to have power supply behavior.

[0116] Under the constraints of a feasible injection time window, combined with the behavioral state evolution information of the key energy supply nodes involved in each path in the regulated power supply path set within the corresponding time period, the predicted energy supply capacity, response stability score, behavioral tolerance interval, and uncertainty risk level of each node within the target time window are extracted based on the node state evolution tensor. By comparing the timing docking capability, synchronous triggering possibility, resource overlap risk, and behavioral coordination cost of each path, multiple rounds of combination screening and collaborative adaptation analysis are performed on the regulated power supply path set to identify a set of path combination structures with high feasibility in terms of physical path accessibility, node behavior consistency, and system control logic. Each combination contains two or more power supply paths that can complete injection behavior within the specified time window and meet the collaborative control conditions. Each combination is assigned a scheduling strategy identifier, coordination strategy recommendation, and execution priority level to construct a set of collaborative injection path combinations.

[0117] Based on the structured energy intention information of each train in the intention perception map, specifically the injection behavior requirements expressed by each train at the target lattice point, including injection type (traction, braking feedback, buffer energy replenishment, etc.), injection priority, target time window, spatial location, and behavior duration, the spatiotemporal relationship of injection among multiple trains is comprehensively planned. The coordinated planning process first matches each train's injection intention with the feasible injection time window in the coordinated path combination, analyzing whether there are time overlap sections and path conflict risks. When multiple trains intend to use the same power supply path or associated node in the same injection window, the cross-train injection coordination strategy is implemented by adjusting the behavior start and end times, injection method, or path allocation weight based on the principles of maximizing train intention completion rate, prioritizing behavior consistency, and optimizing path utilization efficiency. During the planning process, the running intervals between trains, track section conflicts, path tolerance allocation and behavior tolerance boundaries are also comprehensively considered to ensure that the power supply behavior execution data finally generated not only meets the energy needs of individual trains, but also meets the coordination and execution controllability of the system as a whole. The power supply behavior execution data is indexed by trains and contains structural fields such as the selected collaborative injection path ID, injection start and end times, injection mode (single point / segmented / synchronous), node resource identifiers used, behavior execution strategy codes, scheduling priorities and feedback signal mechanisms, providing energy routers and train control terminals with a control execution instruction set that can be directly parsed and issued.

[0118] In this embodiment, by identifying the control power supply path set and feasible injection time window from the lattice point efficiency analysis structure data set, combining the node state evolution tensor to construct a collaborative injection path combination set, and based on the intention perception map and the injection time window, the train's injection spatiotemporal coordination is uniformly planned, and finally the power supply behavior execution data corresponding to each train is generated, which can significantly improve the concurrent power supply capability, scheduling flexibility and system-level coordination efficiency of the power supply system, effectively reduce the operational risks caused by injection conflicts, resource congestion and uneven energy distribution, and realize efficient control and intelligent coordination of the flexible DC traction power supply process.

[0119] In an exemplary embodiment, Figure 9 As shown, based on the intention perception map and the feasible injection time window, the injection time and space coordination of each train is planned to obtain the power supply behavior execution data corresponding to each train, including steps 902 to 908. Step 902 : extract features from the intention perception map to obtain the spatiotemporal demand information of the energy injection intention of each train at any lattice point.

[0120] Step 904 : Perform spatiotemporal matching on the spatiotemporal requirement information of the injection intention and the feasible injection time window to identify potential spatiotemporal overlap information and behavioral collaboration opportunity areas.

[0121] Step 906 : Perform cross-train energy injection demand analysis on potential spatiotemporal overlap information and behavior coordination opportunity areas to obtain spatiotemporal coordination data of cross-train energy injection demand.

[0122] Step 908 , based on the intention completion rate maximization condition of each train and the spatiotemporal coordination data of cross-train energy injection requirements, the power supply parameters of each train are calculated to obtain the power supply behavior execution data corresponding to each train.

[0123] The spatiotemporal demand information for injection intent can be a multi-dimensional representation of the expected injection behavior of a specific train at a specific lattice point during its operation, extracted from the intention perception map. This information typically includes the injection type (such as traction power supply, regenerative braking, energy storage activation, etc.), the expected injection time window, the spatial location (track coordinates or lattice point number), the duration of the behavior, the power demand level, the control priority, the path dependency conditions, and the scheduling tolerance, forming a clear expression of the train's injection requirements in both time and space.

[0124] Among them, time-space matching can be achieved by comparing the time window and spatial position in the intention with the pre-obtained feasible injection time window and lattice path distribution after the train injection intention is extracted, so as to determine whether the train's injection demand matches the energy supply capacity of the current system resources in the time and space dimensions.

[0125] Among them, potential spatiotemporal overlap information can be the spatiotemporal intersection area formed by multiple trains proposing injection intentions in similar time periods and adjacent or identical spatial ranges. In this area, path competition, control resource overlap, injection behavior conflict or synchronous execution opportunities may occur.

[0126] Among them, the behavioral collaboration opportunity area can be the area determined by further analyzing the behavioral dependencies, path complementarity, synchronization control possibility and resource reuse potential after identifying the potential time and space overlap, to determine whether the energy injection behaviors between multiple trains have the conditions for collaborative execution.

[0127] Among them, cross-train energy injection demand analysis can be a process of comprehensive comparison and collaborative evaluation of the energy injection requests between multiple trains in the same space-time area from the perspectives of resource allocation, behavior priority, timing coordination and control dependency.

[0128] The spatiotemporal coordination data for inter-train energy injection demand can be derived from the system's structured modeling of spatiotemporal conflict relationships between train behaviors, resource usage overlap, and behavioral coordination capabilities after completing inter-train energy injection demand analysis. This data is typically organized as a tensor or multidimensional table, and includes information such as conflict levels between train pairs, scheduling compatibility, resource priority recommendations, feasible path combination solutions, synchronous energy injection windows, and coordination strategy recommendations.

[0129] Specifically, multi-dimensional features are extracted from each train's behavioral intention at a target lattice point in the intention perception graph. These features include the type of injection behavior the train is expected to perform at that lattice point (such as traction power injection, braking feedback absorption, and auxiliary energy storage charging), the target injection time window, physical location coordinates, scheduling priority, behavior duration, synchronization requirements, and resource dependencies. The extracted results are structured into an injection intention spatiotemporal demand vector, which describes the train's desired injection behavior at a specific spatiotemporal point.

[0130] The spatiotemporal demand information of the injection intention is aligned with the feasible injection time window identified in the early stage in the spatiotemporal dimension. The injection time period in the train intention is compared with the available time window of the path resource to see if there is any intersection. Combined with the track geographic location mapping, the injection behavior areas that may overlap or be adjacent in space are identified to form potential spatiotemporal overlap information. The potential spatiotemporal overlap information marks which trains may apply for power supply resources at the same time at similar times and locations, and further extracts the behavioral coordination opportunity areas, that is, the regional scope where the multi-train injection behavior has the possibility of coordination, path reuse capability or control logic compatibility.

[0131] Based on potential spatiotemporal overlap information, the spatiotemporal vector of each train's injection intention is analyzed to determine the degree of overlap with other trains' behaviors at target lattice points or in adjacent areas. This is then cross-compared with their respective behavior priorities, injection power requirements, behavior durations, and scheduling tolerances. A behavior conflict detection model and node state evolution tensor are then used to model and analyze the path resource schedulability, node behavior stability, and behavior response conflict probability within the behavior coordination opportunity region. This approach determines whether different behavior combinations can be harmoniously executed under conditions of limited resources or potential control conflicts. A coordination scoring mechanism is also introduced to quantitatively assess path complementarity (e.g., the degree of mutual exclusion between primary and backup paths), behavior dependencies (e.g., simultaneous start-up and sequential execution order), and the potential for improving overall injection efficiency. This results in spatiotemporal coordination data for cross-train injection requirements, which is structured as an injection behavior scheduling compatibility matrix between train pairs. This matrix includes behavior conflict levels, resource coordination possibilities, scheduling priority ranking recommendations, path coordination strategy candidates, and a spatiotemporal coordination index.

[0132] Maximizing the completion rate of each train's injection intent is the global scheduling optimization goal. Based on each train's corresponding priority, resource compatibility level, behavior conflict indicator, and scheduling tolerance parameters in the coordination data, a multi-objective optimization model is constructed. The objective functions are maximizing the injection success rate, minimizing the probability of path conflicts, and improving the overall system power supply efficiency. Scheduling constraints are also set, including power supply path capacity limits, key node control load limits, synchronization behavior window overlap boundaries, and behavior mutual exclusion pair limits. During the solution process, parameters are assigned to each train's injection behavior, including injection start and end times, selected path ID, execution power level, control strategy type (e.g., master / cooperative control), synchronization execution tags, and tolerance fine-tuning configuration. If resource conflicts are difficult to reconcile, the reconciliation strategy priorities provided in the coordination data are referenced to dynamically adjust the injection schedule of lower-priority trains or shift them to a backup injection window. Finally, complete power supply behavior execution data is output for each train, including structured fields for actual scheduling control, such as train number, target lattice point number, execution path combination ID, injection behavior type, control node configuration, injection period, behavior tolerance range, scheduling priority and execution strategy label, etc.

[0133] In this embodiment, feature extraction is performed on the intention perception map to obtain the spatiotemporal demand information of each train's injection intention at each lattice point. This demand information is then matched with the feasible injection time window to identify potential spatiotemporal overlap areas and behavioral coordination opportunities. Cross-train spatiotemporal coordination analysis is then performed on the injection behaviors of multiple trains within the same scheduling cycle. Power supply parameters are accurately calculated under the goal of maximizing the intention completion rate, generating power supply behavior execution data for each train. This effectively avoids path conflicts and resource congestion problems in the concurrent scheduling of multiple trains, improves the coordination between injection behaviors, system power supply efficiency, and task completion quality, and significantly enhances the scheduling adaptability and behavioral contract generation capabilities of the flexible DC traction power supply system in complex operating environments.

[0134] In an exemplary embodiment, Figure 10 As shown, the spatiotemporal information in each structured energy intention is analyzed respectively to obtain the intention perception map corresponding to the train control terminal, including steps 1002 to 1006. Step 1002: perform type recognition and semantic mapping on the fields in any structured energy intent to extract the energy injection spatiotemporal behavior elements.

[0135] Step 1004 , coordinate transformation and serialization expression are performed on the injection spatiotemporal behavior elements to construct an initial spatial behavior spectrum.

[0136] Step 1006: Perform intention aggregation analysis and conflict semantic reasoning on the initial spatial behavior spectrum to obtain an intention perception graph.

[0137] Among them, the energy injection spatiotemporal behavior elements can be extracted from structured energy intention data. They are the basic information units that can describe the energy supply request behavior proposed by the train under specific time and space conditions. They usually include the time or time window of the behavior (such as expected start time, duration, scheduling tolerance), spatial position information (such as track segment number, lattice point coordinates, power supply node ID), behavior type (such as traction power supply, regenerative braking, energy storage activation, etc.), target power level and priority label.

[0138] The initial spatial behavior spectrum can be constructed based on the extracted spatiotemporal elements of energy injection. By sorting each train behavior chronologically, spatially categorizing it by track position or lattice point position, and performing coordinate standardization and unified expression of behavior identifiers, a visual mapping relationship between "behavior event × time × space" is established in a two-dimensional coordinate system. Each node in the spectrum represents a specific instance of energy injection behavior, and its position and time attributes directly reflect the distribution of energy supply demands on the track during train operation.

[0139] Intent aggregation analysis can be the process of identifying, based on the initial spatial behavior spectrum, injection behaviors with spatial proximity, overlapping time windows, or highly consistent target paths, and then classifying and aggregating them to form behavior clusters or intent clusters. Aggregation analysis not only considers the proximity of behaviors in physical space and time, but also incorporates the similarity of behavioral characteristics (such as type, power level, and node dependencies) and scheduling collaboration potential to construct an association structure between behaviors. This reduces analysis complexity, improves semantic clarity, and provides higher-level behavioral units for conflict judgment and resource optimization.

[0140] Among them, conflict semantic reasoning can be based on the results of intention aggregation analysis, combined with the energy supply path topology, resource capacity constraints, control rule model and behavior dependency logic, to conduct knowledge-level logical judgment and reasoning on whether there are scheduling conflicts, path mutual exclusion, resource preemption or control logic inconsistency between multiple aggregated behaviors. The reasoning process outputs the conflict type (such as synchronization failure, timing conflict, resource competition), conflict level, dependency relationship between behaviors and reconciliation possibility suggestions, and expresses them in the form of a graph structure or relationship matrix.

[0141] Specifically, the system traverses the structured energy intent data in the train control terminal, identifying the fields in each record (such as injection target, time stamp, behavior type, power level, node ID, and operating segment number) and analyzing their temporal and spatial characteristics or control behavior categories. Semantic mapping is then performed based on a pre-set semantic rule base, converting the original fields into standardized behavior labels (such as "injection request," "path priority," and "synchronization behavior requirement"). The system then extracts the spatiotemporal injection behavior elements representing the train's energy supply requirements along temporal and spatial dimensions, including the behavior's start and end times, target track area, power supply behavior category, and scheduling tolerance range.

[0142] The spatial location information involved in the injected spatiotemporal behavior elements (such as station number, track section number, and lattice point ID) is uniformly mapped to the standard coordinate system used by the train control terminal. Coordinate transformation is performed to ensure comparability of spatial data between different trains. Next, using the train ID as an index, the injected behaviors are arranged in chronological order to construct a structured behavior sequence. Each train behavior is then partitioned and classified by spatial location, generating a two-dimensional behavior spectrum with track location on the horizontal axis and behavior timeline on the vertical axis. This is called the initial spatial behavior spectrum.

[0143] Based on the track segmentation, lattice point density, and timeline clustering in the initial spatial behavior spectrum, the system aggregates the energized behavior units of multiple trains within adjacent spatial locations or similar time windows. Behavior nodes with behavioral similarity, goal consistency, or potential resource sharing are grouped into the same intention cluster and labeled with a behavior density index, synergy likelihood score, and resource coupling degree. The semantic reasoning engine then uses the path topology, power supply node status, control strategy rules, and behavior scheduling logic to analyze conflicts and logical constraints between the aggregated behavior clusters. This includes, but is not limited to, determining whether behaviors are mutually exclusive (e.g., multiple trains competing for the same power supply path), whether control dependencies exist (e.g., a behavior requires the completion of a preceding behavior to initiate), and whether scheduling priority conflicts exist. Based on these analysis results, a conflict association diagram and reconciliation suggestions are generated. These aggregation results and reasoning outputs are organized into a graph structure to generate an intention-aware map. Nodes in this map represent independent or aggregated energized intention units of trains, and edges represent logical connections, conflicts, or synergy potential between intentions.

[0144] In this embodiment, by performing type identification and semantic mapping on the fields in structured energy intent, the spatiotemporal behavior elements of energy injection are extracted. These extracted behavior elements are then subjected to coordinate transformation and serialized expression to construct an initial spatial behavior spectrum. Based on this, intent aggregation analysis and conflict semantic reasoning are performed to generate an intent-aware spectrum. This approach accurately extracts the energy injection requirements of trains at different spatial locations and scheduling cycles, and systematically identifies potential behavioral conflicts and path resource overlaps. This significantly enhances the Flexible DC power supply system's behavioral cognition depth and coordination prediction capabilities for multi-vehicle scheduling tasks, providing highly reliable data support for energy supply path planning, scheduling priority setting, and behavioral contract generation.

[0145] In one embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0146] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the steps in the above-mentioned method embodiments when executed by a processor.

[0147] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of each of the above-described method embodiments.

[0148] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0149] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0150] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A flexible DC traction power supply method for urban rail transit based on an energy router, characterized in that: The method comprises: constructing structured energy intentions corresponding to the train control terminals according to train operation status data corresponding to the train control terminals; Respectively analyzing the spatiotemporal information in each structured energy intention to obtain an intention perception map corresponding to the train control terminal; Performing energy efficiency analysis on each lattice point in the intention perception map to obtain a lattice point efficiency analysis structure data set; According to the lattice point efficiency analysis structure data set, the power supply behavior of each train corresponding to the train control terminal is analyzed to obtain the power supply behavior execution data corresponding to each train; the power supply behavior execution data is used to provide flexible DC traction power supply to the train.

2. The method according to claim 1, characterized in that The energy efficiency analysis of each lattice point in the intention perception map is performed to obtain a lattice point efficiency analysis structure data set, including: For any of the lattice points, identifying candidate energy supply nodes corresponding to the lattice point from the intention perception map according to the energy supply demand of the lattice point; Constructing a multi-source energy supply transmission structure model according to the energy transmission paths between the lattice points and each of the candidate energy supply nodes; The energy supply parameters and state parameters in the multi-source energy supply transmission structure model are solved to obtain the lattice point efficiency analysis structure data set.

3. The method according to claim 2, characterized in that Solving the energy supply parameters and state parameters in the multi-source energy supply transmission structure model to obtain the lattice point efficiency analysis structure data set includes: Constructing energy behavior trajectories corresponding to each energy supply path in any of the multi-source energy supply transmission structure models; Conducting behavior chain reconstruction analysis on each of the energy behavior trajectories to obtain dynamic response collaborative analysis data; Performing energy supply capability trend analysis and scheduling expectation deviation analysis on each candidate energy supply node in each energy behavior trajectory to obtain a node state evolution tensor; The dynamic response collaborative analysis data and the node state evolution tensor are fused to obtain the lattice point efficiency analysis structure data set.

4. The method according to claim 3, characterized in that The step of performing behavior chain reconstruction analysis on each of the energy behavior trajectories to obtain dynamic response collaborative analysis data includes: Performing response time series analysis on each behavior node in each energy behavior trajectory to obtain an event sequence chain diagram; the event sequence chain diagram includes each behavior chain; Identify the critical path bottleneck nodes, timing conflict windows and resource preemption areas of each of the behavior chains, and build a behavior conflict detection model; Perform synchronization tolerance analysis on each of the behavior chains to obtain a multi-source collaborative energy injection timing matching diagram; The event sequence chain diagram and the multi-source collaborative energy injection timing matching diagram are applied to the behavior conflict detection model to obtain the dynamic response collaborative analysis data.

5. The method according to claim 4, characterized in that Applying the event sequence chain diagram and the multi-source collaborative energy injection timing matching diagram to the behavior conflict detection model to obtain the dynamic response collaborative analysis data includes: Mapping the node response relationship of each behavior chain into a behavior propagation path set; Cross-analyze the path coordination tolerance information in the multi-source coordinated energy injection timing matching graph and the behavior propagation path set to construct an inter-path behavior synchronization conflict table; Inputting the inter-path behavior synchronization conflict table into the behavior conflict detection model to derive the conflict prediction tensor of each lattice point; A tensor coordination analysis is performed on each of the conflict prediction tensors to generate the dynamic response collaborative analysis data.

6. The method according to claim 3, characterized in that The energy supply capability trend analysis and scheduling expectation deviation analysis are performed on each candidate energy supply node in each energy behavior trajectory to obtain a node state evolution tensor, including: Obtaining historical response behavior data, energy supply intention record data and node response trajectory of each candidate energy supply node; Performing time series trend modeling on the historical response behavior data to construct a time scale energy supply capacity trend function; Extracting the energy supply intention record data and the offset error and prediction deviation of the node response trajectory in different scheduling contexts, and constructing a scheduling expectation deviation mapping tensor; The scheduling expected deviation mapping tensor is applied to the time scale energy supply capability trend function to obtain the node state evolution tensor.

7. The method according to claim 1, characterized in that The step of analyzing the power supply behavior of each train corresponding to the train control terminal according to the lattice point efficiency analysis structure data set to obtain power supply behavior execution data corresponding to each train includes: identifying a set of regulated power supply paths and a feasible energy injection time window from the lattice point efficiency analysis structure data set; Under the condition that the feasible injection time window is used as a constraint condition, a collaborative injection path combination set is constructed according to the set of regulated power supply paths and the node state evolution tensor; According to the intention perception map and the feasible injection time window, the spatiotemporal coordination of injection of each train is planned to obtain the power supply behavior execution data corresponding to each train.

8. The method according to claim 7, characterized in that The planning of the spatiotemporal coordination of energy injection for each train based on the intention perception map and the feasible energy injection time window to obtain power supply behavior execution data corresponding to each train includes: Performing feature extraction on the intention perception map to obtain the spatiotemporal demand information of the energy injection intention of each train at any lattice point; Performing spatiotemporal matching of the injection intention spatiotemporal demand information with the feasible injection time window to identify potential spatiotemporal overlap information and behavioral collaboration opportunity areas; Performing cross-train energy injection demand analysis on the potential spatiotemporal overlap information and the behavioral coordination opportunity area to obtain spatiotemporal coordination data of cross-train energy injection demand; Based on the condition of maximizing the intention completion rate of each train, and according to the spatiotemporal coordination data of the cross-train energy injection demand, the power supply parameters of each train are calculated to obtain the power supply behavior execution data corresponding to each train.

9. The method according to claim 1, characterized in that The step of respectively analyzing the spatiotemporal information in each structured energy intention to obtain an intention perception map corresponding to the train control terminal includes: Perform type recognition and semantic mapping on the fields in any structured energy intent to extract the injection spatiotemporal behavior elements; Performing coordinate transformation and serialization expression on the injection spatiotemporal behavior elements to construct an initial spatial behavior spectrum; Perform intention aggregation analysis and conflict semantic reasoning on the initial spatial behavior spectrum to obtain the intention perception spectrum.

10. A flexible DC traction power supply device for urban rail transit based on an energy router, characterized in that: The device comprises: An energy intention construction module is used to construct each structured energy intention corresponding to the train control terminal according to the train operation status data corresponding to the train control terminal; A perception map obtaining module is used to respectively analyze the spatiotemporal information in each structured energy intention to obtain an intention perception map corresponding to the train control terminal; an efficiency analysis execution module, configured to perform energy efficiency analysis on each lattice point in the intention perception map to obtain a lattice point efficiency analysis structure data set; The power supply data acquisition module is used to analyze the power supply behavior of each train corresponding to the train control terminal according to the lattice point efficiency analysis structure data set, and obtain the power supply behavior execution data corresponding to each train; the power supply behavior execution data is used to provide flexible DC traction power supply to the train.

Images