Traffic energy source and load prediction and collaborative optimization method and system based on data fusion

Abstract

The application provides a traffic energy source and load prediction and collaborative optimization method and system based on data fusion, relates to the technical field of traffic energy management, and first accesses multiple-source traffic energy data to generate a traffic energy data correlation pool; generates a dynamic source and load correlation network based on the traffic energy data correlation pool; maps traffic scene features to the dynamic source and load correlation network to generate a multi-dimensional fusion deduction input set; performs source and load change trend deduction on the multi-dimensional fusion deduction input set through a multi-dimensional fusion deduction model adapted to the scene to obtain a traffic energy source and load prediction result; and forms a collaborative optimization scheme based on the traffic energy source and load prediction result and dispatching requirements and feeds back the collaborative optimization scheme to network optimization scene adaptation weights, thereby effectively improving the operation efficiency and reliability of the traffic energy system and reducing energy waste and operation costs.

Description

Technical Field

[0001] This invention relates to the field of transportation energy management technology, and more specifically, to a method and system for predicting and coordinating transportation energy sources and loads based on data fusion. Background Technology

[0002] In the current context of deep integration between transportation and energy, the efficient operation of transportation energy systems is crucial for the sustainable development of cities. During transportation operations, operational data such as traffic flow and speed of different modes of transportation (e.g., road traffic, rail traffic), data on energy supply such as the supply volume and stability of various energy sources (e.g., electricity, natural gas), and environmental factors (e.g., weather, temperature) all influence the supply and demand relationship of transportation energy.

[0003] Currently, traditional methods for predicting and coordinating the energy sources and loads of transportation often have several shortcomings. On the one hand, data acquisition and processing are relatively isolated, focusing only on a single type of data, such as considering only traffic flow data to predict energy demand, or planning energy allocation solely based on energy supply data. This ignores the complex relationships between traffic operation, energy supply, and the environment, leading to inaccurate predictions and a lack of scientific rigor and comprehensiveness in the coordinating optimization schemes. On the other hand, existing methods lack the ability to dynamically adapt to different traffic scenarios, failing to adjust prediction and optimization strategies in real time according to the characteristics of different scenarios (such as peak hours, special weather conditions, etc.), making it difficult to meet the operational needs of transportation energy systems in complex and ever-changing environments. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a method for predicting and coordinating traffic energy source and load based on data fusion, the method comprising:

[0005] Multi-source traffic and energy data, consisting of traffic operation data, energy supply data, and environmental correlation data, are accessed to generate a traffic and energy data association pool. The traffic and energy data association pool includes attribute association information, scenario adaptation identifiers, and data interaction channels for various types of data.

[0006] Based on the attribute association information and scenario adaptation identifier of the traffic energy data association pool, a dynamic source-load association network is generated. The dynamic source-load association network takes data nodes as the core and source-load association relationships as the connection links. The connection links include association transmission logic and scenario adaptation weights.

[0007] Different traffic scenario features are mapped to the dynamic source-load association network, and the data node set and connection link association logic for scenario adaptation are extracted to generate a multi-dimensional fusion inference input set. The multi-dimensional fusion inference input set includes the core data nodes of the scenario, connection link parameters and scenario adaptation constraints.

[0008] By using a multi-dimensional fusion inference model adapted to the scenario, the source load change trend of the multi-dimensional fusion inference input set is inferred to obtain the traffic energy source load prediction result. The traffic energy source load prediction result includes the dynamic curve of supply and demand change, the transmission path of related influences and the scenario adaptation deviation.

[0009] Based on the traffic energy source and load prediction results and traffic energy scheduling requirements, a traffic energy source and load collaborative optimization scheme is formed. The traffic energy source and load collaborative optimization scheme is fed back to the dynamic source and load association network through the data interaction channel to realize the scenario adaptation weight optimization of the dynamic source and load association network.

[0010] Furthermore, embodiments of the present invention also provide a traffic energy source and load prediction and collaborative optimization system based on data fusion, comprising:

[0011] A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to execute the above-described data fusion-based traffic energy source and load prediction and collaborative optimization method by executing the machine-executable instructions.

[0012] Based on the above, a traffic energy data association pool is generated by accessing traffic operation data, energy supply data, and environmental correlation data. This pool integrates various related data and clarifies the attribute association information, scenario adaptation identifiers, and data interaction channels of the data. A dynamic source-load association network is generated based on this pool, constructing a dynamic relationship model between traffic energy sources and loads using data nodes and connection links. The association transmission logic and scenario adaptation weights in the connection links enable the network to accurately reflect the complex relationships between sources and loads under different scenarios. Traffic scenario features are mapped to the dynamic source-load association network to generate a multi-dimensional fusion inference input set, fully considering the impact of scenarios on source-load relationships. The multi-dimensional fusion inference model yields traffic energy source-load prediction results, including dynamic curves of supply and demand changes, correlation impact transmission paths, and scenario adaptation deviations, reflecting the changing trends and mutual influences of traffic energy sources and loads. The collaborative optimization scheme formed based on the traffic energy source-load prediction results and scheduling requirements is fed back to the dynamic source-load association network to optimize scenario adaptation weights, effectively improving the operational efficiency and reliability of the traffic energy system and reducing energy waste and operating costs. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of the execution flow of the traffic energy source and load prediction and collaborative optimization method based on data fusion provided in the embodiments of the present invention.

[0014] Figure 2 This is a schematic diagram of exemplary hardware and software components of the data fusion-based traffic energy source and load prediction and collaborative optimization system provided in this embodiment of the invention. Detailed Implementation

[0015] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a data fusion-based method for predicting and coordinating traffic energy sources and loads, provided in one embodiment of the present invention. The details are as follows.

[0016] Step S110: Access multi-source traffic energy data composed of traffic operation data, energy supply data, and environmental correlation data to generate a traffic energy data correlation pool. The traffic energy data correlation pool includes attribute correlation information, scenario adaptation identifiers, and data interaction channels for various types of data.

[0017] This embodiment uses the coordinated optimization of energy supply and demand in urban rail transit networks as an application scenario for explanation. In this scenario, traffic operation data includes passenger flow at each subway station, passenger flow at each station, train intervals, current train location, passenger capacity, and train speed; energy supply data includes real-time power output of each substation, energy consumption of the traction power supply system, state of charge of energy storage devices, and regenerative braking energy recovery; environmental data includes outdoor temperature, humidity, rainfall, air quality index, holiday information, and weekday peak commuting hours. When accessing the above multi-source data, a distributed data acquisition architecture is adopted. Data is collected in real time through sensors and smart meters deployed in subway stations, trains, and substations. Data transmission uses an encrypted transmission protocol to ensure data security during transmission.

[0018] For data involving user privacy, such as detailed passenger flow trajectories at a specific station within a specific time period, data anonymization is employed to remove personally identifiable information. The data is then aggregated and analyzed, retaining only regional-level statistical data to protect user privacy. When generating the traffic energy data association pool, metadata is labeled for various data types, clearly defining attributes such as data source, collection time, data format, and data accuracy. Attribute association information establishes the inherent connections between data, such as the correlation between train intervals and traction power supply system energy consumption, and the correlation between passenger flow entering stations and station lighting and air conditioning system energy consumption. Scene adaptation identifiers are set according to different operational scenarios, such as weekday morning rush hour scenarios, holiday off-peak scenarios, and severe weather scenarios. Each data entry corresponds to one or more scene adaptation identifiers. The data interaction channel defines the interaction methods between different systems, including data interface specifications, data update frequency, and data synchronization mechanisms, ensuring efficient data flow between traffic dispatching systems, energy management systems, and environmental monitoring systems.

[0019] Step S120: Based on the attribute association information and scenario adaptation identifier of the traffic energy data association pool, a dynamic source-load association network is generated. The dynamic source-load association network takes data nodes as the core and source-load association relationships as the connection links. The connection links include association transmission logic and scenario adaptation weights.

[0020] In this embodiment, data nodes correspond to specific entities or abstract concepts in the transportation energy system, such as train nodes, substation nodes, station passenger flow nodes, and ambient temperature nodes. Each data node contains its corresponding attribute information; for example, the attributes of a train node include vehicle type, rated passenger capacity, traction power, and regenerative braking efficiency. Connection links characterize the source-load relationships between data nodes, such as the link between train operation and substation power supply, or the link between passenger flow changes and station equipment energy consumption. The association transmission logic describes the transmission methods and rules of information or energy between data nodes, such as the transmission logic of train traction energy consumption increasing with passenger capacity, or the transmission logic of regenerative braking energy recovery changing with train braking intensity. Scenario adaptation weights reflect the importance or influence of a certain association in different scenarios. For example, in a weekday morning rush hour scenario, the association weight between train intervals and power supply load is higher, while in a nighttime maintenance scenario, this association weight is relatively lower.

[0021] Step S121: Extract traffic operation data attribute association information, energy supply data attribute association information, and environmental data attribute association information from the traffic energy data association pool to form a multi-source data attribute association set. Each association information in the multi-source data attribute association set corresponds to a unique data node identifier and a scenario adaptation identifier.

[0022] When extracting attribute association information from various types of data from the transportation energy data association pool, a data mining method based on semantic analysis is employed. For traffic operation data, the correlation between train operating status parameters (such as speed, acceleration, and position) and energy consumption parameters (such as traction energy consumption and auxiliary system energy consumption) is analyzed, for example, the positive correlation between train starting acceleration and instantaneous traction energy consumption; the correlation between passenger flow data and train operation plans (such as adding trains and extending operating hours) is also analyzed. For energy supply data, the correlation between substation power supply and total energy consumption of trains on each line is analyzed, as well as the correlation between the state of charge of energy storage devices and regenerative braking energy input and traction energy output, and the correlation between load distribution among different substations. For environmental correlation data, the correlation between outdoor temperature and train air conditioning system energy consumption, the correlation between rainfall and train speed limits, and the correlation between holiday information and overall passenger flow and energy consumption levels are analyzed. The extracted attribute association information is organized into a multi-source data attribute association set. A unique data node identifier is assigned to each association information, such as "train-001", "substation-003", "passenger flow-station A", etc., and its applicable scenario adaptation identifier is also marked, such as "morning peak" and "holiday".

[0023] Step S122: Analyze the direct correlation between traffic operation data nodes and energy supply data nodes. The direct correlation is generated based on the interdependence logic of data attributes and includes data interaction triggering conditions and correlation transmission direction.

[0024] The direct correlation between traffic operation data nodes and energy supply data nodes is a core component of the dynamic source-load correlation network. For example, a train, as a traffic operation data node, directly depends on the electrical energy supplied by the substation for its traction energy consumption; therefore, there is a direct correlation between the train node and the substation node. Data interaction triggering conditions define the conditions under which these correlations are activated. For instance, when a train starts, it triggers a demand for traction power supply from the substation; when the train enters the regenerative braking phase, it triggers an interaction that feeds back electrical energy to energy storage devices or the power grid. The correlation transmission direction characterizes the direction of data or energy flow. During the train traction phase, energy is transmitted from the substation node to the train node, with the correlation transmission direction being substation → train; during the regenerative braking phase, energy is transmitted from the train node to the energy storage device node or the substation node, with the correlation transmission direction being train → energy storage device / substation. When analyzing these direct correlations, it is necessary to consider the interdependence logic of data attributes. For example, the train's traction power demand depends on the train's mass (including passenger capacity) and operating acceleration, while the substation's power supply capacity limits the train's maximum traction power.

[0025] Step S1221: Extract the attribute information of traffic operation data nodes in the traffic energy data association pool. The attribute information of traffic operation data nodes includes data dimensions such as traffic flow attributes, travel route attributes, facility usage attributes, and operation status attributes.

[0026] The attribute information of traffic operation data nodes is crucial for describing their characteristics and behaviors. Traffic flow attributes include data dimensions such as peak hour passenger flow, off-peak hour passenger flow, and transfer passenger flow at each station. These data dimensions reflect the flow characteristics of passengers within the rail transit network. Travel path attributes include passenger route selection from the origin station to the destination station and passenger flow distribution along each transfer route. This travel path attribute dimension affects the passenger capacity distribution of trains and energy consumption in the operating section. Facility usage attributes involve the frequency of use of automatic ticket vending machines within stations, the operating time of elevators and escalators, and the number of times platform screen doors are opened and closed. The usage of these facilities is directly related to the energy consumption of station auxiliary systems. Operational status attributes encompass data dimensions such as train operating speed curves, station dwell time, train formation size, and train fault status. These dimensions determine the traction energy consumption and regenerative braking energy recovery potential of trains.

[0027] Step S1222: Extract the attribute information of energy supply data nodes in the transportation energy data association pool. The attribute information of energy supply data nodes includes data dimensions such as total supply attribute, transmission efficiency attribute, storage status attribute, and supply type attribute.

[0028] The attribute information of energy supply data nodes reflects the supply capacity and operational characteristics of the energy system. Total supply attributes include data dimensions such as the maximum power supply capacity of substations, real-time power load, daily power supply, and monthly power supply, which characterize the overall scale of energy supply. Transmission efficiency attributes cover data such as the transmission loss rate of traction power supply lines, the conversion efficiency of transformer equipment, and the current carrying capacity of cables; this transmission efficiency attribute directly affects the effective utilization of energy from the supply side to the demand side. Storage status attributes include data dimensions such as the rated capacity of energy storage devices, current state of charge, maximum charging and discharging power, and charging and discharging efficiency. As an energy buffer, the status of energy storage devices plays a crucial role in smoothing load fluctuations and improving the utilization rate of regenerative braking energy. Supply type attributes distinguish different forms of energy supply, such as grid power supply, regenerative braking energy recovery power supply, and energy storage device discharge power supply; different supply types have different characteristics and application scenarios.

[0029] Step S1223: Standardize the attribute information of the extracted traffic operation data nodes and energy supply data nodes, compare the standardized attribute information of the traffic operation data nodes and energy supply data nodes, and identify the interdependent attribute dimensions between the two. The interdependent attribute dimensions are those in which changes in one data will cause adjustments in the other data.

[0030] Standardization processing includes dimensionless data conversion and data normalization. For example, passenger flow data is converted into a percentage relative to the line's design capacity, and power supply load is converted into a percentage relative to the substation's rated capacity. Standardization eliminates the influence of different dimensions on attribute comparisons. When comparing standardized attribute information, the focus is on attribute dimensions where changes in one data point lead to corresponding adjustments in the other. For instance, the "peak hour passenger flow" attribute dimension of the traffic operation data node is interdependent with the "real-time power supply load" attribute dimension of the energy supply data node. When peak hour passenger flow increases, train passenger capacity increases, leading to increased traction energy consumption, which in turn increases the real-time power supply load of the energy supply data node; conversely, when passenger flow decreases, the real-time power supply load decreases accordingly. Similarly, the "train interval" attribute dimension of the traffic operation data node is interdependent with the "power supply load fluctuation frequency" attribute dimension of the energy supply data node. The smaller the train interval, the more trains pass through a section per unit time, and the higher the power supply load fluctuation frequency.

[0031] Step S1224: Based on the standardized interdependent attribute dimensions, analyze the transmission logic of data changes. The transmission logic includes the triggering conditions of data changes, the transmission sequence, and the correlation of the degree of impact. The triggering conditions are generated based on the standardized value change threshold of the attribute data.

[0032] For example, step S1224-1: For each set of interdependent attribute dimensions, obtain the historical data records of that set of attribute dimensions. The historical data records include the change values ​​of the attribute data, the change time, and the corresponding related data responses.

[0033] For each set of interdependent attribute dimensions, historical data records are retrieved from the database. These records contain the changes in attribute data over different times, the specific time when the data changed, and the response of other related data when the attribute data changed. By analyzing the historical data, the patterns and relationships of data changes between attribute dimensions can be identified.

[0034] Step S1224-2: Calculate the propagation time of data changes based on the attribute data change time and related data response time in historical data records.

[0035] In historical data records, the time interval between determining the start time of a change in attribute data and the time when related data begins to respond is called the propagation time of the data change. By analyzing and calculating multiple sets of historical data, the distribution characteristics and statistical patterns of propagation time under different circumstances can be obtained, revealing the time required for a data change to propagate from one attribute dimension to another.

[0036] Step S1224-3: Analyze the time difference between attribute data changes and related data responses in historical data records to determine the propagation timeliness of data changes. The propagation timeliness is the time interval between attribute data changes and related data responses.

[0037] Further analysis of the time lag between attribute data changes and related data responses in historical data records clarifies the propagation timeliness of data changes. Propagation timeliness accurately reflects the time interval from when attribute data changes to when related data produces an observable response. This parameter is crucial for understanding and modeling the dynamic relationships between data, and helps predict and respond to changes in related data in practical applications.

[0038] Step S1224-4: Based on the transmission timeliness, divide the transmission time sequence into stages, which include the initial triggering stage, the transmission diffusion stage, and the response stabilization stage. Each stage corresponds to different change characteristics and related logic.

[0039] Based on the duration of transmission and the process of data change, the transmission sequence is divided into different stages. The initial triggering stage is when attribute data begins to change and triggers responses in related data; at this stage, changes are relatively rapid, and the correlation logic mainly manifests as a direct causal relationship. The transmission and diffusion stage is when data changes gradually spread and expand their impact along the correlation dimensions; the rate of change slows down, and the correlation logic may involve multiple intermediate links. The response stabilization stage is when the response of the related data reaches a stable state and no longer changes significantly with further changes in attribute data; at this stage, the correlation logic tends to stabilize. Each stage has its unique characteristics of change and correlation logic, requiring separate analysis and description.

[0040] Step S1224-5: Extract the minimum change value of the attribute data change that triggers the response of the associated data in the historical data record, and set the minimum change value as the data change trigger threshold of the attribute dimension. The data change trigger threshold is matched with the change characteristics of the attribute data.

[0041] Identify the smallest change in attribute data from historical data records that triggers an observable response in related data, and set this as the data change trigger threshold for that attribute dimension. The trigger threshold should be matched to the change characteristics of the attribute data. For example, for attribute dimensions that are more sensitive to change, the trigger threshold can be set lower; for attribute dimensions that are relatively insensitive to change, the trigger threshold can be appropriately increased. The trigger threshold is an important basis for determining whether data changes will trigger a related response.

[0042] Step S1224-6: Combine the data change trigger threshold and the transmission timing stage to determine the triggering conditions for data change. The triggering conditions include the judgment criteria for attribute data reaching the data change trigger threshold, the transmission start timing after triggering, and the trigger failure conditions.

[0043] By combining data change trigger thresholds and propagation timing stages, the triggering conditions for data changes are determined. Triggering conditions include specific criteria for determining when attribute data reaches the trigger threshold, such as determining a trigger when the change in attribute data exceeds the threshold; the propagation initiation timing after triggering, specifying the start time and order of different propagation timing stages after triggering; and trigger failure conditions, specifying under what circumstances the trigger state ends, such as when the attribute data change returns to below the trigger threshold or after a certain period of time, the trigger state automatically fails.

[0044] Step S1224-7: Analyze the correspondence between the change range of attribute data and the response range of related data in historical data records, and determine the influence degree correlation model. The influence degree correlation model reflects the change ratio and correlation law between the two.

[0045] By analyzing historical data records, the correlation between the magnitude of changes in attribute data and the magnitude of responses in related data can be studied. Statistical analysis methods, such as regression analysis, can be used to establish a correlation model of influence degree. This model can reflect the proportion and correlation pattern between the magnitude of changes in attribute data and the magnitude of responses in related data, such as whether the relationship is linear, non-linear, or another specific function. The correlation model of influence degree is an important tool for quantitatively analyzing the impact of data changes.

[0046] Step S1224-8: Based on the influence degree correlation model, generate an influence degree level division. The influence degree level division includes multiple levels, and each level corresponds to a different range of attribute data change amplitude and a range of correlation data response amplitude.

[0047] Based on the impact degree correlation model, the magnitude of attribute data change and the magnitude of related data response are divided into different levels, generating an impact degree level classification. Each level corresponds to a specific range of attribute data change and a corresponding range of related data response. This level classification allows for a more intuitive description of the magnitude and scope of the impact of data changes, facilitating the adoption of appropriate measures and strategies based on different impact levels in practical applications.

[0048] Step S1224-9: Integrate the transmission timeliness, transmission sequence stages, triggering conditions, influence degree correlation model and influence degree level classification to form the transmission logic framework.

[0049] By integrating elements such as transmission timeliness, transmission sequence stages, triggering conditions, impact degree correlation model, and impact degree level classification, a transmission logic framework is formed, which describes the transmission process, triggering mechanism, impact degree, and time sequence characteristics of data changes between interdependent attribute dimensions.

[0050] Step S1224-10: Supplement the special case handling logic in the transmission logic framework. The special case handling logic includes transmission rules for scenarios such as attribute data mutation, unresponsive related data, and simultaneous changes in multiple sets of attribute data, forming the final transmission logic. The transmission logic is used to describe the triggering conditions, transmission sequence, and degree of influence of data changes between interdependent attribute dimensions.

[0051] Building upon the existing transmission logic framework, special case handling logic is added. These special cases include: abrupt changes in attribute data (abnormally rapid or large changes); unresponsive related data (a change in attribute data without the expected response from related data); and simultaneous changes in multiple sets of attribute data, potentially leading to complex interrelationships. For these special cases, corresponding transmission rules are developed, such as using specific trigger thresholds and transmission sequences for attribute data abrupt changes, and performing fault diagnosis and handling when related data fails to respond. After adding the special case handling logic, the final transmission logic is formed, capable of describing the triggering conditions, transmission sequence, and degree of impact of data changes between interdependent attribute dimensions.

[0052] Step S1225: Determine the data interaction trigger condition. The data interaction trigger condition is the trigger rule when the interdependent attribute dimensions reach a preset change state. The trigger rule is consistent with the change characteristics of the attribute data.

[0053] Data interaction triggering conditions are the basis for realizing dynamic interaction between data nodes. Preset change states can be caused by the standardized value of attribute data reaching a certain threshold, the rate of change of attribute data exceeding a certain limit, or the attribute data entering a specific range. For example, for the state of charge attribute dimension of an energy storage device, when the second normalized value is below 0.2, an interaction to charge from the grid is triggered; when the normalized value is above 0.8, an interaction to discharge to the grid or supply power to other loads is triggered. The triggering rules need to be consistent with the change characteristics of the attribute data. For data with continuously changing characteristics, such as train speed, the triggering rule can be set as a gradient threshold for speed change; for data with discrete event characteristics, such as train arrival and departure, the triggering rule can be set at the instant the event occurs. Determining data interaction triggering conditions requires comprehensive consideration of system safety, economy, and stability to avoid frequent triggering that could cause system fluctuations.

[0054] Step S1226: Based on the transmission timing in the transmission logic, determine the associated transmission direction. The associated transmission direction is the flow of data changes from one data node to another data node, and the flow direction corresponds to the attribute dependency logic.

[0055] The direction of correlation propagation is determined by the propagation timing in the propagation logic and corresponds to the attribute dependency logic. In scenarios where data is propagated from energy supply data nodes to traffic operation data nodes, such as a substation supplying power to a train, the propagation timing is such that the substation's power supply parameter adjustments precede changes in the train's energy consumption; the direction of correlation propagation is energy supply data node → traffic operation data node. Conversely, in scenarios where data is propagated from traffic operation data nodes to energy supply data nodes, such as train regenerative braking energy feedback, the propagation timing is such that the train's braking behavior precedes the energy reception of the energy supply data node (e.g., an energy storage device); the direction of correlation propagation is traffic operation data node → energy supply data node. In complex relationships, bidirectional propagation may exist. For example, changes in train operating status affect the substation's power supply strategy, while the substation's power supply capacity limitations, in turn, affect adjustments to the train's operating plan; in this case, the direction of correlation propagation is bidirectional. Determining the direction of correlation propagation helps clarify the causal relationships between data nodes.

[0056] Step S1227: Combine the data interaction triggering conditions and the direction of association transmission to form a single direct association relationship. The single direct association relationship includes the data node identifiers of the two parties involved, the data interaction triggering conditions, the direction of association transmission, and the attribute dependency logic.

[0057] A single direct association is the basic unit connecting two data nodes. The identifiers of the two data nodes uniquely identify the two nodes involved in the association, such as "Train-05 Line-01" and "Substation-220kV-03#". Data interaction triggering conditions specify when the association is initiated; for example, when the standardized value of the traction power demand of "Train-05 Line-01" exceeds 0.7, the power supply association with "Substation-220kV-03#" is initiated. The association transmission direction indicates the flow path of data or energy, such as "Substation-220kV-03#" → "Train-05 Line-01". Attribute dependency logic details the dependency relationship between the attribute dimensions of the two nodes; for example, the traction power attribute of "Train-05 Line-01" depends on the available power supply capacity attribute of "Substation-220kV-03#", as well as the train's own passenger capacity and operating speed attributes. Combining these elements forms a complete description of a direct association.

[0058] Step S1228: Traverse all traffic operation data nodes and energy supply data nodes. For each pair of traffic operation data nodes and energy supply data nodes, repeatedly perform the steps of extracting attribute information, performing standardization processing, comparing and identifying mutually dependent attribute dimensions, analyzing the transmission logic of data changes, determining data interaction trigger conditions, determining the direction of correlation transmission, and forming a single direct correlation relationship to generate all direct correlation relationships between pairs, forming a set of direct correlation relationships.

[0059] The rail transit system contains a large number of traffic operation data nodes and energy supply data nodes, requiring the traversal of all possible node pairs to generate direct relationships between them. For example, there is a direct relationship between all trains on each line and their corresponding power supply substations; there is a direct relationship between passenger flow nodes at each station and the station's energy supply nodes (such as in-station distribution transformers). During the traversal, a complete process needs to be performed on each node pair, from extracting attribute information, identifying interdependent dimensions, analyzing transmission logic, to determining trigger conditions and transmission directions. The generated direct relationships may contain duplicates or redundancies, requiring deduplication and merging to ensure that only the most critical and direct relationships are retained between each node pair. The final set of direct relationships encompasses all basic and direct interactions between traffic operation and energy supply.

[0060] Step S1229: Analyze the attribute dependency logic of each relationship in the set of direct relationships, merge logically consistent relationships, eliminate duplicate relationships, and optimize the structure of the set of direct relationships.

[0061] After the set of direct relationships is generated, its structure needs to be optimized. First, the attribute dependency logic of each relationship is analyzed. Relationships with completely identical or highly similar attribute dependency logic can be merged. For example, the relationship between trains of the same model and the same substation on the same line has essentially the same attribute dependency logic and can be merged into a universal relationship template, which can then be parameterized to adapt to the individual differences of different trains. Eliminating duplicate relationships means removing those relationships that are repeatedly defined between node pairs, ensuring that the relationship between each node pair is unique. Optimizing the structure of the set of direct relationships also includes classifying and hierarchically dividing the relationships, such as classifying them according to energy type (traction power supply, power lighting) and equipment type (train, substation, energy storage device), to facilitate subsequent network management and application.

[0062] Step S12210: Supplement the association description information in the direct association set. The association description information includes the attribute dependency details and data interaction characteristics of the association, forming the direct association set.

[0063] To make the set of direct relationships more complete and easier to understand, additional relationship description information is needed. Attribute dependency details describe the specific dependency between the attribute dimensions of the related parties. For example, the calculation formula for train traction energy consumption and substation power supply voltage and current (describing the physical relationship in words, such as traction energy consumption equals the product of power supply voltage / current and running time, considering line resistance loss), or the statistical regression equation for passenger flow and station lighting energy consumption (describing the statistical law in words, such as lighting energy consumption increasing linearly with increasing passenger flow, with a slope of a certain empirical value). Data interaction characteristics include the data interaction protocol type, data transmission format, data update frequency, and maximum data interaction latency. This information ensures that the relationships can be executed correctly and efficiently in the actual system. After supplementing the relationship description information, the set of direct relationships not only contains the structural information of the relationships but also the behavioral details of the relationships.

[0064] Step S123: Combining the indirect influence logic of environmental data nodes on traffic operation data nodes and energy supply data nodes, supplement the correlation dimension in the direct correlation relationship to form a full source-load correlation relationship set, which includes direct correlation relationships and indirect correlation relationships.

[0065] While environmentally related data nodes do not directly participate in energy production and consumption, they have a significant indirect impact on traffic operations and energy supply. For example, outdoor temperature, as an environmentally related data node, can affect the energy consumption of train air conditioning systems (an attribute of traffic operation data nodes), and thus indirectly affect the power supply load of substations (an attribute of energy supply data nodes). This indirect impact logic needs to be added to the direct relationships to form new relational dimensions. For instance, in the direct relationship between train nodes and substation nodes, adding the dimension of the impact of ambient temperature on train air conditioning energy consumption would provide a more comprehensive description of changes in power supply load. The complete set of source-load relations therefore includes both direct and indirect relations. Indirect relations, mediated by environmentally related data nodes, connect previously unconnected traffic operation data nodes and energy supply data nodes, enriching the network's relational and expressive capabilities.

[0066] Step S124: Using data nodes as the core carrier, the relationships in the full set of source-load relationships are transformed into connection links between data nodes. Each connection link corresponds to a set of relationships, and the connection link embeds the relationship transmission logic and the initial scene adaptation weight.

[0067] Data nodes are the core carriers of the dynamic source-load correlation network, with each data node representing an entity or concept within the transportation energy system. When transforming the correlations in the full set of source-load correlations into connection links, each connection link corresponds to a specific set of correlations, connecting two or more data nodes. Correlation transmission logic is embedded in the connection links, defining how data or energy is transmitted between nodes along the link, including the direction, path, timing, and rules of transmission. Initial scenario adaptation weights are pre-assigned based on the importance of the correlations in different scenarios. For example, in a high-temperature weather scenario, the connection link between the ambient temperature node and the train air conditioning energy consumption node has a higher weight; in a weekday morning rush hour scenario, the connection link between the passenger flow node and the train traction energy consumption node has a higher weight. Initial scenario adaptation weights can be set based on expert experience or statistical analysis of historical data.

[0068] Step S125: Generate the initial architecture of the dynamic source-load association network. The initial architecture includes all data nodes, connection links, association transmission logic, and initial scene adaptation weights. The data nodes include complete attribute information and scene adaptation identifiers.

[0069] The initial architecture of the dynamic source-load association network forms the foundational framework for network construction. This architecture integrates all data nodes and connection links. Data nodes have clearly defined locations and identifiers within the architecture, and their complete attribute information and scenario adaptation identifiers are stored in the network's node attribute library. Connection links connect nodes according to the type and strength of their association relationships, forming the network topology. Association propagation logic and initial scenario adaptation weights are stored as attributes of the connection links, describing their behavioral characteristics and scenario adaptability. The generation of the initial architecture must adhere to certain principles, such as network connectivity, uniform node distribution, and load balancing of links, to ensure that the network accurately reflects the actual situation of the transportation energy system.

[0070] Step S126: Embed a scene adaptation weight dynamic adjustment mechanism, wherein the scene adaptation weight dynamic adjustment mechanism generates adjustment logic based on the scene matching degree of data nodes and the transmission efficiency of connection links.

[0071] The dynamic adjustment mechanism for scene adaptation weights is key to achieving the adaptive capability of the dynamic source-load association network. Scene matching degree is an indicator that measures how well a data node's current state matches the characteristics of its respective scene. For example, in a weekday morning rush hour scenario, the closer the actual passenger flow of a station's passenger flow data node is to the typical passenger flow in that scenario, the higher the scene matching degree. The transmission efficiency of the connection link reflects the smoothness and effectiveness of data or energy transmission in the link. High transmission efficiency means low latency, low loss, and high reliability of the link. The adjustment logic dynamically adjusts the scene adaptation weights of the connection links based on changes in scene matching degree and transmission efficiency. When the scene matching degree of a data node increases and the transmission efficiency of the connection link increases, the scene adaptation weight of the corresponding link increases; conversely, it decreases. This dynamic adjustment mechanism enables the network to automatically optimize resource allocation and energy management strategies according to changes in the actual operating scenario.

[0072] Step S127: Drive the scenario adaptation weight dynamic adjustment mechanism through historical scenario data in multi-source traffic energy data, optimize the initial scenario adaptation weight, and form a dynamically adjusted connection link.

[0073] A dynamic adjustment mechanism for scenario adaptation weights is trained and driven using historical scenario data from multi-source transportation and energy data. This historical scenario data includes information such as data node status, connection link performance, and energy supply and demand under different scenarios. For example, historical data from weekday morning and evening rush hours and holidays over the past year is collected. Through data mining and machine learning methods, the relationship between scenario matching degree, transmission efficiency, and scenario adaptation weights under different scenarios is analyzed to establish an adjustment model. This model is integrated into the dynamic adjustment mechanism. When the network is running, the mechanism can optimize the initial scenario adaptation weights based on the current scenario matching degree and transmission efficiency, referring to optimal experience from historical data, forming dynamically adjusted connection links. This makes the scenario adaptation weights more consistent with the actual operating rules of the system, improving the network's decision-making accuracy.

[0074] Step S128: Based on the dynamically adjusted connection links, optimize the association transmission logic between data nodes to ensure that the association transmission logic is consistent with the scene adaptation weight, thereby improving the scene adaptation accuracy of the association relationship.

[0075] After the scenario adaptation weights of connection links are dynamically adjusted, the correlation and transmission logic between data nodes needs to be optimized accordingly to ensure consistency. For example, when the scenario adaptation weight of a connection link increases in a high-temperature scenario, it indicates that the link's importance in that scenario has increased. In this case, its correlation and transmission logic needs to be optimized to improve the efficiency and reliability of data or energy transmission. Specifically, this can be achieved by shortening transmission delay, increasing transmission bandwidth, and optimizing transmission paths. Optimizing the correlation and transmission logic also requires considering the overall distribution of scenario adaptation weights to avoid network imbalance caused by excessively high or low weights for individual links. By improving the scenario adaptation accuracy of correlation relationships, the dynamic source-load correlation network can better adapt to the needs of different scenarios and improve the effect of energy supply and demand coordination optimization.

[0076] Step S129: Integrate the optimized data nodes, connection links, associated transmission logic, and dynamically adjusted scenario adaptation weights to form the dynamic source-load association network.

[0077] After perfecting the data node attribute information, dynamically adjusting the connection links, and optimizing the associated transmission logic, the above elements are integrated to form the final dynamic source-load association network. The integration process includes storing the optimized node attributes, link parameters, and transmission logic in the network's database, establishing a mapping relationship between nodes and links, constructing a network topology index, and enabling network visualization. The integrated dynamic source-load association network reflects the dynamic model of the source-load relationship in the transportation energy system in real time, and it can automatically adjust the network structure and parameters as the scenario changes.

[0078] Step S1210: Real-time data synchronization between the dynamic source-load association network and the transportation energy data association pool is achieved through the data interaction channel. The real-time data synchronization mechanism ensures that the attribute information of data nodes in the dynamic source-load association network is consistent with the attribute information of data nodes in the transportation energy data association pool.

[0079] To ensure that the dynamic source-load association network accurately reflects the current state of the transportation energy system, a real-time data synchronization mechanism with the transportation energy data association pool needs to be established. The data interaction channel provides the physical and logical links for data transmission between the network and the data association pool. The real-time data synchronization mechanism defines the frequency, method, priority, and conflict resolution strategy for data synchronization. For example, for the attribute information of critical data nodes (such as the real-time power supply load of substations and the current location of trains), high-frequency (e.g., millisecond-level) real-time synchronization is used; for non-critical data (such as historical passenger flow statistics), low-frequency (e.g., minute-level) batch synchronization is used. When the attribute information of data nodes in the network is inconsistent with the information in the data association pool, the synchronization mechanism can adjust according to the preset priority and conflict resolution strategy (e.g., using the latest data in the data association pool as the standard) to ensure consistency between the two.

[0080] Step S130: Map the features of different traffic scenarios to the dynamic source-load association network, extract the data node set and connection link association logic for scenario adaptation, and generate a multi-dimensional fusion inference input set. The multi-dimensional fusion inference input set includes the core data nodes of the scenario, connection link parameters and scenario adaptation constraints.

[0081] Different traffic scenarios (such as weekday rush hour commutes, weekend leisure travel, and emergency evacuations during severe weather) have different characteristics, which need to be accurately mapped into a dynamic source-load correlation network. Scenario characteristics include passenger flow distribution, train operation plans, environmental conditions, and energy prices. The mapping process decomposes scenario characteristics into parameters related to network data nodes and connection links. For example, the characteristic of "surge in passenger flow during weekday morning rush hour" is mapped to an increase in the attribute value of passenger flow data nodes and a decrease in the attribute value of train interval data nodes. Extracting the scenario-adaptive data node set involves selecting the key data nodes that have the greatest impact on energy supply and demand in the current scenario, such as key station passenger flow nodes during peak hours, train nodes on busy lines, and substation nodes that are the main power supply sources. The connection link correlation logic is the correlation relationship and transmission rules between these core nodes. Scenario adaptation constraints include various restrictions under the scenario, such as maximum allowable power supply load, minimum train interval, and charging and discharging limits of energy storage devices. Integrating the core data nodes, connection link parameters, and scenario adaptation constraints generates a multi-dimensional fusion inference input set.

[0082] Step S131: Extract various traffic scene features from the traffic scene feature library. Each traffic scene feature includes scene operation parameters, environmental constraints, energy demand features, and traffic flow features. The scene operation parameters correspond to the data node attribute information dimensions in the dynamic source-load association network.

[0083] The traffic scenario feature library is a knowledge base for storing and managing various traffic scenario features. Each traffic scenario feature extracted from this library is a comprehensive description of a specific scenario. Scenario operation parameters include train departure intervals, operating speeds, stop times, and train formation numbers, which directly correspond to the operational status attribute dimensions of train data nodes in the dynamic source-load association network. Environmental constraints cover temperature, humidity, weather conditions, and holiday arrangements within the scenario, corresponding to the attribute dimensions of environmental data nodes. Energy demand features describe the total demand, demand curve shape, and demand distribution characteristics of the transportation system in this scenario, and are associated with the attribute dimensions of energy supply data nodes. Traffic flow features include the spatiotemporal distribution of passenger flow, transfer volume, and average travel distance, corresponding to the traffic flow attribute dimension of traffic operation data nodes. By mapping scenario operation parameters to the attribute information dimensions of network data nodes, an organic connection between scenario features and the dynamic source-load association network is achieved.

[0084] Step S132: Decompose each traffic scenario feature into multiple scenario core elements. The scenario core elements include scenario key parameters, constraint boundary conditions, and demand priority information. The scenario key parameters are matched with the data attribute association information in the traffic energy data association pool.

[0085] To facilitate the application of scene features in dynamic source-load association networks, they need to be broken down into core scene elements. Key scene parameters are the core data describing scene features, such as maximum hourly passenger flow, average train occupancy rate, and maximum substation power supply load during morning rush hour. These parameters need to be matched with data attribute association information in the transportation energy data association pool to ensure data consistency and availability. Constraint boundary conditions define the limitations of scene operation, such as the minimum safe operating interval for trains, the minimum state-of-charge protection value for energy storage devices, and the maximum power consumption of station facilities. Demand priority information clarifies the priority order of different energy and transportation demands within the scene. For example, in emergency scenarios, the priority of train safety operation demand is higher than energy efficiency demand; in normal operation scenarios, the priority of energy supply and demand balance demand is higher than the punctuality rate demand of individual trains. This makes scene features more structured and refined.

[0086] Step S133: Traverse all data nodes in the dynamic source-load association network, and based on the key parameters of the scene in the core elements of the scene, filter data nodes that are compatible with the scene features to form a set of candidate data nodes for the scene.

[0087] Traversing all data nodes in the dynamic source-load association network aims to filter out data nodes relevant to the current scenario characteristics from a massive number of nodes. The filtering process is based on key scenario parameters within the core elements of the scenario, comparing the matching degree between the attribute information of data nodes and these key parameters. For example, if the key scenario parameter includes "maximum hourly passenger flow of 50,000 people," then all station passenger flow data nodes are traversed, and data nodes with historical maximum hourly passenger flows close to or exceeding this value are included in the scenario candidate data node set. For energy supply data nodes, such as substations, substation nodes capable of meeting this load demand are selected based on the "maximum power supply load" key scenario parameter. The formation of the scenario candidate data node set initially narrows the scope of network analysis, focusing on nodes that have a significant impact on the current scenario.

[0088] Step S134: Analyze the scene adaptation identifier of each data node in the scene candidate data node set, and combine it with the constraint boundary conditions in the core elements of the scene to further filter the scene candidate data node set and obtain the scene core data node set.

[0089] After the candidate data node set for the scenario is generated, further filtering is required to obtain the core data node set for the scenario. The scenario adaptation identifier of each data node is analyzed to determine its applicability to the current scenario. For example, if a data node's scenario adaptation identifier is "holiday only," then that node will be excluded in a weekday scenario. The candidate nodes are then filtered again based on the constraint boundary conditions in the core elements of the scenario. For example, if the constraint boundary condition stipulates that "the power supply load of a substation shall not exceed 90% of its rated capacity," then substation nodes whose power supply load may exceed this limit under the current scenario's key parameters, even if their scenario adaptation identifiers meet the requirements, need to be excluded or specially processed. Through these two steps of filtering, the final core data node set for the scenario accurately reflects the key nodes that play a decisive role in energy supply and demand under the current scenario.

[0090] Step S135: Extract all connection links between the core data node set of the scene in the dynamic source-load association network, obtain the association transmission logic and scene adaptation weight of each connection link, and form a scene association connection link set.

[0091] Once the set of core data nodes for the scenario is determined, all connection links between these nodes need to be extracted. These connection links constitute the channels for interaction between the core data nodes. For each connection link, its associated transmission logic and current scenario adaptation weight need to be obtained. The associated transmission logic describes the transmission method and rules of data or energy in the link, such as the regenerative braking energy transmission logic from the train node to the energy storage node. The scenario adaptation weight reflects the importance of the link in the current scenario. Integrating these connection links and their attributes forms a set of scenario-related connection links, which defines the interaction relationships and influence strength between the core data nodes of the scenario.

[0092] Step S136: Based on the demand priority information in the core elements of the scene, calculate the new scene adaptation weight of each connection link in the scene-related connection link set, and assign the new scene adaptation weight to the corresponding connection link.

[0093] The priority information of the core elements of a scenario determines the relative importance of different connection links in the current scenario. For example, in the priority of requirements, "train traction power supply guarantee" has a higher priority than "station lighting energy saving," so the scenario adaptation weight of the link connecting the train node and the substation node should be higher than that of the link connecting the station lighting node and the substation node. When calculating the new scenario adaptation weight, a basic weight is first assigned to different types of connection links based on the priority information of the requirements, and then adjusted by combining factors such as the current transmission efficiency of the link and the scenario matching degree. For example, for a high-priority link, if its current transmission efficiency is low, its weight is appropriately reduced; if its transmission efficiency is high, its weight is further increased. The calculated new scenario adaptation weight is assigned to the corresponding connection link to realize the dynamic update of the link weight, enabling the network to pay more attention to meeting high-priority requirements.

[0094] Step S1361: Extract the requirement priority information from the core elements of the scene. The requirement priority information includes the priority ranking of multiple requirement dimensions and the weight ratio of each priority. The requirement dimensions correspond to the association characteristics of the scene-related connection links.

[0095] Demand priority information is a crucial component of the core elements of a scenario. Demand dimensions typically include train traction power supply, station lighting, regenerative braking energy recovery, and energy storage device charging / discharging, with each dimension corresponding to a specific type of energy demand or supply activity. Prioritization clarifies the order in which these demand dimensions occur within the current scenario, such as "train traction power supply" > "regenerative braking energy recovery" > "station lighting" > "energy storage device charging / discharging." The weight percentage for each priority quantifies its importance within the overall demand; for example, "train traction power supply" has a weight percentage of 40%, "regenerative braking energy recovery" 30%, "station lighting" 20%, and "energy storage device charging / discharging" 10%. The demand dimensions correspond to the characteristics of the connection links between the scenario and the demand. For instance, the "train traction power supply" demand dimension corresponds to the connection link between the train node and the substation node, characterized by high-power transmission and high real-time requirements.

[0096] Step S1362: Analyze the association transmission logic of each connection link in the scenario association connection link set, identify the requirement dimension corresponding to each connection link, and determine the matching between the association transmission logic and the requirement dimension based on the association attributes of the connection link.

[0097] The associated transmission logic includes information such as the type of energy activity served by the link, the flow of data or energy, and the equipment involved. This information serves as the basis for determining the demand dimension. For example, if the associated transmission logic of a connection link is described as "the train feeds electrical energy back to the energy storage device during braking," then its corresponding demand dimension is "regenerative braking energy recovery." Matching the associated transmission logic with the demand dimension is based on the associated attributes of the connection link, such as the link's power level, transmission direction, and the types of nodes involved. For connection links that simultaneously serve multiple demand dimensions, it is necessary to determine their primary corresponding demand dimension based on their main function and energy proportion, or to perform multi-dimensional decomposition and weight allocation.

[0098] Step S1363: Match the requirement dimension corresponding to each connection link with the requirement dimension in the requirement priority information to obtain the corresponding priority ranking and weight ratio.

[0099] The required dimensions identified in the above steps for each connection link are matched with the required dimensions defined in the required priority information. Upon successful matching, the priority ranking and weight percentage of that required dimension in the current scenario can be obtained. For example, if the required dimension for a connection link is "train traction power supply," matching the required priority information reveals its priority ranking to be 1 and its weight percentage to be 40%. This step closely links the connection links to the required priorities of the scenario. For connection links that cannot be directly matched, manual intervention or a fuzzy matching algorithm may be necessary to ensure that each link obtains the corresponding priority information.

[0100] Step S1364: Based on priority sorting, assign an initial adjustment weight to each connection link in the scene-related connection link set. The initial adjustment weight is positively correlated with the priority sorting, and the higher the priority, the greater the initial adjustment weight.

[0101] For example, priority can be divided into 1 to 5 levels, with an initial adjustment weight of 0.9 for priority 1, 0.7 for priority 2, 0.5 for priority 3, 0.3 for priority 4, and 0.1 for priority 5. This positive correlation ensures that links corresponding to high-priority needs receive more attention and resources in the network. The allocation of initial adjustment weights can use methods such as linear mapping or exponential mapping, depending on the steepness of the priority difference required.

[0102] Step S1365: Combine the weight ratio to correct the initial adjustment weight. The corrected weight ratio is consistent with the weight ratio in the requirement priority information, so that the adjusted connection link scenario adaptation weight conforms to the requirement priority distribution.

[0103] Based on the initial weight adjustments, the weights are further refined by considering the weight percentage of each demand dimension. The goal of this refinement is to ensure that the weight distribution of all connection links aligns with the weight distribution in the demand priority information. For example, assuming the initial weight adjustment for all "train traction power supply" links is A, and its weight percentage is 40%, the refinement factor is (total target weight × 40%) / A. The initial weight adjustment for each link in this demand dimension is multiplied by this refinement factor to obtain the refined weight. This method ensures that links corresponding to demand dimensions with high weight percentages have a larger overall weight, thus dominating the network and guaranteeing that the adjusted connection link scenario adaptation weights conform to the overall distribution of demand priorities.

[0104] Step S1366: Analyze the basic value of the association strength of each connection link. The basic value of the association strength is the initial scenario adaptation weight of the connection link in the dynamic source-load association network. The corrected weight is calibrated in combination with the basic value of the association strength.

[0105] After obtaining the corrected weights, calibration is required in conjunction with the baseline correlation strength value to avoid relying solely on priority information and ignoring the inherent characteristics of the links themselves. The calibration method can be a weighted average of the corrected weights and the baseline correlation strength value. For example, the new scenario adaptation weight = α × corrected weight + (1-α) × baseline correlation strength value, where α is the weight coefficient, adjusted according to the dynamics of the scenario and the importance of the priority. For links with particularly high or low baseline correlation strength values, the calibration process can appropriately increase or decrease their new scenario adaptation weights to reflect their special status in the network.

[0106] Step S1367: Through the weight calibration logic, the corrected weights are fused with the basic value of the association strength to generate the final connection link scene adaptation weights. The weight calibration logic is generated based on the transmission efficiency of the connection link and the scene adaptation accuracy.

[0107] The weight calibration logic is the specific rule governing the fusion of the corrected weights and the baseline association strength value. This logic comprehensively considers two factors: the transmission efficiency of the connection link and the scene adaptation accuracy. Links with high transmission efficiency indicate strong data or energy transmission capabilities, and their weights can be appropriately increased during fusion. Links with high scene adaptation accuracy indicate good matching with the current scene and should also be given higher weights. The weight calibration logic can be implemented by establishing a functional relationship between transmission efficiency, scene adaptation accuracy, and calibration coefficients. For example, for every 10% increase in transmission efficiency, the calibration coefficient increases by 0.05; for every 10% increase in scene adaptation accuracy, the calibration coefficient increases by 0.05. These coefficients are applied to the fusion process of the corrected weights and the baseline association strength value to generate the final connection link scene adaptation weights.

[0108] Step S1368: Update the final connection link scenario adaptation weight to the corresponding connection link in the scenario-associated connection link set, replacing the original scenario adaptation weight, and realize the dynamic adjustment of connection link weight.

[0109] The calculated final connection link scenario adaptation weights are then updated to the corresponding connection link attributes in the scenario-associated connection link set, replacing the original initial scenario adaptation weights or the weights adjusted in the last iteration. This process is the final step in the dynamic adjustment of connection link weights, ensuring that the link weights can reflect the current scenario's demand priority, the link's own characteristics, and its operational status in real time. The updated scenario-associated connection link set can more accurately guide subsequent energy supply and demand trend projections and optimization decisions.

[0110] Step S1369: Apply the verification logic to verify the scenario adaptation weights of all connection links in the adjusted scenario-related connection link set, and record the weight adjustment parameters, as well as all parameters and logic in the weight adjustment process, including the requirement dimension matching results, initial adjustment weights, correction basis, and calibration logic, to form a weight adjustment record as supplementary information for the multi-dimensional fusion inference input set.

[0111] The verification logic includes verifying the weight's value range (e.g., whether it's between 0 and 1), verifying the total weight (e.g., whether the sum of all link weights equals 1 or a certain set value), and verifying the weights of critical links (e.g., whether the link weights corresponding to high-priority requirements are sufficiently large). For weights that fail verification, the parameters and logic in the adjustment process need to be re-examined and corrected. Simultaneously, detailed records of weight adjustment parameters are kept, including the requirement dimension matching results, initial adjusted weights, correction basis (e.g., weight percentage, transmission efficiency), and various coefficients in the calibration logic, forming a weight adjustment record. This weight adjustment record serves as supplementary information to the multi-dimensional fusion simulation input set, helping to trace the weight adjustment process and analyze the reliability of the simulation results.

[0112] Step S137: Integrate the core data node set of the scene, the adjusted scene association connection link set, and the core elements of the scene to form the intermediate result of scene mapping. The intermediate result of scene mapping includes node attributes, connection link parameters, and scene constraint information.

[0113] The intermediate result of scene mapping is a phased product after scene features are mapped to the dynamic source-load association network. It integrates the core data node set of the scene, incorporating its node attributes (such as current passenger flow, train speed, and substation power) into the intermediate result. The adjusted set of scene association connection links, including the link association transmission logic, new scene adaptation weights, transmission efficiency, and other connection link parameters, is also integrated. Scene constraint information, such as key scene parameters, constraint boundary conditions, and demand priority information from the core elements of the scene, is also an important component of the intermediate result.

[0114] Step S138: Based on the connection link parameters in the intermediate results of scene mapping, extract the key path of the associated transmission logic. The key path is the connection link transmission path with the highest association strength between the core data nodes of the scene.

[0115] The critical path is the main channel for energy or information transmission between core data nodes in a scenario. It has the highest correlation strength and the greatest impact on the energy supply and demand outcomes of the scenario. Critical paths are extracted based on the connection link parameters in the intermediate results of scenario mapping, particularly scenario adaptation weights and transmission efficiency. Extraction methods can employ variations of the shortest path algorithm or maximum flow algorithm from graph theory, using the combination of scenario adaptation weights and transmission efficiency as the "strength" index of the link to find the path with the highest strength from the source node (e.g., energy supply node) to the sink node (e.g., energy demand node). For example, in a morning rush hour scenario, the path from the main substation node, through the traction power supply link to the busy line train node has the highest correlation strength and is identified as the critical path. Extracting the critical path helps focus network analysis, simplifies subsequent deduction processes, and improves deduction efficiency.

[0116] Step S139: Supplement the data interaction rules in the critical path. The data interaction rules are generated based on the characteristics of the data interaction channels in the dynamic source-load association network, and include data transmission format, interaction timing and exception handling logic.

[0117] Once the critical path is determined, its data interaction rules need to be supplemented to ensure efficient and reliable transmission of data or energy along the critical path. These rules are generated based on the characteristics of the data interaction channels in the dynamic source-load association network. The data transmission format specifies the encoding method, field definitions, and verification methods for data during the interaction process. The interaction sequence defines the time order, intervals, and synchronization methods for data transmission and reception. The exception handling logic describes the response measures for abnormal situations such as data transmission errors, link failures, and data timeouts, including data retransmission mechanisms, link switching strategies, and degraded operation modes.

[0118] Step S1310: Integrate the core data node set of the scene, key paths, related transmission logic, adjusted scene adaptation weights and data interaction rules to form the multi-dimensional fusion inference input set.

[0119] The multi-dimensional fusion simulation input set is the final input data for extrapolating energy supply and demand trends. It integrates the core data nodes of the scenario, providing the main objects of the simulation; the critical path clarifies the main direction and channels of the simulation; the correlation and transmission logic describes the interaction rules between nodes; the adjusted scenario adaptation weights quantify the impact of each link; and the data interaction rules ensure data flow during the simulation process. Integrating these elements forms the multi-dimensional fusion simulation input set. This multi-dimensional fusion simulation input set can comprehensively reflect the source-load relationship and operational constraints of the transportation energy system under the current scenario, ensuring the accuracy and reliability of the simulation results.

[0120] Step S140: Using a multi-dimensional fusion inference model adapted to the scenario, the source load change trend of the multi-dimensional fusion inference input set is inferred to obtain the traffic energy source load prediction result. The traffic energy source load prediction result includes the dynamic curve of supply and demand change, the transmission path of related influences, and the scenario adaptation deviation.

[0121] The multi-dimensional fusion simulation model for scenario adaptation constructs a dynamic simulation model of the transportation energy system based on information such as core data nodes, critical paths, and related transmission logic from the multi-dimensional fusion simulation input set. The model simulates the interaction relationships between different data nodes, energy flow processes, and the impact of scenario constraints to predict the changing trends of energy supply and demand over a future period. The dynamic curves of supply and demand change, with time as the horizontal axis, show the changes in energy supply and demand over time, including peak values, trough values, and rates of change. The transmission path of related impacts refers to how changes at one node affect other nodes through connecting links during the simulation process, ultimately leading to changes in supply and demand, such as the transmission path of increased passenger flow → increased train traction energy consumption → increased substation power supply load. Scenario adaptation deviation refers to the difference between the simulation results and the scenario constraints or expected targets, such as the actual simulated power supply load exceeding the maximum allowable value in the scenario constraints, or the energy utilization rate being lower than the expected target value. The transportation energy source and load prediction results comprehensively reflect the future operating status of the transportation energy system under the current scenario.

[0122] In this embodiment, the multi-dimensional fusion inference model adopts a hybrid architecture combining deep learning and system dynamics, including a node parsing module, a connection link enhancement module, a scene adaptation module, and a trend inference module. Each module achieves feature interaction through a fully connected approach. The node parsing module uses a 3-layer fully connected neural network. The input layer receives 256-dimensional attribute feature vectors from the core data nodes of the scene. The intermediate layer performs nonlinear transformation using the ReLU activation function, and the output layer extracts 128-dimensional dynamic change features and 64-dimensional association response features. The connection link enhancement module adopts a graph attention network (GAT) structure, constructing a weighted adjacency matrix by combining node features and connection link parameters (scene adaptation weights, transmission efficiency). It calculates the association strength between nodes through an 8-head attention mechanism, generating 192-dimensional association enhancement features. The scene adaptation module uses a gated recurrent unit (GRU) network, temporally fusing the association enhancement features with scene constraint information. It dynamically adjusts the dimensional weights of the 128-dimensional scene adaptation features through a gating mechanism. The trend inference module adopts a Transformer encoder-decoder architecture. The encoder performs spatiotemporal modeling of scene adaptation features, and the decoder outputs the energy supply and demand sequence for the next 24 hours. Each encoder / decoder contains 6 attention sub-layers, and each layer has a hidden dimension of 256.

[0123] Model training is divided into two stages: pre-training and fine-tuning. The pre-training stage uses 12 months of historical traffic and energy data (including 8760 hours of traffic operation, energy supply, and environmental correlation data). Data is standardized before input (Z-score normalization), and the labels are the actual energy supply and demand values ​​for the next 1 hour, 3 hours, and 24 hours. The loss function is a weighted combination of mean squared error (MSE) and dynamic time warping (DTW) loss (weight ratio 7:3). The optimizer is AdamW, with an initial learning rate of 0.001, decaying by 50% every 20 epochs. Pre-training iterates for 300 epochs, stopping when the validation set loss shows no decrease for 20 consecutive epochs. The fine-tuning stage, targeting specific scenarios (such as weekday morning rush hour), uses data from the past 3 months to adjust parameters. The parameters of the feature extraction layer are frozen, and only the fully connected layer of the trend inference module is fine-tuned. The learning rate is reduced to 0.0001, and the iterations are performed for 50 epochs. Early stopping is used during training to prevent overfitting, and training loss and validation metrics (MAE, RMSE) are monitored in real time using TensorBoard.

[0124] When applied to urban rail transit scenarios, the model input is set as the current core data node set (passenger flow at 15 key stations, train status on 8 lines, parameters of 5 substations), connection link parameters (correlation and transmission logic of 32 key links), and scenario constraint information (power supply load limit, energy storage SOC threshold). The input data sampling frequency is 5 minutes / time. The output is a dynamic curve of supply and demand changes over the next 24 hours (sampling points at 15-minute intervals), the top 5 correlation impact transmission paths, and a list of scenario adaptation deviations. The contribution of key features is analyzed using model interpretability tools (SHAP values). For example, the average SHAP value of train traction energy consumption during the morning peak period is 0.72, significantly higher than other features. TensorRT is used for inference acceleration during model deployment, with a single inference time controlled within 2 seconds, meeting real-time scheduling requirements. Privacy-sensitive data such as passenger flow and train location involved in the training data are protected using differential privacy technology (adding Laplace noise, privacy budget ε=1.5) and a federated learning framework (each substation / station acts as a local node, only uploading the model parameter gradient) to ensure that the original data does not leave the local network.

[0125] Step S141: Input the set of core data nodes, key paths and related transmission logic of the scene from the multi-dimensional fusion inference input set into the node parsing module of the multi-dimensional fusion inference model, and extract the dynamic change features and related response features of each core data node of the scene. The dynamic change features reflect the evolution law of the data node over time, and the related response features reflect the feedback logic of the data node to the connection link.

[0126] The node parsing module of the multi-dimensional fusion inference model is responsible for analyzing and extracting features from the input set of core data nodes in the scenario. For each core data node, the node parsing module first analyzes its historical data and current state to extract dynamic change features. Dynamic change features are the patterns of data node evolution over time, such as the periodic fluctuations in passenger flow (morning peak, evening peak), the speed-energy consumption characteristic curve of train energy consumption, and the daily variation trend of substation power supply load. These features can be extracted using methods such as time series analysis and curve fitting. Correlated response features reflect the feedback logic of data nodes to changes in other nodes in the connection link. For example, the response characteristics of the train traction converter when the substation power supply voltage fluctuates; the charging and discharging response strategy of the energy storage device when the train undergoes regenerative braking. Correlated response features can be obtained by parsing the node's control strategy, physical model, or historical interaction data.

[0127] Step S142: Through the connection link strengthening module of the multi-dimensional fusion inference model, based on the scenario adaptation weight in the critical path, strengthen the correlation between dynamic change features and associated response features, and generate a set of associated strengthening features. Each feature in the set of associated strengthening features contains dual information of node attributes and connection link association.

[0128] The role of the link enhancement module is to strengthen the correlation between dynamic change features and associated response features to better reflect the interaction between nodes on the critical path. Based on the scenario adaptation weights in the critical path, the node features associated with link connections that have high scenario adaptation weights are enhanced. For example, for a link on the critical path connecting a train node and a substation node, which has a high scenario adaptation weight, the correlation between the train's dynamic change features (such as speed changes) and the substation's associated response features (such as power supply adjustments) is enhanced. Enhancement methods can include increasing the coupling coefficient between features or introducing new cross-features to describe their combined effect. Each feature in the generated set of enhanced features contains both node attributes (such as train speed and substation voltage) and link association information (such as link weight and transmission efficiency). This dual information allows the features to more comprehensively reflect the synergistic effect between nodes and links.

[0129] Step S143: Input the association enhancement feature set into the scene adaptation module of the multi-dimensional fusion inference model, combine the scene constraint information in the multi-dimensional fusion inference input set, adjust the dimension weights of the association enhancement feature set, and generate a scene adaptation feature set. The dimension weights are consistent with the requirement priority in the scene constraint information.

[0130] The scene adaptation module adjusts the dimensional weights of the associated reinforcement feature set based on scene constraint information in the multi-dimensional fusion inference input set. The priority of requirements in the scene constraint information determines the importance of different types of features. For example, if "train traction energy consumption" has a higher priority than "station lighting energy consumption" in the requirement priority, then in the associated reinforcement feature set, the weights of features related to train traction are increased, while the weights of features related to station lighting are decreased. This adjustment of dimensional weights can be achieved through a mapping function that converts requirement priority information into weight coefficients for feature dimensions. The generated scene-adaptive feature set maintains consistency between the importance of its feature dimensions and the scene requirement priority, enabling the model to focus more on high-priority features during inference and improving the scene adaptability of the inference results.

[0131] Step S144: Through the trend inference module of the multi-dimensional fusion inference model, based on the dynamic change features in the scenario adaptation feature set, the short-term change trend and long-term evolution trend of transportation energy source load are inferred, and preliminary trend inference results are generated. The preliminary trend inference results include the predicted trend of supply and demand changes and key change nodes.

[0132] The trend projection module is the core execution unit of the multi-dimensional fusion projection model. Based on the dynamic changes in the scenario-adaptive feature set, the module uses methods such as time series forecasting, system dynamics simulation, and machine learning prediction to project the short-term changes (e.g., in the next 1 hour or 3 hours) and long-term evolution trends (e.g., in the next 12 hours or 24 hours) of transportation energy sources and loads. The short-term trend projection focuses more on details and real-time performance, considering fluctuations at the minute or hour level; the long-term evolution trend projection focuses more on overall trends and periodicity, considering daily or weekly patterns of change. In the preliminary trend projection results, the predicted trends of supply and demand changes are given in the form of curves or textual descriptions, such as "Energy demand will show a rapid upward trend in the next 3 hours, while supply capacity will remain basically stable." Key change nodes refer to the time points or event points where the supply and demand relationship undergoes significant turning points, such as "Demand will exceed the supply warning threshold in the next 2 hours" or "Regenerative braking energy recovery will reach its peak in the next 6 hours."

[0133] Step S145: Combining the associated response features in the scene adaptation feature set with the preliminary trend inference results, determine the impact of changes in the connection links on the inference results according to the associated transmission logic. By comparing the inference results with the scene constraint information, identify the parts of the inference results that do not conform to the scene constraint information and generate the scene adaptation deviation.

[0134] Identifying scenario adaptation bias is a crucial step in ensuring the feasibility of the simulation results. First, by combining the correlated response features in the scenario adaptation feature set with the preliminary trend simulation results, the impact of changes in connection links on the simulation results is analyzed. For example, a decrease in the transmission efficiency of a critical link may lead to an inability to meet demand in a timely manner, resulting in a lower supply curve in the preliminary simulation results. Based on the correlation transmission logic, the propagation path and extent of this impact can be traced. Next, the preliminary trend simulation results are compared with the scenario constraint information to check whether the simulation results meet all constraints. For example, if the scenario constraint stipulates that "the state of charge of the energy storage device must not be lower than 20%", and the preliminary simulation results show that the state of charge will drop to 15% at some future time, then this part of the result does not conform to the scenario constraint information and is identified as a scenario adaptation bias. Scenario adaptation bias needs to be recorded in detail, including the time of occurrence, the bias value, and the nodes and links involved.

[0135] Step S1451: Extract the associated response features from the scene adaptation feature set. The associated response features include the response pattern, response time and response magnitude of each scene core data node to changes in the connection link. The response pattern is generated based on historical data interaction records.

[0136] The associated response features extracted from the scenario adaptation feature set describe the inherent patterns of how the attribute values ​​of core data nodes in a scenario adjust when the connection link changes. For example, when the transmission power of the link connecting the train and the substation increases by 10%, the train's acceleration response initially increases linearly and then tends to stabilize. This response pattern is generated based on historical data interaction records (such as train acceleration data under similar link changes in the past) through statistical analysis or machine learning model fitting. Response timeliness refers to the time delay from the occurrence of a link change to the node attribute value starting to respond, and the time required for the response to reach stability. For example, after a link power change, the train acceleration begins to change after 2 seconds and reaches a stable value after 5 seconds. Response magnitude is the magnitude of the change in node attribute value; for example, a 10% increase in link power leads to an 8% increase in train acceleration.

[0137] Step S1452: Analyze the supply and demand changes and key change nodes in the preliminary trend projection results, and determine the set of connection links corresponding to the preliminary trend projection results. The set of connection links is the core transmission path that supports the projection results.

[0138] Analyze the preliminary trend projection results to identify the direction of supply and demand changes (such as rising demand, declining supply, and supply-demand equilibrium) and key change points (such as supply-demand crossover points, peak points, and trough points). Based on this information, deduce the core transmission path supporting the projection results, i.e., the set of connection links. For example, if the preliminary projection results show a surge in energy demand in a certain region within the next two hours, the set of connection links supporting this result might include links between passenger flow nodes and train nodes at major stations in the region, links between train nodes and substation nodes, etc. Determining the set of connection links requires analyzing how changes in the attribute values ​​of each data node in the projection results are transmitted and influenced through the connection links, thereby identifying those links that play a decisive role in the results.

[0139] Step S1453: Match the associated response features with the corresponding set of connection links, analyze the impact of changes in each connection link on the response features of the core data nodes in the scenario, and conduct the impact analysis based on the associated transmission logic and response patterns.

[0140] The extracted correlation response features are matched with the determined set of connection links, so that each connection link corresponds to the correlation response features of its related nodes. Then, the impact of changes in each connection link (such as changes in scene adaptation weights and transmission efficiency) on the response features of the core data nodes in the scene is analyzed. The impact analysis is based on the correlation transmission logic (how the link transmits changes) and the response pattern (how the node responds to changes). For example, if the scene adaptation weight of a certain connection link increases, according to the correlation transmission logic, more energy will be transmitted through this link. Combined with the response pattern of the nodes (such as the acceleration response magnitude and timeliness of a train node after receiving more energy), the changes in attributes such as train speed and energy consumption can be analyzed, and then the impact on the overall supply and demand simulation results can be assessed.

[0141] Step S1454: Based on the correlation transmission logic and the impact results, generate the correlation impact transmission path from the change in the connection link through the response of the data node to the adjustment of the inference results.

[0142] Based on the correlation transmission logic and the results of the above impact analysis, a correlation impact transmission path is generated. This path details the entire process from the change in the connection link, through the response of the data nodes, to the final adjustment of the projection results. For example, "the scenario adaptation weight of connection link L1 increases → the traction power response of train node N1 increases (response pattern: the power increase is 1.2 times the weight increase, and the response time is 5 seconds) → the energy consumption of train N1 increases → the demand curve in the preliminary trend projection results adjusts upward." The correlation impact transmission path demonstrates the causal relationship between the link, nodes, and projection results.

[0143] Step S1455: Extract scene constraint information from the multi-dimensional fusion simulation input set. The scene constraint information includes the boundary conditions, requirement standards and adaptation requirements of the scene operation. The scene constraint information is the basis for judging the adaptation deviation.

[0144] The scenario constraint information from the multi-dimensional fusion simulation input set serves as the basis for judging the rationality of the simulation results. This information is extracted from various sources, including: boundary conditions for scenario operation, such as maximum equipment capacity, minimum operating parameters, and safety thresholds; demand standards, such as energy supply reliability requirements (e.g., power outage time must not exceed 5 minutes) and energy efficiency targets (e.g., regenerative braking energy recovery rate must not be less than 30%); and adaptation requirements, such as the deviation range between the simulation results and historical data from the same period, and differences in performance indicators under different scenarios. This scenario constraint information is then organized into judgment criteria to verify each part of the preliminary trend simulation results.

[0145] Step S1456: Compare the preliminary trend projection results with the scenario constraint information, and identify the parts of the projection results that exceed the scenario constraint boundaries, do not meet the requirement standards, or do not meet the adaptation requirements, as potential scenario adaptation deviations.

[0146] The preliminary trend projection results are compared item by item with the extracted scenario constraint information. For numerical projection results, it is checked whether they exceed the scenario constraint boundaries, such as whether "substation power supply load" exceeds "maximum allowable power supply load". For descriptive projection results, it is checked whether they meet the demand standards, such as whether "energy supply interruption time" meets the standard of "not exceeding 5 minutes". At the same time, it is assessed whether the projection results meet the adaptation requirements, such as whether the deviation from historical data is within the allowable range. All parts that exceed the boundaries, do not meet the standards, or fail to meet the requirements are identified as potential scenario adaptation deviations, and their specific locations, deviation degrees, and related node link information are recorded.

[0147] Step S1457: Combine the influence transmission path of the connection link on the simulation results, and associate the potential scenario adaptation deviation with the specific connection link, the associated transmission logic, or the scenario adaptation weight.

[0148] By leveraging the previously generated correlation and influence propagation paths, potential scenario adaptation deviations can be traced back to their root causes. For example, a potential scenario adaptation deviation manifests as "the energy storage device's state of charge is below the constraint threshold." Through the correlation and influence propagation paths, this can be traced back to "the transmission efficiency of the regenerative braking energy recovery link L2 connecting the energy storage device and the train is too low," or "the scenario adaptation weights of this link are set unreasonably," or "there is an error in the calculation model for regenerative braking energy in the correlation and propagation logic." Associating potential scenario adaptation deviations with specific connection links, correlation and propagation logic, or scenario adaptation weights helps to accurately identify the causes of the deviations.

[0149] Step S1458: Classify potential scenario adaptation deviations based on the attribute information and connection link parameters of the core data nodes in the scenario. The classification criteria include data source, logical rules, or scenario constraints.

[0150] Classifying potential scenario adaptation biases helps in implementing targeted corrective measures. Based on data source, biases can be categorized as follows: biases originating from traffic operation data (e.g., demand biases due to inaccurate passenger flow forecasts), biases originating from energy supply data (e.g., supply biases due to overestimation of substation power supply capacity), and biases originating from environmental data (e.g., energy consumption biases due to incorrect temperature forecasts). Based on logical rules, biases can be categorized as those caused by errors in the associated transmission logic (e.g., biases due to incorrect energy transfer models) and biases due to unreasonable scenario adaptation weights. Based on scenario constraints, biases can be categorized as biases violating boundary conditions, biases failing to meet demand standards, and biases not meeting adaptation requirements. The categorized potential scenario adaptation biases have clearer characteristics, facilitating subsequent processing.

[0151] Step S1459: Determine the specific manifestations, scope of impact, and causes of each scene adaptation deviation, and form a scene adaptation deviation list. The structure of the scene adaptation deviation list is consistent with the preliminary trend projection results.

[0152] The scenario adaptation deviation list is a systematic compilation of all potential scenario adaptation deviations. For each deviation, its specific manifestation is clearly defined, such as "the minimum state of charge of the energy storage device in the next 3 hours is 15%, which is lower than the constraint threshold of 20%"; its scope of impact is defined, such as "this deviation will result in the inability to meet the energy recovery needs of train regenerative braking in the following 5 hours, thus affecting the overall energy supply and demand balance"; and its cause is defined, such as "the actual conduction efficiency of the connection link L2 is lower than the model input value, resulting in insufficient energy recovery." The structure of the scenario adaptation deviation list is consistent with the preliminary trend projection results, organized by time sequence or node / link category, facilitating comparative analysis and correction.

[0153] Step S14510: Supplement the scenario adaptation deviation list with correction direction suggestions, which are generated based on associated response features and associated transmission logic.

[0154] Based on the associated response characteristics and associated transmission logic, suggested correction directions are provided for each deviation in the scenario adaptation deviation list. These correction direction suggestions should be targeted and actionable. For example, for the deviation of "insufficient energy recovery due to low transmission efficiency," a suggested direction could be "improving the transmission efficiency of the L2 connection link, such as adjusting the link's control parameters or performing equipment maintenance." For the deviation of "unreasonable scenario adaptation weights," a suggested direction could be "recalculating the scenario adaptation weights of the L2 link, considering the current priority of regenerative braking energy demand." These correction direction suggestions provide initial ideas and options for the subsequent deviation correction strategy development, ensuring that the correction work is targeted and effective.

[0155] Step S146: Based on the associated transmission logic and the scene adaptation weight, generate a deviation correction strategy. The deviation correction strategy includes a correction direction and correction logic, and the correction logic is consistent with the transmission characteristics of the connection link.

[0156] Deviation correction strategies are specific solutions to address scenario adaptation deviations. The correction direction is determined based on suggestions in the scenario adaptation deviation list, such as adjusting the conduction efficiency of connection links, optimizing scenario adaptation weights, and correcting associated conduction logic. The correction logic details how to implement the correction direction. For example, for adjusting conduction efficiency, the correction logic could be "to increase the transmission voltage of the link to reduce resistance loss, thereby improving conduction efficiency." This correction logic needs to be consistent with the conduction characteristics of the connection link (such as resistance, reactance, rated voltage, etc.) to ensure the correction measures are physically feasible. Deviation correction strategies may involve the coordinated adjustment of multiple links and nodes, requiring comprehensive consideration of various influences to avoid generating new deviations.

[0157] Step S147: Apply the deviation correction strategy to adjust the preliminary trend projection results to obtain the corrected trend projection results, which eliminate the discrepancies between the projection results and the scenario constraint information.

[0158] The deviation correction strategy is applied to the initial trend projection results to adjust any parts that do not conform to the scenario constraints. For example, after improving the transmission efficiency of a link according to the correction strategy, the energy transmission volume of that link is recalculated, and the energy supply and demand attributes of the relevant nodes are adjusted accordingly, updating the entire trend projection results. This adjustment process may require iteration; that is, the trend projection is re-run after applying the correction strategy to check for any new deviations, until all scenario constraints are met. The corrected trend projection result is the final projection result that eliminates all known scenario adaptation deviations. It more accurately reflects the future operating state of the transportation energy system under the current scenario, and has higher reliability and practicality.

[0159] Step S148: Extract the key change nodes in the corrected trend projection results, analyze the associated transmission paths and scenario constraints corresponding to the key change nodes, and generate the associated influence transmission paths.

[0160] Key change nodes in the revised trend projection results, such as the energy supply-demand balance point, the maximum demand point, and the minimum supply point, are important turning points in the system's operation. These key change nodes are extracted, and their corresponding transmission paths are analyzed—that is, the transmission process of all relevant connecting links and data nodes leading to the change at that node. Simultaneously, the scenario constraints corresponding to the key change nodes are analyzed, such as which scenario constraints play a dominant role at that node and their specific values. Integrating the key change nodes, their corresponding transmission paths, and scenario constraints generates a correlation influence transmission path. This path reveals the causes and influencing factors of the key change nodes, contributing to a deeper understanding of the system's dynamic behavior.

[0161] Step S149: Integrate the corrected trend projection results, the transmission path of related influences, and the scenario adaptation deviation to form the traffic energy source load prediction results.

[0162] The integrated and corrected trend projection results provide energy supply and demand change curves and information on key change nodes. The transmission paths of related impacts serve as an explanation and supplement to the projection results, illustrating the causes and processes of key changes. Although scenario adaptation bias has been corrected, its record remains important, reflecting problems encountered during the projection process and limitations of the model. The integrated transportation energy source and load forecast results comprehensively and accurately predict the source and load change trends of the transportation energy system.

[0163] Step S150: Based on the traffic energy source and load prediction results and traffic energy scheduling requirements, a traffic energy source and load collaborative optimization scheme is formed. The traffic energy source and load collaborative optimization scheme is fed back to the dynamic source and load association network through the data interaction channel to realize the scenario adaptation weight optimization of the dynamic source and load association network.

[0164] Transportation energy dispatching requirements include targets for energy supply reliability, economy, and environmental friendliness; traffic operation punctuality and comfort; and system safety and stability. Based on source-load forecasting results, future energy supply and demand trends and potential problems are analyzed. Combined with dispatching requirements, optimization strategies are formulated, such as adjusting train operation plans to balance energy demand, optimizing the operation of substations and energy storage devices to improve energy utilization efficiency, and developing contingency plans to address potential supply-demand imbalances. The resulting collaborative optimization schemes are fed back to the dynamic source-load correlation network through data interaction channels, primarily used to optimize the network's scenario adaptation weights, enabling it to better reflect the optimized source-load relationships and ensuring the continuous optimization of the dynamic source-load correlation network and the efficient operation of the transportation energy system.

[0165] Step S151: Extract energy supply constraints, traffic operation demand, supply and demand balance target and dispatch response time from the traffic energy dispatch demand to form the core elements of dispatch demand. The core elements of dispatch demand correspond to the dimensions of the traffic energy source and load prediction results.

[0166] Traffic energy dispatching demand serves as the basis for formulating collaborative optimization schemes, from which core elements of dispatching demand are extracted. Energy supply constraints include the maximum power output of each substation, capacity limitations of energy storage devices, and the lower limit of regenerative braking energy recovery efficiency; traffic operation demands include the minimum train interval, maximum allowable delay time, and service quality standards of station facilities; supply and demand balance targets define the allowable deviation range between energy supply and demand, energy utilization rate target values, peak-valley difference control targets, etc.; dispatching response timeliness stipulates the time limit for formulating and implementing optimization schemes, such as a response time of no more than 10 minutes in emergency situations and an optimization cycle of 1 hour under normal circumstances. The dimensional design of the core elements of dispatching demand corresponds to the dimensions of traffic energy source and load forecasting results. For example, the "substation power output" dimension in the forecasting results corresponds to the "maximum substation power output" constraint in the dispatching demand, ensuring that the two can be effectively compared and integrated.

[0167] Step S152: Analyze the corrected trend projection results, related impact transmission paths, and scenario adaptation deviations in the traffic energy source and load prediction results. Based on the constraints of the core elements of scheduling demand, divide the analysis results into parts that meet the constraints and parts that do not meet the constraints.

[0168] The analysis of traffic energy source and load forecasts aims to gain a deeper understanding of the energy supply and demand curves, key change nodes, major influencing factors and transmission mechanisms in the revised trend projection results, and the specific details of scenario adaptation deviations. Then, based on the constraints in the core elements of dispatch requirements, the analysis results are evaluated and categorized. For example, the portion of the revised trend projection results stating "the power supply of substation A will be lower than its maximum allowable value within the next 3 hours" is categorized as meeting the constraints; the portion stating "the state of charge of the energy storage device will be lower than the minimum protection value within the next 5 hours" is categorized as not meeting the constraints. The transmission paths of related impacts and scenario adaptation deviations are also categorized according to whether they meet the potential requirements of dispatch requirements. For instance, a related path leading to supply and demand imbalance is considered an unmet requirement, while the original analysis results corresponding to the corrected deviations may be considered met.

[0169] Step S153: For the part that meets the constraints, extract the corresponding correlation influence transmission path and scenario constraints as the basic support basis for collaborative optimization. The basic support basis includes data node association logic and scenario adaptation rules.

[0170] For the analysis results that satisfy the constraints, the underlying correlation and influence transmission paths and scenario constraints represent successful system operation experiences and should be extracted as the foundational support for collaborative optimization. The aforementioned correlation and influence transmission paths demonstrate how data nodes interact positively through connection links while meeting scheduling requirements, and their data node association logic (such as dependencies between node attributes and link transmission rules) is effective. Scenario constraints, on the other hand, ensure that the system does not operate out of bounds, and their scenario adaptation rules (such as applying which constraint thresholds in which scenarios) are reasonable.

[0171] Step S154: For the part that does not meet the constraints, determine the type of reason for not meeting the constraints based on the source load prediction results and scheduling requirements. The type of reason includes source load changes exceeding the scheduling constraint range or inconsistencies between the associated transmission path parameters and the scheduling requirement parameters.

[0172] The unmet constraints are the key focus of collaborative optimization. Based on the specific manifestations of the unmet constraints in the source-load forecast results (such as excessive demand, insufficient supply, and parameter exceeding limits) and the specific requirements of the scheduling requirements, the types of reasons for the unmet constraints are analyzed and determined. Source-load changes exceeding the scheduling constraints mean that the predicted values ​​of energy demand or supply themselves exceed the constraint boundaries specified in the scheduling requirements, such as "the predicted maximum traction energy consumption during the morning peak exceeds the substation's maximum power supply capacity." Inconsistencies between the parameters of the associated transmission path and the parameters of the scheduling requirements mean that although the total source-load may be within the constraints, the parameters in the associated transmission process between data nodes do not meet the scheduling requirements, such as "the efficiency parameter of regenerative braking energy recovery is lower than the target value set in the scheduling requirements, resulting in the overall energy utilization rate not meeting the standard." Accurately identifying the type of cause is a prerequisite for formulating effective correction strategies.

[0173] Step S155: Based on the basic support criteria and cause type, generate an energy supply adjustment strategy. The energy supply adjustment strategy includes supply allocation optimization, supply timing adjustment, and energy storage resource allocation. The supply allocation optimization is consistent with the transmission path of related impacts.

[0174] Step S155: Based on the underlying support criteria and cause type, generate an energy supply adjustment strategy, including:

[0175] Step S1551: Analyze the data node association logic and scenario adaptation rules in the basic support basis, and extract the core data that can support energy supply adjustment. The core data includes the connection link parameters, scenario adaptation weights and data node attributes corresponding to the supply and demand balance period.

[0176] When analyzing the underlying support, focus on the data node association logic during the supply-demand balance period, such as the dependencies and operational patterns between different types of data node attributes, and the rules governing data interaction and association under different scenarios in the scenario adaptation rules. Extract core data for this period from the transportation and energy data association pool, including connection link parameters, such as transmission characteristics and efficiency-related parameters of the links; scenario adaptation weights, reflecting the importance of different links in the current scenario; and data node attributes, describing the characteristics and status of the data nodes themselves.

[0177] Step S1552: Based on the core data and transportation energy scheduling needs, generate energy supply adjustment targets, including supply structure optimization, supply efficiency improvement, and effective supply-demand matching.

[0178] By combining core data with transportation energy dispatching needs, such as requirements for the reliability, economy, and environmental friendliness of energy supply, multi-dimensional energy supply adjustment targets are generated. The supply structure optimization target aims to adjust the proportion and distribution relationship between different energy supply sources to achieve a more rational energy allocation; the supply efficiency improvement target focuses on the efficiency of energy production, transmission, and distribution processes, reducing energy loss and waste; and the effective supply-demand matching target aims to ensure that energy supply and demand are consistent in time and quantity, avoiding supply-demand imbalances.

[0179] Step S1553: For the type of reason why the source load change exceeds the scheduling constraint range, generate a total supply adjustment plan based on the time period of the excess, the corresponding connection link and data node. The total supply adjustment plan includes increasing the supply, decreasing the supply, or transferring the supply area.

[0180] When changes in energy sources exceed scheduling constraints, the specific time period in which the excess occurred, the affected connection links, and the relevant data nodes are identified. Based on this information, a total supply adjustment plan is formulated. Increasing supply can be achieved by activating backup energy supply equipment or increasing the output of existing equipment; decreasing supply may require reducing non-critical or adjustable energy consumption to keep total demand within constraints; regional supply transfer involves allocating energy from areas with surplus supply to areas with insufficient supply to achieve inter-regional energy balance.

[0181] Step S1554: For the cause type of inconsistency between the associated transmission path parameters and the scheduling demand parameters, a supply allocation optimization scheme is generated based on the comparison results between the transmission logic parameters and the scheduling demand parameters. The supply allocation optimization scheme includes the adjustment of the allocation ratio of supply resources in the regions or time periods corresponding to different connection links.

[0182] When the parameters of the associated transmission path are inconsistent with the scheduling demand parameters, the transmission logic parameters and scheduling demand parameters are compared to identify the discrepancies. Based on the comparison results, an optimized supply allocation scheme is generated, adjusting the allocation ratio of supply resources in different regions or time periods corresponding to different connection links. For example, for important connection links or critical time periods, the allocation ratio of energy supply can be appropriately increased to meet scheduling demands; for non-critical links or time periods, the allocation ratio can be appropriately decreased.

[0183] Step S1555: Based on the long-term evolution trend in the traffic energy source and load forecast results, a supply timing adjustment strategy is generated. The supply timing adjustment strategy includes a timing arrangement of advance arrangement of supply during peak periods, reasonable reduction during off-peak periods, and stable supply during off-peak periods.

[0184] By referencing the long-term evolution trends in transportation energy source and load forecasts, we can understand the changing patterns of energy demand at different times. Based on this, we can generate supply timing adjustment strategies: before peak supply periods arrive, we can prepare energy reserves and supply in advance to cope with peak demand; during off-peak periods, we can reasonably reduce energy supply to avoid energy waste; and during off-peak periods, we can maintain stable energy supply to ensure the smooth operation of the system.

[0185] Step S1556: Based on the short-term change trend and scenario adaptation deviation in the traffic energy source and load prediction results, generate an energy storage resource allocation plan. The energy storage resource allocation plan includes the charging and discharging sequence of energy storage devices, energy storage capacity allocation, and complementary allocation of energy storage resources among different connection links.

[0186] Based on short-term trends and scenario adaptation deviations in traffic energy source and load forecasts, an energy storage resource allocation plan is formulated. The charging and discharging sequence of energy storage devices is determined, charging occurs during periods of ample energy supply and low prices, and discharging occurs during periods of peak energy demand and tight supply, thus mitigating energy supply and demand fluctuations. Energy storage capacity is rationally allocated to ensure that energy storage needs in different regions or links are met. Simultaneously, complementary allocation of energy storage resources across different connection links is achieved to improve the utilization efficiency of energy storage resources and the overall reliability of the system.

[0187] Step S1557: Analyze the associated constraints in the energy supply adjustment process. The associated constraints include energy production capacity, transmission channel capacity, energy storage equipment capacity, and scenario operation limitations, so that the adjustment strategy operates within the constraints.

[0188] During the adjustment of energy supply, various related constraints are analyzed. Energy production capacity limits the maximum possible output of energy supply; the capacity of transmission channels determines the upper limit of energy transmission capacity between different regions; the capacity of energy storage devices limits the charging and discharging capacity and duration of energy storage resources; and scenario operation constraints include requirements for energy quality, power supply reliability, etc.

[0189] Step S1558: Integrate the total supply adjustment plan, the supply allocation optimization plan, the supply timing adjustment strategy, and the energy storage resource allocation plan to form a preliminary framework for the energy supply adjustment strategy. The preliminary framework includes the strategy objectives, core content, and execution logic.

[0190] By integrating the previously generated total supply adjustment plan, supply allocation optimization plan, supply timing adjustment strategy, and energy storage resource allocation plan, a preliminary framework for an energy supply adjustment strategy is formed. This framework clearly defines the strategy objectives, i.e., the desired effects to be achieved through adjustments; its core content includes the main measures and parameters of each specific plan; and its execution logic defines the steps, triggering conditions, responsible parties, and coordination mechanisms for strategy implementation. The formation of this preliminary framework makes the energy supply adjustment strategy more systematic and organized.

[0191] Step S1559: Refine each strategy content in the preliminary framework, determine the specific execution steps, operation parameters, responsible parties and collaboration requirements, wherein the operation parameters are generated based on core data and scenario constraint information.

[0192] Each strategy element in the initial framework is refined and broken down into specific execution steps, clearly defining the operational content and sequence of each step. Based on core data and scenario constraints, parameters for each operation are determined, such as the adjustment magnitude and time intervals. The responsible party for each execution step is assigned, and the collaboration requirements and communication mechanisms between different parties are clarified to ensure the smooth implementation of the strategy.

[0193] Step S15510: Confirm the compatibility between the energy supply adjustment strategy and the core elements of transportation energy dispatch demand through the demand matching verification logic, so that the energy supply adjustment strategy meets the energy supply constraints, supply and demand balance objectives and dispatch response time requirements, and form the final energy supply adjustment strategy.

[0194] Using demand-match verification logic, we examine whether the energy supply adjustment strategy aligns with the core elements of transportation energy dispatch demand. We verify whether the strategy meets energy supply constraints, such as maximum and minimum supply; whether it can achieve the supply-demand balance target, matching energy supply with demand; and whether it meets dispatch response time requirements, enabling adjustments to be made and effective within the specified timeframe. Through verification, we make necessary modifications and improvements to the strategy, ultimately forming an energy supply adjustment strategy that meets all requirements.

[0195] Step S156: Combine the correlation impact transmission path in the traffic energy source and load prediction results to generate a traffic operation adaptation strategy. The traffic operation adaptation strategy includes traffic flow control direction, path guidance logic, and facility usage optimization. The traffic flow control direction matches the source and load change trend.

[0196] Traffic operation adaptation strategies adjust demand-side behavior to adapt to energy supply constraints and source load trends. Traffic flow control is determined based on energy demand trends in source load forecasts. For example, if a future energy shortage is predicted for a certain area, passenger flow entering that area is controlled, guiding passengers to choose alternative routes or modes of transportation, thus aligning traffic flow changes with source load trends (such as declining demand). Route guidance logic optimizes train routes or passenger transfer routes to reduce traffic flow in high-energy-consuming sections, such as guiding trains to avoid energy-intensive uphill sections or guiding passengers to choose routes with fewer transfers and lower energy consumption. Facility utilization optimization adjusts the operation of station facilities, such as appropriately reducing the brightness of unnecessary lighting and adjusting air conditioning temperature settings during periods of energy shortage to reduce energy consumption of auxiliary systems.

[0197] Step S157: Based on the energy supply adjustment strategy and the traffic operation adaptation strategy, a supply and demand balance guarantee strategy is generated. The supply and demand balance guarantee strategy includes emergency response logic, resource complementarity mechanism, and dynamic monitoring rules. The emergency response logic corresponds to the scenario adaptation deviation.

[0198] The supply and demand balance guarantee strategy is a supportive strategy to ensure the effective implementation of energy supply adjustment strategies and traffic operation adaptation strategies, and to respond to emergencies. The emergency response logic is formulated to address potential risks and unmet constraints identified in the scenario adaptation deviation list. For example, when a scenario adaptation deviation occurs (e.g., "energy storage device state of charge is too low"), the emergency response logic activates backup power or reduces non-critical loads. The resource complementarity mechanism establishes complementary and mutually supportive relationships between different energy resources and different transportation lines. For example, when a substation experiences a power shortage, power is transferred from other substations through inter-regional energy transfer lines; when train energy consumption on a certain line is too high, backup trains are drawn from other lines to replace high-energy-consuming trains. The dynamic monitoring rules specify the monitoring content, monitoring frequency, and data reporting requirements for energy supply and demand status, traffic operation status, and strategy implementation effectiveness. For example, real-time monitoring of the power supply of key substations (once per second) and minute-level monitoring of train operation energy consumption ensure that problems can be detected and strategies adjusted in a timely manner.

[0199] Step S158: Integrate energy supply adjustment strategies, traffic operation adaptation strategies, and supply and demand balance guarantee strategies to form a traffic energy source and load coordinated optimization scheme. The traffic energy source and load coordinated optimization scheme includes specific execution logic and operation sequence.

[0200] The coordinated optimization scheme for transportation energy sources and loads is an organic integration of energy supply adjustment strategies, transportation operation adaptation strategies, and supply-demand balance guarantee strategies. The specific execution logic details the implementation steps, responsible parties, required resources, and technical means for each measure within each strategy. For example, the execution logic for the measure "adjusting the power distribution of substation A" in the energy supply adjustment strategy is: "The dispatch center sends a power adjustment command to substation A → the control system of substation A adjusts the tap changer of the main transformer → real-time monitoring of output power → feedback of adjustment results to the dispatch center." The operational sequence specifies the time order and time nodes for the implementation of each measure, such as "completing the charging capacity of energy storage device B to 80% within the next 30 minutes" and "implementing a new timetable (train interval adjustment) starting from the next train."

[0201] Step S159: Feed back the strategy parameters in the traffic energy source-load collaborative optimization scheme to the dynamic source-load association network through the data interaction channel. The strategy parameters include the adaptation weight adjustment value and the association transmission logic optimization direction.

[0202] After the transportation energy source-load coordinated optimization scheme is formulated, its core strategy parameters need to be fed back to the dynamic source-load association network to achieve continuous network optimization. The adaptation weight adjustment value in the strategy parameters is determined based on the reassessment of the importance of connection links in the optimization scheme. For example, to prioritize train traction power supply, the scenario adaptation weight of the train-substation link is adjusted from 0.7 to 0.9. The optimization direction of the association transmission logic is determined according to the improvement suggestions for node interaction rules in the optimization scheme. For example, to improve the recovery efficiency of regenerative braking energy, the energy feedback logic of the energy storage device-train link is optimized to make it respond more quickly to the braking behavior of the train. Through the data interaction channel, the above strategy parameters are transmitted to the parameter configuration module of the dynamic source-load association network to update the network's link weights and transmission rules.

[0203] Step S1510: After receiving the policy parameters, the dynamic source-load association network optimizes the scene adaptation weight of the corresponding connection link through the scene adaptation weight dynamic adjustment mechanism, and realizes the self-optimization of the dynamic source-load association network through the optimization of scene adaptation weight.

[0204] After receiving the policy parameters, the dynamic source-load association network initiates a dynamic adjustment mechanism for scenario adaptation weights. This mechanism precisely adjusts the scenario adaptation weights of corresponding connection links based on the received adaptation weight adjustment values ​​and the network's current operating status (such as the scenario matching degree of data nodes and the transmission efficiency of connection links). For example, applying "train-substation link adaptation weight adjustment value + 0.2" from the policy parameters to the current weight of 0.7 results in a new weight of 0.9. Through this method, the dynamic source-load association network can absorb the results of collaborative optimization schemes, update its own network structure and parameters, and achieve network self-optimization. The optimized network, in the next source-load prediction and collaborative optimization, can more accurately reflect the actual operating conditions and optimization objectives of the system, forming a closed-loop mechanism for continuous improvement.

[0205] Based on the same inventive concept, please refer to Figure 2 The diagram shows a schematic block diagram of a data fusion-based traffic energy source load prediction and collaborative optimization system 100 provided in this application embodiment for executing the above-described data fusion-based traffic energy source load prediction and collaborative optimization method. The data fusion-based traffic energy source load prediction and collaborative optimization system 100 may include a communication unit 110, a machine-readable storage medium 120, and a processor 130.

[0206] In this embodiment, both the machine-readable storage medium 120 and the processor 130 are located in the data fusion-based traffic energy source and load prediction and collaborative optimization system 100, and are separately configured. Alternatively, the machine-readable storage medium 120 can also be integrated into the processor 130 and can communicate and interact with external systems through the communication unit 110. The machine-readable storage medium 120 is used to store machine-executable instructions for executing the scheme of this application, and the processor 130 is used to execute the machine-executable instructions stored in the machine-readable storage medium 120 to implement the data fusion-based traffic energy source and load prediction and collaborative optimization method provided in the aforementioned method embodiments.

[0207] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A method for predicting and co-optimizing traffic energy sources and loads based on data fusion, characterized in that, The method includes: Multi-source traffic and energy data, consisting of traffic operation data, energy supply data, and environmental correlation data, are accessed to generate a traffic and energy data association pool. The traffic and energy data association pool includes attribute association information, scenario adaptation identifiers, and data interaction channels for various types of data. Based on the attribute association information and scenario adaptation identifier of the traffic energy data association pool, a dynamic source-load association network is generated. The dynamic source-load association network takes data nodes as the core and source-load association relationships as the connection links. The connection links include association transmission logic and scenario adaptation weights. Different traffic scenario features are mapped to the dynamic source-load association network, and the data node set and connection link association logic for scenario adaptation are extracted to generate a multi-dimensional fusion inference input set. The multi-dimensional fusion inference input set includes the core data nodes of the scenario, connection link parameters and scenario adaptation constraints. By using a multi-dimensional fusion inference model adapted to the scenario, the source load change trend of the multi-dimensional fusion inference input set is inferred to obtain the traffic energy source load prediction result. The traffic energy source load prediction result includes the dynamic curve of supply and demand change, the transmission path of related influences and the scenario adaptation deviation. Based on the traffic energy source and load prediction results and traffic energy scheduling requirements, a traffic energy source and load collaborative optimization scheme is formed. The traffic energy source and load collaborative optimization scheme is fed back to the dynamic source and load association network through the data interaction channel to realize the scenario adaptation weight optimization of the dynamic source and load association network.

2. The method for predicting and coordinating traffic energy sources and loads based on data fusion according to claim 1, characterized in that, The generation of a dynamic source-load association network based on the attribute association information and scenario adaptation identifiers of the traffic energy data association pool includes: Extract traffic operation data attribute association information, energy supply data attribute association information, and environmental data attribute association information from the traffic energy data association pool to form a multi-source data attribute association set. Each association information in the multi-source data attribute association set corresponds to a unique data node identifier and a scenario adaptation identifier. The direct correlation between traffic operation data nodes and energy supply data nodes is analyzed. The direct correlation is generated based on the interdependence logic of data attributes and includes data interaction triggering conditions and correlation transmission direction. By combining the indirect influence logic of environmental data nodes on traffic operation data nodes and energy supply data nodes, the correlation dimension in the direct correlation is supplemented to form a complete source-load correlation set, which includes direct correlation and indirect correlation. Using data nodes as the core carrier, the relationships in the full set of source-load relationships are transformed into connection links between data nodes. Each connection link corresponds to a set of relationships, and the connection link embeds the relationship transmission logic and the initial scenario adaptation weight. The initial architecture of the dynamic source-load association network is generated. The initial architecture includes all data nodes, connection links, association transmission logic and initial scene adaptation weights. The data nodes include complete attribute information and scene adaptation identifiers. An embedded scene adaptation weight dynamic adjustment mechanism is used, which generates adjustment logic based on the scene matching degree of data nodes and the transmission efficiency of connection links. By using historical scenario data from multi-source transportation energy data, the scenario adaptation weight dynamic adjustment mechanism is driven to optimize the initial scenario adaptation weight and form a dynamically adjusted connection link. Based on the dynamically adjusted connection links, the association transmission logic between data nodes is optimized to ensure that the association transmission logic is consistent with the scene adaptation weight, thereby improving the scene adaptation accuracy of the association relationship. The integrated and optimized data nodes, connection links, associated transmission logic, and dynamically adjusted scenario adaptation weights form the dynamic source-load association network. The data interaction channel enables real-time data synchronization between the dynamic source-load association network and the transportation energy data association pool. The real-time data synchronization mechanism ensures that the attribute information of data nodes in the dynamic source-load association network is consistent with the attribute information of data nodes in the transportation energy data association pool.

3. The method for predicting and coordinating traffic energy sources and loads based on data fusion according to claim 1, characterized in that, The process of mapping different traffic scenario features to the dynamic source-load association network, extracting the data node set and connection link association logic for scenario adaptation, and generating a multi-dimensional fusion inference input set includes: Extract various traffic scene features from the traffic scene feature library. Each traffic scene feature includes scene operation parameters, environmental constraints, energy demand features, and traffic flow features. The scene operation parameters correspond to the data node attribute information dimensions in the dynamic source-load association network. Each traffic scenario feature is broken down into multiple core scenario elements. The core scenario elements include key scenario parameters, constraint boundary conditions, and demand priority information. The key scenario parameters are matched with data attribute association information in the traffic energy data association pool. Traverse all data nodes in the dynamic source-load association network, and based on the key parameters of the scene in the core elements of the scene, filter data nodes that are compatible with the scene features to form a set of candidate data nodes for the scene. The scenario adaptation identifier of each data node in the scenario candidate data node set is analyzed, and the constraint boundary conditions in the core elements of the scenario are combined to further filter the scenario candidate data node set to obtain the scenario core data node set. Extract all connection links between the core data node set of the scene in the dynamic source-load association network, obtain the association transmission logic and scene adaptation weight of each connection link, and form a scene association connection link set; Based on the demand priority information in the core elements of the scenario, calculate the new scenario adaptation weight for each connection link in the scenario-related connection link set, and assign the new scenario adaptation weight to the corresponding connection link. The core data node set of the scene, the adjusted scene association connection link set, and the core elements of the scene are integrated to form the intermediate result of scene mapping. The intermediate result of scene mapping includes node attributes, connection link parameters, and scene constraint information. Based on the connection link parameters in the intermediate results of scene mapping, the key path of the associated transmission logic is extracted. The key path is the connection link transmission path with the highest association strength between the core data nodes of the scene. Supplement the data interaction rules in the critical path. The data interaction rules are generated based on the characteristics of the data interaction channels in the dynamic source-load association network, including data transmission format, interaction timing and exception handling logic; The core data node set of the scenario, key paths, related transmission logic, adjusted scenario adaptation weights and data interaction rules are integrated to form the multi-dimensional fusion inference input set.

4. The method for predicting and coordinating traffic energy sources and loads based on data fusion according to claim 1, characterized in that, The multi-dimensional fusion inference model adapted to the scenario extrapolates the source load change trend of the multi-dimensional fusion inference input set, and obtains the traffic energy source load prediction results, including: The set of core data nodes, key paths, and related transmission logic of the scene from the multi-dimensional fusion inference input set are input into the node parsing module of the multi-dimensional fusion inference model to extract the dynamic change features and related response features of each core data node in the scene. The dynamic change features reflect the evolution law of the data node over time, and the related response features reflect the feedback logic of the data node to the connection link. The connection link strengthening module of the multi-dimensional fusion inference model strengthens the correlation between dynamic change features and associated response features based on the scenario adaptation weight in the critical path, and generates a set of associated strengthening features. Each feature in the set of associated strengthening features contains dual information of node attributes and connection link association. The association enhancement feature set is input into the scene adaptation module of the multi-dimensional fusion inference model. Combined with the scene constraint information in the multi-dimensional fusion inference input set, the dimensional weights of the association enhancement feature set are adjusted to generate the scene adaptation feature set. The dimensional weights are consistent with the priority of requirements in the scene constraint information. The trend inference module of the multi-dimensional fusion inference model infers the short-term change pattern and long-term evolution trend of transportation energy source load based on the dynamic change features in the scenario adaptation feature set, and generates preliminary trend inference results. The preliminary trend inference results include the predicted trend of supply and demand changes and key change nodes. By combining the associated response features in the scene adaptation feature set, we can analyze the impact of connection links on the preliminary trend projection results and identify the scene adaptation deviations in the preliminary trend projection results. Based on the associated transmission logic and scenario adaptation weight, a deviation correction strategy is generated. The deviation correction strategy includes a correction direction and correction logic, and the correction logic is consistent with the transmission characteristics of the connection link. The deviation correction strategy is applied to adjust the preliminary trend projection results to obtain the corrected trend projection results, which eliminate the discrepancies between the projection results and the scenario constraint information. Extract key change nodes from the corrected trend projection results, analyze the associated transmission paths and scenario constraints corresponding to the key change nodes, and generate associated influence transmission paths. The traffic energy source and load prediction results are formed by integrating and correcting the trend projection results, the transmission paths of related influences, and the scenario adaptation deviations.

5. The method for predicting and coordinating traffic energy sources and loads based on data fusion according to claim 1, characterized in that, Based on the traffic energy source and load prediction results and traffic energy scheduling requirements, a traffic energy source and load collaborative optimization scheme is formed. This scheme is fed back to the dynamic source and load association network through a data interaction channel, enabling scenario adaptation weight optimization of the dynamic source and load association network, including: Extract energy supply constraints, traffic operation demands, supply-demand balance targets, and dispatch response time from traffic energy dispatch demand to form core elements of dispatch demand. These core elements of dispatch demand correspond to the dimensions of traffic energy source and load forecast results. The revised trend projection results, the transmission path of related impacts, and the scenario adaptation deviation in the traffic energy source and load forecast results are analyzed. Based on the constraints of the core elements of scheduling demand, the analysis results are divided into the part that meets the constraints and the part that does not meet the constraints. For the parts that meet the constraints, the corresponding related influence transmission paths and scenario constraints are extracted as the basic support basis for collaborative optimization. The basic support basis includes data node association logic and scenario adaptation rules. For the parts that do not meet the constraints, the cause of the failure to meet the constraints is determined based on the source load prediction results and scheduling requirements. The cause types include source load changes exceeding the scheduling constraint range or inconsistencies between the associated transmission path parameters and the scheduling requirement parameters. Based on the fundamental support and cause types, an energy supply adjustment strategy is generated. The energy supply adjustment strategy includes supply allocation optimization, supply timing adjustment, and energy storage resource allocation. The supply allocation optimization is consistent with the transmission path of related impacts. By combining the correlation and transmission paths of traffic energy source and load prediction results, a traffic operation adaptation strategy is generated. The traffic operation adaptation strategy includes traffic flow control direction, path guidance logic, and facility usage optimization. The traffic flow control direction matches the source and load change trend. Based on energy supply adjustment strategies and traffic operation adaptation strategies, a supply and demand balance guarantee strategy is generated. The supply and demand balance guarantee strategy includes emergency response logic, resource complementarity mechanism, and dynamic monitoring rules. The emergency response logic corresponds to the scenario adaptation deviation. By integrating energy supply adjustment strategies, traffic operation adaptation strategies, and supply and demand balance guarantee strategies, a traffic energy source and load coordinated optimization scheme is formed, which includes specific execution logic and operation sequence. The strategy parameters in the traffic energy source-load collaborative optimization scheme are fed back to the dynamic source-load association network through the data interaction channel. The strategy parameters include the adaptation weight adjustment value and the association transmission logic optimization direction. After receiving the policy parameters, the dynamic source-load association network optimizes the scene adaptation weights of the corresponding connection links through a scene adaptation weight dynamic adjustment mechanism, thereby achieving self-optimization of the dynamic source-load association network through the optimization of scene adaptation weights.

6. The method for predicting and coordinating traffic energy sources and loads based on data fusion according to claim 2, characterized in that, The analysis of the direct correlation between traffic operation data nodes and energy supply data nodes includes: Extract the attribute information of traffic operation data nodes from the traffic energy data association pool. The attribute information of traffic operation data nodes includes data dimensions such as traffic flow attributes, travel route attributes, facility usage attributes, and operation status attributes. Extract the attribute information of energy supply data nodes in the transportation energy data association pool. The attribute information of energy supply data nodes includes data dimensions such as total supply attribute, transmission efficiency attribute, storage status attribute, and supply type attribute. The attribute information of the extracted traffic operation data nodes and the attribute information of the energy supply data nodes are standardized. The attribute information of the extracted traffic operation data nodes and the attribute information of the energy supply data nodes are compared after standardization. The attribute dimensions that are interdependent between the two are identified. The interdependent attribute dimensions are those in which changes in one data will cause adjustments in the other data. Based on the standardized interdependent attribute dimensions, the transmission logic of data changes is analyzed. The transmission logic includes the triggering conditions of data changes, the transmission sequence, and the correlation of the degree of impact. The triggering conditions are generated based on the threshold of the standardized value change of the attribute data. Determine the triggering conditions for data interaction. The triggering conditions for data interaction are the triggering rules when the interdependent attribute dimensions reach a preset change state. The triggering rules are consistent with the change characteristics of the attribute data. Based on the transmission sequence in the transmission logic, the direction of association transmission is determined. The direction of association transmission is the flow of data change from one data node to another, and the flow corresponds to the attribute dependency logic. By combining the data interaction triggering conditions and the direction of association transmission, a single direct association relationship is formed. A single direct association relationship includes the data node identifiers of the two parties involved, the data interaction triggering conditions, the direction of association transmission, and the attribute dependency logic. Traverse all traffic operation data nodes and energy supply data nodes. For each pair of traffic operation data nodes and energy supply data nodes, repeatedly perform the following steps: extract attribute information, perform standardization processing, compare and identify interdependent attribute dimensions, analyze the transmission logic of data changes, determine the triggering conditions for data interaction, determine the direction of correlation transmission, and form a single direct correlation relationship. Generate all direct correlation relationships between pairs of nodes and form a set of direct correlation relationships. Analyze the attribute dependency logic of each relationship in the set of direct relationships, merge logically consistent relationships, eliminate duplicate relationships, and optimize the structure of the set of direct relationships; Supplement the set of direct relationships with association description information, which includes attribute dependency details and data interaction characteristics of the relationships, to form the set of direct relationships.

7. The method for predicting and coordinating traffic energy sources and loads based on data fusion according to claim 3, characterized in that, The step of adjusting the scene adaptation weight of each connection link in the scene-related connection link set based on the demand priority information in the core elements of the scene, so that the scene adaptation weight of the connection link matches the demand priority information, includes: Extract the requirement priority information from the core elements of the scenario. The requirement priority information includes the priority ranking of multiple requirement dimensions and the weight ratio of each priority. The requirement dimensions correspond to the association characteristics of the connection link between the scenario and the scenario. The association transmission logic of each connection link in the set of scenario-related connection links is analyzed, and the corresponding requirement dimension of each connection link is identified. The matching between the association transmission logic and the requirement dimension is determined based on the association attributes of the connection link. Match the requirement dimension corresponding to each connection link with the requirement dimension in the requirement priority information to obtain the corresponding priority ranking and weight ratio; Based on priority ranking, an initial adjustment weight is assigned to each connection link in the set of scene-related connection links. The initial adjustment weight is positively correlated with the priority ranking, with higher priority resulting in a larger initial adjustment weight. The initial weights are adjusted based on the weight ratios, and the adjusted weight ratios are consistent with the weight ratios in the requirement priority information, so that the adjusted connection link scenario adaptation weights conform to the requirement priority distribution. The basic value of the association strength of each connection link is analyzed. The basic value of the association strength is the initial scenario adaptation weight of the connection link in the dynamic source-load association network. The corrected weight is calibrated in combination with the basic value of the association strength. The corrected weights are fused with the basic value of the association strength through the weight calibration logic to generate the final connection link scenario adaptation weights. The weight calibration logic is generated based on the transmission efficiency of the connection link and the scenario adaptation accuracy. The final connection link scenario adaptation weight is updated to the corresponding connection link in the scenario-associated connection link set, replacing the original scenario adaptation weight, thereby realizing the dynamic adjustment of connection link weight; The application verification logic verifies the scenario adaptation weights of all connection links in the adjusted scenario-related connection link set, and records the weight adjustment parameters, as well as all parameters and logic in the weight adjustment process, including the requirement dimension matching results, initial adjustment weights, correction basis, and calibration logic, forming a weight adjustment record as supplementary information to the multi-dimensional fusion inference input set.

8. The method for predicting and coordinating traffic energy sources and loads based on data fusion according to claim 4, characterized in that, The method of combining the associated response features in the scenario adaptation feature set to analyze the impact of connection links on the preliminary trend projection results and identify scenario adaptation deviations in the preliminary trend projection results includes: Extract the associated response features from the scene adaptation feature set. The associated response features include the response pattern, response time and response magnitude of each scene's core data node to changes in the connection link. The response pattern is generated based on historical data interaction records. Analyze the supply and demand changes and key change nodes in the preliminary trend projection results, and determine the set of connection links corresponding to the preliminary trend projection results. The set of connection links is the core transmission path that supports the projection results. The associated response features are matched with the corresponding set of connection links, and the impact of changes in each connection link on the response features of the core data nodes in the scenario is analyzed. The impact analysis is based on the associated transmission logic and response patterns. Based on the correlation transmission logic and the impact results, a correlation impact transmission path is generated from the change in the connection link through the response of the data node to the adjustment of the inference results; Extract scenario constraint information from the multi-dimensional fusion and inference input set. The scenario constraint information includes the boundary conditions, demand standards and adaptation requirements of the scenario operation. The scenario constraint information serves as the basis for judging the adaptation deviation. By comparing the preliminary trend projection results with the scenario constraint information, the parts of the projection results that exceed the scenario constraint boundaries, do not meet the demand standards, or do not meet the adaptation requirements are identified as potential scenario adaptation deviations. By combining the impact transmission path of connection links on the simulation results, potential scenario adaptation deviations are associated with specific connection links, associated transmission logic, or scenario adaptation weights. Based on the attribute information and connection link parameters of the core data nodes in the scenario, potential scenario adaptation deviations are classified, and the classification criteria include data source, logical rules or scenario constraints. The specific manifestations, scope of impact, and causes of each scenario adaptation deviation are determined, and a scenario adaptation deviation list is formed. The structure of the scenario adaptation deviation list is consistent with that of the preliminary trend projection results. The proposed correction directions are supplemented to the scenario adaptation deviation list. These correction directions are generated based on associated response features and associated transmission logic.

9. The method for predicting and coordinating traffic energy sources and loads based on data fusion according to claim 5, characterized in that, The energy supply adjustment strategy generated based on the underlying support criteria and cause types includes: The data node association logic and scenario adaptation rules in the basic support are analyzed to extract the core data that can support energy supply adjustment. The core data includes the connection link parameters, scenario adaptation weights and data node attributes corresponding to the supply and demand balance period. Based on the core data and transportation energy scheduling needs, energy supply adjustment targets are generated, including supply structure optimization, supply efficiency improvement, and effective supply-demand matching. For the type of reason why the source load change exceeds the scheduling constraint range, a total supply adjustment plan is generated based on the time period of the excess, the corresponding connection link and data node. The total supply adjustment plan includes increasing the supply, decreasing the supply, or transferring the supply area. For the reasons why the transmission path parameters and scheduling requirement parameters are inconsistent, a supply allocation optimization scheme is generated based on the comparison results between the transmission logic parameters and the scheduling requirement parameters. The supply allocation optimization scheme includes the adjustment of the allocation ratio of supply resources in the regions or time periods corresponding to different connection links. Based on the long-term evolution trend in the traffic energy source and load forecast results, a supply timing adjustment strategy is generated. The supply timing adjustment strategy includes the timing arrangement of advance arrangement of supply during peak periods, reasonable reduction during off-peak periods, and stable supply during off-peak periods. Based on the short-term changing trends and scenario adaptation deviations in the traffic energy source and load forecast results, an energy storage resource allocation scheme is generated. The energy storage resource allocation scheme includes the charging and discharging sequence of energy storage devices, energy storage capacity allocation, and complementary allocation of energy storage resources among different connection links. The analysis of the associated constraints in the energy supply adjustment process includes energy production capacity, transmission channel capacity, energy storage equipment capacity, and scenario operation limitations, so that the adjustment strategy operates within the constraints. By integrating the total supply adjustment plan, the supply allocation optimization plan, the supply timing adjustment strategy, and the energy storage resource allocation plan, a preliminary framework for the energy supply adjustment strategy is formed. The preliminary framework includes the strategy objectives, core content, and execution logic. Each strategy in the preliminary framework is refined to determine the specific execution steps, operation parameters, responsible parties, and collaboration requirements. The operation parameters are generated based on core data and scenario constraint information. The energy supply adjustment strategy is confirmed to be compatible with the core elements of transportation energy dispatch demand by the demand matching verification logic, so that the energy supply adjustment strategy meets the energy supply constraints, supply and demand balance objectives and dispatch response time requirements, thus forming the final energy supply adjustment strategy.

10. A traffic energy source and load prediction and collaborative optimization system based on data fusion, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the data fusion-based traffic energy source and load prediction and collaborative optimization method according to any one of claims 1 to 9 by executing the machine-executable instructions.

Images