Partitioned power carbon footprint dataset dynamic updating method and system based on power grid topology propagation model

Abstract

The application relates to the field of power system data processing, and discloses a partition power carbon footprint dataset dynamic updating method and system based on a power grid topology propagation model. A partition power grid topology graph is constructed according to time slices, and time-sharing power carbon footprint factors of nodes in each region are calculated; when power grid operation data changes, a disturbance event containing a disturbance source node and a change amplitude is generated, an influence score is calculated according to a weighted adjacency matrix and a propagation attenuation coefficient, nodes and time slices with an influence score not less than an updating threshold are written into an incremental recalculation queue and are recalculated only, a new version of the disturbance event and a propagation path is written, the un-affected nodes use a parent version, and a product calling object is generated for recalculation or archiving along with a blood relationship. The scheme helps to reduce the recalculation range and improve version traceability.

Description

Technical Field

[0001] This application relates to the field of power system data processing and product carbon footprint accounting technology, and in particular to a dynamic update technology for partitioned power carbon footprint datasets based on a power grid topology propagation model. Background Technology

[0002] Product carbon footprint accounting and life cycle assessment are increasingly widely used in export-oriented manufacturing, green electricity trading, and carbon emission disclosure. Taking companies in industries such as electrolytic aluminum, data centers, power batteries, and steel as examples, their electricity consumption periods vary significantly, their cross-regional power reception ratio is high, and their renewable energy output fluctuates considerably. For instance, manufacturing companies with a high proportion of cross-provincial power reception may have different power source structures in different production shifts. When wind and solar power output fluctuates significantly, the industrial park-level product database needs to differentiate the power supply structure for different time slots. After high-energy-consuming processes such as aluminum processing and electronics manufacturing generate product batch datasets, subsequent tie-line power flow corrections, generator output supplementation, or regional boundary adjustments may require determining whether published data has been affected. In these scenarios, product carbon footprint accounting typically requires multiplying the company's electricity consumption activity data with the corresponding regional and time-slot electricity carbon emission factors, and using versioned time-of-use and regional electricity carbon footprint factors as the repeatedly accessible base dataset for the product LCA background database.

[0003] However, existing technologies still have many shortcomings in the aforementioned scenarios. Annual or monthly average electricity emission factors are insufficient to reflect the differences in carbon intensity within the same region under different grid operating conditions. While some methods for calculating electricity carbon emission flows or node carbon intensity can calculate carbon intensity under a specific grid condition, they often remain at the level of one-time calculations or scheduling analysis, making it difficult to directly generate versioned background datasets for product LCA calls. Existing database update methods often struggle to distinguish which region-time slice factors are affected by disturbance propagation after changes in grid operating data, leading to problems such as excessive recalculation scope, duplicate copying of unaffected records, and difficulty in locating product batches that have already called older versions of factors. Especially when cross-regional tie-line power flow, topology interruptions, power generation output supplementation, and data quality status change simultaneously, traditional methods lack a mechanism to uniformly link the grid physical propagation path, factor version changes, and product call lineage. Therefore, a technical solution is urgently needed that can identify affected region-time slices based on grid topology propagation relationships and perform traceable incremental updates to the regional electricity carbon footprint dataset to address these issues. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for dynamically updating a partitioned power carbon footprint dataset based on a power grid topology propagation model, in order to solve the problems of difficulty in determining the affected region-time slice after changes in power grid operation data, excessive recalculation range of the partitioned power carbon footprint dataset, duplicate copying of unaffected records, and difficulty in tracing and locating product calling objects that have called old version factors.

[0005] This application provides a method for dynamically updating a partitioned electricity carbon footprint dataset based on a power grid topology propagation model, including: The power grid operation data is acquired and grouped into time slices according to a preset time granularity. For each time slice, a partitioned power grid topology map is constructed with power grid nodes or regions as vertices and transmission lines or inter-regional exchange relationships as directed edges. Based on the partitioned power grid topology map and the power grid operation data, the time-sharing electricity carbon footprint factor of each regional node in the corresponding time slice is calculated. When the power grid operation data changes, a disturbance event is generated that includes the disturbance source node and the magnitude of the change; For the time slice affected by the disturbance event, a weighted adjacency matrix is ​​generated based on the power flow direction and power consumption ratio of the partitioned power grid topology. Based on the weighted adjacency matrix and the propagation attenuation coefficient, the propagation result of the disturbance source node on each candidate node is calculated. The influence score of each candidate node is calculated by combining the change magnitude and the influence propagation result, and candidate nodes and time slices with influence scores not less than the update threshold are written into the incremental recalculation queue. Only for candidate nodes and time slices in the incremental recalculation queue, the time-sharing electricity carbon footprint factor is recalculated in the same way as the time-sharing electricity carbon footprint factor. The new version is written and the version lineage record of the parent version, the disturbance event, the propagation path, and the impact score is bound. For candidate nodes and time slices not written to the incremental recalculation queue, the original version is used and the inheritance relationship is recorded. Based on the call lineage query, the product call object that has called the time-of-use electricity carbon footprint factor before the update is called, calculates the product carbon footprint difference based on the electricity consumption of the product call object and the time-of-use electricity carbon footprint factor before and after the update, and generates a recalculation task or archive record of the product call object based on the product carbon footprint difference.

[0006] In a preferred embodiment, generating a weighted adjacency matrix based on the power flow direction and energy percentage of the partitioned power grid topology includes: Generate time slice Weighted adjacency matrix Its elements Calculate using the following formula: In the formula, For time slices internal nodes To the node The input power is at the node The proportion of the total received power, i.e., the elements of the weighted adjacency matrix; For time slices internal nodes To node Input power; For nodes In time slice The sum of all received input charges, of which To the node Input the upstream node index of the power level; To avoid extremely small positive values ​​with a denominator of zero; When the direction of the trend is uncertain, candidate directions are generated based on the preceding and following time slices, the transaction plan, and the net exchange direction of the tie line, and the direction confidence is written.

[0007] In a preferred embodiment, calculating the propagation result of the influence of the disturbance source node on each candidate node based on the weighted adjacency matrix and the propagation attenuation coefficient includes: The maximum order calculated by time slice is Influence propagation matrix As a result of the aforementioned impact propagation: In the formula, For time slices The influence propagation matrix; It is the identity matrix; The propagation attenuation coefficient is mentioned above; For time slices The weighted adjacency matrix; For the weighted adjacency matrix Power; The maximum propagation order; When satisfied When the propagation calculation stops, The preset cutoff threshold, This is a function that takes the maximum value of the matrix elements. In calculating each order During the process, the node sequence and topological edges that the disturbance source node passes through to each candidate node are extracted as the propagation path.

[0008] In a preferred embodiment, the influence score of each candidate node is calculated by combining the magnitude of the change and the result of the influence propagation, using the following formula: In the formula, Candidate nodes In time slice The impact score; For the disturbance source node In time slice The range of change; The influence propagation matrix The middle is composed of disturbance source nodes to candidate node The propagation coefficient; Candidate nodes The coverage weight is used to reflect whether the node is supplying power to external sources or being accessed by the product database. This is the quality weight for the disturbance source, used to reflect the data source and completeness of the disturbance source; When the same candidate node and time slice are affected by multiple disturbance events, its recalculation priority in the incremental recalculation queue is determined by the following formula: In the formula, Priority for recalculation; As a product hit factor, it is used when the candidate node and time slice match the published product call lineage. Otherwise take .

[0009] In a preferred embodiment, deduplication is also included: For the aforementioned disturbance event, an event deduplication key is generated based on the disturbance source node, source field, affected time window, and model version. Disturbance data with the same event deduplication key are merged without generating independent disturbance events repeatedly. When writing candidate nodes and time slices into the incremental recalculation queue, a queue deduplication key is generated based on the identifier of the disturbance event, the identifier of the regional node, the identifier of the time slice, and the model version. When there are incomplete recalculation tasks with the same queue deduplication key, the corresponding impact score and propagation path are merged in the incremental recalculation queue without repeated recalculation.

[0010] In a preferred embodiment, the new version also records the solution status and the computational model version; The step of using the original version and recording the inheritance relationship for candidate nodes and time slices that are not written to the incremental recalculation queue includes: not copying the full factor data for the candidate node and time slice, but writing the inherited parent version identifier, target dataset version identifier and inheritance reason into the version inheritance index. The reasons for inheritance include at least one of the following: the impact score is lower than the update threshold, the call lineage is not hit, the impact propagation path is not covered, or the time slice is frozen.

[0011] In a preferred embodiment, abnormal operating condition handling is also included: When Time Slice When the power flow of the connecting line is missing, candidate power flow values ​​are generated based on the cross-regional transaction plan and power flow data of adjacent time slices. : In the formula, For nodes To node In time slice Candidate power flow values; The current flow value given for cross-regional transaction plans; , These are the actual power flow values ​​for the previous and next time slices, respectively. The interpolation weights are determined based on the confidence of the transaction plan and the completeness of adjacent time slice data; topological edges with the candidate power flow values ​​are used as candidate edges, and their direction confidence and distribution confidence are reduced; When there are closed-loop mutual supply nodes in the weighted adjacency matrix, strongly connected components are identified and the time-sharing electricity carbon footprint factor is iteratively solved within the strongly connected components. When the number of iterations reaches the maximum number of iterations and the factor difference between two iterations is still not less than the convergence threshold, it is marked as a low confidence state, the generated time-sharing electricity carbon footprint factor is written into the isolation area and a review task is triggered, without directly overwriting the published valid version.

[0012] In a preferred embodiment, the step of calculating the product carbon footprint difference based on the electricity consumption of the product call object and the time-of-use electricity carbon footprint factor before and after the update includes: Calculate the product carbon footprint difference using the following formula : In the formula, For product call object Differences in product carbon footprint; For product call object In the affected time slice Electricity consumption; For the corresponding area In time slice Updated time-of-use electricity carbon footprint factor; For the corresponding area In time slice The time-of-use electricity carbon footprint factor before the update; when When the product call object's dataset status is set to require recalculation, the recalculation task is generated; when the ratio is less than the product difference threshold, it is set to low-impact archiving, and the archived record is generated; where... The original total electricity carbon footprint of the product being invoked. The preset product difference threshold; when When the value is zero or less than a preset minimum positive value, the relative difference judgment is not performed, but rather based on... Whether the absolute difference threshold is reached determines whether the recalculation task, the archived record, or the review task is generated.

[0013] This application provides a dynamic update system for partitioned power carbon footprint datasets based on a power grid topology propagation model, including modules corresponding to the methods described above: The data access and factor calculation module is used to acquire power grid operation data and group it into time slices according to a preset time granularity. For each time slice, it constructs a partitioned power grid topology map with power grid nodes or regions as vertices and transmission lines or inter-regional exchange relationships as directed edges. Based on the partitioned power grid topology map and the power grid operation data, it calculates the time-sharing electricity carbon footprint factor of each regional node in the corresponding time slice. A disturbance identification module is used to generate a disturbance event containing the disturbance source node and the magnitude of the change when the power grid operation data changes; The influence propagation module is used to generate a weighted adjacency matrix based on the power flow direction and power consumption ratio of the partitioned power grid topology for the time slice affected by the disturbance event, and to calculate the influence propagation result of the disturbance source node on each candidate node based on the weighted adjacency matrix and the propagation attenuation coefficient. The influence score and queue module is used to calculate the influence score of each candidate node by combining the change magnitude and the influence propagation result, and write the candidate nodes and time slices with influence scores not less than the update threshold into the incremental recalculation queue. The incremental recalculation and lineage writing module is used to recalculate the time-sharing electricity carbon footprint factor only for candidate nodes and time slices in the incremental recalculation queue, write the new version and bind the version lineage record of the parent version, the disturbance event, the propagation path and the impact score, and retain the original version for candidate nodes and time slices not written to the incremental recalculation queue and record the inheritance relationship. The product call linkage module is used to query the product call object with the time-of-use electricity carbon footprint factor before the update based on the call lineage, calculate the product carbon footprint difference based on the electricity consumption of the product call object and the time-of-use electricity carbon footprint factor before and after the update, and generate a recalculation task or archive record of the product call object based on the product carbon footprint difference.

[0014] This application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0015] The above method groups power grid operation data into time slices according to a preset time granularity, and constructs a regional power grid topology map with power grid nodes or regions as vertices and transmission lines or inter-regional exchange relationships as directed edges. This allows power system physical quantities such as power generation unit output, regional load, tie line power flow, line loss, and topology interruption to participate in the calculation under the same time base and the same topology structure. This reduces the problem of unclear factor update boundaries caused by inconsistent data sampling periods, inconsistent regional boundaries, or inconsistent power flow directions, and provides a calculable data foundation for subsequent disturbance propagation analysis.

[0016] When power grid operation data changes, a disturbance event containing the disturbance source node and the magnitude of the change is generated, and a weighted adjacency matrix is ​​further generated based on the power flow direction and the power consumption ratio. It can convert changes in a single data field into a computable signal that propagates along the power grid's receiving structure; affecting the propagation matrix. And the impact score With the cooperation of [unclear], the system no longer updates mechanically based on administrative regions or the entire database. Instead, it selects candidate nodes and time slices based on disturbance amplitude, propagation coefficient, coverage weight, and disturbance source quality weight. This reduces the recalculation range when the disturbance impact range is limited and improves the accuracy of affected area-time slice positioning.

[0017] By writing candidate nodes and time slices with impact scores no less than the update threshold into the incremental recalculation queue, and recalculating the time-of-use electricity carbon footprint factor only for objects in the queue, the need for unaffected regions and time slices to participate in repeated calculations can be reduced. The new version binds the parent version, disturbance event, propagation path, impact score, solution status, and calculation model version when writing, so that each factor change can be traced back to the corresponding power grid physical disturbance and propagation path. Objects that do not enter the recalculation queue use the original version through the version inheritance index, which helps to reduce the storage and maintenance overhead caused by full copying and makes "recalculation" and "inheritance" have interpretable boundaries.

[0018] The setting of event deduplication keys and queue deduplication keys enables duplicate data under the same disturbance source node, source field, affected time window, and model version to be merged and processed. The same regional nodes and time slices hit by the same event can also be merged at the queue level to affect the score and propagation path, thereby reducing the risk of repeatedly triggering recalculation tasks. For situations where the power flow direction is uncertain, the power flow of the connecting line is missing, the closed-loop mutual supply nodes do not converge, or the critical node missing rate is high, the processing of candidate edges, direction confidence, candidate power flow values, strong connected component iteration, isolation zone, and verification tasks helps to reduce the risk of low confidence factors directly covering the published effective version.

[0019] In the product call linkage process, the product batch, process, enterprise, project or unit process that has called the time-of-use electricity carbon footprint factor before the update is queried by calling the lineage query, and the product carbon footprint difference is calculated based on the electricity consumption and the factor difference before and after the update. This system can further map the propagation results of grid-side disturbances to specific calling objects in the product LCA database; it generates recalculation tasks for objects with differences reaching preset conditions and archives records for low-impact objects, ensuring that subsequent processing of product carbon footprint results is based on technical calculations of factor versions, calling time slices, and electricity consumption, rather than relying on human experience judgment. The system implements the above processing through modules such as data access and factor calculation, disturbance identification, impact propagation, impact scores and queues, incremental recalculation and lineage writing, and product calling linkage. The corresponding storage medium enables the processor to execute this process, thus providing stable data processing support for the operation, maintenance, version tracking, and necessary recalculation of time-sharing and zone-based power background datasets.

[0020] The specification of this application contains numerous technical features distributed across various technical solutions. Listing all possible combinations of these features (i.e., technical solutions) would make the specification excessively lengthy. To avoid this problem, the various technical features disclosed in the above-described invention, the various technical features disclosed in the following embodiments and examples, and the various technical features disclosed in the accompanying drawings can be combined with each other, provided it is technically feasible, to form various new technical solutions (all of which should be considered as described in this specification), unless such a combination of technical features is technically infeasible. For example, one example discloses feature A+B+C, and another example discloses feature A+B+D+E. Features C and D are equivalent technical means that serve the same function, and technically only one needs to be used; they cannot be used simultaneously. Feature E can technically be combined with feature C. Therefore, the solution A+B+C+D should not be considered as described because it is technically infeasible, while the solution A+B+C+E should be considered as described. Attached Figure Description

[0021] Figure 1 is a flowchart illustrating the method for dynamically updating a partitioned power carbon footprint dataset based on a power grid topology propagation model according to an embodiment of this application.

[0022] Figure 2 is a schematic diagram of the structure of a dynamic update system for a partitioned power carbon footprint dataset based on a power grid topology propagation model according to an embodiment of this application. Detailed Implementation

[0023] In the following description, many technical details are presented to help the reader better understand this application. However, those skilled in the art will understand that the technical solutions claimed in this application can be implemented even without these technical details and various variations and modifications based on the following embodiments.

[0024] Explanation of some concepts: A time slice is the smallest calculation period with a unified time base, formed by merging power generation unit output, regional load, tie line power flow, topology interruption, inter-regional exchange plan, and data quality information according to a preset time granularity.

[0025] Time-of-use electricity carbon footprint factor Carbon footprint refers to the unit electricity consumption carbon footprint value determined by a node in a certain region at a certain time slice, after taking into account the carbon intensity of local power generation, the carbon intensity carried by external power input, and line loss correction.

[0026] A regional power grid topology graph is a directed graph with power grid nodes or regions as vertices and transmission lines, tie lines, or identifiable inter-regional exchange relationships as directed edges, which can represent the power flow direction and power receiving structure within a time slice.

[0027] Regional nodes refer to provincial power grid nodes, municipal power grid nodes, substation supply area nodes, transaction settlement area nodes, or other identifiable power supply area nodes used to carry out time-of-use electricity carbon footprint factor calculation and product LCA default mapping.

[0028] Weighted adjacency matrix This refers to a time slice The matrix is ​​generated based on the power flow direction and energy percentage of the regional power grid topology, where elements... Representation Nodes To the node The input power is at the node The proportion of total received input power.

[0029] Influence propagation matrix It refers to the weighted adjacency matrix. The matrix obtained by calculating the propagation attenuation coefficient is used to characterize the finite-order propagation influence of the disturbance source node on each candidate node.

[0030] Disturbance events This refers to an event record that includes the source node of the disturbance and the magnitude of the change, generated when there is a correction of the unit emission factor, a supplementary recording of the power generation output, a correction of the tie line power flow, an adjustment of the regional boundary, a topology break, or a change in the data quality status.

[0031] A disturbance source node refers to a power grid node or region where data changes occur during a disturbance event.

[0032] range of change This refers to the disturbance source node. In time slice The difference between the corrected value and the original value of the corresponding source field.

[0033] Affecting scores It refers to the numerical signal used to determine whether candidate nodes and time slices should enter the incremental recalculation queue, calculated by combining the magnitude of change, the impact of propagation results, the coverage weight, and the quality weight of the disturbance source.

[0034] Coverage weight , refers to the method used to reflect candidate nodes In time slice Weights related to whether there is external power supply or whether it is accessed by the product's LCA database.

[0035] Disturbance source quality weight , refers to the weight used to reflect the source, completeness, delay, review status, or missing status of the disturbance source data.

[0036] transmission path , refers to the path record formed by information such as the sequence of nodes traversed from the disturbance source node to the candidate node, topological edges, propagation order, directional confidence, and topological version.

[0037] The incremental recalculation queue is a data queue used to store candidate nodes and time slices that affect the score and meet the update conditions, and which serve as the objects for subsequent time-sharing electricity carbon footprint factor recalculation.

[0038] Version lineage records refer to records that bind information such as parent version, perturbation event, propagation path, impact score, solution status, and computation model version when a new version factor is written.

[0039] Version inheritance index refers to a data index that records the parent version identifier, target dataset version identifier, and inheritance reason for regions and time slices that have not entered the incremental recalculation queue, without copying the full factor data.

[0040] Calling bloodline It refers to the data record that records the correspondence between the region, time slice, factor version, electricity consumption and calculation results called by a product batch, process, enterprise, project or unit process.

[0041] The product call object refers to the product batch, process, enterprise, project or unit process that has called the time-sharing and zone-based electricity carbon footprint factor.

[0042] Product carbon footprint differences This refers to the change in the electricity carbon footprint calculated by the product calling object based on electricity consumption and the difference in time-of-use electricity carbon footprint factors before and after the update.

[0043] Directional confidence refers to a data signal generated based on the preceding and following time slices, transaction plans, and net exchange direction of tie lines when the direction of power flow is uncertain, used to characterize the credibility of candidate directions.

[0044] The isolation zone refers to the data area used to temporarily store low-confidence factor versions and trigger review tasks when the power flow direction is uncertain, the power flow of the connecting line is missing, the closed-loop mutual supply nodes do not converge, or the missing rate of key nodes is high.

[0045] A strongly connected component is a maximal subset of nodes in a directed graph corresponding to a weighted adjacency matrix, where there is a directed path between any two nodes.

[0046] The following is a brief summary of some of the innovative aspects of this application: In summary, the inventiveness of this application does not lie in providing the time-sharing electricity carbon footprint factor of regional nodes within a certain time slice in isolation. The calculation itself—the determination of nodal carbon intensity—is already disclosed in the art. The key difference lies in steps 100 to 800 as shown in Figure 1, combined with the data access and factor calculation module, disturbance identification module, influence propagation module, influence score and queue module, incremental recalculation and lineage writing module, and product call linkage module shown in Figure 2, to generate the weighted adjacency matrix signal used to characterize the power grid's receiving structure. A propagation matrix signal used to characterize the influence of disturbances propagating along the power flow path with finite-order attenuation. And the influence fraction signal used to couple the disturbance amplitude, propagation coefficient, call coverage weight, and disturbance source quality weight. By embedding it into the versioned update chain of the partitioned power carbon footprint dataset, changes in power grid operation data do not directly trigger a full factor replacement of the background database. Instead, they are first identified by disturbance events. Anchoring the disturbance source node, source field, and affected time window, step 500 then generates a propagation path related to power flow direction, power percentage, and direction confidence. Then, step 600 writes the candidate nodes and time slices that affect the score and meet the update conditions into the incremental recalculation queue, and step 700 generates a new version identifier only for the region-time slice in the queue. At the same time, bind the parent version identifier. Transmission Path Affecting scores With the solution state For regions / time slices not covered by the propagation path or whose affected scores do not meet the update conditions, the original version is used via the version inheritance index without copying the full factor data. Further, step 800 utilizes the pedigree table signal. Reverse lookup of product objects that have invoked the old version of factors, and generate product carbon footprint difference signals based on the factor differences before and after the update. This constructs a continuous technical chain that integrates grid-side physical disturbances, database-side incremental recalculation, and product-side necessary recalculation, ensuring controlled progression, traceable fields, and divertable states. The technical concept transforms the topology propagation relationship from an intermediate quantity in carbon intensity calculation into a trigger for database update range filtering and version lineage constraints. Each step exhibits an inherent dependency: the propagation result determines the queue, the queue limits recalculation, and version lineage triggers the call. It is precisely this interlocking and indispensable collaborative relationship between features, and the resulting synergistic technical effects, that makes it difficult for those skilled in the art to directly obtain the overall technical solution of this application from existing node carbon intensity calculations or ordinary factor version update mechanisms through obvious combinations. This enhances the maintainability and traceability of the time-sharing and zone-based power background dataset.

[0047] More specifically, the inventors of this application, through long-term in-depth research, discovered that the time-sharing and zone-based power carbon footprint dataset is difficult to update incrementally in actual maintenance. This is not simply due to the insufficient precision of the power factor calculation formula, but rather because the existing processing chain typically processes changes in grid operation data, regional-time slice factor versions, and product LCA call records at different levels, lacking intermediate constraints that can span the physical disturbances of the grid and the lifecycle of the database version. After in-depth analysis, the inventors found that although unit output supplementation, tie-line power flow correction, topology interruption, or regional boundary adjustment manifest as changes in a certain source field, their impact does not spread evenly to the entire database. Instead, it forms a path-dependent range of influence along the power receiving ratio, power flow direction, propagation order, and data quality status. If the entire database is recalculated directly, it will unnecessarily increase the processing burden of irrelevant regional-time slices. If only the source node factor is modified, downstream nodes and product objects that have called the old version factor may be missed. Furthermore, through analysis and research of numerous practical maintenance cases, the inventors realized that ordinary version numbers only indicate that data has been replaced, but cannot specify which disturbance event, which topology propagation path, what impact score, or what solution state caused the replacement. Therefore, they are insufficient to support necessary recalculation or low-impact archiving of subsequent product datasets. Based on this understanding, the inventors of this application creatively propose a technical solution that uses disturbance events as entry points, determines affected regions and time slices using weighted adjacency matrices and impact propagation matrices, triggers incremental recalculation queues using impact scores, and binds parent versions, propagation paths, and call lineages in the new version. This allows local physical changes on the power grid side to be transmitted to the product carbon footprint database in a controlled manner.

[0048] To make the technical problem to be solved, the technical solution adopted, and the technical effect achieved by the present invention clearer, the following detailed description of the method and system for dynamic updating of partitioned power carbon footprint dataset based on the power grid topology propagation model provided by the present invention is provided in conjunction with specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention; without departing from the technical concept of the present invention, those skilled in the art can adaptively adjust specific fields, thresholds, and data table structures according to the granularity of power grid nodes, time slice granularity, data sources, and the calling method of the product life cycle assessment (LCA) database. Such adjustments do not change the core processing method of the present invention, which utilizes the power grid topology propagation results to filter the incremental recalculation range.

[0049] It should be noted that the objects processed in this invention are always data with physical meaning in the power system, such as power generation unit output, power generation technology type, unit emission factor, regional load, tie line power flow, line loss, maintenance outages, topology interruptions, and inter-regional exchange plans, rather than abstract financial, certification, or management data. The calculation steps involving weighted adjacency matrices, influence propagation matrices, and influence scores discussed below are all based on the operating laws of the power system, such as power flow direction, power receiving structure, line loss, regional load, and data quality. Their inputs come from the power grid SCADA / EMS system, metering system, dispatching system, and inter-regional transaction plan data sources. Those skilled in the art should understand that the various calculation steps utilize the operating laws of the power system, rather than purely abstract mathematical relationships.

[0050] In various embodiments of the present invention, the regional node can be a provincial power grid node, a municipal power grid node, a substation supply area node, a transaction settlement area node, or other identifiable power supply area node; the time slice can be set according to a preset time granularity such as 15 minutes, 30 minutes, 1 hour, or day; the product calling object can be a product batch, process, enterprise, project, or unit process.

[0051] Example 1: Dynamic Update Method for Partitioned Electricity Carbon Footprint Dataset This embodiment provides a method for dynamically updating a partitioned power carbon footprint dataset based on a power grid topology propagation model. The overall approach is as follows: First, calculate the time-of-use power carbon footprint factor for each regional node in each time slice based on power grid operation data. This serves as a versioned dataset that can be called from the product's LCA background database. When changes occur in power grid operation data, topology, or regional boundaries, the entire database is not recalculated. Instead, the change is first abstracted as a disturbance event. Then, the affected region-time slice is identified using the power grid topology propagation relationship. Only the affected components are incrementally recalculated and written to the new version. Finally, the local disturbance on the power grid side is propagated along the calling lineage to the product calling objects that already reference the old version factor. As shown in Figure 1, the method in this embodiment may include steps 100 to 800, and executes the abnormal operating condition handling in step 900 when situations such as power flow loss, closed-loop mutual supply, uncertain direction, or low-confidence release occur. The following is a step-by-step explanation.

[0052] Step 100: Acquire power grid operation data and perform time-slice standardization. In step 100, the system obtains power grid operation data from the power grid SCADA / EMS, metering system, dispatching planning system, inter-regional trading planning system, or other power grid operation data sources, and groups data from different sources and with different sampling frequencies into time slices according to a preset time granularity. The power grid operation data includes generator output, generator technology type, unit emission factor, regional load, tie-line power flow, line loss, maintenance outages, topology interruptions, inter-regional exchange plans, and information related to data quality such as completeness, latency, and audit status. Specifically, the system generates a time slice record for each time slice, which includes at least a time slice identifier, start time, end time, time granularity, data source, data latency, data completeness, quality identifier, and freeze status. This process unifies power system physical quantities from different sources to the same time base, avoiding inconsistencies in the time boundaries of propagation calculations and factor updates caused by generator output, tie-line power flow, and regional load being in different sampling periods.

[0053] Furthermore, the system maintains a time slice table, regional node table, topology edge table, generation unit emission table, time-sharing and partitioning factor table, disturbance event table, and call lineage table in the database to support subsequent incremental updates and version lineage tracing. Specifically, the time slice table records time slice identifier, start time, end time, time granularity, peak-valley-flatness identifier, and quality fields such as integrity, delay, and freeze status; the regional node table records regional node identifier, administrative region, grid region, node type, latitude and longitude, parent region, product LCA default mapping, and tracing fields such as regional version and boundary effective time; the topology edge table records the start point, end point, line type, rated capacity, power flow direction, loss rate, direction confidence, and version fields such as topology version, effective time, and failure time for transmission lines or inter-regional exchange relationships; the generation unit emission table uses the generation unit identifier and time slice identifier as primary keys to record generation technology, fuel type, grid-connected electricity, direct emission factor, upstream factor, or unit generation LCI. The data sources and vectors are as follows: The time-sharing and partitioned factor table uses the region identifier, time slice identifier, and factor version identifier as primary keys to record time-sharing electricity carbon footprint factors, direct emission factors, upstream factors, renewable energy share, purchased electricity share, confidence level, and version lineage fields such as parent version, disturbance event, propagation path, and calculation model version; the disturbance event table records event type, disturbance source node, source field, value before change, value after change, trigger threshold, review status, and model version; the call lineage table records product batch, process, enterprise, project, call region, call time slice, factor version, electricity consumption, and calculation results. The above tables are linked through time slice identifier, region node identifier, topology version, factor version, and product call object identifier, thereby organizing power grid operation data, region nodes, topology edges, time slices, and product LCA call lineage into a computable and traceable data structure.

[0054] Step 200: Construct the regional power grid topology map In step 200, the system constructs a regional power grid topology map for each time slice. The regional power grid topology map uses regional nodes as vertices and transmission lines, tie lines, or identifiable inter-regional exchange relationships as directed edges. The direction of the directed edges is determined according to the power flow direction within that time slice, enabling the topology map to reflect the physical relationship of power propagation from upstream nodes to downstream nodes, and is accordingly written into the regional node table and the topology edge table. Optionally, the vertices can be set according to provincial power grids, municipal power grids, substation supply areas, or transaction settlement areas, thereby retaining a reasonable scope of application. The regional node table supports the traceability of regional boundary changes through regional version and boundary effective time fields, and the topology edge table supports the traceability of topology structure changes through topology version, effective time, and expiration time fields.

[0055] Step 300: Calculate the time-sharing electricity carbon footprint factor for regional nodes. In step 300, the system calculates the time-of-use electricity carbon footprint factor for each regional node in the corresponding time slice based on the regional power grid topology map, power grid operation data, and power generation unit emission data. For example, for a node... and time slices It can be calculated using the following formula: in, Represents a node In time slice Time-of-use electricity carbon footprint factor; Represents a node A collection of power generation units within the area; Indicates power generation unit In time slice The amount of electricity used for internet access; Indicates power generation unit In time slice The carbon footprint factor of power generation; Represents a node To node Input power; Indicates upstream node In time slice Time-of-use electricity carbon footprint factor; Represents a node To node Line loss rate; Represents a node In time slice Regional load; Represents a node In time slice The sum of the electricity output to other nodes; To avoid extremely small positive values ​​when divided by zero, as can be seen from Formula 1, the node factor considers both the carbon intensity of local generating units and the carbon intensity carried by externally input electricity through upstream node factors and line loss corrections, thus reflecting the carbon intensity of the node under the corresponding grid supply structure.

[0056] It should be noted that Formula 1 is used to explain the calculation method of the time-of-use electricity carbon footprint factor. In specific implementation, the direct emission factor, upstream factor, renewable energy ratio, and purchased electricity ratio can be written into the time-of-use zoning factor table according to the regional node granularity, data availability, and factor scope of the product LCA background database. For topological regions without closed-loop mutual supply relationships, the system can calculate the factors of each node sequentially according to the topological order corresponding to the power flow direction. For regions with ring networks or mutual supply nodes, the system first identifies strongly connected components and iteratively solves the time-of-use electricity carbon footprint factor within the strongly connected components. For the handling of abnormal cases, please refer to step 900.

[0057] Step 400: Generate disturbance events when power grid operation data changes. In step 400, when the system receives data changes related to unit emission factor correction, power generation output supplementation, tie-line power flow correction, regional boundary adjustment, topology interruption, time slice quality status change, or other data changes related to the physical operating status of the power grid, the change is converted into a disturbance event and written into the disturbance event table. It should be noted that the disturbance events referred to in this invention all refer to physical changes in the power system, rather than generalized data changes, thus ensuring that the method always remains within the scope of power system data processing. Specifically, each disturbance event includes at least a disturbance event identifier, a disturbance source node, a source field, a value before the change, a value after the change, an affected time window, data source quality, audit status, topology version, and calculation model version. Disturbance source node In time slice The magnitude of the change is expressed by the following formula: in, Indicates the source node of the disturbance In time slice The range of change; and These represent the values ​​of the disturbance source field after correction and before correction, respectively; the source field can be fields such as tie line power flow, generator unit output, unit emission factor, topology edge status, area boundary version, or data quality status.

[0058] Optionally, when comparing the degree of influence across different disturbance source fields with different dimensions, the magnitude of change can be made dimensionless: in, Indicates the magnitude of change after dimensionless transformation; This represents the reference baseline value corresponding to the disturbance source field, such as the typical or nominal value of the field within a historical time window. This dimensionless processing is an optional implementation and does not change the main process of influence propagation driven by the disturbance source node and the magnitude of change.

[0059] Furthermore, to avoid the same physical disturbance being processed repeatedly, the system generates event deduplication keys based on the disturbance source node, source field, affected time window, and model version: (Formula 4) in, This represents a hash function, whose four input parameters are the source node of the disturbance, the source field that has changed, the affected time window, and the calculation model version. When newly received data has the same event deduplication key as an existing incomplete disturbance event, the system does not generate a new independent disturbance event. Instead, it records a revised version of the event under the original disturbance event and uses the latest data after approval as the input for subsequent impact propagation calculations. Data that fails approval is placed in a pending approval state and does not enter the incremental recalculation queue. This avoids redundant recalculation caused by the same physical disturbance being triggered repeatedly.

[0060] Step 500: Generate the weighted adjacency matrix and calculate the impact propagation results. In step 500, for each time slice affected by the disturbance event, the system generates a weighted adjacency matrix based on the power flow direction and energy percentage of the partitioned power grid topology. Its elements are calculated using the following formula: in, Indicates time slice internal nodes To the node The input power is at the node The proportion of the total received input power, i.e., the elements of the weighted adjacency matrix; Indicates time slice internal nodes To node Input power; Represents a node In time slice The sum of all received input charges, of which To the node Input the upstream node index of the power level; To avoid extremely small positive values ​​with a denominator of zero, it should be noted that the physical meaning of Equation 5 lies in characterizing the nodes. The power receiving structure, i.e., the node What percentage of the power received comes from upstream nodes? This provides a physical basis for subsequently determining the path and magnitude of the disturbance propagation along the power grid. When multiple upstream nodes simultaneously send signals to the node... When supplying power, the denominator of Formula 5 is normalized; when the node load or power receiving capacity is missing or even zero, it is normalized according to... The estimated value is then used as a fallback. More specifically, when the power flow direction is determined and consistent with the cross-regional trading plan or the net exchange direction of the tie line, the system uses the corresponding topological edge as a valid propagation edge; when the power flow direction is uncertain or inconsistent with the trading plan, candidate directions are generated based on the power flow direction, trading plan, and net exchange direction of the tie line in the preceding and following time slices, and these are written into the direction confidence score. When the direction confidence score is lower than the direction confirmation threshold, the system will not accept the candidate direction. When this occurs, the edge is saved as a candidate edge (candidate_edge_flag=1) and is not directly published as a high-confidence propagation edge. Its confidence level is also reduced accordingly in the influence score calculation and factor publication control to avoid the uncertain direction of the trend data directly changing the published time-sharing partition factor version.

[0061] Furthermore, regarding the disturbance source nodes The system is based on a weighted adjacency matrix. The propagation attenuation coefficient is calculated by time slice, with the maximum order being... Influence propagation matrix As a result of the propagation of the aforementioned influence: in, Indicates time slice The influence propagation matrix; This represents the identity matrix, used to represent the impact of the disturbance source node on itself; This represents the propagation attenuation coefficient, used to characterize the degree of attenuation as a disturbance propagates along the power grid topology; Indicates time slice The weighted adjacency matrix, Indicate its The power is used to represent the perturbation along the edge. The result of topological path propagation; This represents the maximum propagation order. It should be noted that Formula 6 elevates the power grid topology matrix from a simple "carbon intensity calculation tool" to a "database update impact range screening tool": through finite-order power series expansion, it can identify candidate nodes that may be affected by the disturbance source node and their influence relationships in advance without performing a full database recalculation.

[0062] Considering that actual power grid structures are often sparse, to avoid redundant propagation calculations with minimal final impact, propagation cutoff conditions can be set in some implementations. Higher-order propagation calculations will stop when the following equation is satisfied: in, Indicates the current propagation order; Representation matrix The maximum value of the elements in the middle; This represents the propagation truncation threshold. The purpose of this truncation is to reduce unnecessary computation while preserving the main propagation path. It should be noted that this is done during step-by-step calculations. During the process, the system simultaneously extracts data from the disturbance source nodes. To each candidate node The sequence of nodes traversed, the propagation order, the topological edges involved, the edge weights, the direction confidence, and the topology version number are recorded as the propagation path. Since the propagation path is stored together with the disturbance event and the factor version, the factor differences of subsequent product calls can not only be traced back to the old and new factor versions, but also to the power grid physical path that caused the version change.

[0063] Step 600: Calculate the impact scores and generate an incremental recalculation queue. In step 600, the system combines the change amplitude of the disturbance source node and the effect propagation result to calculate the influence score of each candidate node in the corresponding time slice, which can be exemplarily calculated by the following formula: in, Indicates candidate nodes In time slice The impact score; Indicates the source node of the disturbance In time slice The magnitude of the change (which can be replaced in the implementation using the dimensionless processing of Formula 3) is And update the threshold (Matching the selected method of representing the magnitude of change); Indicating the influence propagation matrix The middle is composed of disturbance source nodes to candidate node The propagation coefficient; Indicates candidate nodes The coverage weight is used to reflect whether the candidate node has external power supply, whether it is called by the product LCA database, the frequency of call, or the number of product batches covered. This represents the quality weight of the disturbance source, used to reflect data quality factors such as data origin, completeness, review status, delay, and missing data ratio. It should be noted that... Business importance scores are not inexplicable. It is not a subjective credibility score; both are determined by recordable fields. The impact score includes the magnitude of the disturbance itself, the attenuation relationship of topology propagation, and the call coverage and data quality of the affected objects, thus providing a comprehensive measure of the actual impact of the disturbance.

[0064] when Not less than the update threshold If the corresponding time slice is not in a low-confidence isolation state, the system writes the candidate node and time slice into the incremental recalculation queue. The incremental recalculation queue records the queue identifier, disturbance event identifier, region node identifier, time slice identifier, impact score, propagation path, priority, queue deduplication key, queue status, creation time, and computation model version. Through this queue, disturbance events do not directly trigger a full-database recalculation; instead, they are first filtered by the topology propagation impact range, and then targeted recalculation is performed on the affected region nodes and time slices. To avoid the same disturbance event repeatedly triggering recalculation for the same region node and time slice, the system generates a queue deduplication key based on the disturbance event identifier, region node identifier, time slice identifier, and model version. The input parameters are the disturbance event identifier, the region node identifier, the time slice identifier, and the calculation model version, respectively. When there are already recalculation tasks with the same deduplication key in the incremental recalculation queue that have not yet been completed, the system merges the corresponding impact scores and propagation paths instead of generating new recalculation tasks. When the same candidate node and time slice are affected by multiple disturbance events, the system retains multiple disturbance event identifiers and corresponding propagation paths, and determines their recalculation priority in the incremental recalculation queue using the following formula: in, Indicates candidate nodes In time slice Recalculation priority; This represents the product hit factor, which is used when the candidate node and time slice match the published product call lineage. Otherwise take Formula 8 Used to reflect the data quality of the disturbance source in the score calculation stage, in Formula 10 This is used for priority sorting when multiple candidate nodes or multiple disturbance events enter the queue concurrently; when secondary sorting of quality weights is not required during implementation, the priorities in Formula 10 can also be sorted according to... Configure accordingly. As shown in Formula 10, region nodes and time slices already called by the product's LCA database can be prioritized for recalculation, thus aligning the recalculation sequence with the timeliness requirements of the product report without altering the factor physical calculation formula. Optionally, in scenarios with high timeliness requirements for product reports, a hit weight coefficient can also be introduced. Write the priority as This will further improve the recalculation efficiency of cited factors.

[0065] Step 700: Perform incremental recalculation, write the new version, and establish version inheritance relationships. In step 700, the system only uses the factor calculation method (i.e., Formula 1) in step 300 for regional nodes and time slices in the incremental recalculation queue, and recalculates the time-sharing electricity carbon footprint factor based on the latest grid operation data, topology version, and data quality status of the corresponding time slice. After the recalculation is completed, the system writes the new version into the time-sharing partition factor table. The new version is bound to at least the factor version identifier, parent version identifier, disturbance event identifier, propagation path, impact score, solution status, calculation model version, and confidence level, thereby forming a version lineage record strongly bound to physical causes. It should be noted that the version lineage of this invention is not an ordinary version number, but rather saves the factor version together with grid physical disturbances, topology propagation paths, and recalculation status. For example, the factor version of a certain regional node can simultaneously record its parent version, the tie-line power flow correction event that triggered the update, the propagation path from the disturbance source node to the regional node, the impact score, and the solution status, so that each factor update has a white-box auditable basis.

[0066] More specifically, for region nodes and time slices not written to the incremental recalculation queue, the system does not copy the full factor data. Instead, it records the region node identifier, time slice identifier, inherited parent version identifier, target dataset version identifier, inheritance identifier, and inheritance reason in the version inheritance index. The inheritance reason includes at least one of the following: the impact score is lower than the update threshold, the call lineage is not hit, the impact propagation path is not covered, or the time slice is frozen. Therefore, when a new dataset version is formed, affected region nodes and time slices generate new factor records, while unaffected nodes reuse the parent version record through the version inheritance index. This avoids both the storage redundancy caused by full database copying and the computational overhead of full database recalculation. It should be noted that there is a clear synergistic relationship between this step and the preceding steps: the impact propagation results of step 500 determine which region-time slices enter the incremental recalculation queue in step 600; the incremental recalculation queue, in turn, determines which region-time slices are written to the new version and which retain the parent version in this step; the writing of the new version further triggers the product call lineage hit processing in step 800. It is this continuous chain of "impact propagation → queued recalculation → version inheritance → call lineage hit" that enables local disturbances on the power grid side to be transmitted to the product LCA background dataset in an orderly, controllable, and traceable manner, rather than simply stacking functions. In addition, if a time slice experiences situations such as strong connected components not converging, critical link missing rate exceeding the threshold, direction confidence below the release threshold, or solution failure, its new factor is written to the isolation area, without overwriting the already released valid version, and a review task is generated. For details of the relevant processing, please refer to step 900.

[0067] Step 800: Generate the product LCA call impact set. In step 800, the system queries the product calling objects that have called the time-of-use electricity carbon footprint factor before the update based on the calling lineage. The calling lineage records the product batch, process, enterprise, project, unit process, calling region, calling time slice, factor version, electricity consumption, and original calculation results. The system calculates the product carbon footprint difference based on the electricity consumption of the product calling object in the affected time slice and the time-of-use electricity carbon footprint factor before and after the update. in, Indicates the product call object Differences in product carbon footprint; Indicates the product call object In the affected time slice Electricity consumption; and They represent the corresponding regions respectively. In time slice The updated and unupdated time-of-use electricity carbon footprint factors. It should be noted that the input to this step is the difference in the physical quantity of the electricity carbon footprint factor. The processing is based on numerical calculations of electricity consumption and factor version, and the output is the writing of database status markers. The entire process does not involve management judgments made by personnel based on reporting requirements. Furthermore, the system allocates traffic according to the relative magnitude of the product carbon footprint differences, with the allocation criterion as shown in the following formula: in, Indicates the product call object The total original carbon footprint of electricity; This represents the product difference threshold. When Formula 12 is met, the system sets the dataset status of the product call object to "recalculate required" and generates a recalculation task; when the above ratio is less than the product difference threshold, it is set to "low impact archive" and an archive record is generated. It should be noted that when... When the value is zero, missing, or less than a preset minimum positive value, the system does not perform the above relative difference judgment, but instead... Whether the preset absolute difference threshold has been reached, a recalculation task, a low-impact archive record, or a review task will be generated accordingly to avoid [further issues]. Extremely small relative ratios are abnormally amplified. Optionally, to avoid duplicate reporting or multiple arrivals of the same disturbance event leading to repeated task generation for the same product call object, the system generates a deduplication key for product tasks based on the product call object identifier, target dataset version identifier, and disturbance event identifier. When a recalculation task, archived record, or review task with the same deduplication key already exists, a new task or record is not generated. Instead, the product carbon footprint difference is re-determined based on the latest factor version after approval, and the original task or record and the dataset status of the product call object are updated accordingly. Furthermore, for product results that have been frozen or released, the system does not directly overwrite the original results. Instead, it generates a recalculation prompt record with the disturbance event identifier, factor version identifier, propagation path, and product carbon footprint difference. This ensures that while directing the transmission of power grid physical disturbances to specific product call objects, the consistency of data in published reports is also maintained.

[0068] Step 900: Handling Abnormal Operating Conditions In step 900, the system handles abnormal operating conditions such as power flow loss, closed-loop mutual supply, direction uncertainty, load loss, multiple disturbance concurrency, and low-confidence release to reduce the risk of low-confidence data overwriting already published valid versions. Specifically, when the time slice... When the power flow of the connecting line is missing, the system generates candidate power flow values ​​based on the cross-regional transaction plan and the power flow data of adjacent time slices: in, Represents a node To node In time slice Candidate power flow values; This indicates the flow rate value given by the cross-regional transaction plan; and These represent the actual power flow values ​​for the previous and next time slices, respectively. The interpolation weight is determined based on the confidence level of the transaction plan and the completeness of adjacent time slice data. Topological edges with candidate power flow values ​​are written into the candidate edge identifier, and their direction confidence and distribution confidence are reduced, with the confidence level decreasing as the missing proportion increases.

[0069] More specifically, when closed-loop mutual supply nodes exist in the weighted adjacency matrix, the system first identifies strongly connected components and iteratively solves the time-sharing electricity carbon footprint factor within each strongly connected component. For components, the solution is performed in a directed acyclic topological order. If the difference between the factors obtained from two iterations is less than the convergence threshold... If the factor calculation within the corresponding strongly connected component is considered converged, and the number of iterations reaches the maximum number of iterations while the factor difference between two consecutive iterations is still not less than the convergence threshold, then the corresponding region node and time slice are marked as low-confidence state, the generated factors are written to the isolation area and a review task is triggered, without directly overwriting the already published valid version. Furthermore, when the node load is zero or missing, the system adopts... Avoid division by zero and write the estimated identifier; when the missing rate of critical nodes exceeds the missing rate threshold. At that time, the system writes the factor version corresponding to the time slice into the isolation area, retaining the disturbance event, candidate edges, missing fields, and review status. Through the above anomaly handling, the risk of low-confidence data directly entering the official release version can be reduced, forming a fault-tolerant isolation mechanism for sparse and closed-loop large power grids.

[0070] Step 1000 Exemplary Application Process To facilitate understanding of the coordination between the above steps, the following explanation uses a tie-line power flow correction event throughout the entire process. For example, within a certain hourly time slice, the system has already integrated grid operation data from thermal power, wind power, photovoltaic power, and inter-provincial power receiving, and generated an old version of the time-sharing zoning factor based on regional nodes, topology edges, generation unit emission data, and regional load. Subsequently, the system receives a measured power flow from an inter-regional tie-line, which is approximately... MWh revised to approximately For MWh data, if the disturbance source node is N_IMPORT_03, then the power flow change magnitude of this disturbance is: in, This indicates the magnitude of the change corresponding to the power flow correction of the connection line. The system generates an event identifier and event deduplication key for this correction in step 400, reads the weighted adjacency matrix and topology version of the corresponding time slice in step 500, calculates the impact propagation matrix and propagation path in step 600, and forms candidate node impact records in step 600. For example, the impact scores of the three load nodes within the province obtained by the system are approximately... , and When updating the threshold Approximately At that time, the impact score was approximately and The regions and time slices corresponding to the two nodes are written to the incremental recalculation queue; the impact score is approximately The node does not enter the incremental recalculation queue, and the parent version is used as the inheritance reason in the version inheritance index: "impact score is lower than the update threshold".

[0071] Furthermore, for region A entering the incremental recalculation queue, the system recalculates the time-sharing electricity carbon footprint factor for that region in the corresponding time slice according to step 700. For example, the old factor for region A is approximately... The recalculated new factor is approximately The system writes the new version to the time-sharing partition factor table and binds it to the parent version, disturbance event, propagation path, impact score, solution status, and calculation model version. If a product calls the old version factor in region A during this time slice, and the electricity consumption is approximately... MWh (i.e.) If the energy carbon footprint difference of the product's call object is kWh, then the difference is: in, This indicates the product carbon footprint difference caused by the update of the time-of-use electricity carbon footprint factor for the product being invoked. Based on this, the system writes the product invoking object, invoking region, invoking time slice, old and new factor versions, disturbance event, propagation path, and difference value into the product LCA invoking impact set, and then sorts them according to the product difference threshold. Generates tasks requiring recalculation or low-impact archive records. This example illustrates that tie-line flow correction does not directly trigger a full database recalculation, but rather, after topology propagation, impact score filtering, queue recalculation, and version lineage writing, it further targets specific product call objects. It should be noted that the weighted adjacency matrix in the above example... Propagation coefficient Coverage weight With quality weight The node-by-node values ​​are determined by calibration and historical sample statistics. A set of exemplary self-consistent intermediate values ​​can be found in the "Exemplary Parameter Determination and Boundary Case Description" section below, which does not affect the feasibility of the above calculation rules and endpoint data.

[0072] Example 2: Dynamic Update System for Partitioned Electricity Carbon Footprint Dataset This embodiment also provides a dynamic update system for a partitioned electricity carbon footprint dataset based on a power grid topology propagation model. As shown in Figure 2, the system includes a data access and factor calculation module, a disturbance identification module, an impact propagation module, an impact score and queue module, an incremental recalculation and lineage writing module, and a product call linkage module. Each module can be deployed on the same server, database platform, or distributed data processing platform, or it can be deployed in a distributed manner according to the data source and computational tasks.

[0073] The data access and factor calculation module acquires power grid operation data and groups it into time slices according to a preset time granularity. For each time slice, it constructs a partitioned power grid topology map with power grid nodes or regions as vertices and transmission lines or inter-regional exchange relationships as directed edges. Based on the partitioned power grid topology map and power grid operation data, it calculates the time-of-use electricity carbon footprint factor for each regional node in the corresponding time slice. This module maintains a time slice table, a regional node table, a topology edge table, a power generation unit emission table, and a time-of-use partitioned factor table. The disturbance identification module generates disturbance events containing the disturbance source node and the magnitude of the change when power grid operation data changes. This module generates event deduplication keys based on the disturbance source node, source field, affected time window, and model version. It merges disturbance data with the same event deduplication key and keeps the data that fails the review in a pending review state. The impact propagation module generates a weighted adjacency matrix based on the power flow direction and electricity percentage of the regional power grid topology for time slices affected by disturbance events. It then calculates the impact propagation results of the disturbance source node on each candidate node based on the weighted adjacency matrix and propagation attenuation coefficients. This module also extracts propagation paths when calculating propagation matrices of different orders and writes direction confidence and candidate edge identifiers when the power flow direction is uncertain. The impact score and queue module calculates the impact score of each candidate node by combining the change magnitude and impact propagation results. Candidate nodes with impact scores not less than the update threshold and their time slices are written into the incremental recalculation queue. This module generates a queue deduplication key based on the disturbance event identifier, regional node identifier, time slice identifier, and model version. When there are incomplete recalculation tasks with the same queue deduplication key, the impact score and propagation path are merged. The incremental recalculation and lineage writing module is used to recalculate the time-of-use electricity carbon footprint factor only for candidate nodes and time slices in the incremental recalculation queue, write the new version, and bind the version lineage record of the parent version, disturbance event, propagation path, and impact score. For candidate nodes and time slices not written to the incremental recalculation queue, the original version is used and the inheritance relationship is recorded in the version inheritance index. If there is a failure to converge strongly connected components, a critical link missing rate exceeding the threshold, a direction confidence lower than the release threshold, or a solution failure, this module will write the corresponding factor to the isolation area and trigger a review task, without directly overwriting the already released valid version. The product call linkage module is used to query the product call objects that have called the time-of-use electricity carbon footprint factor before the update based on the call lineage. It calculates the product carbon footprint difference based on the electricity consumption of the product call object and the time-of-use electricity carbon footprint factor before and after the update, and generates a recalculation task or archive record for the product call object based on the product carbon footprint difference. For product results that have been frozen or released externally, this module generates a recalculation prompt record without directly overwriting the original result.

[0074] It should be noted that the system modules in this embodiment correspond to the method steps in Embodiment 1: the data access and factor calculation module corresponds to steps 100 to 300, the disturbance identification module corresponds to step 400, the influence propagation module corresponds to step 500, the influence score and queue module corresponds to step 600, the incremental recalculation and lineage writing module corresponds to steps 700 and 900, and the product call linkage module corresponds to step 800. Therefore, each module in the above system is not only a functional name, but also corresponds to specific data input, matrix calculation, queue writing, version lineage saving, and product call influence set generation actions. Those skilled in the art can understand the specific implementation of each module based on this, and will not be elaborated further here.

[0075] Example 3: Computer-readable storage medium This embodiment also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements the method for dynamically updating the partitioned power carbon footprint dataset based on the power grid topology propagation model described in Embodiment 1. The computer-readable storage medium can be a disk, solid-state drive, memory card, read-only memory, random access memory, or other tangible medium capable of storing a computer program. The modules in Figure 2 can be implemented by the processor executing the computer program in the storage medium, or they can be deployed in a database platform or a distributed data processing platform. When the processor executes the computer program, it can complete processes such as power grid operation data access, time slice merging, topology graph construction, time-sharing power carbon footprint factor calculation, disturbance event generation, impact propagation result calculation, impact score calculation, incremental recalculation queue generation, version lineage writing, version inheritance index recording, anomaly isolation, and product LCA call impact set generation.

[0076] The above embodiments have the following technical effects The above embodiments organize power grid operation data, regional nodes, topology edges, time slices, and product LCA call lineage into a computable and traceable data structure. This allows changes in power grid operation status to be converted into disturbance events that include disturbance source nodes and change magnitudes. By using a weighted adjacency matrix and an influence propagation matrix, affected candidate nodes and time slices are identified based on power flow direction and power receiving structure, rather than directly recalculating the entire database after a disturbance occurs. Therefore, when the disturbance's impact range is limited, it helps reduce the need for full database recalculation, thereby reducing database recalculation overhead and narrowing the update scope.

[0077] The above embodiments further generate an incremental recalculation queue by influencing scores and update thresholds, enabling the system to recalculate factors only for affected region nodes and time slices. Through event deduplication and queue deduplication, the system can reduce the repeated triggering of recalculations for the same disturbance or the same region-time slice, thereby improving system efficiency. Since unaffected region nodes and time slices do not copy the full factor data but reuse parent version records through version inheritance indexes, storage efficiency and dataset maintainability can be improved while preserving topology propagation lineage. Furthermore, because the new version simultaneously binds to the parent version, disturbance event, propagation path, influencing scores, solution status, and computation model version, it helps improve the accuracy of locating affected factors and the interpretability of reviewing data sources, propagation paths, and version changes.

[0078] The above embodiments also transmit the updated time-of-use (TOU) electricity carbon footprint factors to specific product calling objects by calling the lineage table, and calculate the product carbon footprint difference based on the product's electricity consumption and the TOU electricity carbon footprint factors before and after the update, thereby providing data lineage support for necessary recalculation or low-impact archiving of product calling objects. For abnormal operating conditions such as missing power flow, closed-loop mutual supply, uncertain direction, and low-confidence solutions, candidate edges, iterative solutions, isolation zones, and verification tasks help reduce the risk of low-confidence factors directly covering published valid versions. Overall, the above technical means help to narrow the recalculation scope while preserving the topology propagation lineage, and improve the version traceability and maintenance efficiency of the partitioned electricity carbon footprint dataset.

[0079] Exemplary parameter determination and boundary case description In the implementation using the dimensionless processing described in Formula 3, the magnitude of change can be represented by the ratio of the disturbance source field to the reference value. When using a non-dimensionally-free magnitude of change, the update threshold is set according to the source field type to prevent disturbance source fields with different dimensions from being directly compared under the same threshold. For example, a reference value for the disturbance source field is taken. MWh, then the normalized variation Take the quality weight of the disturbance source. Three candidate load nodes , , All involve external power supply and product usage; coverage weight is applied accordingly. ; via propagation matrix After unfolding, it is generated by the disturbance source node. The propagation coefficients to the three candidate nodes are approximately , , Substituting into Formula 8, we can obtain the factors affecting the score. , , This is consistent with the endpoint data in the previous example.

[0080] Propagation attenuation coefficient The factor change rate can be calibrated based on the historical tidal disturbance propagated along adjacent regions, and the value is located in... to Between, for example, about to Maximum propagation order The maximum effective propagation order can be set to a positive integer based on the regional power grid topology, for example, approximately to Propagation cutoff threshold The threshold can be set based on the contribution ratio of higher-order propagation terms to the score; update the threshold. The parameters can be set comprehensively based on historical factor change rates, regional electricity consumption scale, and product call frequency. For example, in the aforementioned calculation example, it is approximately... Coverage weight Can be generated by nodes In time slice The weighting is determined by factors including whether there is external power supply, whether there are product LCA call records, the number of product batches called, and the percentage of electricity consumed during the call. For example, a higher weight is given when there is a call lineage of published products, and a lower weight is given when there is only internal power supply and no product calls. The specific weights are pre-calibrated by the database's historical call statistics. Disturbance source quality weighting. The quality weight can be determined based on the data source level, collection completeness, review status, data delay, and missing ratio; for example, data from the actual measurement system that has passed the review corresponds to a higher quality weight, while data from candidate completion values ​​or data with high delay corresponds to a lower quality weight.

[0081] Formula 8 Used to reflect the data quality of the disturbance source in the score calculation stage, in Formula 10 This is used for priority sorting when multiple candidate nodes or multiple disturbance events enter the queue concurrently; when secondary sorting of quality weights is not required during implementation, the priorities in Formula 10 can also be sorted according to... Configure accordingly. Directional confidence can be determined by the consistency between the measured power flow direction in the current time slice, the power flow directions in the preceding and following time slices, the direction of the trading plan, and the net exchange direction of the tie line; when the directional confidence is lower than... When this happens, the edge is written into the candidate edge list with candidate_edge_flag=1, and is not directly published as a high-confidence propagation edge. The confidence level is also reduced in the factor version.

[0082] Product Difference Threshold The determination can be made based on the product's LCA database update strategy, historical power factor fluctuations, and the electricity consumption ratio of the product's target users; when A recalculation task is generated when the threshold is reached or exceeded; a low-impact archive record is generated when the threshold is below. If the value is zero, missing, or less than a preset minimum positive value, the system will not perform a relative difference judgment, but will instead determine the relative difference based on... Whether the absolute difference threshold is reached determines whether a recalculation task, low-impact archived record, or review task is generated. Missing rate threshold. The convergence threshold can be set based on the proportion of missing data in key connection lines or key nodes. The interpolation weight can be set based on the allowable difference between the time-sharing electricity carbon footprint factors of two consecutive iterations; The settings can be adjusted based on the credibility of the transaction plan and the completeness of the actual power flow data in adjacent time slices. The setting is increased when the credibility of the transaction plan is high. When the measured data of adjacent time slices are more complete, the reduction is lower. .

[0083] It should be understood that the weighted adjacency matrix in this application Influence propagation matrix And the impact score The computational granularity is always a region node-time slice hierarchy, consisting of region nodes and time slices. This affects the score. It is a screening signal used to determine whether candidate nodes and time slices should enter the incremental recalculation queue. It is generally not interpreted as the actual change in the time-sharing electricity carbon footprint factor of the candidate node. It does not replace product carbon footprint differences. The calculation of the changes in the propagation process is as follows. Therefore, the dimensionless transformation of the magnitude of changes, the finite-order truncation, and the calibration of coverage and quality weights are mainly used to determine whether candidate nodes and time slices enter the incremental recalculation queue. For regional nodes and time slices that have entered the incremental recalculation queue and completed recalculation, their new time-of-use electricity carbon footprint factors are still calculated according to Formula 1 based on the latest grid operation data, topology version, and data quality status of the corresponding time slice. The product carbon footprint difference of the corresponding product call object is still calculated based on the call lineage, electricity consumption, and factors before and after the actual update. The above approximate screening does not directly participate in the numerical calculation of the new factor value and the product carbon footprint difference value, but it will affect the scope of objects covered by this incremental update task. Update threshold. When a smaller value is taken, the range of entries is expanded accordingly, and the screening tends to be conservative. The choice between the strictness and leniency of this threshold is an implementation trade-off between the recalculation range and the computational cost.

[0084] It should also be understood that the use of the term "affected" in this application has different meanings in two contexts: First, "time slice affected by the disturbance event" refers to the time slice that is within the affected time window of the disturbance event and needs to participate in the weighted adjacency matrix generation and influence propagation calculation; Second, when describing the update results, "affected regional nodes and time slices" and "unaffected" refer to regional nodes and time slices that meet the update conditions after influence propagation calculation and enter the incremental recalculation queue, respectively, and regional nodes and time slices that do not enter the queue but use the parent version through version inheritance index. For regional nodes and time slices that have a non-zero propagation coefficient with the disturbance source node but whose impact score is lower than the update threshold, the version inheritance index uses "impact score lower than the update threshold" as the inheritance reason, so that no new formal factor record is generated in this version of the target dataset. This is the design result of the impact propagation screening mechanism, and does not mean that the recalculated factor value is necessarily strictly the same as the parent version mathematically. The coverage and conservatism of this screening are jointly determined by the propagation order, the propagation truncation condition (see Formula 17 for the rough upper bound of the propagation coefficient tail term), the update threshold, and the labeling of related weights. When the regional node and time slice are subsequently affected by new disturbance events, hit by product calls, or meet new update conditions, they will still participate in the impact propagation judgment again and can enter the subsequent incremental recalculation queue.

[0085] It should be noted that the propagation coefficient in the propagation matrix is ​​affected. It is caused by the disturbance source node to candidate node The aggregate of all effective propagation contributions, i.e.: In the formula, Indicates time slice The partitioned power grid topology diagram consists of nodes To the node , length is A set of directed propagation sequences, which are directed paths in the topological part without closed loops, and may contain repeated nodes or topological edges in strongly connected components. This represents one of the propagation sequences; Represents the propagation sequence A topological edge on; The terms correspond to the identity matrix This refers to the impact of the disturbance source node on itself. Therefore, the propagation path record is not determined by... The solution is obtained through inverse calculation, not through step-by-step computation. Synchronously generated during the process: For example, the system uses perturbation source nodes Starting from the effective propagation order determined by Equation 7, the propagation is expanded step-by-step on the sparse partitioned power grid topology, contributing to the single line. Not less than the path record threshold The propagation sequence is recorded, including its node sequence, the topological edges traversed, edge weights, propagation order, direction confidence, and topological version, serving as propagation path entries. When multiple propagation sequences satisfying the conditions exist from the perturbation source node to the same candidate node, the first one is recorded in descending order of contribution. 100 propagation path entries, of which Reserves a preset number of paths and can simultaneously record the aggregated contribution value of parts not expanded path by path. Path recording threshold. Independent of the cutoff threshold used to stop higher-order propagation calculations in Equation 7. The propagation coefficient is used only to control the number of propagation path entries stored in the version lineage, and does not change the calculation results of the influence propagation matrix and influence score. Therefore, the propagation coefficient is used for the aggregation and screening of influence scores, and the propagation path entries are used for path-level tracing of version lineage. The two have different levels of function and do not replace each other. This application traces the factor version change back to the set of corresponding disturbance events and related propagation paths, rather than using the value of the propagation coefficient itself as the sole basis for determining a certain propagation path or a single root cause.

[0086] It should be noted that, as shown in Formula 5, the weighted adjacency matrix... The elements of the array are non-negative and the sum of the elements in each column does not exceed 1, therefore its powers are... All elements therein do not exceed 1, therefore in At that time, taking infinite propagation as a reference, exceeding the maximum propagation order. The propagation coefficient tail terms satisfy: Formula 17 is only used to give a rough upper bound for the tail terms of the propagation coefficient, and does not directly represent the upper bound for the calculation error of the time-of-use electricity carbon footprint factor or the product carbon footprint difference; those skilled in the art can set the maximum propagation order according to the degree of contribution of the propagation coefficient to be retained. With propagation cutoff threshold This is to adjust the conservatism of the spread range screening and the computational cost.

[0087] For example, for candidate nodes and time slices in the incremental recalculation queue, the system performs recalculation according to the topological dependency order of the corresponding time slice topology graph. For multiple nodes within a strongly connected component, joint iterative solution is performed as described in step 900. When an enqueued node, when calculated according to Formula 1, also depends on an upstream node affected by the propagation of the same disturbance event, the system can include the upstream node in the dependency closure of this recalculation, so that the upstream time-sharing electricity carbon footprint factor required for calculation is updated before the downstream enqueued node. The upstream node included in the dependency closure participates in the intermediate calculation of this task, and may not generate a separate official release version, or may be written into the new version and bound to the version lineage record when needed. For upstream nodes that are not covered by the propagation path of the disturbance event, or are confirmed by the preset dependency judgment not to need recalculation, their current valid version is used to participate in the calculation of Formula 1. Thus, the incremental recalculation queue is used to determine the regional nodes and time slices that need to generate a new official version, and the dependency closure is used to ensure the integrity of the calculation of the enqueued node. The two have different functions.

[0088] For example, when the same candidate node and time slices When affected by multiple independent disturbance events, if these disturbance events correspond to source fields of different types or dimensions, then before calculating the event-level impact score and performing cross-event comparisons or summations, the change amplitude is converted into a comparable dimensionless change amplitude according to Formula 3 or the preset benchmark value corresponding to each source field. Event-level impact scores that have not undergone unified dimension processing are not directly compared or summed across events. Let the set of disturbance events affecting the candidate node and time slice be . The system handles each disturbance event. The dimensionless variation magnitude of the event itself, and the perturbation source node of the event. to candidate node The propagation coefficient and the disturbance source mass weight of the event are used to calculate the event-level impact score according to Formula 8. The maximum value among them is used as the queue determination influence score for the candidate node and time slice: Alternatively, the sum of the impact scores of each event level can be used as the queue's impact score. The specific aggregation rule—whether to take the maximum value or summation—is pre-configured and recorded in the computational model version. A consistent aggregation rule is used within the same computational model version to ensure that the same input has a unique, recalcible result within the same computational model version. Correspondingly, the recalculation priority is calculated separately for each disturbance event according to Formula 10, and the maximum value is taken. In this case, the value in Formula 10... This refers to the quality weight of the disturbance source for the corresponding disturbance event, and its meaning is determined in the case of multiple events. The incremental recalculation queue retains all disturbance event identifiers, impact scores for each event level, and corresponding propagation path entries. The version lineage information of each disturbance event is not deleted during aggregation processing and is jointly bound to the version lineage record of the new version after recalculation. It should be noted that duplicate reported data with the same event deduplication key or revised versions of the same disturbance event do not constitute independent disturbance events. The system does not accumulate their impact scores but uses the latest event version after approval in the calculation to avoid the same physical disturbance being counted repeatedly.

[0089] For example, coverage weight This is used to reflect the external power supply status and product call coverage of candidate nodes, rather than to determine whether a disturbance can propagate to the candidate node in terms of the physical relationship of the power grid. To avoid candidate nodes with temporarily empty call records being directly set to zero during the screening process, a basic weight greater than zero and less than 1 can be set for candidate nodes that have been covered by the propagation path. Based on this, the adjustment component is determined according to the external power supply situation and product call situation. : Similarly, the perturbation source mass weight Used to adjust the triggering degree of low completeness, high latency, candidate completion values, or incompletely reviewed data on the incremental recalculation queue; when the quality of the disturbance source data is lower than the quality condition that allows direct release, the corresponding factor is written to the isolation area or a review task is generated according to step 900, without interpreting the lower quality weight as the disturbance having no physical impact.

[0090] For example, for topology breakage disturbances, the magnitude of the change can be taken as the difference in charge carried by the affected topology edge before and after the breakage, i.e. After the break, the charge on that edge is set to zero. When the state of a topology edge has changed from open to closed, but the measured power flow in the corresponding time slice is zero or missing, the state change can be mapped to a preset dimensionless state change amplitude, and the equivalent change amplitude can be determined by combining the rated capacity of the topology edge, historical typical power flow, or power flow in adjacent time slices. For disturbances such as regional boundary adjustments, the change amplitude can be taken as the charge received by the nodes in the region or the change in load affiliation before and after the boundary adjustment. For disturbances such as data quality status changes, the change amplitude can be taken as the difference between the source field before and after the quality status change, or the value can be taken according to the preset mapping relationship between quality status and equivalent amplitude. It should be noted that topology breakage and regional boundary adjustments will change the topology map of the regional power grid itself. In this case, the impact propagation calculation uses the weighted adjacency matrix corresponding to the topology version after the disturbance. Alternatively, the impact propagation results can be calculated for the two topology versions before and after the disturbance, and the larger of the two impact scores can be taken for the same candidate node and time slice to cover the case where the power receiving structure itself changes.

[0091] For example, the direction confidence is lower than the direction confirmation threshold. When generating the weighted adjacency matrix, candidate edges can be weighted according to their directional confidence, that is, the edges corresponding to each other can be weighted according to their directional confidence. The factor is multiplied by its directional confidence score before being used in the propagation calculation; alternatively, weighted adjacency matrices can be generated for each of the two candidate directions, and the influence propagation results can be calculated separately. For the same candidate node and time slice, the larger of the two influence scores is used for queuing. For candidate nodes and time slices that are hit through candidate edge propagation, the recalculated factor version has its deployment confidence score reduced accordingly.

[0092] For example, whenever a product's LCA calculation calls a time-of-use electricity carbon footprint factor for a node in a certain region under a certain time slice, the system writes a call record into the call lineage table. This call record includes the product call object identifier, call region, call time slice, the called factor version identifier, electricity consumption, and calculation result. The reverse lookup in step 800 is based on this call record. For historical calls that occurred before the call lineage table was enabled, they can be supplemented by data import interface according to the same field structure.

[0093] For example, when a product calling object has multiple calling records in its calling lineage, such as multiple records corresponding to different processes, different unit processes, different calling regions, or different time slices, the product carbon footprint difference is accumulated one by one for each calling record in the calling lineage. Let... Indicates the product call object The collection of bloodline records is invoked. This indicates one of the call records. , These represent the call region and call time slice recorded in the call log, respectively. This indicates that the new version of the region node-time slice set is written in this incremental update, then: In the formula, Indicates call record The recorded electricity consumption; Indicates call record The time-of-use electricity carbon footprint factor value corresponding to the recorded factor version is the factor before the update that was actually called in this record; This represents the time-of-use electricity carbon footprint factor value for the corresponding region and time slice after this update. This incremental update did not include the call records corresponding to the new version's calling region and time slice, and is therefore not included in the calculation of the product's carbon footprint difference. Formula 11 is a simplified expression for the case where each affected time slice corresponds to only one call record, and the factor before the update is the same as the factor version called by that record.

[0094] It should also be understood that the location and identification of product call objects are achieved through the correspondence between the product call object identifier, call region, call time slice, and factor version in the call lineage, rather than relying on the numerical value of the product carbon footprint difference. Different product call objects may have the same product carbon footprint difference value when their electricity consumption and the called factor are exactly the same, but they still generate recalculation tasks or archive records separately based on their respective product call object identifiers and call lineage records. Therefore, the term "hitting or locating a specific product call object" in this application refers to finding the specific database object that previously called the factor version before the update based on the call lineage, and does not mean solely based on the product carbon footprint difference value. The numerical value can be used to uniquely identify a physical product.

[0095] For example, the upgrade of the computational model version can be treated as a dataset maintenance trigger condition distinct from power grid operation data disturbance events: for regional nodes and time slices that need to be regenerated based on the new computational model, the system establishes an independent model migration and recalculation task according to the applicable scope of the new computational model, and records the original computational model version, the new computational model version, and the parent version identifier in the generated new factor version; the model migration and recalculation task does not rely on the disturbance source node and the influence propagation matrix for range filtering, nor does it change the processing flow of this application, which performs topology propagation filtering based on disturbance events when power grid operation data changes. Accordingly, the factor change referred to in this application can be traced back to the corresponding power grid physical disturbance and propagation path, and is aimed at factor changes triggered by disturbance events due to changes in power grid operation data; the factor version generated by the model migration and recalculation task is traced back through the recorded old and new computational model versions and the parent version identifier.

[0096] It should be noted that in this application, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. In this application, if it refers to performing an action according to an element, it means performing the action at least according to that element, including two cases: performing the action only according to that element, and performing the action according to that element and other elements. Expressions such as "multiple," "repeatedly," and "various" include two, two times, two kinds, and more than two, more than two times, and more than two kinds.

[0097] Furthermore, it should be understood that after reading the above disclosure of this application, those skilled in the art can make various alterations or modifications to this application, and these equivalent forms also fall within the scope of protection claimed in this application.

Claims

1. A method for dynamic updating of partitioned power carbon footprint dataset based on grid topology propagation model, characterized in that, include: The power grid operation data is acquired and grouped into time slices according to a preset time granularity. For each time slice, a partitioned power grid topology map is constructed with power grid nodes or regions as vertices and transmission lines or inter-regional exchange relationships as directed edges. Based on the partitioned power grid topology map and the power grid operation data, the time-sharing electricity carbon footprint factor of each regional node in the corresponding time slice is calculated. When the power grid operation data changes, a disturbance event is generated that includes the disturbance source node and the magnitude of the change; For the time slice affected by the disturbance event, a weighted adjacency matrix is ​​generated based on the power flow direction and power consumption ratio of the partitioned power grid topology. Based on the weighted adjacency matrix and the propagation attenuation coefficient, the propagation result of the disturbance source node on each candidate node is calculated. The influence score of each candidate node is calculated by combining the change magnitude and the influence propagation result, and candidate nodes and time slices with influence scores not less than the update threshold are written into the incremental recalculation queue. Only for candidate nodes and time slices in the incremental recalculation queue, the time-sharing electricity carbon footprint factor is recalculated in the same way as the time-sharing electricity carbon footprint factor. The new version is written and the version lineage record of the parent version, the disturbance event, the propagation path, and the impact score is bound. For candidate nodes and time slices not written to the incremental recalculation queue, the original version is used and the inheritance relationship is recorded. Based on the call lineage query, the product call object that has called the time-of-use electricity carbon footprint factor before the update is called, calculates the product carbon footprint difference based on the electricity consumption of the product call object and the time-of-use electricity carbon footprint factor before and after the update, and generates a recalculation task or archive record of the product call object based on the product carbon footprint difference.

2. The method of claim 1, wherein, The step of generating a weighted adjacency matrix based on the power flow direction and energy percentage of the partitioned power grid topology includes: Generation time slice weighted adjacency matrix whose elements are calculated as follows: In the formula, For time slices internal nodes To the node The input power is at the node The proportion of the total received power, i.e., the elements of the weighted adjacency matrix; For time slices internal nodes To node Input power; For nodes In time slice The sum of all received input charges, of which To the node Input the upstream node index of the power level; To avoid extremely small positive values ​​with a denominator of zero; When the direction of the trend is uncertain, candidate directions are generated based on the preceding and following time slices, the transaction plan, and the net exchange direction of the tie line, and the direction confidence is written.

3. The method according to claim 2, characterized in that, The calculation of the influence propagation result of the disturbance source node on each candidate node based on the weighted adjacency matrix and propagation attenuation coefficient includes: The maximum order calculated by time slice is Influence propagation matrix As a result of the aforementioned impact propagation: In the formula, For time slices The influence propagation matrix; It is the identity matrix; The propagation attenuation coefficient is mentioned above; For time slices The weighted adjacency matrix; For the weighted adjacency matrix Power; The maximum propagation order; When satisfied When the propagation calculation stops, The preset cutoff threshold, This is a function that takes the maximum value of the matrix elements. In calculating each order During the process, the node sequence and topological edges that the disturbance source node passes through to each candidate node are extracted as the propagation path.

4. The method according to claim 3, characterized in that, The influence score of each candidate node is calculated by combining the change magnitude and the influence propagation result, using the following formula: In the formula, Candidate nodes In time slice The impact score; For the disturbance source node In time slice The range of change; The influence propagation matrix The middle is composed of disturbance source nodes to candidate node The propagation coefficient; Candidate nodes The coverage weight is used to reflect whether the node is supplying power to external sources or being accessed by the product database. This is the quality weight for the disturbance source, used to reflect the data source and completeness of the disturbance source; When the same candidate node and time slice are affected by multiple disturbance events, its recalculation priority in the incremental recalculation queue is determined by the following formula: In the formula, Priority for recalculation; As a product hit factor, it is used when the candidate node and time slice match the published product call lineage. Otherwise take .

5. The method according to claim 1, characterized in that, It also includes deduplication: For the aforementioned disturbance event, an event deduplication key is generated based on the disturbance source node, source field, affected time window, and model version. Disturbance data with the same event deduplication key are merged without generating independent disturbance events repeatedly. When writing candidate nodes and time slices into the incremental recalculation queue, a queue deduplication key is generated based on the identifier of the disturbance event, the identifier of the regional node, the identifier of the time slice, and the model version. When there are incomplete recalculation tasks with the same queue deduplication key, the corresponding impact score and propagation path are merged in the incremental recalculation queue without repeated recalculation.

6. The method according to claim 1, characterized in that: The new version also records the solution status and the computational model version; The step of using the original version and recording the inheritance relationship for candidate nodes and time slices that are not written to the incremental recalculation queue includes: not copying the full factor data for the candidate node and time slice, but writing the inherited parent version identifier, target dataset version identifier and inheritance reason into the version inheritance index. The reasons for inheritance include at least one of the following: the impact score is lower than the update threshold, the call lineage is not hit, the impact propagation path is not covered, or the time slice is frozen.

7. The method according to claim 2, characterized in that, It also includes handling abnormal operating conditions: When Time Slice When the power flow of the connecting line is missing, candidate power flow values ​​are generated based on the cross-regional transaction plan and power flow data of adjacent time slices. : In the formula, For nodes To node In time slice Candidate power flow values; The current flow value given for cross-regional transaction plans; , These are the actual power flow values ​​for the previous and next time slices, respectively. The interpolation weights are determined based on the credibility of the transaction plan and the completeness of data in adjacent time slices. The topological edges with the aforementioned candidate power flow values ​​are used as candidate edges, and their direction confidence and distribution confidence are reduced. When there are closed-loop mutual supply nodes in the weighted adjacency matrix, identify the strongly connected components and iteratively solve the time-sharing power carbon footprint factor within the strongly connected components; When the number of iterations reaches the maximum number of iterations and the factor difference between two consecutive iterations is still not less than the convergence threshold, it is marked as a low confidence state. The generated time-sharing electricity carbon footprint factor is written into the isolation area and a review task is triggered, without directly overwriting the published valid version.

8. The method according to claim 1, characterized in that, The calculation of the product carbon footprint difference based on the electricity consumption of the product call object and the time-of-use electricity carbon footprint factor before and after the update includes: Calculate the product carbon footprint difference using the following formula : In the formula, For product call object Differences in product carbon footprint; For product call object In the affected time slice Electricity consumption; For the corresponding area In time slice Updated time-of-use electricity carbon footprint factor; For the corresponding area In time slice The time-of-use electricity carbon footprint factor before the update; when When the product call object's dataset status is set to require recalculation, the recalculation task is generated; when the ratio is less than the product difference threshold, it is set to low-impact archiving, and the archived record is generated; where... The original total electricity carbon footprint of the product being invoked. The preset product difference threshold; when When the value is zero or less than a preset minimum positive value, the relative difference judgment is not performed, but rather based on... Whether the absolute difference threshold is reached determines whether the recalculation task, the archived record, or the review task is generated.

9. A dynamic update system for a partitioned electricity carbon footprint dataset based on a power grid topology propagation model, characterized in that, Includes modules corresponding to the method described in any one of claims 1 to 8: The data access and factor calculation module is used to acquire power grid operation data and group it into time slices according to a preset time granularity. For each time slice, it constructs a partitioned power grid topology map with power grid nodes or regions as vertices and transmission lines or inter-regional exchange relationships as directed edges. Based on the partitioned power grid topology map and the power grid operation data, it calculates the time-sharing electricity carbon footprint factor of each regional node in the corresponding time slice. A disturbance identification module is used to generate a disturbance event containing the disturbance source node and the magnitude of the change when the power grid operation data changes; The influence propagation module is used to generate a weighted adjacency matrix based on the power flow direction and power consumption ratio of the partitioned power grid topology for the time slice affected by the disturbance event, and to calculate the influence propagation result of the disturbance source node on each candidate node based on the weighted adjacency matrix and the propagation attenuation coefficient. The influence score and queue module is used to calculate the influence score of each candidate node by combining the change magnitude and the influence propagation result, and write the candidate nodes and time slices with influence scores not less than the update threshold into the incremental recalculation queue. The incremental recalculation and lineage writing module is used to recalculate the time-sharing electricity carbon footprint factor only for candidate nodes and time slices in the incremental recalculation queue, write the new version and bind the version lineage record of the parent version, the disturbance event, the propagation path and the impact score, and retain the original version for candidate nodes and time slices not written to the incremental recalculation queue and record the inheritance relationship. The product call linkage module is used to query the product call object with the time-of-use electricity carbon footprint factor before the update based on the call lineage, calculate the product carbon footprint difference based on the electricity consumption of the product call object and the time-of-use electricity carbon footprint factor before and after the update, and generate a recalculation task or archive record of the product call object based on the product carbon footprint difference.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 8.

Images