Zero-carbon industrial park full life-cycle carbon footprint tracking and accounting methods and systems

Abstract

This application discloses a method and system for tracking and calculating the carbon footprint of a zero-carbon industrial park throughout its entire lifecycle, belonging to the field of carbon emissions. This application is applicable to zero-carbon industrial parks with multi-energy coupling power supply from photovoltaics, energy storage, and grid power. By acquiring spatial topology data, real-time power data, and carbon intensity factors of each energy supply unit, an energy network topology is constructed, defining energy-related units as nodes and physical lines as weighted directed edges. A dynamic energy flow digital model is then generated by combining this with real-time power data. This model is then input into a graph neural network to calculate an energy flow direction matrix representing the power allocation ratio from each energy supply unit to the energy consumption unit. Finally, based on this matrix, the carbon intensity factors of each energy supply unit are converted into source-end carbon intensity vectors and allocated to each energy consumption unit. Combined with real-time power data, the carbon footprint value is calculated, enabling accurate tracking and calculation of the carbon footprint of end-users in the zero-carbon industrial park.

Description

Technical Field

[0001] This application pertains to the field of carbon emissions, and in particular relates to a method and system for tracking and calculating the carbon footprint of a zero-carbon industrial park throughout its entire life cycle. Background Technology

[0002] As a crucial vehicle for achieving energy conservation and carbon reduction, zero-carbon industrial parks have made carbon footprint tracking and accounting essential. Accurate carbon footprint accounting methods are not only a scientific basis for assessing the park's green and low-carbon level and managing carbon assets, but also a key technological support for promoting green electricity trading within the park, guiding users to conserve energy and reduce carbon emissions, and achieving optimized energy system scheduling, thus possessing broad application prospects.

[0003] In existing technologies, the carbon footprint accounting method for zero-carbon parks usually adopts a macro-level measurement method based on total energy consumption. This method calculates the carbon emissions of electricity used in the entire park by obtaining the total electricity purchased by the power grid company and the average carbon emission factor of the regional power grid.

[0004] However, when existing technologies are applied to modern zero-carbon parks where multiple energy sources such as photovoltaics and energy storage are deeply integrated with the public power grid, they fail to track the specific flow and distribution ratio of the mixed electricity along various transmission and distribution paths within the park's internal power grid. This results in insufficient accuracy in calculating the carbon footprint of end users. Therefore, existing technologies suffer from the technical problem of being unable to accurately track and calculate the carbon footprint of end users in zero-carbon parks. Summary of the Invention

[0005] This application provides a method, system, equipment, and computer storage medium for tracking and calculating the carbon footprint of end users in a zero-carbon park throughout its entire life cycle, which can accurately track and calculate the carbon footprint of end users in a zero-carbon park.

[0006] Firstly, this application provides a method for tracking and calculating the carbon footprint of a zero-carbon industrial park throughout its entire life cycle, applicable to zero-carbon industrial parks powered by multi-energy coupling. The method includes:

[0007] Acquire spatial topology data of energy-related units within the zero-carbon park, real-time power data of energy-related units, and carbon intensity factor of each energy supply unit. Each energy-related unit includes at least one energy supply unit and at least one energy consumption unit. The spatial topology data includes the physical parameters of the physical lines between energy-related units.

[0008] Based on spatial topology data, energy-related units are defined as nodes, physical lines connecting nodes are defined as directed edges, and physical parameters of physical lines are defined as weights of directed edges, thus constructing an energy network topology graph.

[0009] Real-time power data is mapped to the node attributes of the corresponding nodes in the energy network topology diagram to generate a digital energy flow model that represents the energy flow status of the zero-carbon park.

[0010] The energy flow digital model is input into the trained graph neural network. By calculating the power distribution relationship between nodes, the energy flow direction matrix is ​​obtained. The row dimension of the energy flow direction matrix corresponds to the number of energy supply units, the column dimension corresponds to the number of application energy units, and the matrix elements represent the power distribution ratio coefficient.

[0011] The carbon intensity factor of each energy supply unit is converted into a source-end carbon intensity vector. Through the energy flow matrix, each source-end carbon intensity vector is allocated to the energy consumption unit. Based on the real-time power data of the energy consumption unit, the carbon footprint value of the energy consumption unit is calculated.

[0012] In one feasible implementation, the energy supply unit includes a photovoltaic power generation unit, an energy storage unit, and a public grid unit.

[0013] In one feasible implementation, the method further includes:

[0014] When the real-time power data of the energy storage unit corresponds to the charging state, the energy storage unit is identified as the energy consumption unit.

[0015] Based on the energy flow matrix, the source carbon intensity vector allocated to the energy storage unit is calculated, and based on the real-time power data of the energy storage unit, the charging capacity of the energy storage unit and the corresponding carbon attribute data block are calculated.

[0016] A first-in-first-out carbon accounting queue is established for the energy storage unit, and the carbon attribute data blocks of the energy storage unit are stored in the tail of the carbon accounting queue in chronological order. When the energy storage unit is the power supply unit, the carbon attribute data block corresponding to the target discharge capacity of the energy storage unit is extracted from the head of the carbon accounting queue according to the target discharge capacity of the energy storage unit, so as to generate the carbon intensity factor of the energy storage unit.

[0017] In one feasible implementation, the method further includes:

[0018] Acquire grid synchronization phasor data for target energy-related units within the zero-carbon industrial park;

[0019] The power grid synchronization phasor data is mapped to the node attributes of the corresponding nodes in the energy network topology diagram, and the energy flow digital model is updated.

[0020] In one feasible implementation, the energy flow digital model is input into a trained graph neural network, and the energy flow direction matrix is ​​obtained by calculating the power distribution relationship between nodes, including:

[0021] The energy flow digital model is input into a trained graph neural network to calculate the initial power distribution relationship between nodes. The initial power distribution relationship is then corrected using grid synchronization phasor data as physical constraints to obtain the energy flow direction matrix.

[0022] In one feasible implementation, the energy flow digital model is input into a trained graph neural network to calculate the initial power allocation relationship between nodes. The initial power allocation relationship is then corrected using grid synchronization phasor data as physical constraints to obtain the energy flow direction matrix, including:

[0023] Based on the node attributes of each node in the energy flow digital model and the weights of the directed edges connecting the nodes, the initial power allocation relationship between each node is determined using a trained graph neural network.

[0024] Based on the grid synchronization phasor data of the target energy-related units, calculate the voltage magnitude and phase angle difference between the two target nodes corresponding to the two target energy-related units, and combine the weight of the edge between the two target nodes to determine the verification power allocation relationship including the power magnitude and flow direction between the two target nodes.

[0025] By comparing the initial target power allocation relationship between two target nodes with the verified power allocation relationship, a correction amount for adjusting the initial target power allocation relationship between target nodes is determined.

[0026] By using the allocation relationship correction, the target initial power allocation relationship between two target nodes is corrected, and all initial power allocation relationships are updated using the corrected target initial power allocation relationship to obtain the energy flow matrix.

[0027] In one feasible implementation, the carbon intensity factor of each energy supply unit is converted into a source-end carbon intensity vector. This source-end carbon intensity vector is then allocated to the energy-consuming units using an energy flow matrix. Finally, based on the real-time power data of the energy-consuming units, the carbon footprint of each unit is calculated, including:

[0028] Arrange the carbon intensity factors of each energy supply unit according to the row order in the energy flow matrix to form the source carbon intensity vector;

[0029] The carbon intensity vector at the source end is multiplied by the energy flow direction matrix to obtain the carbon attribute vector of the energy-consuming unit. Each element of the carbon attribute vector of the energy-consuming unit represents the unit power carbon attribute value allocated to the corresponding energy-consuming unit in the energy flow digital model.

[0030] The carbon footprint of the energy-consuming unit is obtained by performing a scalar multiplication operation between the active power component in the real-time power data of the energy-consuming unit and the corresponding element in the carbon attribute vector of the energy-consuming unit.

[0031] Secondly, this application provides a zero-carbon industrial park full life-cycle carbon footprint tracking and accounting system, applicable to zero-carbon industrial parks powered by multi-energy coupling, the system comprising:

[0032] The acquisition module is used to acquire spatial topology data of energy-related units in the zero-carbon park, real-time power data of energy-related units, and carbon intensity factor of each energy supply unit. The energy-related units include at least one energy supply unit and at least one energy consumption unit. The spatial topology data includes the physical parameters of the physical lines between energy-related units.

[0033] The module is used to construct an energy network topology graph based on spatial topology data, defining energy-related units as nodes, physical lines connecting nodes as directed edges, and physical parameters of physical lines as weights of directed edges.

[0034] The generation module is used to map real-time power data to the node attributes of the corresponding nodes in the energy network topology diagram, and generate a digital energy flow model that represents the energy flow status of the zero-carbon park.

[0035] The generation module is also used to input the energy flow digital model into the trained graph neural network, and obtain the energy flow direction matrix by calculating the power distribution relationship between nodes. The row dimension of the energy flow direction matrix corresponds to the number of energy supply units, the column dimension corresponds to the number of application energy units, and the matrix elements represent the power distribution ratio coefficient.

[0036] The calculation module is used to convert the carbon intensity factor of each energy supply unit into a source-end carbon intensity vector, allocate each source-end carbon intensity vector to the energy consumption unit through the energy flow matrix, and calculate the carbon footprint value of the energy consumption unit based on the real-time power data of the energy consumption unit.

[0037] Thirdly, this application provides an electronic device, the device including: a processor, and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement the zero-carbon park full life cycle carbon footprint tracking and accounting method as described in any embodiment of the first aspect.

[0038] Fourthly, this application provides a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the zero-carbon park full life-cycle carbon footprint tracking and accounting method as described in any embodiment of the first aspect.

[0039] This application discloses a method, system, equipment, and computer storage medium for tracking and calculating the carbon footprint of a zero-carbon industrial park throughout its entire life cycle. It accurately models the physical connections of a multi-energy coupled power grid by constructing a weighted energy network topology graph. Combined with real-time power data, it generates a dynamic energy flow digital model, accurately simulating the energy transmission path in a deeply hybrid scenario of photovoltaics, energy storage, and grid power. Furthermore, it utilizes a graph neural network to solve the nonlinear power distribution relationship between nodes, generating an energy flow matrix to accurately quantify the dynamic power allocation ratio from energy supply units to energy consumption units. Based on this ratio, the real-time carbon intensity factor is losslessly transmitted to the terminal through matrix operations. This solves the problem of carbon footprint calculation distortion caused by the inability to track hybrid power transmission and distribution paths in existing technologies, achieving high-precision dynamic traceability and calculation of the terminal-level carbon footprint in the complex power grid scenario of a zero-carbon industrial park.

[0040] Furthermore, by acquiring grid synchronization phasor data as high-precision physical measurements, a definite physical verification benchmark is provided for model calculations. First, these physical measurements are used to calculate the actual power flow direction and magnitude on the critical path. Then, based on this, the initial allocation relationship derived by the graph neural network is quantitatively corrected, ensuring that the final generated energy flow matrix is ​​consistent with the physical operating state of the grid. Therefore, by introducing strong physical constraints, the accuracy of end-user carbon footprint tracking and accounting is further enhanced. Attached Figure Description

[0041] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a flowchart illustrating a method for tracking and calculating the carbon footprint of a zero-carbon industrial park throughout its entire life cycle, provided in one embodiment of this application.

[0043] Figure 2 This is a flowchart illustrating a method for determining the carbon intensity factor of an energy storage unit according to an embodiment of this application;

[0044] Figure 3 This is a flowchart illustrating a method for generating an energy flow direction matrix according to an embodiment of this application;

[0045] Figure 4 This is a schematic diagram of the structure of a zero-carbon park full life-cycle carbon footprint tracking and accounting system provided in one embodiment of this application;

[0046] Figure 5 This is a schematic diagram of the hardware structure of an electronic device provided in one embodiment of this application. Detailed Implementation

[0047] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0048] It should be noted that, in this document, 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..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0049] However, when existing technologies are applied to modern zero-carbon parks where multiple energy sources such as photovoltaics and energy storage are deeply integrated with the public power grid, they fail to track the specific flow and distribution ratio of the mixed electricity along various transmission and distribution paths within the park's internal power grid. This results in insufficient accuracy in calculating the carbon footprint of end users. In particular, the inability to distinguish the differences in the proportion of low-carbon electricity consumed by different electricity users leads to insufficient fairness and accuracy in the calculation results, making it difficult to implement differentiated carbon management and incentive strategies. Therefore, existing technologies suffer from the technical problem of failing to accurately track and calculate the carbon footprint of end users in zero-carbon parks.

[0050] To address the problems of existing technologies, embodiments of this application provide a method, system, device, and computer storage medium for tracking and calculating the carbon footprint of a zero-carbon industrial park throughout its entire life cycle. The method for tracking and calculating the carbon footprint of a zero-carbon industrial park throughout its entire life cycle, as provided in this application embodiment, will be described first below.

[0051] Figure 1 This illustration shows a flowchart of a method for tracking and calculating the carbon footprint of a zero-carbon industrial park throughout its entire lifecycle, according to an embodiment of this application. This method is applicable to zero-carbon industrial parks powered by multi-energy coupling, such as... Figure 1 As shown, the method includes steps S110 to S150.

[0052] S110: Acquire spatial topology data of energy-related units within the zero-carbon park, real-time power data of energy-related units, and carbon intensity factor of each energy supply unit. Energy-related units include at least one energy supply unit and at least one energy consumption unit. Spatial topology data includes physical parameters of physical lines between energy-related units.

[0053] A zero-carbon park powered by multi-energy coupling refers to a specific functional area that integrates multiple energy supply methods within a specific geographical region and coordinates power supply through an internal power grid, aiming to achieve net-zero carbon emissions. Energy-related units are the collective term for all physical entities within the park related to the generation, transmission, conversion, storage, and consumption of energy. They are the basic elements constituting the energy network and can include photovoltaic power generation arrays, energy storage power stations, the park's main substation, various levels of distribution boxes, electric vehicle charging stations, and buildings. Energy supply units and energy consumption units are dynamic classifications of the functional roles of energy-related units at specific times. Energy supply units refer to energy-related units that output power to the network as energy sources; these can include photovoltaic power generation units, energy storage units in a discharging state, and public power grid units. Energy consumption units refer to energy-related units that are the final stage of energy consumption; these can include buildings, production equipment, charging stations, and energy storage units in a charging state.

[0054] Spatial topology data refers to a dataset describing the location and physical connections of various energy-related units in three-dimensional space. The physical parameters it includes refer to the inherent physical properties of the connecting lines, such as cable material, length, wire diameter, and theoretical impedance. Real-time power data refers to the electrical energy operation status data collected in real time at each energy-related unit by sensors. This can include active power, reactive power, voltage, current, and, for energy storage units, the charging and discharging status. The carbon intensity factor is a quantitative indicator used to characterize the carbon emissions contained in a unit of electrical energy. For photovoltaic power generation units, the carbon intensity factor is a fixed value of zero or close to zero. For public grid units, the carbon intensity factor changes in real time. For energy storage units, the carbon intensity factor when acting as an energy supplier is dynamically determined by the historical sources of their charging power.

[0055] Spatial topology data can be acquired by parsing Building Information Modeling (BIM) data or Geographic Information System (GIS) data from the park's construction phase, combined with on-site surveys and engineering drawings. Data processing extracts the precise coordinates of all energy-related units and the physical parameters of connecting cables. Real-time power data acquisition relies on smart meters, power sensors, or microgrid energy management systems deployed on various energy-related units within the park. These devices continuously upload the collected power operation data streams to the data processing platform via industrial IoT communication protocols at a preset time frequency, such as once per minute. Carbon intensity factors are acquired in diverse ways. Carbon intensity factors for public grid units are obtained by calling real-time application programming interfaces (APIs) released by external authoritative institutions or grid companies, while carbon intensity factors for energy supply units such as photovoltaic power generation units within the park are preset in the system according to their technology type. In some specific implementations, to achieve higher-precision power distribution relationship correction later, grid synchronization phasor data deployed at key target energy-related units can also be acquired simultaneously. Finally, these three types of data are integrated as the data inputs required for subsequent steps, laying the foundation for building a dynamic and accurate energy network model.

[0056] First, the location information of all energy-related units, such as photovoltaic power generation arrays, energy storage power stations, and buildings, is extracted using the building information model (BIM) of the zero-carbon park. Physical parameters of the physical lines connecting these units, such as cable material, length, and theoretical impedance, are also extracted. This information collectively constitutes the spatial topology data of the zero-carbon park. Using smart meters and power sensors deployed on each energy-related unit, operational status information is uploaded at a fixed frequency of once per minute via an Industrial Internet of Things (IIoT) communication protocol. This information includes the active and reactive power of each unit. For energy storage units, their charging and discharging status is also obtained. These uploaded data streams together form real-time power data. Simultaneously, the carbon intensity factor of the public power grid, which serves as the energy supply unit, is obtained from relevant institutions. For power supply units such as photovoltaic power generation units, their carbon intensity factor is pre-set based on their zero-emission technology type. For an energy storage unit, its carbon intensity factor when acting as an energy supply unit is dynamically determined by the composition of its historical charging power sources and the corresponding carbon intensity factor. For example, based on the proportion of electricity absorbed by the energy storage unit from the photovoltaic power generation unit and the public grid unit during the charging period, the carbon intensity factors of the two energy supply units are calculated by weighted average, and the result is the carbon intensity factor of the energy storage unit during this discharge.

[0057] For example, a 5 MW rooftop photovoltaic power generation system and a 2 MW / 4 MWh containerized lithium battery energy storage power station are built in a zero-carbon smart industrial park, connected to the municipal power grid via a 10 kV dedicated line. The main energy-related units in the park include the grid-connected inverter for the photovoltaic power generation system, the bidirectional converter PCS for the energy storage power station, the park's 10 kV main substation, the power distribution room of Building A office building, the power distribution cabinet of Building B production workshop, and three 120 kW DC fast charging piles in the parking lot.

[0058] First, by parsing and calling the pre-imported as-built BIM 3D model of the park, the geographical coordinates of all the energy-related units in the park and the models of all the power cables connecting them were obtained. Based on the cable model and laying length, the theoretical impedance value was calculated to obtain a complete spatial topology data.

[0059] Meanwhile, utilizing multi-functional smart energy meters and sensor arrays deployed at key measuring points in the aforementioned energy-related units, real-time power data is transmitted to the data server of the park's energy management system at a frequency of once per minute via the Modbus-TCP industrial Ethernet protocol. The data server received the following data: the active power output of the photovoltaic grid-connected inverter was 2.0 MW; the active power of the bidirectional converter PCS of the energy storage power station was 0.5 MW, with its status indicating a discharge state; the metering gate meter of the park's main substation showed an active power purchased from the public grid of 1.1 MW; the active power of the main incoming line of Building A office building was 1.5 MW; the active power of the main incoming line of Building B production workshop was 2 MW; and the total active power of the three charging piles was 0.1 MW. The server requested and obtained the real-time carbon intensity factor of the regional power grid at the current moment through the power grid data service platform, which was 0.58 kg CO2 per kilowatt-hour. Based on this real-time data, the photovoltaic power generation unit, the public grid unit, and the energy storage power station are currently acting as energy supply units, while the office building A, the production workshop B, and the charging piles are all acting as energy consumption units.

[0060] S120: Based on spatial topology data, energy-related units are defined as nodes, physical lines connecting nodes are defined as directed edges, and physical parameters of physical lines are defined as weights of directed edges, thus constructing an energy network topology graph.

[0061] An energy network topology graph is a data model that uses graph theory to represent the physical connections of an energy system. It consists of three core elements: nodes, directed edges, and weights. A node represents an independent energy-involved unit, such as a photovoltaic power generation array or a building. A directed edge represents a physical line connecting two nodes, such as a cable running from a substation to a distribution box; its direction indicates the main direction of energy flow in the design. A weight is one or more values ​​assigned to a directed edge to quantify the physical parameters of that line, such as the theoretical impedance of the cable. The magnitude of the weight directly reflects the physical impact on energy as it travels along that path.

[0062] First, for each energy-related unit in the spatial topology data, such as photovoltaic power generation arrays, energy storage power stations, and buildings, a corresponding node is created in the data structure. Then, based on the physical line connections recorded in the spatial topology data, a directed edge is established between the corresponding two nodes. The direction of the edge is determined according to the normal power flow design direction of the power grid or the master-slave relationship, for example, from the main substation node of the park to the distribution room node of the office building. Finally, the physical parameters of each physical line in the spatial topology data, such as the theoretical impedance value calculated using cable material, length, and diameter, are assigned as numerical values ​​to the weight attributes of the corresponding directed edges. After processing all energy-related units and physical lines, the energy network topology map reflecting the static physical characteristics of the park's energy network is constructed.

[0063] For example, firstly, for each energy-related unit identified in the spatial topology data, such as the grid-connected inverter of the photovoltaic power generation system, the bidirectional converter PCS of the energy storage power station, the 10 kV main substation of the park, the power distribution room of Building A office building, the power distribution cabinet of Building B production workshop, and the three charging piles, a corresponding node is created in the data structure. Then, based on the physical line connection relationships recorded in the spatial topology data, a directed edge is established between the corresponding two nodes. For example, based on the wiring diagram information, a directed edge is established between the 10 kV main substation node and the power distribution room node of Building A office building, with its direction determined according to the design principle of power flowing from the main station to the branch station, pointing from the main substation to the power distribution room of the office building. Similarly, a directed edge is also established between the power distribution room node of Building A office building and one of the charging pile nodes.

[0064] Finally, the physical parameters of each physical line in the spatial topology data are assigned as numerical values ​​to the weight attributes of the corresponding directed edges. For example, the theoretical impedance value of the cable between the main substation and the power distribution room of Building A is 0.05 + 0.02 ohms, which is used as the weight of the directed edge connecting these two nodes. After processing all energy-related units and physical lines, the energy network topology is completed.

[0065] S130: Map real-time power data to the node attributes of the corresponding nodes in the energy network topology diagram to generate a digital energy flow model that characterizes the energy flow status of the zero-carbon park.

[0066] A digital energy flow model refers to adding the current real-time power data of each node as a node attribute to the static skeleton of the energy network topology. This model not only includes the physical connections and line parameters of the park's energy network, but also synchronously reflects the power generation, consumption, or transmission status of each energy-related unit, thus dynamically representing the true energy flow status of the entire zero-carbon park.

[0067] First, a data mapping relationship is established, associating each data source in the acquired real-time power data stream, such as the unique identifier of a smart meter, with a specific node in the energy network topology, such as the node representing the energy-related unit where the smart meter is located. Based on the mapping relationship, the latest active power, reactive power, and charging / discharging status indicators of each node are updated in the node attributes of its corresponding node.

[0068] For example, based on the established mapping relationship, the active power value at the output of the photovoltaic grid-connected inverter, i.e., 2.0 MW, is updated in the node attributes of the node representing the photovoltaic power generation system grid-connected inverter in the energy network topology diagram. The active power value of 0.5 MW and the discharge state identifier are updated in the node attributes of the node representing the energy storage power station; the active power value of 1.1 MW is updated in the node attributes of the node representing the park's main substation; and the power values ​​of 1.5 MW, 2.0 MW, and 0.1 MW are updated in the node attributes of the nodes representing office building A, production workshop B, and three charging piles, respectively. Finally, an energy flow digital model containing the current power state of each node is constructed.

[0069] S140: Input the energy flow digital model into the trained graph neural network, and obtain the energy flow matrix by calculating the power distribution relationship between nodes. The row dimension of the energy flow matrix corresponds to the number of energy supply units, the column dimension corresponds to the number of application energy units, and the matrix elements represent the power distribution ratio coefficient.

[0070] Graph Neural Networks (GNNs) are artificial intelligence models used to process graph-structured data. They learn the representation of nodes in a graph by aggregating information from neighboring nodes, effectively capturing the complex dependencies between the graph's topology and node features. Power allocation between nodes refers to the power exchange between any two directly connected nodes, determined by the weight of their connecting edges and their respective node attributes. For example, a distribution box node acting as an intermediate transmission node will distribute the total power received from the upstream main substation node according to the different line impedances connecting the downstream office building node A and the production workshop node B, according to a certain proportion. All these local power exchanges interact to constitute the end-to-end power allocation relationship from all energy supply units to all energy consumption units in the entire energy network. An energy flow matrix is ​​a two-dimensional data structure used to quantitatively represent power allocation relationships. The rows of this matrix correspond to all energy supply units within the park, and the columns correspond to all energy consumption units. Each element in the matrix, i.e., a power allocation ratio coefficient, represents the proportion of power generated by a specific energy supply unit that flows to a specific energy consumption unit. The ratio coefficient ranges from 0 to 1.

[0071] A pre-built graph neural network (Graph Neural Network) model is trained using a large amount of historical energy flow digital model data as training samples. During training, the Graph Neural Network learns to map the input energy flow digital model—a graph structure containing node power attributes and edge weights—to a baseline end-to-end power allocation relationship. This baseline end-to-end power allocation relationship can be pre-calculated using power grid power flow simulation software. The trained Graph Neural Network then possesses the ability to infer power allocation relationships. The latest generated energy flow digital model is fed into this trained Graph Neural Network. Through its internal multi-layer graph convolution operations, the Graph Neural Network iteratively calculates the local power exchange between any two directly connected nodes based on the weights of their connecting edges and their respective node attributes. Through the stacking of multiple layers of computation, this local information is propagated and aggregated throughout the network, enabling the Graph Neural Network to ultimately understand and calculate the end-to-end power allocation relationship from all energy supply units to all energy consumption units, and finally output this relationship as an energy flow matrix.

[0072] S150: Convert the carbon intensity factor of each power supply unit into a source-end carbon intensity vector, allocate each source-end carbon intensity vector to the power consumption unit through the energy flow matrix, and calculate the carbon footprint value of the power consumption unit based on the real-time power data of the power consumption unit.

[0073] The source-end carbon intensity vector is constructed by arranging the carbon intensity factors of all energy-supplying units within the park at the current moment in the same order as the row dimensions of the energy flow matrix, thus centrally representing the source carbon attributes of all electricity entering the park's energy network. The carbon footprint value is a quantitative figure used to characterize the total greenhouse gas emissions directly or indirectly generated by a single energy-consuming unit within a specific time period due to electricity consumption; its unit is typically kilograms of carbon dioxide equivalent.

[0074] First, the carbon intensity factor of each energy supply unit is organized into a source-end carbon intensity vector according to its row order in the energy flow matrix. Then, through a matrix multiplication operation, this source-end carbon intensity vector is multiplied by the energy flow matrix; that is, the carbon intensity of each source end is weighted and summed according to the power allocation ratio coefficient, resulting in a carbon attribute vector for each energy-consuming unit. Each element in this vector represents the comprehensive carbon attribute of the unit's power consumption per unit of electricity. Finally, each element in this carbon attribute vector is multiplied by the active power component in the real-time power data of the corresponding energy-consuming unit and the calculation time period to calculate the carbon footprint value of the energy-consuming unit within that time period.

[0075] This embodiment accurately models the physical connections of multi-energy coupled power grids by constructing a weighted energy network topology graph. It generates a dynamic energy flow digital model by combining real-time power data, accurately simulating the energy transmission path in a deeply hybrid scenario of photovoltaics, energy storage, and grid power. Furthermore, it uses a graph neural network to solve the nonlinear power distribution relationship between nodes to generate an energy flow matrix, accurately quantifying the dynamic power distribution ratio from the energy supply unit to the energy consumption unit. Based on this ratio, the real-time carbon intensity factor is losslessly transmitted to the terminal through matrix operations, solving the problem of carbon footprint accounting distortion caused by the inability to track hybrid power transmission and distribution paths in existing technologies. This enables high-precision dynamic traceability and accounting of the terminal-level carbon footprint in complex power grid scenarios in zero-carbon parks.

[0076] In one feasible implementation, the energy supply unit includes a photovoltaic power generation unit, an energy storage unit, and a public grid unit.

[0077] Photovoltaic power generation units refer to power generation facilities that directly convert solar energy into electrical energy. Within the park, these can be rooftop photovoltaic arrays or photovoltaic carports, and the electricity they output is generally considered zero-carbon or low-carbon energy. Energy storage units are devices or systems that can store electrical energy and release it when needed. Within the park, these can be containerized lithium battery energy storage power stations, which can function as energy-consuming units during charging and as energy-supplying units during discharging. Public grid units refer to the gateway connecting the park to the external power grid. The park can purchase electricity from the external grid to supplement local power generation. The carbon intensity factor of this electricity is determined by the overall power generation structure of the regional power grid.

[0078] Figure 2 A flowchart illustrating a method for determining the carbon intensity factor of an energy storage unit according to an embodiment of this application is shown. Figure 2 As shown, the method includes steps S210 to S230.

[0079] In one feasible implementation, the method further includes:

[0080] S210: When the charging / discharging state corresponding to the real-time power data of the energy storage unit is the charging state, the energy storage unit is identified as the energy consumption unit.

[0081] The system monitors the real-time power data of the energy storage unit and checks the charge / discharge status indicators contained in the data. When an indicator clearly indicates that the current state is charging, the energy storage unit is classified as an energy-consuming unit within the current calculation cycle.

[0082] S220: Calculate the source-end carbon intensity vector allocated to the energy storage unit based on the energy flow matrix, and calculate the charging capacity and corresponding carbon attribute data block of the energy storage unit based on the real-time power data of the energy storage unit.

[0083] A carbon attribute data block is a structured data unit used to record the amount of electricity in a batch of charging and its corresponding calculated composite carbon intensity value, which reflects the mixing ratio of different energy supply units in that batch of electricity.

[0084] The column vectors corresponding to the energy storage units are extracted from the matrix. Each element in this column vector, i.e., the power allocation ratio coefficient, represents the proportion of electricity charged into the energy storage unit that comes from each energy supply unit. This column vector is then multiplied by the constructed source-side carbon intensity vector, i.e., a weighted summation of the carbon intensity factors of each energy supply unit, to obtain a carbon intensity value representing the overall attribute of the charging power for that batch. Simultaneously, based on the active power component in the real-time power data of the energy storage unit and the calculation time period, the charging capacity for that period is calculated. Finally, the calculated charging capacity and the overall carbon intensity value are combined to construct a new carbon attribute data block.

[0085] S230: Establish a first-in-first-out carbon accounting queue for the energy storage unit, and store the carbon attribute data block of the energy storage unit into the tail of the carbon accounting queue in chronological order. When the energy storage unit is the power supply unit, the carbon attribute data block corresponding to the target discharge capacity of the energy storage unit is extracted from the head of the carbon accounting queue according to the target discharge capacity of the energy storage unit to generate the carbon intensity factor of the energy storage unit.

[0086] A carbon accounting queue is a digital data queue established for each energy storage unit, following a first-in, first-out (FIFO) principle. It stores a series of carbon attribute data blocks in chronological order, acting like a digital ledger that records the history of each charging cycle of the energy storage unit. The target discharge capacity refers to the total electrical energy that the energy storage unit plans to release or actually releases to the grid within the current calculation cycle.

[0087] A dedicated carbon accounting queue is created for each energy storage unit within the zero-carbon park. Whenever a new carbon attribute data block is generated, it is added to the tail of the corresponding energy storage unit's carbon accounting queue. This addition process is continuous, and the queue grows as the energy storage units are charged. When it is necessary to determine the carbon intensity factor of an energy storage unit in a discharging state, it extracts one or more carbon attribute data blocks sequentially from the head of its carbon accounting queue, based on the target discharge capacity of that energy storage unit, until the total capacity contained in the extracted carbon attribute data blocks meets or exceeds the target discharge capacity. Subsequently, the carbon intensity of these extracted data blocks is weighted and averaged according to the proportion of their respective capacities to the target discharge capacity, ultimately generating a carbon intensity factor for the energy storage unit that accurately reflects the true carbon attributes of the discharged power.

[0088] For example, at a certain moment, the photovoltaic power generation of the zero-carbon smart industrial park is sufficient, and the energy storage station begins charging with a charging power of 1 MW. By monitoring the real-time power data of the energy storage station, its charging / discharging state is identified as charging. Therefore, within the current calculation cycle, the energy storage unit is identified as an energy-consuming unit. The latest energy flow matrix is ​​retrieved, and the column vector corresponding to the energy storage unit is extracted. This vector shows that 80% of the electricity being charged into the energy storage station at this moment comes from the zero-carbon photovoltaic power generation unit, and 20% comes from the public grid unit with a carbon intensity factor of 0.58. The comprehensive carbon intensity of this batch of charging electricity is calculated to be 0.116 kg CO2 per kilowatt-hour. Simultaneously, based on the energy storage unit's 1000 kW active power and a calculation time cycle of 0.1 hours, the charging amount for this cycle is calculated to be 100 kilowatt-hours. Finally, the calculated charging amount and the comprehensive carbon intensity value are combined to construct a new carbon attribute data block.

[0089] Simultaneously, a dedicated carbon accounting queue was created for the energy storage power station, and the newly generated carbon attribute data block was added to the tail of the queue. During the evening peak electricity consumption period, assuming the energy storage power station's target discharge capacity is 50 kWh to support the park's load, data is extracted from the head of its carbon accounting queue. A carbon attribute data block with a capacity of 100 kWh and a carbon intensity of 0.116, stored at 2 PM, is identified. Since the capacity of this data block exceeds the target discharge capacity, 0.116 is directly used as the carbon intensity factor for this discharge, and the data block is updated, modifying its remaining capacity to 50 kWh for use in the next discharge. This carbon intensity factor of 0.116 will serve as the input value for the carbon intensity of the energy storage unit when constructing the source-side carbon intensity vector.

[0090] In one feasible implementation, the method further includes:

[0091] Acquire grid synchronization phasor data for target energy-related units within the zero-carbon industrial park.

[0092] Target energy-related units refer to energy-related units within the industrial park that hold a significant position as important power grid hubs or have a substantial impact on power grid stability. These can include the park's main substation, the grid connection point of a large energy storage power station, or the main feeder outlet. Power grid synchronization phasor data refers to power grid operation data with GPS timestamps. It includes the amplitude of voltage and current, as well as corresponding phase angle information. By comparing the phase angles at different locations, the direction and magnitude of power flow can be determined.

[0093] Power grid synchronization phasor data is acquired through phasor measurement units (PMUs) deployed at the target energy-involved units. The PMUs sample voltage and current waveforms at high speed and timestamp each measurement result using an integrated GPS module. The acquired data stream, containing information such as voltage amplitude, voltage phase angle, current amplitude, and current phase angle, constitutes the power grid synchronization phasor data.

[0094] The power grid synchronization phasor data is mapped to the node attributes of the corresponding nodes in the energy network topology diagram, and the energy flow digital model is updated.

[0095] Based on the established mapping relationship, the latest power grid synchronization phasor data, such as voltage amplitude and voltage phase angle, are obtained as a new set of values ​​and updated to the node attributes of the corresponding target node, generating an energy flow digital model that not only includes the power magnitude of each node, but also includes phase angle information at key nodes.

[0096] For example, phasor measurement units (PMUs) were installed at two key energy-related units: the 10 kV main substation node and the power distribution room node of Building A office building. While acquiring regular real-time power data, the PMUs deployed at the main substation and the power distribution room of Building A simultaneously collected grid synchronization phasor data at that moment. For instance, the voltage amplitude at the main substation node was 10.1 kV and the phase angle was 0 degrees, while the voltage amplitude at the power distribution room node of Building A office building was 10.05 kV and the phase angle was -0.5 degrees. Subsequently, based on the established mapping relationship, this grid synchronization phasor data, containing voltage amplitude and phase angle, was added as new node attributes and updated to the corresponding target nodes in the energy flow digital model, thereby generating a physically constrained energy flow digital model.

[0097] In one feasible implementation, step S140 involves inputting the energy flow digital model into a trained graph neural network, and obtaining the energy flow direction matrix by calculating the power distribution relationship between nodes, including:

[0098] The energy flow digital model is input into a trained graph neural network to calculate the initial power distribution relationship between nodes. The initial power distribution relationship is then corrected using grid synchronization phasor data as physical constraints to obtain the energy flow direction matrix.

[0099] The initial power allocation relationship refers to the preliminary power allocation network of all nodes directly derived from a trained graph neural network. It includes the end-to-end power allocation relationship from the energy supply unit to the energy consumption unit, as well as the detailed power allocation relationship when energy flows through various intermediate transmission nodes, such as substations and distribution boxes. Physical constraints refer to a set of deterministic relationships extracted from grid synchronization phasor data that represent the physical laws governing the actual operation of the power grid. For example, the magnitude and direction of the power flow between two target nodes equipped with PMUs can be calculated using the voltage magnitude, phase angle difference, and impedance of the line between the two nodes, through the basic physical formulas of power flow, resulting in a strong physical constraint.

[0100] The grid synchronization phasor data of the target energy-related units with deployed PMUs are extracted, and the actual power flow values ​​between these key nodes are calculated according to the physical formula for power flow calculation, which serve as physical constraints. The portion of the initial power allocation relationship corresponding to these key nodes is compared with the calculated physical constraints, and the entire initial power allocation relationship network is adjusted based on the comparison results so that the adjusted network conforms to the physical constraints at the key nodes.

[0101] Figure 3 A flowchart illustrating a method for generating an energy flow direction matrix according to an embodiment of this application is shown. Figure 3As shown, the method includes steps S310 to S340.

[0102] In one feasible implementation, the energy flow digital model is input into a trained graph neural network to calculate the initial power allocation relationship between nodes. The initial power allocation relationship is then corrected using grid synchronization phasor data as physical constraints to obtain the energy flow direction matrix, including:

[0103] S310: Based on the node attributes of each node in the energy flow digital model and the weights of the directed edges connecting the nodes, the initial power allocation relationship between each node is determined using a trained graph neural network.

[0104] A pre-built and trained graph neural network can consist of an encoder with multiple graph convolutional layers and a decoder for predicting relationships. Graph convolutional layers, such as Graph Convolutional Networks (GCNs) or Graph Attention Networks (GATs), are used to aggregate neighbor node information to update the feature representation of the central node. The generated digital energy flow model, a graph structure containing the real-time power attributes of all nodes and the weights of connecting edges, is input to the encoder of the graph neural network. The encoder iteratively passes and aggregates information between connected nodes through its multi-layer graph convolutional operations, learning layer by layer and generating a high-dimensional feature vector for each node. This vector encodes the node's own state and its position and environment information in the network topology. Subsequently, the decoder receives the high-dimensional feature vectors of all nodes and processes any pair of nodes, ultimately outputting an initial power allocation relationship describing the power exchange between any two nodes in the network.

[0105] For example, the energy flow digital model is input into a trained graph neural network. The encoder of this graph neural network consists of a three-layer graph attention network (GAT layer), and the decoder is a simple multilayer perceptron. At the start of inference, each node in the energy flow digital model, such as the office building A node, has an initial feature vector consisting of its own real-time power value of 1.5 MW. In the first layer of GAT operation, the office building A node selectively aggregates information from its direct upstream node, namely the 10 kV main substation node in the park, according to the attention weights. After three layers of information aggregation and updating, the office building A node generates a new high-dimensional feature vector. This vector not only contains its own power information but also incorporates the influence from global sources such as photovoltaics, energy storage, and the public power grid, as well as the network topology.

[0106] Finally, the decoder receives the high-dimensional feature vectors of all nodes, including the node in office building A, and calculates the power exchange relationships between each pair of nodes, forming a complete initial power allocation relationship. This relationship contains detailed information about the power flow between all nodes in the network. For example, the allocation result could be that of the total power flowing out from the grid-connected inverter node of the photovoltaic power generation system, 1.2 MW flows to the 10 kV main substation node in the park; of the total power flowing out from the 10 kV main substation node in the park, 0.75 MW flows to the power distribution room node in office building A, and 1.5 MW flows to the power distribution cabinet node in the production workshop of building B; and of the total power flowing out from the power distribution room node in office building A, 0.08 MW flows to a charging pile node connected to it.

[0107] In one embodiment, before inputting the energy flow digital model into the trained graph neural network in step S140 to determine the initial power allocation relationship between each node, the method further includes: acquiring a training sample set containing a large amount of historical data. Each training sample in the training sample set contains an energy flow digital model at a historical moment and a corresponding end-to-end power allocation relationship as a real label. The end-to-end power allocation relationship as a real label is pre-calculated using power grid power flow simulation software based on the park operation data at that historical moment. In each iteration of training, the energy flow digital model from a training sample is input into the graph neural network to obtain a predicted power allocation relationship derived by the model. Subsequently, by comparing this predicted power allocation relationship with the real upgrade configuration information label corresponding to the sample, a loss function value is calculated to measure the difference between the two. If the loss function value does not meet the preset training stopping conditions, such as reaching a minimum threshold or completing a specified number of iterations, the model parameters inside the graph neural network, such as the weight coefficients of the graph convolutional layer, are adjusted through the backpropagation algorithm to reduce the loss function value. This process will be repeated until the prediction results of the graph neural network are highly consistent with the true labels, satisfying the training stopping condition, and finally a well-trained graph neural network is obtained.

[0108] S320: Based on the grid synchronization phasor data of the target energy-related units, calculate the voltage magnitude and phase angle difference between the two target nodes corresponding to the two target energy-related units, and determine the verification power allocation relationship, including the power magnitude and flow direction between the two target nodes, by combining the weight of the edge between the two target nodes.

[0109] Verifying the power distribution relationship is a power distribution relationship directly calculated from physical formulas. It refers to the specific information on the magnitude and direction of power flow between two target nodes equipped with PMUs. This information is used as a benchmark to verify and correct the power distribution relationship derived from the model.

[0110] The grid synchronization phasor data of the nodes of the target energy-related units are extracted from the energy flow digital model, namely the voltage magnitude and phase angle of the first node and the voltage magnitude and phase angle of the last node. First, the voltage magnitude difference and voltage phase angle difference between the two target nodes are calculated. Then, the weights of the directed edges connecting the two target nodes, i.e., the theoretical impedance values ​​of the lines, are combined. These parameters are substituted into the commonly used line power transmission equations in power systems. Through calculation, a verification power distribution relationship in terms of power magnitude and flow direction between the two target nodes can be obtained.

[0111] For example, it is necessary to calculate the verification power distribution relationship between the main substation node (head end, labeled i) and the power distribution room node of Building A office building (end end, labeled j). First, the active power value is calculated using the line active power transmission equation as shown in formula (1):

[0112] (1)

[0113] in, This represents the active power flowing from node i to node j. and Let i and j represent the voltage magnitudes at nodes i and j, respectively. and Let i and j represent the voltage phase angles at nodes i and j, respectively. This represents the reactance value of the line connecting node i and node j.

[0114] The following parameters are extracted from the energy flow digital model: First-end voltage amplitude It is 10.1 kV, and the phase angle of the first-terminal voltage is... 0 degrees; terminal voltage amplitude It is 10.05 kV, and the phase angle of the terminal voltage is... -0.5 degrees; the reactance value in the line impedance connecting the two nodes. The value is 0.02 ohms. Substituting these values ​​into the above formula, we can calculate the active power flowing from the main substation node to the office building A node. The value is approximately 0.8 MW. Since the phase angle at the beginning (0 degrees) leads the phase angle at the end (-0.5 degrees), the power flow direction is determined to be from the main substation to office building A. Therefore, the final determined verification power distribution relationship is that the active power flowing from the main substation node to the office building A node is 0.8 MW.

[0115] S330: By comparing the target initial power allocation relationship between two target nodes with the verification power allocation relationship, a correction amount for adjusting the target initial power allocation relationship between target nodes is determined.

[0116] The target initial power allocation relationship refers to the initial power allocation relationship between the two target nodes in the generated initial power allocation relationship, that is, the model's preliminary prediction of the power exchange between these two nodes. The allocation relationship correction amount represents the difference between the target initial power allocation relationship and the verification power allocation relationship. The correction amount includes the magnitude and direction of the adjustment to the initial allocation relationship in order to make the model prediction consistent with physical reality.

[0117] Extract the target initial power allocation relationship corresponding to the two target nodes from the initial power allocation relationship. Compare this target initial power allocation relationship with the verification power allocation relationship used for physical verification. For example, subtract the target initial power allocation relationship value from the verification power allocation relationship value. The result, the difference between the two, serves as the adjustment amount for adjusting the initial power allocation relationship between the target nodes.

[0118] First, from the generated initial power allocation relationship containing power flows between all nodes, the target initial power allocation relationship corresponding to the main substation node and the office building A node is extracted. This target initial power allocation relationship, predicted by the model, represents the active power flowing from the main substation node to the office building A node, and its value is 0.75 MW. Then, this target initial power allocation relationship is compared with the calculated verification power allocation relationship, which is 0.8 MW. Specifically, the value of the target initial power allocation relationship is subtracted from the value of the verification power allocation relationship; that is, a subtraction operation is performed: 0.8 MW minus 0.75 MW. The result, 0.05 MW, is determined as the allocation relationship correction amount for this calculation. This correction amount explicitly indicates that the initially predicted power allocation value needs to be increased by 0.05 MW, and it will be used to guide the global correction of the entire power allocation network.

[0119] In one embodiment, the model may incorrectly predict reversed power flow. For example, the model might predict 0.1 MW of active power flowing from the office building node to the main substation node, meaning the target initial power distribution is -0.1 MW. However, the verified power distribution calculated from PMU data is 0.1 MW, indicating that the actual power flow is from the main substation node to the office building node. In this case, subtracting the target initial power distribution from the verified power distribution value (0.1 MW minus -0.1 MW) yields a correction of 0.2 MW. This correction not only indicates an adjustment in magnitude but, more importantly, its positive sign signifies the need to reverse the initially predicted power flow. The calculated correction will be used to guide the global correction of the entire power distribution network.

[0120] S340: By using the allocation relationship correction amount, the target initial power allocation relationship between two target nodes is corrected, and all initial power allocation relationships are updated using the corrected target initial power allocation relationship to obtain the energy flow direction matrix.

[0121] First, the power allocation correction is applied to the initial target power allocation relationship. For example, through addition, the power allocation relationship between two target nodes is directly corrected to ensure it is completely consistent with the verified power allocation relationship. Next, this locally precisely corrected power allocation relationship is used as a new anchor point or hard constraint. The initial power allocation relationship network is then globally updated using the Constrained Least Squares (CLS) algorithm. The goal of this algorithm is to minimize the overall deviation between the updated and initial allocation relationships while satisfying two core constraints: the first constraint is that the updated relationship must be strictly equal to the corrected value between the target nodes; the second constraint is that the sum of the power inputs of all nodes in the network must equal the sum of their outputs, i.e., satisfying the power balance law. After optimizing and updating the entire initial power allocation relationship network, the end-to-end relationships from all power supply units to all power consumption units are extracted and organized into the final output energy flow matrix.

[0122] For example, the power distribution correction value is 0.05 MW. The power distribution correction value is applied to the target initial power distribution relationship. Through addition, the power distribution relationship predicted by the model from the main substation node to the office building node of Building A is directly corrected from the initial 0.75 MW to 0.8 MW, so that it is completely consistent with the verification power distribution relationship.

[0123] Next, the condition that the active power flowing from the main substation node to the office building A node must be 0.8 MW is used as a hard constraint. The initial power allocation network is then globally updated using a constrained least squares algorithm. The optimization objective of this method is to minimize the deviation from the initial allocation relationship while satisfying two core constraints. The first constraint is that the power from the main substation to the office building A must be 0.8 MW; the second constraint is that the total input power of all nodes in the network, such as the main substation node, must equal the total output power. To satisfy these constraints, this method might reduce the initial allocation of 1.5 MW of power flowing from the main substation to the distribution cabinet node in the production workshop of building B by 0.05 MW, adjusting it to 1.45 MW. After optimizing and updating the entire initial power allocation network, the end-to-end allocation ratios of the three energy supply units (photovoltaics, energy storage, and the grid) to the three energy consumption units (building A, workshop B, and charging piles) are extracted, and an energy flow matrix is ​​constructed.

[0124] For example, the energy flow matrix is ​​shown in Table 1. The rows of the matrix correspond to three energy supply units: a photovoltaic power generation unit, an energy storage power station, and a public power grid unit; the columns correspond to three energy consumption units: Building A (office building), Building B (production workshop), and three charging piles. Each element in the matrix, i.e., the power allocation ratio coefficient, represents the proportion of the output power of the corresponding energy supply unit that flows to the corresponding energy consumption unit. Taking the element in the first row and first column as an example, the ratio coefficient 0.5333 indicates that of the total 1.5 MW of electricity used in Building A (office building), 53.33% of the electricity, i.e., 0.8 MW, comes from the clean photovoltaic power generation unit.

[0125]

[0126] Table 1

[0127] This embodiment uses grid synchronization phasor data as high-precision physical measurements, providing a definite physical verification benchmark for model calculations. First, these physical measurements are used to calculate the actual power flow direction and magnitude on the critical path. Then, based on this, the initial allocation relationship derived by the graph neural network is quantitatively corrected, ensuring that the final generated energy flow matrix is ​​consistent with the physical operating state of the power grid. Therefore, by introducing strong physical constraints, the accuracy of end-user carbon footprint tracking and accounting is further enhanced.

[0128] In one feasible implementation, step S150: converting the carbon intensity factor of each power supply unit into a source-end carbon intensity vector, allocating each source-end carbon intensity vector to the power consumption unit through an energy flow matrix, and calculating the carbon footprint value of the power consumption unit based on the real-time power data of the power consumption unit, including:

[0129] Arrange the carbon intensity factors of each energy supply unit according to the row order in the energy flow matrix to form the source carbon intensity vector.

[0130] First, determine the order of the energy supply units corresponding to the row dimensions of the energy flow matrix. For example, the first row is photovoltaic power generation units, the second row is energy storage units, and the third row is public grid units. Then, arrange the carbon intensity factors of each energy supply unit at the current moment according to this order to form a source-end carbon intensity vector. For example, if the carbon intensity factors of photovoltaic, energy storage, and grid are 0, 0.232, and 0.58, respectively, then the source-end carbon intensity vector is [0, 0.232, 0.58].

[0131] The carbon intensity vector at the source end is multiplied by the energy flow direction matrix to obtain the carbon attribute vector of the energy-consuming unit. Each element of the carbon attribute vector of the energy-consuming unit represents the unit power carbon attribute value allocated to the corresponding energy-consuming unit in the energy flow digital model.

[0132] The dimension of the carbon attribute vector of energy consumption units is the same as the column dimension of the energy flow matrix, that is, the number of energy consumption units corresponds to each other. Each element in the vector represents the total carbon emissions contained in the unit of electricity consumed by the energy consumption unit corresponding to its location, after weighted averaging of all energy supply unit sources. This value can also be regarded as the unit power carbon attribute value of that energy consumption unit.

[0133] Multiply the source-end carbon intensity vector by the energy flow matrix. For each energy-consuming unit (each column of the matrix), perform a dot product operation on its corresponding power allocation ratio coefficient group (column vector) with the source-end carbon intensity vector. For example, performing a dot product operation on the vector [0, 0.232, 0.58] with the column vector [0.5333, 0.2000, 0.2667] representing office building A, i.e., performing a weighted summation of the carbon intensities at each source end, yields the unit power carbon attribute value of office building A. After completing the calculation for all energy-consuming units, these unit power carbon attribute values ​​together constitute an energy-consuming unit carbon attribute vector.

[0134] The carbon footprint of the energy-consuming unit is obtained by performing a scalar multiplication operation between the active power component in the real-time power data of the energy-consuming unit and the corresponding element in the carbon attribute vector of the energy-consuming unit.

[0135] The process iterates through each element in the generated carbon attribute vector of energy-consuming units. For each element, it identifies the corresponding energy-consuming unit and extracts its active power component within the current calculation period from real-time power data. This active power component is then multiplied by the calculation time period to obtain the total electricity consumption. Finally, the total electricity consumption is multiplied by the corresponding unit power carbon attribute value. For example, if Building A's unit power carbon attribute value is 0.201086, its active power is 1500 kW, and the calculation period is 1 hour, then the final carbon footprint value is 1500 multiplied by 1 and then multiplied by 0.201086. The result is the final carbon footprint value for that energy-consuming unit within this time period. After calculating for all energy-consuming units, a dynamically updated end-user carbon footprint list is obtained.

[0136] For example, firstly, according to the row order of the energy flow matrix—the first row being photovoltaic power generation units, the second row being energy storage power stations, and the third row being public grid units—the corresponding carbon intensity factors are arranged into a source-end carbon intensity vector with values ​​of [0, 0.232, 0.58]. Next, this source-end carbon intensity vector is multiplied by the energy flow matrix. Taking office building A as an example, its corresponding column vector [0.5333, 0.2000, 0.2667] is multiplied by the source-end carbon intensity vector to obtain the carbon attribute value per unit power for office building A: 0.201086 kg of carbon dioxide per kilowatt-hour. Similarly, the carbon attribute values ​​per unit power for the production workshop in building B and the three charging piles are also calculated, collectively forming a carbon attribute vector for each energy-consuming unit.

[0137] Next, the final carbon footprint value is calculated based on the carbon attribute vector of this energy-consuming unit. For Building A, its active power is extracted from real-time power data as 1500 kW, and the calculation time period is set to 1 hour, so its total electricity consumption is 1500 kWh. By multiplying the total electricity consumption by the carbon attribute value per unit power of Building A (0.201086), the carbon footprint value of Building A in that hour is 301.629 kg of carbon dioxide. For the three charging piles, their carbon attribute value per unit power is 0, so their carbon footprint value is 0 regardless of their electricity consumption. By performing the same calculation on all energy-consuming units, the carbon footprint of all end users in the park for that hour is accurately tracked and calculated, resulting in a complete list.

[0138] Based on the same concept, this application provides a zero-carbon park full life-cycle carbon footprint tracking and accounting system, which is described below in conjunction with... Figure 4 This application provides a detailed description of the zero-carbon industrial park full life-cycle carbon footprint tracking and accounting system provided in its embodiments.

[0139] Figure 4 This is a structural block diagram of a zero-carbon park full life cycle carbon footprint tracking and accounting system shown in an embodiment of this application.

[0140] like Figure 4 As shown, this zero-carbon park life-cycle carbon footprint tracking and accounting system is applicable to zero-carbon parks powered by multi-energy coupling. The system may include:

[0141] The acquisition module 410 is used to acquire spatial topology data of energy-related units in the zero-carbon park, real-time power data of energy-related units, and carbon intensity factor of each energy supply unit. The energy-related units include at least one energy supply unit and at least one energy consumption unit. The spatial topology data includes physical parameters of the physical lines between energy-related units.

[0142] Module 420 is used to construct an energy network topology graph based on spatial topology data, defining energy-related units as nodes, physical lines connecting nodes as directed edges, and physical parameters of physical lines as weights of directed edges.

[0143] The generation module 430 is used to map real-time power data to the node attributes of the corresponding nodes in the energy network topology diagram, and generate an energy flow digital model that represents the energy flow status of the zero-carbon park.

[0144] The generation module 430 is also used to input the energy flow digital model into the trained graph neural network, and obtain the energy flow direction matrix by calculating the power distribution relationship between nodes. The row dimension of the energy flow direction matrix corresponds to the number of energy supply units, the column dimension corresponds to the number of application energy units, and the matrix elements represent the power distribution ratio coefficient.

[0145] The calculation module 440 is used to convert the carbon intensity factor of each energy supply unit into a source carbon intensity vector, allocate each source carbon intensity vector to the energy consumption unit through the energy flow matrix, and calculate the carbon footprint value of the energy consumption unit based on the real-time power data of the energy consumption unit.

[0146] In one embodiment, the power supply unit includes a photovoltaic power generation unit, an energy storage unit, and a public grid unit.

[0147] In one embodiment, the generation module 430 is further configured to determine the energy storage unit as an energy-consuming unit when the charging / discharging state corresponding to the real-time power data of the energy storage unit is a charging state; calculate the source-end carbon intensity vector allocated to the energy storage unit according to the energy flow matrix, and calculate the charging capacity and corresponding carbon attribute data block of the energy storage unit according to the real-time power data of the energy storage unit; establish a first-in-first-out carbon accounting queue for the energy storage unit, and store the carbon attribute data block of the energy storage unit in the tail of the carbon accounting queue in chronological order, so as to extract the carbon attribute data block corresponding to the target discharge capacity from the head of the carbon accounting queue when the energy storage unit is an energy-supplying unit, and generate the carbon intensity factor of the energy storage unit.

[0148] In one embodiment, the acquisition module 410 is further configured to acquire grid synchronization phasor data of the target energy-related units within the zero-carbon park; map the grid synchronization phasor data to the node attributes of the corresponding nodes in the energy network topology diagram; and update the energy flow digital model.

[0149] In one feasible implementation, the energy flow digital model is input into a trained graph neural network, and the energy flow direction matrix is ​​obtained by calculating the power distribution relationship between nodes, including:

[0150] The energy flow digital model is input into a trained graph neural network to calculate the initial power distribution relationship between nodes. The initial power distribution relationship is then corrected using grid synchronization phasor data as physical constraints to obtain the energy flow direction matrix.

[0151] In one embodiment, the generation module 430 is specifically used to determine the initial power allocation relationship between each node based on the node attributes of each node in the energy flow digital model and the weights of the directed edges connecting the nodes, using a trained graph neural network; calculate the voltage magnitude and phase angle difference between the two target nodes corresponding to the two target energy-related units based on the grid synchronization phasor data of the target energy-related units, and determine the verification power allocation relationship including the power magnitude and flow direction between the two target nodes by combining the weights of the edges between the two target nodes; determine an allocation relationship correction amount for adjusting the target initial power allocation relationship between the target nodes by comparing the target initial power allocation relationship between the two target nodes with the verification power allocation relationship; and obtain the energy flow direction matrix by correcting the target initial power allocation relationship between the two target nodes using the allocation relationship correction amount and updating all initial power allocation relationships using the corrected target initial power allocation relationship.

[0152] In one embodiment, the calculation module 440 is further configured to arrange the carbon intensity factors of each energy supply unit according to the row dimension arrangement order in the energy flow matrix to form a source-end carbon intensity vector; perform matrix multiplication operation on the source-end carbon intensity vector and the energy flow matrix to obtain the energy-consuming unit carbon attribute vector, where each element of the energy-consuming unit carbon attribute vector represents the unit power carbon attribute value allocated to the corresponding energy-consuming unit in the energy flow digital model; and perform scalar multiplication operation on the active power component in the real-time power data of the energy-consuming unit and the value of the corresponding element in the energy-consuming unit carbon attribute vector to obtain the carbon footprint value of the energy-consuming unit.

[0153] Figure 4 Each module in the system shown has an implementation Figures 1 to 3 The functions of each step in the process and their corresponding technical effects are described in detail here for the sake of brevity.

[0154] Figure 5 A schematic diagram of the hardware structure of an electronic device provided in one embodiment of this application is shown.

[0155] The electronic device may include a processor 510 and a memory 520 storing computer program instructions.

[0156] Specifically, the processor 510 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0157] Memory 520 may include mass storage for data or instructions. For example, and not limitingly, memory 520 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 520 may include removable or non-removable (or fixed) media. Where appropriate, memory 520 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 520 is non-volatile solid-state memory.

[0158] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to the first aspect of this disclosure.

[0159] The processor 510 reads and executes computer program instructions stored in the memory 520 to implement any of the zero-carbon park full life cycle carbon footprint tracking and accounting methods in the above embodiments.

[0160] In one example, the electronic device may also include a communication interface 530 and a bus 540. Wherein, such as Figure 5 As shown, the processor 510, memory 520, and communication interface 530 are connected through bus 540 and complete communication with each other.

[0161] The communication interface 530 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0162] Bus 540 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 540 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0163] This electronic device can execute the zero-carbon park full life-cycle carbon footprint tracking and accounting method in the embodiments of this application, thereby achieving a combination of Figures 1 to 3 The method described is for tracking and calculating the carbon footprint of a zero-carbon industrial park throughout its entire life cycle.

[0164] Furthermore, in conjunction with the zero-carbon park full life-cycle carbon footprint tracking and accounting method in the above embodiments, this application embodiment can provide a computer-readable storage medium for implementation. This computer-readable storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the zero-carbon park full life-cycle carbon footprint tracking and accounting methods in the above embodiments.

[0165] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0166] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0167] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0168] The aspects of this application have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by dedicated hardware performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0169] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A method for tracking and calculating the carbon footprint of a zero-carbon industrial park throughout its entire life cycle, applicable to zero-carbon industrial parks powered by multi-energy coupling, characterized in that, include: Acquire spatial topology data of energy-related units within the zero-carbon park, real-time power data of the energy-related units, and carbon intensity factor of each energy supply unit. The energy-related units include at least one energy supply unit and at least one energy consumption unit. The spatial topology data includes physical parameters of the physical lines between the energy-related units. Based on the spatial topology data, the energy-related units are defined as nodes, the physical lines connecting the nodes are defined as directed edges, and the physical parameters of the physical lines are defined as the weights of the directed edges, thereby constructing an energy network topology graph. The real-time power data is mapped to the node attributes of the corresponding nodes in the energy network topology diagram to generate an energy flow digital model that characterizes the energy flow status of the zero-carbon park. The energy flow digital model is input into the trained graph neural network. By calculating the power distribution relationship between the nodes, an energy flow matrix is ​​obtained. The row dimension of the energy flow matrix corresponds to the number of energy supply units, the column dimension corresponds to the number of energy consumption units, and the matrix elements represent the power distribution ratio coefficient. The carbon intensity factor of each energy supply unit is converted into a source-end carbon intensity vector. Through the energy flow matrix, each source-end carbon intensity vector is allocated to the energy consumption unit. Based on the real-time power data of the energy consumption unit, the carbon footprint value of the energy consumption unit is calculated. The source-end carbon intensity vector is formed by arranging the carbon intensity factors of each energy supply unit according to the row dimension arrangement order in the energy flow matrix.

2. The method according to claim 1, characterized in that, The energy supply unit includes a photovoltaic power generation unit, an energy storage unit, and a public power grid unit.

3. The method according to claim 2, characterized in that, The method further includes: When the charging / discharging state corresponding to the real-time power data of the energy storage unit is the charging state, the energy storage unit is identified as the energy consumption unit. Based on the energy flow matrix, the source carbon intensity vector allocated to the energy storage unit is calculated, and based on the real-time power data of the energy storage unit, the charging capacity of the energy storage unit and the corresponding carbon attribute data block are calculated. A first-in-first-out carbon accounting queue is established for the energy storage unit, and the carbon attribute data block of the energy storage unit is stored in the tail of the carbon accounting queue in chronological order. When the energy storage unit is the energy supply unit, the carbon attribute data block corresponding to the target discharge capacity of the energy storage unit is extracted from the head of the carbon accounting queue according to the target discharge capacity of the energy storage unit, and the carbon intensity factor of the energy storage unit is generated.

4. The method according to claim 1, characterized in that, The method further includes: Obtain grid synchronization phasor data for the target energy-related units within the zero-carbon park; The power grid synchronization phasor data is mapped to the node attributes of the corresponding nodes in the energy network topology diagram, and the energy flow digital model is updated.

5. The method according to claim 4, characterized in that, The step of inputting the energy flow digital model into the trained graph neural network and obtaining the energy flow direction matrix by calculating the power distribution relationship between the nodes includes: The energy flow digital model is input into a trained graph neural network to calculate the initial power allocation relationship between the nodes. The initial power allocation relationship is then corrected using the power grid synchronization phasor data as physical constraints to obtain the energy flow direction matrix.

6. The method according to claim 5, characterized in that, The process involves inputting the energy flow digital model into a trained graph neural network, calculating the initial power allocation relationship between the nodes, and correcting the initial power allocation relationship using the power grid synchronization phasor data as physical constraints to obtain the energy flow direction matrix, including: Based on the node attributes of each node in the energy flow digital model and the weights of the directed edges connecting the nodes, the initial power allocation relationship between each node is determined using the trained graph neural network. Based on the grid synchronization phasor data of the target energy-related unit, calculate the voltage amplitude and phase angle difference between the two target nodes corresponding to the two target energy-related units, and combine the weight of the edge between the two target nodes to determine the verification power allocation relationship including the power magnitude and flow direction between the two target nodes; By comparing the target initial power allocation relationship between the two target nodes with the verification power allocation relationship, a correction amount for adjusting the target initial power allocation relationship between the target nodes is determined; By using the allocation relationship correction amount, the target initial power allocation relationship between the two target nodes is corrected, and all the initial power allocation relationships are updated using the corrected target initial power allocation relationship, thus obtaining the energy flow direction matrix.

7. The method according to claim 1, characterized in that, The step of converting the carbon intensity factor of each energy supply unit into a source-end carbon intensity vector, allocating each source-end carbon intensity vector to the energy consumption unit through the energy flow matrix, and calculating the carbon footprint value of the energy consumption unit based on the real-time power data of the energy consumption unit includes: The source carbon intensity vector is multiplied by the energy flow direction matrix to obtain the energy-consuming unit carbon attribute vector. Each element of the energy-consuming unit carbon attribute vector represents the unit power carbon attribute value allocated to the corresponding energy-consuming unit in the energy flow digital model. The carbon footprint value of the energy-consuming unit is obtained by performing a scalar multiplication operation between the active power component in the real-time power data of the energy-consuming unit and the corresponding element in the carbon attribute vector of the energy-consuming unit.

8. A zero-carbon industrial park lifecycle carbon footprint tracking and accounting system, applicable to zero-carbon industrial parks powered by multi-energy coupling, characterized in that, The system includes: The acquisition module is used to acquire spatial topology data of energy-related units in the zero-carbon park, real-time power data of the energy-related units, and carbon intensity factor of each energy supply unit. The energy-related units include at least one energy supply unit and at least one energy consumption unit. The spatial topology data includes physical parameters of the physical lines between the energy-related units. A construction module is used to define the energy-related units as nodes, the physical lines connecting the nodes as directed edges, and the physical parameters of the physical lines as the weights of the directed edges based on the spatial topology data, thereby constructing an energy network topology graph. The generation module is used to map the real-time power data to the node attributes of the corresponding nodes in the energy network topology diagram, and generate an energy flow digital model that characterizes the energy flow status of the zero-carbon park. The generation module is also used to input the energy flow digital model into the trained graph neural network, and obtain the energy flow direction matrix by calculating the power distribution relationship between the nodes. The row dimension of the energy flow direction matrix corresponds to the number of the energy supply units, the column dimension corresponds to the number of the energy consumption units, and the matrix elements represent the power distribution ratio coefficient. The calculation module is used to convert the carbon intensity factor of each energy supply unit into a source-end carbon intensity vector, allocate each source-end carbon intensity vector to the energy consumption unit through the energy flow matrix, and calculate the carbon footprint value of the energy consumption unit based on the real-time power data of the energy consumption unit. The source-end carbon intensity vector is formed by arranging the carbon intensity factors of each energy supply unit according to the row dimension arrangement order in the energy flow matrix.

9. An electronic device, characterized in that, The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the zero-carbon park full life cycle carbon footprint tracking and accounting method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the zero-carbon park full life-cycle carbon footprint tracking and accounting method as described in any one of claims 1-7.

Images