Secure multi-party computing system and method for energy-carbon multi-source data in private computing park

Abstract

The invention relates to the technical field of privacy computing and energy-carbon management, and discloses a secure multi-party computing system and method for energy-carbon multi-source data of a privacy computing park. The method comprises the following steps: constructing a dynamic park energy carbon data carrier containing a space grid topology and a time sequence topology; original energy carbon data streams are obtained from a plurality of data source nodes, the original energy carbon data streams are injected into corresponding positions of carriers to generate initialized carriers, and each grid unit comprises energy consumption, carbon emission and a data quality attribute field; on the initialized carrier, privacy calculation constraint force is applied to adjacent grid units of different data source nodes, attribute fields of the grid units are driven to conduct directional data interaction and fusion, and a carrier after privacy fusion is generated; and by taking the carrier as an initial state, performing energy-carbon state deduction on the space grid topology according to energy conduction and carbon emission diffusion rules, and generating an energy-carbon state deduction carrier of multiple time slices in the future. According to the method, deep fusion and high-fidelity space-time deduction of park energy carbon data under privacy protection are realized.

Description

Technical Field

[0001] This invention relates to the fields of privacy computing and energy and carbon management technology, specifically to a secure multi-party computation system and method for multi-source energy and carbon data in privacy computing parks. Background Technology

[0002] In park-level energy and carbon emission management, it is necessary to aggregate and analyze widely distributed data on the consumption of various energy sources such as electricity, gas, and heat, and their corresponding carbon emissions. Existing technologies typically employ a centralized data aggregation model, directly transmitting raw readings from each data source node to a central platform for unified processing and calculation. Another common approach is to use a distributed acquisition architecture, but the data is only stored locally or reported after simple summation, and the correlation calculations between different data sources rely on the central server's processing of plaintext data.

[0003] Existing centralized or simple distributed processing solutions have shortcomings. Directly aggregating raw data poses a privacy risk, as data owners may be unwilling to provide details. Centralized computing models cannot meet the security and compliance requirements of data remaining within its domain. Existing methods treat data as discrete points with weak correlations to physical location and time, making it difficult to support refined modeling and extrapolation of the dynamic transmission processes of energy consumption and carbon emissions within the industrial park. The inherent spatiotemporal connections of data are fragmented, limiting the depth of analysis and prediction.

[0004] This invention aims to address how to achieve deep fusion and collaborative computing of multi-source energy and carbon data in industrial parks while protecting the privacy of various data sources. Simultaneously, this invention addresses how to construct a data structure capable of carrying the spatiotemporal attributes of data and supporting dynamic process simulation on spatial topology, thereby surpassing traditional data aggregation and statistics to achieve process-oriented energy and carbon state analysis and prediction. Summary of the Invention

[0005] The purpose of this invention is to provide a secure multi-party computation system and method for carbon multi-source data in privacy computing parks, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, this invention provides a secure multi-party computation method for multi-source energy and carbon data in privacy computing parks, the method comprising: A dynamic energy and carbon data carrier for the park is constructed. The energy and carbon data carrier includes a spatial grid topology and a time series topology. The spatial grid topology is divided according to the geographical boundaries and functional zones of the park, and the time series topology is synchronized with the time series period of data collection. Raw energy and carbon data streams are obtained from multiple independent data source nodes, with each data source node corresponding to a spatial grid. The original energy and carbon data stream is injected into the corresponding spatial grid and time series position in the park's energy and carbon data carrier to generate an initialized energy and carbon data carrier. Each grid cell in the initialized energy and carbon data carrier contains an energy consumption data attribute field, a carbon emission data attribute field, and a data quality attribute field. On the initialized energy and carbon data carrier, privacy computing constraints are applied between grid cells belonging to different data source nodes, so that the attribute fields of adjacent grid cells undergo directional data interaction and fusion under the action of the privacy computing constraints, thereby generating a privacy-fused energy and carbon data carrier. Using the energy and carbon data carrier as the initial state, an energy and carbon state deduction process is executed on the spatial grid topology. The energy and carbon state deduction process is calculated according to predefined energy conduction rules and carbon emission diffusion rules to generate an energy and carbon state deduction carrier containing multiple future time slices.

[0007] Preferably, constructing a dynamic energy and carbon data carrier for the industrial park includes: Based on the vector boundaries of the park map, the park is divided into a set of grid cells of equal area or weighted by functional area, and each grid cell is assigned a unique spatial coordinate code; A local coordinate system is established for each grid cell, with the origin of the local coordinate system located at the center of the grid, which is used to locate the data source nodes inside the grid; Define a continuous time axis and discretize the time axis into a sequence of time steps with the same time step as the minimum interval of data reporting, with each time step corresponding to a time slice; Establish a mapping relationship between the spatial grid topology and the time series topology, so that each grid cell corresponds to a data storage unit in each time slice. The data storage unit is used to store the energy consumption data attribute field, carbon emission data attribute field and data quality attribute field. The values ​​of the energy consumption data attribute field, carbon emission data attribute field, and data quality attribute field are initialized to zero or a preset reference field.

[0008] Preferably, injecting the original energy and carbon data stream into the park's energy and carbon data carrier includes: Receive real-time data packets from the energy data source nodes of the enterprise's production, and parse the instantaneous power value and geographical location tag from the real-time data packets; The spatial coordinate code of the target grid cell is determined based on the geographic location label, and the instantaneous power value is written as the field strength value into the energy consumption data attribute field of the target grid cell in the corresponding time slice. Receive a sequence of sampled data from a carbon emission monitoring data source node, and parse the concentration value and sampling point coordinates from the sampled data sequence; The sampling point coordinates are associated with one or more neighboring grid cells, and the concentration values ​​are assigned to the carbon emission data attribute fields of the one or more neighboring grid cells according to distance weights. At the end of each time slice, a quality evaluation value is calculated based on the data reporting success rate and integrity of each data source node within that time slice, and the quality evaluation value is written into the data quality attribute field of all grid cells.

[0009] Preferably, applying privacy-preserving computational constraints to enable directional data interaction and fusion of attribute fields between adjacent grid cells includes: On the boundary of adjacent grid cells of the initialized energy carbon data carrier, a privacy computing constraint is defined, the strength of which is inversely proportional to the privacy protocol level between the data source nodes to which the adjacent grid cells belong; The privacy computing constraint acts on the property field boundary of adjacent grid cells, driving the property field value of high field strength grid cells to diffuse into low field strength grid cells through gradient diffusion. The diffusion flux is determined by the strength of the privacy computing constraint and the gradient of the property field itself. During the diffusion process, a homomorphic transformation operation is performed on the attribute field values ​​leaving the source grid cell to keep them encrypted during transmission until they enter the target grid cell and are homomorphically aggregated with the attribute field values ​​of the target grid cell. After each fusion cycle, the property field balance of each grid cell is recalculated. If the property field balance does not reach the convergence threshold, the strength parameter of the privacy computing constraint is adjusted, and the next fusion cycle is started until the convergence condition is met, generating a privacy-fused energy and carbon data carrier.

[0010] Preferably, the energy-carbon state deduction process, calculated based on predefined energy conduction rules and carbon emission diffusion rules, includes: An initial "energy consumption potential" is assigned to the energy consumption data attribute field of each grid cell of the privacy-fused energy and carbon data carrier, and an initial "carbon emission concentration" is assigned to the carbon emission data attribute field. The energy transfer rule is defined as follows: a grid cell with a high energy potential will transfer energy load to an adjacent grid cell with a low energy potential along the gradient direction of the energy data attribute field. The amount of load transferred is determined by the energy admittance between the two grid cells. The energy admittance is calculated from the device type of the grid cell and the historical rate of change of the energy data attribute field. The carbon emission diffusion rule is defined, which describes the migration of "carbon emission concentration" following the diffusion equation on the spatial grid topology. The diffusion coefficient is associated with the spatial properties of the grid cell and the value of the data quality attribute field. Based on the energy conduction rules and carbon emission diffusion rules, the evolution process of "energy potential" and "carbon emission concentration" of each grid cell is solved step by step on the time series topology to generate a series of intermediate energy and carbon data carriers describing future states. All intermediate energy and carbon data carriers are connected in chronological order to form an energy and carbon state inference carrier, which records the complete trajectory of the attribute field of each grid cell changing over time.

[0011] Preferably, the method further includes: During the simulation process of the energy and carbon state simulation carrier, the property field gradient change of each grid cell is monitored in real time. When the property field gradient exceeds the safety threshold, a local data field compensation process is triggered to dynamically correct the local grid properties of the energy and carbon state simulation carrier. Extract the final state of the energy and carbon state projection carrier, and analyze the spatial coordinates of the abnormal energy consumption accumulation area, the trajectory of the high carbon emission load path, and the topological structure of the energy efficiency weak area from the final state; Based on the spatial coordinates of the abnormal energy consumption cluster, the trajectory of the high carbon emission load path, and the topology of the energy efficiency weak area, a set of control instructions are calculated in reverse on the spatial grid topology. The control instructions include energy flow redirection instructions, carbon emission path blocking instructions, and energy efficiency field enhancement instructions. The control command is sent to the corresponding spatial grid of the park's energy and carbon data carrier, driving the park's energy and carbon data carrier to perform an iterative adjustment and generate an optimized park energy and carbon data carrier.

[0012] Preferably, triggering a local data field compensation process includes: At each time step in the simulation process, the magnitude of the gradient of the energy consumption data attribute field and the magnitude of the gradient of the carbon emission data attribute field are calculated for each grid cell. When the magnitude of the gradient of the energy consumption data attribute field of a certain grid cell exceeds the energy consumption safety threshold, the grid cell is determined to be an energy consumption field distortion point, and energy consumption field compensation is initiated. The energy consumption field compensation is achieved by virtually introducing a "load sinking" grid cell. The "load sinking" grid cell is adjacent to the energy consumption field distortion point and has an extremely low initial "energy consumption potential". According to the energy consumption conduction rule, the excess load at the energy consumption field distortion point will be quickly guided to the "load sinking" grid cell, thereby smoothing the field gradient. When the magnitude of the gradient of the carbon emission data attribute field of a certain grid cell exceeds the carbon emission safety threshold, the grid cell is determined to be a carbon emission field distortion point, and carbon emission field compensation is initiated. The carbon emission field compensation is achieved by temporarily increasing the virtual diffusion coefficient between the carbon emission field distortion point and its upwind grid cell, thereby accelerating the migration of high-concentration carbon emissions and reducing the field gradient. After the compensation process is completed, the virtual "load sinking" grid cells are removed and the diffusion coefficient is restored, and the energy carbon state deduction process continues.

[0013] Preferably, the step of resolving the spatial coordinates of the abnormal energy consumption cluster, the trajectory of the high carbon emission load path, and the topology of the energy-inefficient region from the final state includes: In the final state of the energy carbon state projection carrier, all grid cells whose "energy consumption potential" exceeds a preset critical value are identified, and the set of spatial coordinate codes of these grid cells is marked as an abnormal energy consumption cluster area. On the spatial grid topology, starting from the grid cell with the highest carbon emission concentration, the path formed along the gradient descent direction of the carbon emission data attribute field is traced, and the grid cell sequence traversed by the path is connected to form the trajectory of the high carbon emission load path. Calculate the ratio of the average "energy potential" to the average "carbon emission concentration" of each grid cell during the simulation process, and define the ratio as the energy efficiency field strength; On the spatial grid topology, the energy efficiency field intensity value is divided into regions by growth. Continuous grid cell regions with energy efficiency field intensities below the intensity threshold are merged to form a topological structure of energy-inefficient regions. The topological structure is described by the region boundary coordinates and internal connectivity.

[0014] Preferably, the step of reverse-calculating a set of control instructions on the spatial grid topology includes: For the set of spatial coordinates of the abnormal energy consumption accumulation area, calculate one or more energy diversion paths. The energy diversion paths connect the abnormal energy consumption accumulation area with the grid area with low energy load. The energy flow redirection command includes adjusting the parameter value of "energy admittance" along the energy diversion path. For the trajectory of the high carbon emission load path, key node grid cells on the trajectory are identified. The key node grid cells are hubs connecting different segments of the path. The carbon emission path blocking command includes setting a virtual diffusion barrier at the key node grid cell. The virtual diffusion barrier is implemented by setting the carbon emission diffusion coefficient of the node to zero. For the topology of the energy-inefficient region, the distribution of energy efficiency field intensity inside it is analyzed, and the core sub-region with the lowest intensity is identified. The energy efficiency field enhancement command includes injecting virtual energy efficiency enhancement incentives into the grid cells of the core sub-region. The incentive is reflected in the calculation coefficient that temporarily increases the ratio of "energy use potential" to "carbon emission concentration" of the grid cell.

[0015] Preferably, the present invention also includes a secure multi-party computation system for energy and carbon multi-source data in a privacy computing park. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the secure multi-party computation method for energy and carbon multi-source data in a privacy computing park as described above.

[0016] Compared with the prior art, the beneficial effects of the present invention are: The constructed energy and carbon data carrier for the park uses a spatial grid divided according to actual geographical boundaries and functional zones, with time series data strictly synchronized with the collection cycle. Energy consumption, carbon emissions, and data quality information are structured into attribute fields of grid cells, forming a spatiotemporally aligned field data foundation. This design preserves the intrinsic spatiotemporal correlations of the data, and all calculations are performed on grids with clear physical meaning and adjacency relationships, providing a built-in spatiotemporal framework for analysis and avoiding the correlation breaks caused by traditional point-based data management.

[0017] On the initialization platform, privacy-preserving computation constraints are applied to adjacent grid cells belonging to different data source nodes. These constraints drive directed interaction and fusion between the attribute fields of adjacent grids. The data interaction path is entirely defined by the adjacent topological relationships of the grids, and the original data does not leave the local area. This transforms privacy-preserving computation from a general protocol into a directed fusion process based on neighbor relationships that fits the physical spatial structure, achieving collaborative computation and compensation of energy and carbon information in space while protecting the privacy of each data source.

[0018] Using the fused carrier as the initial state, energy and carbon state simulations are performed on a spatial grid topology. Since the carrier itself is already a complete spatiotemporal field integrating data from multiple sources, the energy conduction and carbon emission diffusion rules upon which the simulation is based can be directly simulated and calculated on the grid and its adjacency relationships. The simulation process is based on a global, privacy-preserving, fused high-resolution data field, rather than extrapolating from data from isolated nodes. This allows the simulation of the future energy and carbon state to more accurately reflect the complex spatial interactions and dynamic processes within the park. Attached Figure Description

[0019] Figure 1 This is a schematic diagram illustrating the working principle of the secure multi-party computation method for carbon multi-source data in the privacy computing park described in this invention. Figure 2 A flowchart for constructing a dynamic energy and carbon data carrier for the industrial park; Figure 3 A diagram showing the factors influencing the diffusion flux of grid cell pairs; Figure 4 A flowchart illustrating the process of extrapolating the carbon state. Figure 5 This is a graph showing the evolution trend of the carbon state over time. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 This invention provides a secure multi-party computation method for privacy-preserving energy and carbon multi-source data in a park. The method includes: constructing a dynamic park energy and carbon data carrier, which comprises a spatial grid topology and a time-series topology. The spatial grid topology is divided according to the park's geographical boundaries and internal functional zones, while the time-series topology is synchronized with the data acquisition time series. The method obtains raw energy and carbon data streams from multiple independent data source nodes, each corresponding to a specific spatial grid. The raw energy and carbon data streams are then injected into the corresponding spatial grid and time-series position in the park's energy and carbon data carrier, thereby generating an initialized energy and carbon data carrier. In this initialized carrier, each grid cell contains an energy consumption data attribute field, a carbon emission data attribute field, and a data quality attribute field. Subsequently, on the initialized energy and carbon data carrier, a privacy-preserving computation constraint is applied between grid cells belonging to different data source nodes. This constraint prompts directional data interaction and fusion of the attribute fields of adjacent grid cells, thereby generating a privacy-preserving fused energy and carbon data carrier. Finally, using the privacy-integrated energy and carbon data carrier as the initial state, an energy and carbon state deduction process is executed on the spatial grid topology. This deduction process is calculated based on predefined energy conduction rules and carbon emission diffusion rules to generate an energy and carbon state deduction carrier containing multiple future time slices.

[0022] Example 1: See Figure 2When constructing a dynamic energy and carbon data carrier for the industrial park, the entire park is divided into a set of grid cells of equal area or weighted by functional zones based on the vector boundaries of the park map, and each grid cell is assigned a unique spatial coordinate code. A local coordinate system is established for each grid cell, with its origin located at the center of the grid, used to locate the data source nodes within the grid. A continuous time axis is defined and discretized into a time step sequence with the same time interval as the minimum data reporting interval, with each time step corresponding to a time slice. A mapping relationship is established between the spatial grid topology and the time series topology, so that each grid cell corresponds to a data storage unit on each time slice. This data storage unit is used to store the energy consumption data attribute field, carbon emission data attribute field, and data quality attribute field. During initialization, the values ​​of the energy consumption data attribute field, carbon emission data attribute field, and data quality attribute field are set to the zero field or a preset reference field. When injecting the raw energy and carbon data stream into the park's energy and carbon data carrier, real-time data packets from enterprise production energy data source nodes are received. Instantaneous power values ​​and geographic location tags are parsed from these data packets. The spatial coordinate encoding of the target grid cell is determined based on the geographic location tags, and the instantaneous power value is written as the field strength value into the energy consumption data attribute field of the target grid cell in the corresponding time slice. Simultaneously, sampled data sequences from carbon emission monitoring data source nodes are received. Concentration values ​​and sampling point coordinates are parsed from these sequences. The sampling point coordinates are associated with one or more neighboring grid cells, and the concentration values ​​are distributed to the carbon emission data attribute fields of these neighboring grid cells according to distance weights. At the end of each time slice, a quality evaluation value is calculated based on the data reporting success rate and completeness of each data source node within that time slice, and this quality evaluation value is written into the data quality attribute field of all grid cells.

[0023] In practical implementation, when constructing a dynamic energy and carbon data carrier for the industrial park, a high-precision park map vector boundary file is used. This file defines the precise geographical outline and internal functional zoning polygons of the park. The spatial grid topology is established by regularly dividing the entire geographical area of ​​the park into multiple sets of grid cells. The grid cells are square in shape, and their size is pre-set according to the required spatial granularity of analysis. Each generated grid cell is assigned a globally unique spatial coordinate code in the format of "region identifier-row number-column number," used to uniquely locate each grid cell in the carrier. In some embodiments, the division of grid cells is not strictly equal in area, but rather based on area weighting according to the functional zoning within the park. Specifically, the area of ​​grid cells in the industrial production functional area will be smaller than that in the office and leisure functional area, in order to achieve a higher spatial resolution description of key areas. A local coordinate system is established for each grid cell. The origin of the local coordinate system is set at the geographic center of the grid cell. The X and Y axes of the local coordinate system are parallel to the global coordinate axes of the park map. The local coordinate system is used to accurately calibrate and map the relative positions of various data source nodes located within the grid cell. The data source nodes include the physical installation locations of electricity meters, gas meters, distributed energy sensors, and carbon emission monitors.

[0024] Define a continuous timeline, starting at the system initialization time, and incrementing continuously with standard timestamps. Discretize this continuous timeline into a sequence of time steps, each corresponding to a time slice. The length of each time step is synchronized with the minimum time interval between data reports from all data source nodes; for example, if the fastest data source node has a reporting cycle of 5 minutes, the time step is set to 5 minutes. Establish a mapping between the spatial grid topology and the time series topology, specifically represented by a three-dimensional data cube. The two horizontal dimensions of the data cube correspond to the row and column indices of the spatial grid topology, and the vertical dimension corresponds to the time slice index of the time series topology. Each intersection determined by a specific grid cell and a specific time slice is a data storage unit, which stores three attribute fields: energy consumption data attribute field, carbon emission data attribute field, and data quality attribute field. During the carrier initialization phase, the three attribute fields in all data storage units are assigned initial values. The initial value of the energy data attribute field is zero, indicating that there is no energy flow at the initial moment. The initial value of the carbon emission data attribute field is a benchmark concentration field preset according to the historical average level of the park. The initial value of the data quality attribute field is a specific identifier value indicating that the data is completely missing or has not been verified.

[0025] In practical implementation, when the raw energy and carbon data stream is injected into the park's energy and carbon data carrier, the system continuously receives data packets from different data source nodes. Real-time data packets from enterprise production energy data source nodes are received. These real-time data packets follow IoT communication protocols, and two key fields are parsed from their payload: instantaneous power value and geographic location tag. Based on the parsed geographic location tag, a coordinate-grid matching algorithm is used to determine the spatial coordinate code of the target grid cell. The matching algorithm calculates the distance between the coordinates of the geographic location tag and the center point of all grid cells, identifying the grid cell with the smallest distance as the target grid cell. After determining the target grid cell and its corresponding current time slice, the parsed instantaneous power value is used as the field strength value and directly written into the data storage unit corresponding to the energy consumption data attribute location of the target grid cell in the current time slice. It can be understood that within the same time slice, the same grid cell may receive instantaneous power values ​​from multiple data source nodes. The processing method is to accumulate these values, using the sum as the final energy consumption data attribute field strength value for that grid cell in that time slice.

[0026] Simultaneously, the system receives sampling data sequences from fixed or mobile carbon emission monitoring data source nodes. These sequences are typically uploaded in batches. The system parses the carbon dioxide concentration value and sampling point coordinates for each set of sampling data from these sequences. Since the spatial distribution of carbon emission monitoring points may not coincide with the grid center, the concentration value of a single sampling point needs to be correlated with and influence multiple surrounding grid cells. Based on the sampling point coordinates, the Euclidean distance from the sampling point to the center point of its neighboring grid cells is calculated. The concentration values ​​are then assigned to the carbon emission data attribute fields of one or more neighboring grid cells according to distance weights. The distance weights are calculated following an inverse distance weighting principle; the closer the grid cell is to the sampling point, the greater the assigned concentration weight. One possible weight calculation formula is:

[0027] in: This represents the weight assigned to the concentration value of the j-th sampling point in the i-th grid cell. represents the distance from the center of the i-th grid cell to the j-th sampling point, k is the distance attenuation coefficient, and N is the total number of neighboring grid cells involved in the allocation. Finally, the sum of the concentration values ​​allocated to the i-th grid cell from all relevant sampling points within a certain time slice will be accumulated in the corresponding data storage unit of its carbon emission data attribute field.

[0028] At the end of each time slice, the system compiles statistics on the data reporting status of all data source nodes within that time slice. The statistics include the data reporting success rate and data packet integrity verification results for each data source node. Based on preset evaluation rules, such as assigning weights to the reporting success rate and data integrity and calculating weighted scores, a global quality assessment value is ultimately calculated. This quality assessment value is written into the data storage unit corresponding to the data quality attribute location of all grid cells within that time slice. This means that within the same time slice, all spatial locations share the same preliminary data quality assessment result, which reflects the overall reliability level of data acquisition during that time period. Example 2: When applying privacy-preserving computation constraints and enabling directional data interaction and fusion of attribute fields in adjacent grid cells, a privacy-preserving computation constraint is defined on the boundary of adjacent grid cells in the initialized energy and carbon data carrier. The strength of this constraint is inversely proportional to the privacy protocol level between the data source nodes of the adjacent grid cells. The privacy-preserving computation constraint acts on the boundary of the attribute fields of adjacent grid cells, driving the attribute field values ​​of high-field-strength grid cells to diffuse gradients to low-field-strength grid cells. The diffusion flux is determined by the strength of the privacy-preserving computation constraint and the gradient of the attribute fields themselves. During the diffusion process, a homomorphic transformation operation is performed on the attribute field values ​​leaving the source grid cell, keeping them encrypted during transmission until they enter the target grid cell and are homomorphically aggregated with the attribute field values ​​of the target grid cell. After each fusion cycle, the attribute field balance of each grid cell is recalculated. If the attribute field balance does not reach the convergence threshold, the strength parameter of the privacy-preserving computation constraint is adjusted, and the next fusion cycle is started until the convergence condition is met, generating the privacy-fused energy and carbon data carrier.

[0029] In practical implementation, the process of imposing privacy-preserving computation constraints and enabling directional data interaction and fusion between the attribute fields of adjacent grid cells begins with defining the privacy relationships between the data source nodes of the grid cells in the initialized energy and carbon data carrier. Before participating in computation, each data source node signs a privacy agreement with the system or other nodes at different levels. The privacy agreement level quantifies the willingness and scope of data sharing. In some embodiments, the privacy agreement levels are divided into integer levels from 1 to 5, where level 1 represents allowing the maximum degree of data sharing and fusion, and level 5 represents requiring the highest level of privacy protection and allowing only the minimum necessary data interaction. Privacy-preserving computation constraints are defined on the boundaries of adjacent grid cells in the initialized energy and carbon data carrier. These constraints are virtual forces acting on the boundaries of the grid cell attribute fields, and their strength coefficient is inversely proportional to the privacy agreement level between the data source nodes of adjacent grid cells. Specifically, when the privacy agreement level between the data source nodes of two adjacent grid cells is low, the strength coefficient of the privacy-preserving computation constraint is large, meaning that stronger data interaction is allowed or encouraged; when the privacy agreement level is high, the strength coefficient of the privacy-preserving computation constraint is small, meaning that the strength of data interaction is limited.

[0030] Privacy-preserving computation constraints act directly on the attribute field boundaries of adjacent grid cells, driving the attribute field values ​​of high-field-strength grid cells to diffuse gradients to low-field-strength grid cells. Both energy consumption data attribute fields and carbon emission data attribute fields follow this diffusion mechanism, with the diffusion flux determined by the strength of the privacy-preserving computation constraints and the spatial gradient of the attribute field itself. It can be understood that the diffusion flux describes the amount of attribute field value transferred from one grid cell to an adjacent grid cell through a unit-length boundary within a time step. One possible formula for calculating the diffusion flux is:

[0031] in: This represents the amount of attribute field value transferred from mesh cell a to adjacent mesh cell b within a fusion step. The privacy computing constraint strength coefficient is jointly determined by the privacy protocol levels between the data source nodes to which grid cell a and grid cell b belong. It is the gradient value of the property field in the direction from grid cell a to grid cell b. The negative sign indicates that the diffusion direction is opposite to the gradient increase direction, that is, from high field strength to low field strength.

[0032] In practical implementation, the diffusion process needs to protect the privacy of the original data. A homomorphic transformation operation is performed on the attribute field values ​​leaving the source grid cell. This operation is applied immediately upon the data leaving the source grid cell's data storage unit, ensuring that the attribute field values ​​remain encrypted throughout transmission and before aggregation with the target grid cell's values. In some embodiments, an additive homomorphic encryption algorithm, such as the Paillier encryption algorithm, is used to encrypt the incremental attribute field values ​​to be transmitted. The encrypted data is in ciphertext form during network transmission and when it enters the target grid cell's processing buffer. Only after entering the secure computing environment of the target grid cell does the target grid cell's local encrypted attribute field values ​​undergo a homomorphic aggregation operation with the incoming encrypted incremental values. This aggregation operation is calculated directly on the ciphertext, generating a new ciphertext. This new ciphertext, after decryption, represents the updated attribute field values ​​of the target grid cell.

[0033] After each complete fusion cycle, the system recalculates the attribute field balance of each grid cell. Attribute field balance is an indicator that measures the uniformity of the distribution of a certain attribute field value across all grid cells in the entire carrier space. Optionally, attribute field balance can be defined as the standard deviation of the attribute field value across all grid cells, or as the average of the absolute differences in attribute field values ​​between adjacent grid cells. The calculated attribute field balance is compared with a preset convergence threshold. If the attribute field balance does not reach the convergence threshold, it indicates that the data fusion within the carrier has not yet reached a stable or balanced state. In this case, the system automatically adjusts the strength parameter of the privacy computation constraint. The adjustment strategy can be proportional adjustment based on the deviation between the current fusion rate and the balance, and then the next fusion cycle is initiated. Iterative fusion cycles continue until the calculated attribute field balance reaches or exceeds the preset convergence condition. At this point, the data interaction within the carrier reaches a dynamic equilibrium, generating a stable, fused, privacy-preserving energy and carbon data carrier. The attribute field values ​​in the privacy-preserving energy and carbon data carrier are the result of interaction and balance among multiple data sources under the premise of privacy protection, and are no longer directly equivalent to the original data of any single data source node.

[0034] See Figure 3 This is a bar chart comparing the characteristic values ​​of adjacent grid cell pairs, used to display the core parameters of different grid pairs during the privacy fusion of energy and carbon data in the park. The constraint strength coefficient (green bar): the virtual force strength driving data interaction in privacy computation; attribute field gradient value (light yellow bar): the degree of difference in the distribution of energy and carbon data between grids; diffusion flux (absolute value, red bar): the amount of data interaction between grids (determined by the former two). This type of chart is commonly used for process analysis of privacy fusion of energy and carbon data in the park. It helps identify grid areas with active data interaction (such as grids 3-4) to assist in optimizing privacy protocol configuration; and it verifies the rationality of data diffusion to ensure a balance between privacy protection and data fusion effectiveness.

[0035] Example 3: See Figure 4 The energy and carbon state extrapolation process is calculated based on predefined energy consumption transfer rules and carbon emission diffusion rules. An initial "energy consumption potential" is assigned to the energy consumption data attribute field of each grid cell in the privacy-fused energy and carbon data carrier, and an initial "carbon emission concentration" is assigned to the carbon emission data attribute field. An energy consumption transfer rule is defined, describing how grid cells with high "energy consumption potential" will transfer energy load to adjacent grid cells with low "energy consumption potential" along the gradient direction of the energy consumption data attribute field. The amount of transferred load is determined by the "energy consumption admittance" between the two grid cells, which is calculated from the grid cell's equipment type and the historical rate of change of the energy consumption data attribute field. A carbon emission diffusion rule is defined, describing how "carbon emission concentration" migrates according to a diffusion equation on the spatial grid topology. The diffusion coefficient is correlated with the values ​​of the grid cell's spatial attributes and data quality attribute fields. Based on the energy consumption transfer rule and the carbon emission diffusion rule, the evolution process of the "energy consumption potential" and "carbon emission concentration" of each grid cell is solved step-by-step on the time-series topology, generating a series of intermediate energy and carbon data carriers describing future states. All intermediate energy and carbon data carriers are connected in chronological order to form an energy and carbon state projection carrier, which records the complete trajectory of the attribute field of each grid cell changing over time.

[0036] In practice, the energy and carbon state deduction process is calculated based on predefined energy consumption conduction rules and carbon emission diffusion rules. This process uses the privacy-fused energy and carbon data carrier as the initial state of spatiotemporal evolution. Each grid cell of the privacy-fused energy and carbon data carrier is assigned an initial "energy consumption potential" to its energy consumption data attribute field. The value of the initial "energy consumption potential" is directly taken from the field strength value of that grid cell in the energy consumption data attribute field after privacy fusion, and the unit is kilowatt.

[0037] Simultaneously, an initial "carbon emission concentration" is assigned to the carbon emission data attribute field of each grid cell in the privacy-fused energy-carbon data carrier. The initial "carbon emission concentration" value is directly taken from the field strength value of that grid cell in the carbon emission data attribute field after privacy fusion, and the unit is ppm. An energy transfer rule is defined, which describes how grid cells with high "energy potential" will transfer energy load to adjacent grid cells with low "energy potential" in the spatial grid topology along the gradient direction of the energy data attribute field. The amount of transferred load is determined by the "energy admittance" between the two grid cells. "Energy admittance" is a parameter characterizing the ease of energy transfer between two grid cells. The energy admittance is calculated from the grid cell's device type attribute and the historical rate of change of the energy data attribute field through a predefined function. The equipment type attribute first classifies the equipment installed in the grid cell into types such as adjustable load, energy storage equipment, or base load. Each type corresponds to different energy transfer characteristics, thus providing basic weight parameters for admittance calculation. For example, adjustable load may be assigned a higher admittance value to promote load adjustment, while base load corresponds to a more stable admittance benchmark. The historical rate of change is quantified by analyzing the fluctuation of the energy consumption data attribute field over a period of time, such as calculating variance or fluctuation frequency to reflect the dynamic characteristics of the energy consumption behavior of the grid cell. A predefined function combines the weight of equipment type mapping with the historical rate of change index, and through a dynamic adjustment mechanism, the energy consumption admittance value is positively correlated with the historical rate of change. That is, the more volatile the area, the higher the admittance value, to characterize the ease of energy transfer. In some embodiments, the energy consumption admittance is positively correlated with the historical rate of change, indicating that the load in areas with strong volatility is more easily conducted out. The carbon emission diffusion rule is defined, which describes how the "carbon emission concentration" migrates according to the diffusion equation on the spatial grid topology. The diffusion equation is a partial differential equation that describes the migration of concentration from high concentration range to low concentration range in space. The diffusion coefficient is the key parameter that controls the diffusion rate.

[0038] Based on the energy transfer and carbon emission diffusion rules, the evolution of the "energy potential" and "carbon emission concentration" of each grid cell is solved iteratively over a time-series topology. The solution process proceeds iteratively in discrete time steps. Within each time step, the system first calculates the change in "energy potential" of each grid cell due to energy transfer in the next time step, based on the current distribution of "energy potential" of all grid cells and the "energy admittance" between grid cells, according to the energy transfer rules. Simultaneously, based on the current distribution of "carbon emission concentration" of all grid cells and the diffusion coefficient of the corresponding grid cell, the system calculates the change in "carbon emission concentration" of each grid cell due to carbon emission diffusion in the next time step, according to the carbon emission diffusion rules. An optional discretized formula for calculating the change in "energy potential" caused by the transfer of "energy potential" between adjacent grid cells is as follows:

[0039] in: This represents the change in the "energy potential" of grid cell a within time step t. Let represent the set of all adjacent grid cells of grid cell 'a' in the spatial grid topology. It is the "energy admittance" between grid cell a and its adjacent grid cell b. and These represent the "energy potential" values ​​of grid cells a and b at the end of the previous time step (t-1), respectively. This refers to the time step length used in the extrapolation process. This formula simulates the flow of energy from cells with high "energy use potential" to cells with low "energy use potential". When calculating carbon emission concentration diffusion, a grid-based finite difference method is used to discretize and solve the diffusion equation, calculating the carbon emission exchange caused by the concentration difference between each grid cell and its surrounding cells. After each time step iteration, the "energy use potential" and "carbon emission concentration" values ​​of all grid cells are updated, thus generating a time slice describing the future state at that moment—an intermediate energy-carbon data carrier. By continuously executing iterative calculations for multiple time steps, the system generates a series of intermediate energy-carbon data carriers describing future states, arranged chronologically. Connecting all intermediate energy-carbon data carriers chronologically forms a complete energy-carbon state extrapolation carrier. This carrier is a four-dimensional data structure, adding a time dimension to the three-dimensional spatial grid topology, recording the complete trajectory of the "energy use potential" and "carbon emission concentration" attribute fields of each grid cell over time.

[0040] Example 4: During the simulation process of the energy and carbon state projection carrier, the attribute field gradient change of each grid cell is monitored in real time. When the attribute field gradient exceeds the safety threshold, a local data field compensation process is triggered to dynamically correct the local grid attributes of the energy and carbon state projection carrier. The final state of the energy and carbon state projection carrier is extracted, and the spatial coordinates of the abnormal energy consumption cluster area, the trajectory of the high carbon emission load path, and the topology of the energy efficiency weak area are analyzed from the final state. Based on the analyzed spatial coordinates of the abnormal energy consumption cluster area, the trajectory of the high carbon emission load path, and the topology of the energy efficiency weak area, a set of control instructions is calculated in reverse on the spatial grid topology. The control instructions include energy flow redirection instructions, carbon emission path blocking instructions, and energy efficiency field enhancement instructions. The control instructions are sent to the corresponding spatial grid of the park's energy and carbon data carrier, driving the park's energy and carbon data carrier to perform an iterative adjustment to generate an optimized park energy and carbon data carrier. When the local data field compensation process is triggered, the magnitude of the energy consumption data attribute field gradient and the magnitude of the carbon emission data attribute field gradient of each grid cell are calculated at each time step in the simulation process. When the magnitude of the energy consumption data attribute field gradient of a certain grid cell exceeds the energy consumption safety threshold, the grid cell is identified as an energy consumption field distortion point, and energy consumption field compensation is initiated. Energy consumption field compensation is achieved by virtually introducing a "load sinking" grid cell. The "load sinking" grid cell is adjacent to the energy consumption field distortion point and has an extremely low initial "energy consumption potential." According to the energy consumption conduction rules, the excess load at the energy consumption field distortion point will be quickly directed to the "load sinking" grid cell, thereby smoothing the field gradient. When the magnitude of the carbon emission data attribute field gradient of a certain grid cell exceeds the carbon emission safety threshold, the grid cell is identified as a carbon emission field distortion point, and carbon emission field compensation is initiated. Carbon emission field compensation is achieved by temporarily increasing the virtual diffusion coefficient between the carbon emission field distortion point and its upwind grid cell, accelerating the migration of high-concentration carbon emissions, thereby reducing the field gradient. After the compensation process is completed, the virtual "load sinking" grid cell is removed, the diffusion coefficient is restored, and the energy and carbon state deduction process continues.

[0041] In practical implementation, during the simulation of the energy and carbon state, the system monitors the attribute field gradient changes of each grid cell in real time. These gradient changes are calculated by measuring the magnitudes of the energy consumption data attribute field gradient and the carbon emission data attribute field gradient for each grid cell at each simulation time step. The magnitude of the energy consumption data attribute field gradient characterizes the spatial drastic change in the grid cell's "energy potential," while the magnitude of the carbon emission data attribute field gradient characterizes the spatial drastic change in the grid cell's "carbon emission concentration." The system compares the calculated gradient magnitudes with preset safety thresholds, which are values ​​pre-set based on historical operational data and safety regulations of the industrial park. When the magnitude of the energy consumption data attribute field gradient of a grid cell exceeds the energy consumption safety threshold, the grid cell is identified as an energy consumption field distortion point, and a local data field compensation process for the energy consumption field is immediately initiated. Energy consumption field compensation is achieved by virtually introducing a "load sinking" grid cell, which is temporarily added to the spatial grid topology at a position adjacent to the energy consumption field distortion point. The "load-heavy" grid cell is assigned an extremely low initial "energy consumption potential" value, such as zero. According to the energy consumption conduction rule, a significant gradient will be generated between the high "energy consumption potential" at the energy consumption field distortion point and the low "energy consumption potential" of the "load-heavy" grid cell. This drives the excess load at the energy consumption field distortion point to be rapidly guided to the "load-heavy" grid cell along the energy consumption conduction path. This process smooths out the gradient of the energy consumption data attribute field at the energy consumption field distortion point. When the modulus of the carbon emission data attribute field gradient of a certain grid cell exceeds the carbon emission safety threshold, the grid cell is determined to be a carbon emission field distortion point, and a local data field compensation process for the carbon emission field is initiated. Carbon emission field compensation is achieved by temporarily increasing the virtual diffusion coefficient between the carbon emission field distortion point and its upwind grid cell. The upwind direction is determined based on the dominant wind direction of the park or real-time wind direction data. The temporarily increased virtual diffusion coefficient enhances the effect of the diffusion term in the carbon emission diffusion rule, thereby accelerating the migration of high-concentration carbon emissions from the carbon emission field distortion point to the upwind grid cell, thus reducing the carbon emission data attribute field gradient at the carbon emission field distortion point. In practice, the local data field compensation process is dynamic and instantaneous. After the compensation process is completed, the system immediately removes the virtually introduced "load sinking" grid cell and restores the temporarily increased virtual diffusion coefficient to its original value, then continues the normal energy and carbon state deduction process. Refer to Table 1 for the triggering conditions and operations of energy field compensation and carbon emission field compensation.

[0042] Table 1: Local Data Field Compensation Triggering and Operation Table

[0043] In some embodiments, the intensity of the compensation operation can be adjusted by a compensation factor. For example, in energy consumption field compensation, the actual load transmitted from the energy consumption field distortion point to the "load sinking" grid cell can be calculated by the following formula:

[0044] in: This represents the amount of compensation load transmitted from the distortion point of the energy field within one compensation step. It is a compensation factor between 0 and 1, used to control the compensation rate. This represents the "energy potential" at time t, indicating the point of distortion in the energy field. This is a preset energy safety threshold corresponding to a potential reference level. This formula ensures that the transmitted load is proportional to the degree to which it exceeds the safety threshold.

[0045] In practical implementation, the final state of the energy and carbon state projection carrier is extracted. The final state refers to the intermediate energy and carbon data carrier corresponding to a predetermined future termination time point in the projection process. From the final state, the spatial coordinates of abnormal energy consumption clusters, the trajectory of high carbon emission load paths, and the topological structure of energy-inefficient weak areas are analyzed. Based on the analyzed spatial coordinates of abnormal energy consumption clusters, the trajectory of high carbon emission load paths, and the topological structure of energy-inefficient weak areas, a set of control commands are calculated in reverse on the spatial grid topology. These control commands include energy flow redirection commands, carbon emission path blocking commands, and energy efficiency field enhancement commands. In essence, reverse calculation refers to deriving the adjustment measures that need to be applied to the spatial grid topology based on the problems discovered in the projection. The calculated control commands are then sent to the corresponding spatial grids of the park's energy and carbon data carriers. The energy flow redirection commands adjust the "energy admittance" parameters between grid cells on the target path, the carbon emission path blocking commands modify the carbon emission diffusion coefficient of the target grid cells, and the energy efficiency field enhancement commands change the energy efficiency calculation parameters of the target grid cells. These instructions drive the park's energy and carbon data carrier to undergo an iterative adjustment. The carrier is reinitialized or short-term simulation based on the new parameters, thereby generating a new set of optimized park energy and carbon data carriers that reflect the potential state after optimization measures. In some embodiments, the optimized park energy and carbon data carriers can be compared and analyzed with the original simulation carriers to evaluate the potential effects of different control strategies.

[0046] Example 5: When resolving the spatial coordinates of abnormal energy consumption clusters, the trajectory of high carbon emission load paths, and the topology of energy-inefficient regions from the final state of the energy-carbon state projection carrier, all grid cells whose "energy consumption potential" exceeds a preset critical value are identified in the final state of the energy-carbon state projection carrier. The set of spatial coordinate codes of these grid cells is marked as an abnormal energy consumption cluster. On the spatial grid topology, the path formed by starting from the grid cell with the highest carbon emission concentration and following the gradient descent direction of the carbon emission data attribute field is traced. The sequence of grid cells traversed by the path is connected to form the trajectory of the high carbon emission load path. The ratio of the average "energy consumption potential" to the average "carbon emission concentration" of each grid cell during the projection process is calculated, and this ratio is defined as the energy efficiency field intensity. On the spatial grid topology, the energy efficiency field intensity values ​​are divided into regions by growth. Continuous grid cell regions with energy efficiency field intensities below the intensity threshold are merged to form the topology of energy-inefficient regions. This topology is described by the region boundary coordinates and internal connectivity. When calculating a set of control commands in reverse on a spatial grid topology, one or more energy diversion paths are calculated for the spatial coordinate set of abnormal energy consumption clusters. These paths connect abnormal energy consumption clusters with grid regions with lower energy loads. The energy flow redirection command includes adjusting the parameter value of "energy admittance" along the energy diversion path. For the trajectory of high carbon emission load paths, key node grid cells are identified. These key node grid cells are the hubs connecting different segments of the path. The carbon emission path blocking command includes setting virtual diffusion barriers at key node grid cells, which are achieved by setting the carbon emission diffusion coefficient of that node to zero. For the topology of energy-inefficient regions, the distribution of energy efficiency field intensity within them is analyzed, and the core sub-region with the lowest intensity is identified. The energy efficiency field enhancement command includes injecting virtual energy efficiency enhancement incentives into the grid cells of the core sub-region. This incentive is manifested as a calculated coefficient that temporarily increases the ratio of "energy consumption potential" to "carbon emission concentration" of that grid cell.

[0047] In practical implementation, the spatial coordinates of abnormal energy consumption clusters, the trajectory of high carbon emission load paths, and the topological structure of energy-inefficient areas are extracted from the final state of the energy and carbon state projection carrier. This process first processes the final state data of the energy and carbon state projection carrier. In the final state of the energy and carbon state projection carrier, the system traverses all grid cells and identifies all grid cells whose "energy consumption potential" values ​​exceed a preset threshold. The preset threshold is a threshold set based on the park's historical maximum load or planned design capacity. The spatial coordinates of each grid cell whose "energy consumption potential" exceeds the preset threshold are encoded and recorded. The set of these spatial coordinate codes is collectively marked as an abnormal energy consumption cluster. In some embodiments, the system also calculates the spatial centroid, coverage area, and the sum and average of the "energy consumption potential" within the cluster as auxiliary descriptive information.

[0048] On a spatial grid topology, the system tracks the trajectory of high-carbon-emission paths, starting from the grid cell with the highest "carbon emission concentration" in the carbon emission data attribute field. From this highest-concentration grid cell, the system moves progressively along the gradient descent direction of the carbon emission data attribute field. This gradient descent direction is defined at each grid cell as pointing towards the minimum "carbon emission concentration" value among adjacent grid cells. At each step, the system encodes and records the spatial coordinates of the current grid cell into the path sequence and moves to the selected adjacent grid cell. The tracking process continues until a grid cell with a "carbon emission concentration" below a preset threshold is reached, or the path length reaches a preset upper limit. Finally, the spatial coordinate encoding sequences of the grid cells traversed throughout the entire tracking process are connected sequentially to form the trajectory of the high-carbon-emission path.

[0049] The ratio of the average "energy consumption potential" to the average "carbon emission concentration" for each grid cell throughout the entire simulation process covered by the energy and carbon state simulation carrier is calculated, and this calculated ratio is defined as the energy efficiency field intensity of that grid cell. The energy efficiency field intensity characterizes the energy consumption level supported by a unit of carbon emission within the simulation period for that grid cell. One possible formula for calculating the energy efficiency field intensity is:

[0050] in: This represents the energy efficiency field strength of grid cell i. This represents the average "energy potential" of grid cell i across all time slices during the simulation process. This represents the average "carbon emission concentration" of grid cell i across all time slices during the simulation process. On the spatial grid topology, the calculated energy efficiency field intensity values ​​of all grid cells are divided into regions through growth. The seed point for region growth is selected as the grid cell with the lowest energy efficiency field intensity value. The growth criterion is to merge spatially connected adjacent grid cells with energy efficiency field intensity values ​​below a preset intensity threshold into the same region. Through iterative growth, continuous grid cell regions with energy efficiency field intensities below the intensity threshold are merged to form the topology of energy-inefficient regions. The topology of energy-inefficient regions is described by the coordinate sequence of the region's outer boundary grid and the connectivity between grids within the region.

[0051] In practical implementation, a set of control commands is calculated in reverse on the spatial grid topology, starting with the calculation of the spatial coordinate set of abnormal energy consumption clusters. The system calculates one or more energy diversion paths on the spatial grid topology. The starting point of each energy diversion path is located at the boundary or inside the abnormal energy consumption cluster, and the ending point connects to a grid region with a lower energy load. Grid regions with lower energy loads are defined by their "energy potential" being below another lower threshold. The path planning algorithm needs to consider the spatial adjacency relationship between grid cells and the initial value of the "energy admittance." The energy flow redirection command includes adjusting the parameter value of the "energy admittance" along the calculated energy diversion path. Specifically, the adjustment method is to increase the "energy admittance" value between adjacent grid cells on the energy diversion path, thereby promoting load transfer along this path in subsequent simulations. For the trajectory of high carbon emission load paths, the system analyzes the grid cell sequence on the trajectory and identifies key node grid cells on the trajectory. The criteria for identifying key node grid cells are their connectivity in the trajectory, such as grid cells connecting multiple path branches or located at narrow points in the path. The carbon emission path blocking instruction includes setting virtual diffusion barriers at identified critical node grid cells. These virtual diffusion barriers are implemented by setting the carbon emission diffusion coefficient to zero on the virtual connections between the critical node grid cell and all its adjacent grid cells, thereby blocking carbon emission diffusion along the original path in the model. In some embodiments, the virtual diffusion barrier can also be set to a very small value instead of absolute zero to simulate the effect of incomplete blocking. For the topology of energy-inefficient regions, the system analyzes the distribution of energy efficiency field intensity within them, identifying the core sub-region with the lowest intensity. The core sub-region is a connected subset within the topology of the energy-inefficient region, where the energy efficiency field intensity values ​​of all grid cells are below a core threshold, which is more stringent than the region threshold.

[0052] See Figure 5 This is a double-line graph showing the evolution of the energy and carbon status of the industrial park over time, illustrating the changing patterns of "average energy consumption potential" and "average carbon emission concentration" across different time slices. Average energy consumption potential (red line): initially rises rapidly (time slices 0-7.5), peaking at nearly 29.5; then declines significantly, with a slight rebound after time slice 15; average carbon emission concentration (blue line): initially rises slowly to approximately 22.5 (time slice 5), then declines continuously, with a slight rebound after time slice 15. This graph corresponds to the projected energy and carbon status of the industrial park. Such graphs can be used to predict peak energy and carbon periods in the park, allowing for advance control strategies; verify the rationality of energy and carbon status projection models; and locate time slices with abnormal energy and carbon fluctuations for further analysis of the local state of grid cells.

[0053] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A secure multi-party computation method for multi-source carbon data in a privacy-preserving computing park, characterized in that, The method includes: A dynamic energy and carbon data carrier for the park is constructed. The energy and carbon data carrier includes a spatial grid topology and a time series topology. The spatial grid topology is divided according to the geographical boundaries and functional zones of the park, and the time series topology is synchronized with the time series period of data collection. Raw energy and carbon data streams are obtained from multiple independent data source nodes, with each data source node corresponding to a spatial grid. The original energy and carbon data stream is injected into the corresponding spatial grid and time series position in the park's energy and carbon data carrier to generate an initialized energy and carbon data carrier. Each grid cell in the initialized energy and carbon data carrier contains an energy consumption data attribute field, a carbon emission data attribute field, and a data quality attribute field. On the initialized energy and carbon data carrier, privacy computing constraints are applied between grid cells belonging to different data source nodes, so that the attribute fields of adjacent grid cells undergo directional data interaction and fusion under the action of the privacy computing constraints, thereby generating a privacy-fused energy and carbon data carrier. Using the energy and carbon data carrier as the initial state, an energy and carbon state deduction process is executed on the spatial grid topology. The energy and carbon state deduction process is calculated according to predefined energy conduction rules and carbon emission diffusion rules to generate an energy and carbon state deduction carrier containing multiple future time slices.

2. The secure multi-party computation method for multi-source carbon data in a privacy computing park according to claim 1, characterized in that, The construction of a dynamic energy and carbon data carrier for the industrial park includes: Based on the vector boundaries of the park map, the park is divided into a set of grid cells of equal area or weighted by functional area, and each grid cell is assigned a unique spatial coordinate code; A local coordinate system is established for each grid cell, with the origin of the local coordinate system located at the center of the grid, which is used to locate the data source nodes inside the grid; Define a continuous time axis and discretize the time axis into a sequence of time steps with the same time step as the minimum interval of data reporting, with each time step corresponding to a time slice; Establish a mapping relationship between the spatial grid topology and the time series topology, so that each grid cell corresponds to a data storage unit in each time slice. The data storage unit is used to store the energy consumption data attribute field, carbon emission data attribute field and data quality attribute field. The values ​​of the energy consumption data attribute field, carbon emission data attribute field, and data quality attribute field are initialized to zero or a preset reference field.

3. The secure multi-party computation method for multi-source carbon data in a privacy computing park according to claim 2, characterized in that, The step of injecting the original energy and carbon data stream into the park's energy and carbon data carrier includes: Receive real-time data packets from the energy data source nodes of the enterprise's production, and parse the instantaneous power value and geographical location tag from the real-time data packets; The spatial coordinate code of the target grid cell is determined based on the geographic location label, and the instantaneous power value is written as the field strength value into the energy consumption data attribute field of the target grid cell in the corresponding time slice. Receive a sequence of sampled data from a carbon emission monitoring data source node, and parse the concentration value and sampling point coordinates from the sampled data sequence; The sampling point coordinates are associated with one or more neighboring grid cells, and the concentration values ​​are assigned to the carbon emission data attribute fields of the one or more neighboring grid cells according to distance weights. At the end of each time slice, a quality evaluation value is calculated based on the data reporting success rate and integrity of each data source node within that time slice, and the quality evaluation value is written into the data quality attribute field of all grid cells.

4. The secure multi-party computation method for multi-source carbon data in a privacy computing park according to claim 3, characterized in that, The application of privacy-preserving computation constraints, enabling directional data interaction and fusion of attribute fields between adjacent grid cells, includes: On the boundary of adjacent grid cells of the initialized energy carbon data carrier, a privacy computing constraint is defined, the strength of which is inversely proportional to the privacy protocol level between the data source nodes to which the adjacent grid cells belong; The privacy computing constraint acts on the property field boundary of adjacent grid cells, driving the property field value of high field strength grid cells to diffuse into low field strength grid cells through gradient diffusion. The diffusion flux is determined by the strength of the privacy computing constraint and the gradient of the property field itself. During the diffusion process, a homomorphic transformation operation is performed on the attribute field values ​​leaving the source grid cell to keep them encrypted during transmission until they enter the target grid cell and are homomorphically aggregated with the attribute field values ​​of the target grid cell. After each fusion cycle, the property field balance of each grid cell is recalculated. If the property field balance does not reach the convergence threshold, the strength parameter of the privacy computing constraint is adjusted, and the next fusion cycle is started until the convergence condition is met, generating a privacy-fused energy and carbon data carrier.

5. The secure multi-party computation method for multi-source carbon data in a privacy computing park according to claim 4, characterized in that, The energy-carbon state deduction process, based on predefined energy conduction rules and carbon emission diffusion rules, includes the following calculations: An initial "energy consumption potential" is assigned to the energy consumption data attribute field of each grid cell of the privacy-fused energy and carbon data carrier, and an initial "carbon emission concentration" is assigned to the carbon emission data attribute field. The energy transfer rule is defined as follows: a grid cell with a high energy potential will transfer energy load to an adjacent grid cell with a low energy potential along the gradient direction of the energy data attribute field. The amount of load transferred is determined by the energy admittance between the two grid cells. The energy admittance is calculated from the device type of the grid cell and the historical rate of change of the energy data attribute field. The carbon emission diffusion rule is defined, which describes the migration of "carbon emission concentration" following the diffusion equation on the spatial grid topology. The diffusion coefficient is associated with the spatial properties of the grid cell and the value of the data quality attribute field. Based on the energy conduction rules and carbon emission diffusion rules, the evolution process of "energy potential" and "carbon emission concentration" of each grid cell is solved step by step on the time series topology to generate a series of intermediate energy and carbon data carriers describing future states. All intermediate energy and carbon data carriers are connected in chronological order to form an energy and carbon state inference carrier, which records the complete trajectory of the attribute field of each grid cell changing over time.

6. The secure multi-party computation method for multi-source carbon data in a privacy computing park according to claim 5, characterized in that, The method further includes: During the simulation process of the energy and carbon state simulation carrier, the property field gradient change of each grid cell is monitored in real time. When the property field gradient exceeds the safety threshold, a local data field compensation process is triggered to dynamically correct the local grid properties of the energy and carbon state simulation carrier. Extract the final state of the energy and carbon state projection carrier, and analyze the spatial coordinates of the abnormal energy consumption accumulation area, the trajectory of the high carbon emission load path, and the topological structure of the energy efficiency weak area from the final state; Based on the spatial coordinates of the abnormal energy consumption cluster, the trajectory of the high carbon emission load path, and the topology of the energy efficiency weak area, a set of control instructions are calculated in reverse on the spatial grid topology. The control instructions include energy flow redirection instructions, carbon emission path blocking instructions, and energy efficiency field enhancement instructions. The control command is sent to the corresponding spatial grid of the park's energy and carbon data carrier, driving the park's energy and carbon data carrier to perform an iterative adjustment and generate an optimized park energy and carbon data carrier.

7. The secure multi-party computation method for multi-source carbon data in a privacy computing park according to claim 6, characterized in that, The process of triggering a local data field compensation includes: At each time step in the simulation process, the magnitude of the gradient of the energy consumption data attribute field and the magnitude of the gradient of the carbon emission data attribute field are calculated for each grid cell. When the magnitude of the gradient of the energy consumption data attribute field of a certain grid cell exceeds the energy consumption safety threshold, the grid cell is determined to be an energy consumption field distortion point, and energy consumption field compensation is initiated. The energy consumption field compensation is achieved by virtually introducing a "load sinking" grid cell. The "load sinking" grid cell is adjacent to the energy consumption field distortion point and has an extremely low initial "energy consumption potential". According to the energy consumption conduction rule, the excess load at the energy consumption field distortion point will be quickly guided to the "load sinking" grid cell, thereby smoothing the field gradient. When the magnitude of the gradient of the carbon emission data attribute field of a certain grid cell exceeds the carbon emission safety threshold, the grid cell is determined to be a carbon emission field distortion point, and carbon emission field compensation is initiated. The carbon emission field compensation is achieved by temporarily increasing the virtual diffusion coefficient between the carbon emission field distortion point and its upwind grid cell, thereby accelerating the migration of high-concentration carbon emissions and reducing the field gradient. After the compensation process is completed, the virtual "load sink" grid cells are removed and the diffusion coefficient is restored, and the energy carbon state deduction process continues.

8. The secure multi-party computation method for multi-source carbon data in a privacy computing park according to claim 7, characterized in that, The process of resolving the spatial coordinates of the abnormal energy consumption cluster, the trajectory of the high carbon emission load path, and the topology of the energy-inefficient region from the final state includes: In the final state of the energy carbon state projection carrier, all grid cells whose "energy consumption potential" exceeds a preset critical value are identified, and the set of spatial coordinate codes of these grid cells is marked as an abnormal energy consumption cluster area. On the spatial grid topology, starting from the grid cell with the highest carbon emission concentration, the path formed along the gradient descent direction of the carbon emission data attribute field is traced, and the grid cell sequence traversed by the path is connected to form the trajectory of the high carbon emission load path. Calculate the ratio of the average "energy utilization potential" to the average "carbon emission concentration" of each grid cell during the simulation process, and define the ratio as the energy efficiency field strength; On the spatial grid topology, the energy efficiency field intensity value is divided into regions by growth. Continuous grid cell regions with energy efficiency field intensities below the intensity threshold are merged to form a topological structure of energy-inefficient regions. The topological structure is described by the region boundary coordinates and internal connectivity.

9. The secure multi-party computation method for multi-source carbon data in a privacy computing park according to claim 8, characterized in that, The reverse calculation of a set of control instructions on the spatial grid topology includes: For the set of spatial coordinates of the abnormal energy consumption accumulation area, calculate one or more energy diversion paths. The energy diversion paths connect the abnormal energy consumption accumulation area with the grid area with low energy load. The energy flow redirection command includes adjusting the parameter value of "energy admittance" along the energy diversion path. For the trajectory of the high carbon emission load path, key node grid cells on the trajectory are identified. The key node grid cells are hubs connecting different segments of the path. The carbon emission path blocking command includes setting a virtual diffusion barrier at the key node grid cell. The virtual diffusion barrier is implemented by setting the carbon emission diffusion coefficient of the node to zero. For the topology of the energy-inefficient region, the distribution of energy efficiency field intensity inside it is analyzed, and the core sub-region with the lowest intensity is identified. The energy efficiency field enhancement command includes injecting virtual energy efficiency enhancement incentives into the grid cells of the core sub-region. The incentive is reflected in the calculation coefficient that temporarily increases the ratio of "energy use potential" to "carbon emission concentration" of the grid cell.

10. A secure multi-party computation system for carbon multi-source data in a privacy computing park, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the secure multi-party computation method for carbon multi-source data in the privacy computing park as described in any one of claims 1 to 9.

Images