Carbon flow data dynamic analysis method and system based on digital twinning

Abstract

The invention discloses a carbon flow data dynamic analysis method and system based on digital twinning, and solves the problems that in the prior art, carbon flow data analysis accuracy is low, and accurate carbon emission prediction cannot be carried out, the method comprises the steps that multi-dimensional data are collected and fused, and a fused data set is obtained; performing carbon emission analysis on the fused data set; constructing a digital twinborn model of the warehousing and logistics process, integrating a carbon emission analysis result into the digital twinborn model, and performing carbon emission prediction based on the digital twinborn model; generating a carbon emission optimization strategy based on a multi-objective optimization algorithm according to the carbon emission prediction result and the carbon emission analysis result; and executing the carbon emission optimization strategy, and continuously optimizing the digital twin model according to an execution result. The limitation that traditional carbon emission management depends on manual statistics and simple data analysis is overcome, the dynamics and complexity of the warehouse logistics process are effectively dealt with, and accurate carbon emission prediction can be carried out.

Description

Technical Field

[0001] This invention relates to the field of carbon emission analysis technology, and in particular to a method and system for dynamic analysis of carbon flow data based on digital twins. Background Technology

[0002] As global climate change becomes increasingly severe, a low-carbon economy has become a crucial development direction for countries worldwide. Against this backdrop, businesses face immense pressure and challenges in reducing carbon emissions and achieving green operations. This is particularly true in the warehousing and logistics industry, where complex operational processes and high energy consumption make carbon emission management especially critical. Traditional carbon emission management methods rely heavily on manual statistics and simple data analysis, rarely considering the dynamic and complex nature of warehousing and logistics processes. This lack of consideration hinders accurate analysis of carbon flow data, resulting in low-precision carbon emission predictions and an inability to provide accurate data support for carbon emission management. Summary of the Invention

[0003] The purpose of this invention is to address the problem that existing carbon flow data analysis does not consider the dynamics and complexity of warehousing and logistics processes, resulting in low accuracy. This invention provides a dynamic carbon flow data analysis method and system based on digital twins, overcoming the limitations of traditional carbon emission management that relies on manual statistics and simple data analysis. It effectively addresses the dynamics and complexity of warehousing and logistics processes, enabling accurate carbon emission prediction and multi-objective optimization based on the prediction results, generating carbon emission optimization strategies. This provides strong technical support for enterprises to achieve low-carbon and green operations.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: A dynamic analysis method for carbon flow data based on digital twins includes the following steps: S1: Collect multi-dimensional data and perform fusion processing to obtain a fused dataset; S2: Encrypt and store the fused dataset, and perform carbon emission analysis on the fused dataset; S3: Construct a digital twin model of the warehousing and logistics process, integrate carbon emission analysis results into the digital twin model, and predict carbon emissions based on the digital twin model; S4: Based on the carbon emission prediction results and carbon emission analysis results, generate a carbon emission optimization strategy using a multi-objective optimization algorithm; S5: Execute the carbon emission optimization strategy and continuously optimize the digital twin model based on the execution results.

[0005] The method of this invention constructs a digital twin model reflecting warehousing and logistics processes by collecting and fusing multi-dimensional data. This model enables carbon emission prediction and, based on the prediction results, performs multi-objective optimization to generate carbon emission optimization strategies. Furthermore, it allows for continuous optimization of the digital twin model based on the execution results of the carbon emission optimization strategies, forming a closed-loop dynamic analysis process. This overcomes the limitations of traditional carbon emission management, which relies on manual statistics and simple data analysis, and effectively addresses the dynamic and complex nature of warehousing and logistics processes. Through real-time data collection, dynamic model updates, and continuous optimization, it provides strong technical support for enterprises to achieve low-carbon and green operations.

[0006] Preferably, step S3 includes: S3.1: Construct a multi-layered digital twin model that includes warehouse layout, IoT device distribution, and logistics routes; integrate the carbon emission intensity heat map into the multi-layered digital twin model; S3.2: Use historical carbon emission data and multi-dimensional data as input; use a long short-term memory network to extract time series features; S3.3: Integrate spatial and temporal features through an attention mechanism; S3.4: Utilize ensemble learning methods to predict carbon emissions at multiple scales, including short-term, medium-term, and long-term. S3.5: Output the expected carbon emissions and carbon emission trends for a specified future period.

[0007] Preferably, step S5, which involves continuously optimizing the digital twin model, includes: S5.1: Record the results of the optimization strategy execution to the blockchain network via a smart contract; S5.2: Utilize historical data from the blockchain network to construct a reputation assessment model and dynamically adjust the credibility of different data sources; S5.3: Based on the execution results, use reinforcement learning methods to optimize the parameters of the digital twin model; S5.4: Update the carbon emission intensity heat map and carbon emission hotspots based on the optimized digital twin model; S5.5: Based on the updated carbon emission intensity heat map and carbon emission hotspots, evaluate the long-term effects of the optimization strategy and adjust the optimization strategy accordingly.

[0008] Preferably, step S1, which involves fusing multi-dimensional data, includes: S1.1: Utilize the spatiotemporal tags of IoT devices to align warehouse management system data with IoT device data in time and space, and build a unified data view; S1.2: Based on the collected multi-dimensional data, calculate the energy efficiency index of IoT devices, construct a comprehensive environmental index, and calculate the transportation carbon emission factor; S1.3: An attention mechanism is used to fuse the calculation results in S1.2 with the collected original multi-dimensional data to obtain a comprehensive feature representation.

[0009] Preferably, in step S2, the carbon emission analysis includes: S2.1: Allocate total carbon emissions to each batch of materials using the activity-based costing method based on time percentage, weight percentage, and path characteristics; S2.2: Use spatial autocorrelation analysis to identify carbon emission hotspots and construct a carbon emission intensity heat map; S2.3: Calculate the carbon emission contribution rate of each region, and set the weights of the multi-objective optimization algorithm based on the carbon emission contribution rate.

[0010] Preferably, the multi-objective optimization algorithm takes minimizing total carbon emissions and minimizing operating costs as objective functions, and carbon emission caps and budget constraints as constraints. The decision variables of the multi-objective optimization algorithm include equipment energy efficiency optimization parameters, intelligent scheduling parameters, and path planning parameters.

[0011] As a preferred option, the encrypted fusion dataset is stored through a blockchain network, and the execution results of the carbon emission optimization strategy are recorded through the blockchain network.

[0012] As a preferred option, homomorphic encryption algorithm is used to encrypt the fused dataset; the blockchain network adopts a consortium blockchain structure, and participating nodes include warehousing enterprises, logistics companies and regulatory agencies.

[0013] Preferably, the multi-dimensional data includes warehouse management system data, IoT device data, and transportation system data.

[0014] A dynamic analysis system for carbon flow data based on digital twins, comprising: The data processing module collects multi-dimensional data, performs fusion processing on the multi-dimensional data based on the attention mechanism, encrypts the fused dataset, and stores the encrypted fused dataset using a blockchain network. The digital twin model module constructs a digital twin model of the warehousing and logistics process, integrates carbon emission analysis results, performs carbon emission prediction, and continuously optimizes the data twin model based on the execution results of carbon emission optimization strategies. The carbon emission analysis module performs carbon emission analysis based on the fused dataset, and generates carbon emission optimization strategies based on the carbon emission analysis results and carbon emission prediction results using a multi-objective optimization algorithm.

[0015] Therefore, the present invention has the following beneficial effects: 1. By collecting and fusing multi-dimensional data, a digital twin model reflecting the warehousing and logistics process was constructed. This model enables carbon emission prediction and, based on the prediction results, performs multi-objective optimization to generate carbon emission optimization strategies. The execution results of these optimization strategies are recorded in a blockchain network, ensuring the credibility and security of the data.

[0016] 2. It can continuously optimize the digital twin model based on the execution results of carbon emission optimization strategies, forming a closed-loop dynamic analysis process. This overcomes the limitations of traditional carbon emission management, which relies on manual statistics and simple data analysis, and effectively addresses the dynamic and complex nature of warehousing and logistics processes. Through real-time data collection, dynamic model updates, and continuous optimization, it provides strong technical support for enterprises to achieve low-carbon and green operations. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the overall steps of the dynamic analysis method for carbon flow data based on digital twins in this invention. Detailed Implementation

[0018] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments: Example 1: This embodiment provides a method for dynamic analysis of carbon flow data based on digital twins, such as... Figure 1 As shown, the operation process is as follows: Step 1, collect multi-dimensional data and perform fusion processing to obtain a fused dataset; Step 2, encrypt and store the fused dataset, and perform carbon emission analysis on the fused dataset; Step 3, construct a digital twin model of the warehousing and logistics process, integrate the carbon emission analysis results into the digital twin model, and predict carbon emissions based on the digital twin model; Step 4, generate a carbon emission optimization strategy based on a multi-objective optimization algorithm according to the carbon emission prediction results and carbon emission analysis results; Step 5, execute the carbon emission optimization strategy, and continuously optimize the digital twin model according to the execution results.

[0019] The carbon flow data dynamic analysis method based on digital twins provided in this embodiment constructs a digital twin model reflecting warehousing and logistics processes by collecting and fusing multi-dimensional data. This model enables carbon emission prediction and, based on the prediction results, performs multi-objective optimization to generate carbon emission optimization strategies. The execution results of the optimization strategies are recorded in a blockchain network, ensuring data credibility and security. This method also continuously optimizes the digital twin model based on the execution results, forming a closed-loop dynamic analysis process. It overcomes the limitations of traditional carbon emission management, which relies on manual statistics and simple data analysis, and effectively addresses the dynamic and complex nature of warehousing and logistics processes. Through real-time data acquisition, dynamic model updates, and continuous optimization, it provides strong technical support for enterprises to achieve low-carbon and green operations.

[0020] The following examples and specific application scenarios further illustrate the technical solution and effects of the present invention. The following examples are explanations of the present invention, but the present invention is not limited to the following examples.

[0021] Specifically, this manifests as follows: Step 1: Collect multi-dimensional data, including data from the warehouse management system, IoT devices, and transportation systems.

[0022] Multi-dimensional data is collected through the data acquisition module. This multi-dimensional data includes data from the warehouse management system, such as information on incoming materials, outgoing materials, and inventory; data from IoT devices, such as equipment energy consumption and environmental data; and data from the transportation system, such as vehicle identification, routes, transportation distances, and fuel consumption.

[0023] Furthermore, for multi-dimensional data, including: 1. Warehouse management system data: such as inbound, outbound, and inventory information.

[0024] Specifically, regarding warehouse management system data: A warehouse management system (WMS) is one of the key sources of data collection in this step. Specifically, it includes: a) Inbound Material Information: Record all material details entering the warehouse, including but not limited to: material number, material name, material category, quantity, weight, volume, inbound time, supplier information, and storage location.

[0025] b) Outbound Material Information: Record all material details leaving the warehouse, including but not limited to: material number, material name, material category, quantity, weight, volume, outbound time, destination information, and transportation method.

[0026] c) Inventory information: Records the real-time status of all materials in the current warehouse, including but not limited to: material number, material name, material category, current quantity, storage location, and date of receipt.

[0027] 2. IoT device data: such as energy consumption data and environmental monitoring data.

[0028] For IoT device data, these devices are deployed in warehousing and logistics processes to monitor environmental parameters and energy consumption in real time. Specifically, this includes: a) Equipment energy consumption data: Electricity consumption: Electricity consumption of various electrical equipment (such as lighting systems, refrigeration equipment, conveyor belts, etc.); Fuel consumption: Fuel consumption of equipment such as forklifts and generators; Water consumption: Clean water consumption.

[0029] b) Environmental data: Temperature: Real-time temperature in different areas of the warehouse; Humidity: Real-time humidity in different areas of the warehouse; Air quality index: Including CO2 concentration, particulate matter concentration, etc.

[0030] 3. Transportation system data: such as vehicle information, routes, and fuel consumption.

[0031] In this embodiment, the transportation system data mainly comes from the logistics management system (TMS), specifically including: a) Vehicle identification: including license plate number and vehicle type (e.g., refrigerated truck, regular truck, etc.); b) Route information: origin and destination, waypoints, and estimated route; c) Transportation distance: The actual mileage traveled for each transportation mission; d) Fuel consumption: Fuel usage for each transport mission; for electric vehicles, record the amount of charge.

[0032] These data are transmitted to the data acquisition module in real time through their respective data interfaces. By collecting these multi-dimensional data from the above embodiments, the dynamic carbon and sulfur data analysis method provided in this embodiment can comprehensively grasp various key information in the warehousing and logistics process, providing a solid data foundation for subsequent data fusion, carbon emission analysis, and optimization strategy formulation.

[0033] The second step is to fuse the multi-dimensional data to obtain a fused dataset.

[0034] The collected multi-dimensional data is fused to form a unified fused dataset. The purpose of the fusion process is to integrate data from different sources into a unified dataset to facilitate subsequent analysis.

[0035] Furthermore, the fusion of multi-dimensional data includes: By leveraging the spatiotemporal tags of IoT devices, warehouse management system data and IoT device data are spatiotemporally aligned to build a unified data view; Calculate the equipment energy efficiency index based on the energy consumption value and equipment utilization rate in the equipment energy consumption data; A comprehensive environmental index is constructed by combining temperature, humidity, and air quality index data from environmental data. Based on the origin of goods entering the warehouse and the destination of goods leaving the warehouse, combined with transportation system data, the transportation carbon emission factor is calculated. A multi-source data fusion method based on attention mechanism is adopted to fuse equipment energy efficiency indicators, comprehensive environmental index, and transportation carbon emission factors with data from warehouse management system, IoT device data, and transportation system data to obtain a comprehensive feature representation; the comprehensive feature representation is used as input data for the carbon emission analysis and the carbon emission prediction.

[0036] Step 3: Encrypt the fused dataset and store the encrypted fused dataset through a blockchain network.

[0037] To ensure data security and credibility, this method uses encryption algorithms to encrypt the merged dataset, and the encrypted data is stored through a blockchain network.

[0038] Homomorphic encryption allows specific computational operations to be performed on ciphertext without decryption, thus protecting data privacy. The use of blockchain networks ensures the immutability and traceability of data. Specifically: blockchain networks adopt a consortium blockchain structure, with participating nodes including warehousing companies, logistics companies, and regulatory agencies; blockchain networks use consensus mechanisms to ensure data consistency and implement automated data verification and access control through smart contracts. Each participant acts as a blockchain node, collectively maintaining data consistency and integrity.

[0039] Step 4: Perform carbon emission analysis based on the fused dataset to obtain the carbon emission analysis results.

[0040] This step utilizes a fused dataset encrypted and stored on a blockchain to conduct carbon emissions analysis. The analysis process considers carbon emission factors at various stages of warehousing and logistics, yielding carbon emissions analysis results. These results provide data support for subsequent decision-making.

[0041] Further carbon emission analysis includes: The total carbon emissions are allocated to each batch of materials using the activity-based costing method that takes into account time percentage, weight percentage, and path characteristics. Spatial autocorrelation analysis was used to identify carbon emission hotspots. Based on the above analysis results, a carbon emission intensity heat map was constructed. Calculate the carbon emission contribution rate of each region, which is used to set the weights for the multi-objective optimization algorithm.

[0042] Step 5: Build a digital twin model of the warehousing and logistics process and integrate the carbon emission analysis results into the digital twin model.

[0043] Based on the data and analysis results from the preceding steps, this step constructs a digital twin model of the warehousing and logistics processes. The digital twin model is a digital mapping of the physical world, containing key information such as warehouse layout and logistics routes.

[0044] Specifically, this includes: constructing a multi-layered digital twin model that includes warehouse layout, IoT device distribution, and logistics routes, where IoT devices include energy consumption monitoring devices and environmental monitoring devices; and integrating carbon emission intensity heat maps into the multi-layered digital twin model.

[0045] Step 6: Based on the digital twin model, predict carbon emissions to obtain the carbon emission prediction results.

[0046] Carbon emissions are predicted using a constructed digital twin model that combines historical and real-time data. The predictions include expected carbon emissions and trends for a future period.

[0047] Furthermore, carbon emission predictions are made based on digital twin models, yielding carbon emission prediction results, including: Historical carbon emission data, real-time warehouse management system data, IoT device data, and transportation system data are used as inputs; Use a long short-term memory network to extract time series features; Integrating spatial and temporal features through attention mechanisms; Using ensemble learning methods, we can predict carbon emissions at multiple scales, including short-term, medium-term, and long-term. Output the prediction results.

[0048] Step 7: Based on the carbon emission analysis results and carbon emission prediction results, execute a multi-objective optimization algorithm to generate a carbon emission optimization strategy.

[0049] Based on the carbon emission analysis and prediction results, a multi-objective optimization algorithm is executed. This algorithm considers multiple objectives, such as minimizing carbon emissions and minimizing operating costs, and generates a series of feasible carbon emission optimization strategies.

[0050] Specifically, multi-objective optimization algorithms include the following objectives and constraints: Objectives: Minimize total carbon emissions and minimize operating costs; Constraints: Carbon emission caps, budget restrictions; Decision variables include: equipment energy efficiency optimization parameters; intelligent scheduling parameters; and path planning parameters.

[0051] Step 8: Implement carbon emission optimization strategies and record the results via a blockchain network.

[0052] The selected optimization strategies are implemented in the actual warehousing and logistics system. The execution process and results are recorded through a blockchain network, ensuring the authenticity and traceability of the data.

[0053] Specifically, recording the results of implementing carbon emission optimization strategies includes: a) Smart contract design: Define the record structure, including policy ID, execution time, execution parameters, actual carbon emissions, actual cost, etc.; implement the record write function to ensure that only authorized nodes can write data; implement the data query function to facilitate subsequent analysis.

[0054] b) Blockchain recording process: When the strategy execution begins, the smart contract is triggered to record the initial state; during the strategy execution, the intermediate state is updated periodically; when the strategy execution ends, the final result is recorded.

[0055] Step 9: Continuously optimize the digital twin model based on the execution results.

[0056] Finally, the digital twin model is continuously optimized based on the actual results of strategy execution. This process includes updating model parameters and adjusting prediction algorithms to ensure that the model can more accurately reflect the actual situation.

[0057] Continuous optimization of digital twin models specifically includes: The results of the optimization strategy execution will be recorded on the blockchain network via smart contracts. By utilizing historical data in the blockchain network, a reputation assessment model can be built to dynamically adjust the credibility of different data sources; Based on the execution results, reinforcement learning methods are used to continuously optimize the parameters of the digital twin model; Based on the optimized digital twin model, update the carbon emission intensity heat map and carbon emission hotspot areas; Based on the updated carbon emission intensity heatmap and carbon emission hotspots, the long-term effects of the optimization strategy are evaluated, and the optimization strategy is adjusted accordingly.

[0058] Through the above steps, the digital twin-based dynamic carbon flow data analysis method provided in this embodiment enables dynamic analysis and continuous optimization of carbon emissions during warehousing and logistics processes, providing strong support for enterprises to achieve green and low-carbon operations. It also combines the data security of blockchain technology with the high-fidelity simulation capabilities of digital twin technology, achieving reliable analysis of carbon emission data and precise formulation of optimization strategies. Furthermore, it enables continuous optimization of the digital twin model and dynamic adjustment of optimization strategies. This closed-loop optimization mechanism ensures the accuracy of carbon emission analysis and prediction, while adapting to environmental changes and evolving enterprise needs, providing continuous decision support for long-term low-carbon operations.

[0059] Example 2: This embodiment provides a dynamic analysis method for carbon flow data based on digital twins. Based on Embodiment 1, and combined with specific application scenarios, the multi-dimensional data fusion method in Embodiment 1 is further optimized.

[0060] In this embodiment, the fusion of multi-dimensional data includes: Step (1): Spatiotemporal alignment.

[0061] By leveraging the spatiotemporal tags of IoT devices, warehouse management system data is spatiotemporally aligned with IoT device data to construct a unified data view. Specific steps include: a) Time synchronization: unify the timestamps of all data sources to the same time zone and precision (e.g., unify to UTC time, accurate to the second); b) Spatial mapping: Establish a spatial coordinate system for warehouses and transportation routes, and map the location information of various data sources to this unified coordinate system; c) Data matching: Based on time and space information, data from different sources are matched to form a unified spatiotemporal data structure.

[0062] Step (2): Calculate energy efficiency indicators.

[0063] Based on the energy consumption values ​​and equipment utilization rate in the equipment energy consumption data, the equipment energy efficiency index is calculated. The specific method is as follows: a) Calculate energy consumption per unit time: energy consumption value / time; b) Calculate equipment utilization rate: actual working time / total time; c) Energy efficiency index = Energy consumption per unit time / Equipment utilization rate. This index reflects the energy utilization efficiency of the equipment during actual use.

[0064] Step (3): Construction of comprehensive environmental index.

[0065] A comprehensive environmental index is constructed by combining temperature, humidity, and air quality index data. The calculation method is as follows: Comprehensive Environmental Index = w1 * Temperature Index + w2 * Humidity Index + w3 * Air Quality Index. Where w1, w2, and w3 are weighting coefficients, which can be adjusted according to specific application scenarios.

[0066] Step (4): Calculation of transport carbon emission factor.

[0067] Based on the origin of goods entering the warehouse and the destination of goods leaving the warehouse, combined with transportation system data, the transportation carbon emission factor is calculated. The calculation method is as follows: a) Calculate the transportation distance: Based on the origin and destination information, combined with the actual transportation route, calculate the transportation distance; b) Determine the carbon emission coefficient of the transportation vehicle: Determine its carbon emission coefficient based on the energy type and efficiency of different transportation vehicles (such as trucks, trains, ships, etc.); c) Transportation carbon emission factor = transportation distance * transportation vehicle carbon emission coefficient.

[0068] Step (5): Multi-source data fusion.

[0069] A multi-source data fusion method based on attention mechanism is adopted to fuse the equipment energy efficiency index, environmental comprehensive index and transportation carbon emission factor calculated above with the original warehouse management system data, IoT device data and transportation system data to obtain a comprehensive feature representation.

[0070] The specific integration process is as follows: a) Feature vector construction: Construct feature vectors from various indicators (indices, transportation carbon emission factors) and raw data.

[0071] b) Attention weight calculation: The importance weight of each feature is calculated using the attention mechanism.

[0072] c) Weighted fusion: Based on the attention weights, the features are weighted and summed to obtain the final comprehensive feature representation.

[0073] Step (6): Data output.

[0074] The resulting comprehensive feature representation is used as input data for carbon emission analysis and carbon emission prediction. This comprehensive feature representation contains multi-dimensional and multi-scale information, which can comprehensively reflect various key factors in the warehousing and logistics process.

[0075] Through the above steps, the carbon flow data dynamic analysis method based on digital twins provided in this embodiment achieves efficient fusion processing of multi-dimensional data, providing high-quality data input for subsequent carbon emission analysis and prediction, thereby improving the accuracy and reliability of carbon emission analysis and optimization.

[0076] In this embodiment, the raw data includes warehouse management system data, IoT device data, and transportation system data.

[0077] Example 3: This embodiment provides a dynamic analysis method for carbon flow data based on digital twins. Building upon Embodiment 1, this embodiment further optimizes the carbon emission analysis method in Embodiment 1 in conjunction with specific application scenarios.

[0078] In this embodiment, carbon emission analysis specifically includes: (1) Carbon emission allocation using the activity-based costing method.

[0079] The total carbon emissions are allocated to each batch of materials using the activity-based costing method, which considers time percentage, weight percentage, and path characteristics. The specific steps are as follows: Step a): Calculation of time percentage: Time percentage = Time spent on storage and transportation of a single batch of materials / Total observation time.

[0080] Step b): Weight percentage calculation: Weight percentage = weight of single batch of materials / total weight of materials.

[0081] Step c): Calculation of route feature factors: Route feature factor = f(transportation distance, transportation mode, road condition complexity), where f is a weighting function that can be adjusted according to the actual situation.

[0082] Step d): Carbon emission allocation: Carbon emission of a single batch of materials = Total carbon emission * (w1 * Time proportion + w2 * Weight proportion + w3 * Path characteristic factor), where w1, w2, and w3 are weighting coefficients, satisfying w1 + w2 + w3 = 1.

[0083] This method allows for a more accurate allocation of carbon emissions to each batch of goods, reflecting the actual carbon emissions of different goods during warehousing and logistics.

[0084] (2) Spatial autocorrelation analysis.

[0085] Spatial autocorrelation analysis was used to identify carbon emission hotspots. The specific steps are as follows: Step a): Spatial unit division: Divide the warehousing and logistics network into several spatial units.

[0086] Step b): Calculation of local spatial autocorrelation index.

[0087] For each spatial cell, calculate the local Moran's I index: Where Ii is the local Moran's I index of the i-th spatial unit, and xi is the carbon emissions of the i-th unit. Let σi be the average carbon emissions of all units, σ2 be the variance of carbon emissions, wij be the weight between units i and j in the spatial weight matrix, and xj be the carbon emissions of unit j.

[0088] Step c): Calculation of the global spatial autocorrelation index.

[0089] Calculate the global Moran's I exponent: Where n is the total number of spatial units, wij is the same as above, xi and xj are the same as above. Same as above.

[0090] Step d): Hotspot identification: Based on the local Moran's I index values ​​and significance levels, identify carbon emission hotspots and coldspots.

[0091] (3) Construction of carbon emission intensity heat map. Based on the above analysis results, a carbon emission intensity heatmap is constructed. The specific steps are as follows: Step a): Gridding: Divide the warehousing and logistics network into a fine grid.

[0092] Step b): Interpolation: Using Kriging interpolation, estimate the carbon emission intensity of unknown points in the grid based on known carbon emission data points.

[0093] Step c): Color mapping: Map carbon emission intensity values ​​to a color gradient, using warm colors (such as red) for high intensity areas and cool colors (such as blue) for low intensity areas.

[0094] Step d): Visualization: Generate a heat map to visually display the carbon emissions in different regions.

[0095] (4) Calculation of carbon emission contribution rate. Calculate the carbon emission contribution rate of each region for weight setting in the multi-objective optimization algorithm. The specific steps are as follows: Step a): Regional carbon emissions calculation: For each identified region, calculate its total carbon emissions.

[0096] Step b): Contribution rate calculation: Regional carbon emission contribution rate = Regional carbon emissions / Total carbon emissions.

[0097] Step c): Weight setting: Based on the contribution rate, set the optimization weights for each region in the multi-objective optimization algorithm. The higher the contribution rate, the greater the optimization weight.

[0098] Through the above steps, a comprehensive analysis of carbon emissions was achieved. This not only accurately quantifies the carbon emissions of each batch of materials but also identifies carbon emission hotspots, providing strong support for the formulation of subsequent optimization strategies. The construction of a carbon emission intensity heat map makes the carbon emission situation more intuitive and visible, helping decision-makers quickly grasp the overall carbon emission status. Furthermore, the calculation of the carbon emission contribution rate provides a basis for optimizing the rational allocation of resources, ensuring that the optimization effect is maximized.

[0099] Example 4: This embodiment provides a dynamic analysis method for carbon flow data based on digital twins. Based on Embodiment 1, the process of constructing the digital twin model in Embodiment 1 is further optimized in combination with specific application scenarios.

[0100] Specifically, in this embodiment, constructing a digital twin model of the warehousing and logistics process includes the following steps: Step (1): Construct a multi-layer digital twin model.

[0101] Construct a multi-layered digital twin model that includes warehouse layout, IoT device distribution, and logistics routes. The specific implementation is as follows: a) Physical layer.

[0102] Using 3D modeling technology, a three-dimensional model of the warehouse is constructed based on the CAD drawings of the actual storage facilities; the location of key facilities such as shelves, conveyor belts, and loading and unloading areas is marked in the model; and logistics paths are drawn, including the material flow routes inside the warehouse and external transportation routes.

[0103] b) Equipment layer.

[0104] Mark the distribution locations of IoT devices in the physical layer model.

[0105] Energy consumption monitoring equipment: such as smart meters, fuel consumption sensors, etc.

[0106] Environmental monitoring equipment: such as temperature and humidity sensors, air quality monitors, etc.

[0107] c) Data layer.

[0108] Establish a real-time data interface to map data collected by IoT devices into the digital twin model in real time; design a data visualization interface to intuitively display the real-time status and data of each device.

[0109] d) Analysis layer.

[0110] It integrates a data analysis module to process and analyze real-time data; it also enables carbon emission calculation, dynamically calculating carbon emissions based on real-time data.

[0111] Step (2): Integrate the carbon emission intensity heat map.

[0112] The carbon emission intensity heatmap generated in the preceding steps is integrated into a multi-layered digital twin model: a) Overlay of heatmaps.

[0113] The two-dimensional heatmap is converted into a three-dimensional surface and overlaid on the physical layer of the digital twin model; a semi-transparent effect is used to ensure that the underlying physical structure remains visible.

[0114] b) Dynamic updates.

[0115] The design incorporates a dynamic heatmap update mechanism to adjust the color intensity of the heatmap based on real-time data; it also implements timeline control, allowing users to view changes in carbon emission intensity over different time periods.

[0116] c) Interactive functions.

[0117] Implement the function of displaying specific carbon emission values ​​when the mouse hovers over the data; add a region selection tool to allow users to select specific regions for in-depth analysis.

[0118] Through the above steps, high-fidelity digital twin modeling of warehousing and logistics processes was achieved, and carbon emission analysis was deeply integrated into the model.

[0119] Example 5: This embodiment provides a dynamic analysis method for carbon flow data based on digital twins. Building upon Embodiment 1, and considering specific application scenarios, the process of predicting carbon emissions using a digital twin model in Embodiment 1 is further optimized.

[0120] Specifically, carbon emission prediction is performed using the constructed digital twin model, and the specific steps are as follows: Step (1): Data preparation.

[0121] Historical carbon emission data: Extract carbon emission records from a database for a certain period of time (such as one year).

[0122] Real-time warehouse management system data includes current inventory levels, inbound and outbound plans, etc.

[0123] IoT device data: including real-time energy consumption data, environmental data, etc.; transportation system data: including current and planned transportation task information.

[0124] Step (2): Time series feature extraction.

[0125] Long Short-Term Memory (LSTM) networks are used to process time-series data; long-term dependencies and periodic patterns in the data are learned through LSTM networks.

[0126] LSTM network structure: input layer, LSTM layer (multiple layers are possible), fully connected layer, output layer.

[0127] Step (3): Integrating spatial and temporal features.

[0128] Design a multi-head attention mechanism that focuses on spatial and temporal features separately. The multi-head attention mechanism includes: Spatial attention: Focusing on the carbon emission contributions of different regions.

[0129] Time attention: Focusing on carbon emission patterns across different time scales.

[0130] The outputs of the two attention mechanisms are weighted and fused.

[0131] Step (4): Multi-scale carbon emission prediction.

[0132] Employ ensemble learning methods, such as random forests or gradient boosting decision trees (e.g., XGBoost).

[0133] Construct three prediction models: Short-term forecasting model: Predicts carbon emissions over the next 24 hours.

[0134] Medium-term forecasting model: Predicts carbon emissions for the next 1-4 weeks.

[0135] Long-term forecasting model: Predicts carbon emissions for the next 1-12 months.

[0136] Each model is trained independently, but they share basic features.

[0137] Step (5): Output the prediction results.

[0138] Generate forecast reports, including: expected carbon emissions for a specified future period (short-term, medium-term, and long-term); carbon emission trend graphs showing the trend of carbon emission changes during the forecast period; uncertainty estimates, such as forecast ranges or confidence levels; visualization of forecast results and integration into the interface of the digital twin model; and provision of interpretive analysis of forecast results, such as the main factors affecting the forecast.

[0139] Example 6: This embodiment provides a dynamic analysis method for carbon flow data based on digital twins. Building upon Embodiment 1, and considering specific application scenarios, the method for generating carbon emission optimization strategies using the multi-objective optimization algorithm in Embodiment 1 is further optimized.

[0140] In this embodiment, a multi-objective optimization algorithm is used to generate a carbon emission optimization strategy. The specific implementation of the algorithm is as follows: Step (1): Determine the optimization objective.

[0141] a) Minimize total carbon emissions: min F1=Σ(Ei+Ti); Where Ei represents carbon emissions from the warehousing process, which can include emissions from warehouse lighting, air conditioning, equipment operation, etc.; Ti represents carbon emissions from the transportation process, which mainly comes from fuel consumption of transport vehicles; and F1 represents the total carbon emissions that need to be minimized.

[0142] b) Minimize operating costs: min F2=Σ(Ci+Mi+Li); Where Ci is the equipment operating cost, Mi is the labor cost, Li is the logistics cost, and F2 is the total operating cost that needs to be minimized.

[0143] Step (2): Determine the constraints.

[0144] a) Carbon emission cap.

[0145] Σ(Ei+Ti)≤Emax; where Emax is the carbon emission cap set by the government or enterprise.

[0146] b) Budget constraints.

[0147] Σ(Ci+Mi+Li)≤Bmax; where Bmax is the maximum budget set by the enterprise.

[0148] Step (3): Determine the decision variables.

[0149] a) Equipment energy efficiency optimization parameters.

[0150] Air conditioning temperature setting: x1∈[18℃, 28℃]; Lighting system brightness: x2∈[0%, 100%]; Refrigeration equipment operating frequency: x3∈[0, 1].

[0151] b) Intelligent scheduling parameters.

[0152] Warehouse space allocation strategy: x4∈{proximity priority, turnover rate priority, hybrid strategy}; Staffing scheduling strategy: x5∈{fixed shifts, flexible working hours, demand-driven}; Equipment start-up and shutdown strategies: x6∈{on-demand start-up and shutdown, timed start-up and shutdown, predictive start-up and shutdown}.

[0153] c) Path planning parameters.

[0154] Delivery route selection: x7∈{shortest distance, least time, lowest energy consumption}; Vehicle loading rate threshold: x8∈[0.6,1]; Multimodal transport ratio: x9∈[0,1].

[0155] In this embodiment, an improved non-dominated sorting genetic algorithm (NSGA-III) is used to solve this multi-objective optimization problem. The specific steps are as follows: Step (1): Problem coding.

[0156] Encode decision variables x1 to x9 as chromosomes; use real number encoding for continuous variables (such as x1, x2, x3, x8, x9); and use integer encoding for discrete variables (such as x4, x5, x6, x7).

[0157] Step (2): Initialize the population.

[0158] Randomly generate initial solutions that satisfy the constraints, ensuring that the generated solutions satisfy the carbon emission cap (Σ(Ei+Ti)≤Emax) and the budget limit (Σ(Ci+Mi+Li)≤Bmax).

[0159] Step (3): Fitness evaluation.

[0160] Calculate the two objective function values ​​F1 (total carbon emissions) and F2 (operating costs) for each individual; check whether the constraints are met, and penalize solutions that violate the constraints.

[0161] Step (4): Non-dominated sorting.

[0162] The solutions in the population are non-dominated and ordered based on the values ​​of F1 and F2; the rank of each solution is determined.

[0163] Step (5): Reference point generation.

[0164] Generate a set of uniformly distributed reference points in the target space; the number of reference points is comparable to the population size.

[0165] Step (6): Calculate the correlation degree.

[0166] Calculate the perpendicular distance between each solution and the reference point; associate the solution with the nearest reference point.

[0167] Step (7): Select operation.

[0168] The tournament selection method was used to select parent individuals; the non-dominant ranking and the correlation with the reference point were considered during the selection.

[0169] Step (8): Cross operation.

[0170] For continuous variables (x1, x2, x3, x8, x9), the simulated binary crossover (SBX) method is used; for discrete variables (x4, x5, x6, x7), single-point or double-point crossover is used.

[0171] Step (9): Mutation operation.

[0172] For continuous variables, a polynomial mutation method is used; for discrete variables, a random mutation method is used; ensuring that the mutated solution still satisfies the constraints.

[0173] Step (10): Repair operation.

[0174] If the generated offspring violates the constraints, a repair operation is performed to make them satisfy the constraints. Specifically: For carbon emission ceiling constraints, emissions can be reduced by adjusting equipment energy efficiency parameters (x1, x2, x3); for budget constraints, costs can be controlled by adjusting intelligent scheduling parameters (x4, x5, x6).

[0175] Step (11): Elite retention. The parent and offspring generations are merged; based on non-dominated ordering and reference point correlation, outstanding individuals are selected to form a new generation of population.

[0176] Step (12): Termination condition. The maximum number of iterations is reached; there is no significant improvement in the Pareto front over multiple generations (e.g., 20 generations).

[0177] Step (13): Output the results. Output the final Pareto optimal solution set. For each solution, provide the specific decision variable values ​​(x1 to x9) and their corresponding objective function values ​​(F1 and F2).

[0178] Through the above steps, the algorithm fully utilizes the parameters of the optimization objective, constraints, and decision variables to output a set of Pareto optimal solutions. Decision-makers can then choose the most suitable optimization strategy from this set of solutions based on their actual needs and preferences. For example, they can choose the solution with the lowest carbon emissions, or the solution with the lowest cost within an acceptable range of carbon emissions.

[0179] This improved algorithm better reflects the characteristics of multi-objective optimization problems and ensures that the optimization results meet all constraints, while achieving a balance between carbon emission reduction and cost control.

[0180] Example 7: This embodiment provides a dynamic analysis method for carbon flow data based on digital twins. Building upon Embodiment 1, and considering specific application scenarios, the method for continuously optimizing the digital twin model in Embodiment 1 is further optimized.

[0181] Specifically, in this embodiment, continuous optimization of the digital twin model includes: Step (1): Construction of the reputation assessment model.

[0182] a) Data preparation.

[0183] Historical data is extracted from the blockchain network, including data source ID, predicted value, and actual value. Here, data source ID refers to the aforementioned raw data.

[0184] b) Reputation calculation.

[0185] Reputation score of the data source: ReputationScore = 1 / (1+α*MAPE); Among them, MAPE is the mean absolute percentage error, which measures the difference between the predicted value and the actual value; α is a adjustment factor used to control the degree of influence of MAPE on the score.

[0186] c) Dynamic adjustment.

[0187] Update credit rating using the Exponential Moving Average (EMA) method: New reputation score: NewScore = β*CurrentScore + (1-β)*HistoricalScore; Where β is a smoothing factor that controls the weight of new and old data, CurrentScore is the reputation score for the current calculation period, and HistoricalScore is the historical reputation score.

[0188] Step (2): Optimize the parameters of the digital twin model.

[0189] Optimize digital twin model parameters using policy gradient-based reinforcement learning methods (such as the DDPG algorithm): Environment definition a) State space: including current carbon emissions, operating costs, equipment status, inventory levels, transportation tasks, etc.

[0190] b) Action space: Adjustable parameters corresponding to the digital twin model, such as equipment energy efficiency parameters, scheduling strategies, path planning, etc.

[0191] c) Reward function: R = -w1*NormalizedCarbonEmission - w2*NormalizedCost; where w1 and w2 are the weighting coefficients of carbon emissions and costs, respectively, R is the reward value, used to evaluate the degree of good or bad; NormalizedCarbonEmission represents the normalized carbon emissions; NormalizedCost represents the normalized cost.

[0192] The negative sign before the reward function formula indicates that the goal of this step is to minimize this value; in other words, it is to reduce carbon emissions and costs.

[0193] Furthermore, in this embodiment, the design concept of the reward function is: When carbon emissions or costs decrease, the value of R increases (because of the negative sign), indicating a positive reward.

[0194] When carbon emissions or costs increase, the value of R decreases, indicating a negative reward (or penalty).

[0195] By adjusting the values ​​of w1 and w2, the relative importance of the two objectives of reducing carbon emissions and controlling costs can be balanced.

[0196] For example, if w1 > w2, the algorithm will be more inclined to reduce carbon emissions, even if it may slightly increase costs. Conversely, if w2 > w1, the algorithm will focus more on cost control.

[0197] This reward function provides a clear optimization objective for reinforcement learning algorithms, guiding them to find the optimal balance between reducing carbon emissions and controlling costs.

[0198] d) Policy network and value network: Use deep neural networks to construct policy networks and value networks.

[0199] The policy network outputs actions (parameter adjustments), and the value network estimates the state value.

[0200] e) Training process: Collect experience: Execute the current policy and collect state transition samples; Update the network: Update the policy network and value network using the collected samples; Explore: Add noise to the actions to facilitate exploration.

[0201] Step (3): Update the carbon emission intensity heat map and hotspot areas.

[0202] a) Data collection: Extract the latest carbon emission data from the optimized digital twin model.

[0203] b) Heatmap update: Recalculate the carbon emission intensity of the entire region using Kriging interpolation; update the color mapping of the heatmap.

[0204] c) Hotspot identification: Recalculate the local Moran's I index; update hotspot and coldspot areas based on the new index values.

[0205] Step (4): Optimize the long-term effects of the strategy and make adjustments.

[0206] a) Effect evaluation: Calculate the reduction in carbon emissions and cost savings before and after optimization; analyze carbon emission trends and determine whether the expected effect has been achieved.

[0207] b) Strategy adjustment: If the effect is not good, increase the exploration rate and expand the parameter search range; if the effect is good, decrease the learning rate and make fine adjustments.

[0208] c) Feedback loop: Use the evaluation results and adjusted strategies as input for the next round of optimization; conduct a comprehensive strategy evaluation and large-scale adjustment regularly (e.g., monthly).

[0209] Through the steps described above, the digital twin-based dynamic carbon flow data analysis method provided in this embodiment enables dynamic analysis and continuous optimization of carbon emissions during warehousing and logistics processes, providing strong support for enterprises to achieve green and low-carbon operations. This method combines the data security of blockchain technology with the high-fidelity simulation capabilities of digital twin technology, enabling reliable analysis of carbon emission data and precise formulation of optimization strategies.

[0210] Example 8: This embodiment provides a dynamic analysis system for carbon flow data based on digital twins, used to execute a dynamic analysis method for carbon flow data based on digital twins in Embodiment 1.

[0211] Specifically: A dynamic carbon flow data analysis system based on digital twins includes: a data processing module, a digital twin model module, and a carbon emission analysis module. The data processing module is used to collect multi-dimensional data, fuse the multi-dimensional data based on an attention mechanism, encrypt the fused dataset, and store the encrypted fused dataset using a blockchain network. The digital twin model module is used to construct digital twin models of warehousing and logistics processes, integrate carbon emission analysis results, predict carbon emissions, and continuously optimize the data twin model based on the execution results of carbon emission optimization strategies. The carbon emission analysis module is used to perform carbon emission analysis based on the fused dataset, and generate carbon emission optimization strategies based on multi-objective optimization algorithms according to the carbon emission analysis results and carbon emission prediction results.

[0212] Furthermore, the data processing module includes: The data acquisition module is used to collect multi-dimensional data, including data from the warehouse management system, IoT devices, and transportation systems.

[0213] The data fusion processing module is used to fuse multi-dimensional data to obtain a fused dataset.

[0214] The data encryption storage module is used to encrypt the fused dataset and store the encrypted fused dataset through a blockchain network.

[0215] The digital twin model module includes: The digital twin model building module is used to build digital twin models of warehousing and logistics processes and integrate carbon emission analysis results into the digital twin models.

[0216] The model optimization module is used to continuously optimize the digital twin model based on the execution results sent by the strategy execution module.

[0217] The carbon emission prediction module is used to predict carbon emissions based on the digital twin model built by the digital twin model building module, and obtain the carbon emission prediction results.

[0218] The carbon emission analysis module includes: The carbon emission analysis module is used to perform carbon emission analysis based on the fused dataset and obtain the carbon emission analysis results.

[0219] The multi-objective optimization module is used to execute multi-objective optimization algorithms based on carbon emission analysis results and carbon emission prediction results to generate carbon emission optimization strategies.

[0220] The strategy execution module is used to execute carbon emission optimization strategies, record the execution results through a blockchain network, and feed the results back to the model optimization module.

[0221] The carbon emission prediction system based on this digital twin model can fully utilize historical and real-time data, combined with advanced deep learning and ensemble learning technologies, to achieve accurate predictions across multiple scales. This provides strong data support and decision-making basis for enterprises to formulate carbon reduction strategies and optimize operational decisions.

[0222] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Other variations and modifications are possible without departing from the technical solutions described in the claims.

Claims

1. A method for dynamic analysis of carbon flow data based on digital twins, characterized in that, include: S1: Collect multi-dimensional data and perform fusion processing to obtain a fused dataset; S2: Encrypt and store the fused dataset, and perform carbon emission analysis on the fused dataset; S3: Construct a digital twin model of the warehousing and logistics process, integrate carbon emission analysis results into the digital twin model, and predict carbon emissions based on the digital twin model; S4: Based on the carbon emission prediction results and carbon emission analysis results, generate a carbon emission optimization strategy using a multi-objective optimization algorithm; S5: Implement carbon emission optimization strategies and continuously optimize the digital twin model based on the implementation results.

2. The method for dynamic analysis of carbon flow data based on digital twins according to claim 1, characterized in that, Step S3 includes: S3.1: Construct a multi-layered digital twin model that includes warehouse layout, IoT device distribution, and logistics routes; integrate the carbon emission intensity heat map into the multi-layered digital twin model; S3.2: Use historical carbon emission data and multi-dimensional data as input; use a long short-term memory network to extract time series features; S3.3: Integrate spatial and temporal features through an attention mechanism; S3.4: Utilize ensemble learning methods to predict carbon emissions at multiple scales, including short-term, medium-term, and long-term. S3.5: Output the expected carbon emissions and carbon emission trends for a specified future period.

3. A method for dynamic analysis of carbon flow data based on digital twins according to claim 1 or 2, characterized in that, In step S5, the continuous optimization of the digital twin model includes: S5.1: Record the results of the optimization strategy execution to the blockchain network via a smart contract; S5.2: Utilize historical data from the blockchain network to construct a reputation assessment model and dynamically adjust the credibility of different data sources; S5.3: Based on the execution results, use reinforcement learning methods to optimize the parameters of the digital twin model; S5.4: Update the carbon emission intensity heat map and carbon emission hotspots based on the optimized digital twin model; S5.5: Based on the updated carbon emission intensity heat map and carbon emission hotspots, evaluate the long-term effects of the optimization strategy and adjust the optimization strategy accordingly.

4. The method for dynamic analysis of carbon flow data based on digital twins according to claim 1, characterized in that, In step S1, the multi-dimensional data fusion process includes: S1.1: Utilize the spatiotemporal tags of IoT devices to align warehouse management system data with IoT device data in time and space, and build a unified data view; S1.2: Based on the collected multi-dimensional data, calculate the energy efficiency index of IoT devices, construct a comprehensive environmental index, and calculate the transportation carbon emission factor; S1.3: An attention mechanism is used to fuse the calculation results in S1.2 with the collected original multi-dimensional data to obtain a comprehensive feature representation.

5. A method for dynamic analysis of carbon flow data based on digital twins according to claim 1, 2, or 4, characterized in that, In step S2, the carbon emission analysis includes: S2.1: Allocate total carbon emissions to each batch of materials using the activity-based costing method based on time percentage, weight percentage, and path characteristics; S2.2: Use spatial autocorrelation analysis to identify carbon emission hotspots and construct a carbon emission intensity heat map; S2.3: Calculate the carbon emission contribution rate of each region, and set the weights of the multi-objective optimization algorithm based on the carbon emission contribution rate.

6. A method for dynamic analysis of carbon flow data based on digital twins according to claim 1, 2, or 4, characterized in that, The multi-objective optimization algorithm takes minimizing total carbon emissions and minimizing operating costs as objective functions, and carbon emission caps and budget constraints as constraints. The decision variables of the multi-objective optimization algorithm include equipment energy efficiency optimization parameters, intelligent scheduling parameters, and path planning parameters.

7. A method for dynamic analysis of carbon flow data based on digital twins according to claim 1, 2, or 4, characterized in that, The encrypted fusion dataset is stored through a blockchain network, which also records the execution results of the carbon emission optimization strategy.

8. The method for dynamic analysis of carbon flow data based on digital twins according to claim 7, characterized in that, The fused dataset is encrypted using a homomorphic encryption algorithm; the blockchain network adopts a consortium blockchain structure, with participating nodes including warehousing companies, logistics companies, and regulatory agencies.

9. A method for dynamic analysis of carbon flow data based on digital twins according to claim 1 or 4, characterized in that, The multi-dimensional data includes warehouse management system data, IoT device data, and transportation system data.

10. A dynamic analysis system for carbon flow data based on digital twins, employing the dynamic analysis method for carbon flow data based on digital twins as described in any one of claims 1-9, characterized in that, include: The data processing module collects multi-dimensional data, performs fusion processing on the multi-dimensional data based on the attention mechanism, encrypts the fused dataset, and stores the encrypted fused dataset using a blockchain network. The digital twin model module constructs a digital twin model of the warehousing and logistics process, integrates carbon emission analysis results, performs carbon emission prediction, and continuously optimizes the data twin model based on the execution results of carbon emission optimization strategies. The carbon emission analysis module performs carbon emission analysis based on the fused dataset, and generates carbon emission optimization strategies based on the carbon emission analysis results and carbon emission prediction results using a multi-objective optimization algorithm.

Images