AI-based carbon footprint and supply chain carbon emission collaborative management method and platform

Abstract

The invention provides an AI-based carbon footprint and supply chain carbon emission collaborative management method and platform, and the method comprises the steps: collecting various types of carbon-related data from each link of a supply chain in real time, and carrying out the standardization processing to form a unified data flow. An intelligent sensing technology is combined with a dynamic reference model, and sudden carbon emission fluctuation generated in production operation is automatically recognized. The propagation influence and potential risk of these carbon emission fluctuations in the entire supply chain are analyzed. The risk propagation condition and the processing capacity of each node are comprehensively considered, and an optimal carbon emission distribution processing scheme is formulated through an optimization algorithm. A distributed account book technology and an automatic execution mechanism are utilized, and an excitation system is combined to ensure that each participant completes a corresponding carbon processing task according to an optimal scheme. According to the method, the intelligent level and the overall effect of supply chain carbon management can be effectively improved, and a technical support is provided for an enterprise to realize a carbon emission reduction target.

Description

Technical Field

[0001] The present invention belongs to the field of carbon emission management technology, and specifically relates to an AI-based carbon footprint and supply chain carbon emission collaborative management method and platform. Background Art

[0002] The market's growing demand for smarter, more efficient, and more transparent carbon management solutions is driving the innovative development of related technologies. Existing carbon management systems often employ static, lagging approaches that fail to promptly identify and address sudden fluctuations in carbon emissions. This results in a lack of effective response mechanisms across the entire supply chain when faced with carbon emissions shocks. This passive management model not only hinders carbon reduction efforts but can also lead to irrational resource allocation, increasing carbon costs and compliance risks for businesses. Summary of the Invention

[0003] The present invention provides an AI-based carbon footprint and supply chain carbon emission collaborative management method and platform to solve the above technical problems.

[0004] In a first aspect, the present invention provides an AI-based method for collaboratively managing carbon footprint and supply chain carbon emissions, the method comprising the following steps:

[0005] Real-time access to supply chain carbon-related multi-source heterogeneous data from each supply chain node, and pre-processing of multi-source heterogeneous data to form a carbon footprint data stream;

[0006] Run an event-driven AI perception engine and identify instantaneous carbon pulses associated with production and operation events based on carbon footprint data streams and a dynamic carbon emissions baseline model;

[0007] Analyze the risk of carbon shock transmission in the supply chain caused by instantaneous carbon pulses;

[0008] Combining the carbon shock transmission risk and the carbon absorption capacity of each node in the supply chain and using a dynamic matching algorithm to match, we can obtain the optimal absorption allocation plan for instantaneous carbon pulses.

[0009] Utilize smart contracts and oracles on the blockchain platform associated with the supply chain, and adopt an incentive allocation mechanism to automatically execute the optimal absorption allocation plan, so that each node in the supply chain can complete all carbon impact absorption tasks in the optimal absorption allocation plan.

[0010] Optionally, the carbon footprint data stream includes product carbon footprint data and carbon footprint raw material data. The product carbon footprint data represents the total amount of greenhouse gas emissions generated by the supply chain products throughout their entire life cycle from raw material mining to waste treatment. The carbon footprint raw material data represents the greenhouse gas emissions generated by the raw materials used in the supply chain products during the extraction, primary processing and transportation to the factory.

[0011] Optionally, the real-time access to carbon-related multi-source heterogeneous data from each supply chain node includes the following steps:

[0012] Through pre-deployed edge computing gateways and developed application program interface extractors, the system can connect to IoT devices, enterprise resource planning systems, and supply chain management systems at various supply chain nodes using standard IoT protocols, direct database connections, or application program interface calls.

[0013] Configure data subscription and scheduled pull tasks to capture multi-source heterogeneous data including equipment energy consumption readings, production batch information, material flow records, transportation vehicle carbon emission factors, and external environmental impact parameters;

[0014] Use stream processing technology to buffer and distribute incoming multi-source heterogeneous data in real time;

[0015] Perform preliminary format verification and metadata tagging on multi-source heterogeneous data from different data sources, and establish a data archive for each supply chain node to record the data source type, access frequency, and data quality profile.

[0016] Optionally, the preprocessing of multi-source heterogeneous data to form a carbon footprint data stream includes the following steps:

[0017] Use algorithms including linear interpolation, mean filling, or business rule-based algorithms to fill in missing key carbon-related parameters in multi-source heterogeneous data;

[0018] Use statistical methods to identify and correct abnormal data points in multi-source heterogeneous data caused by sensor failure or data transmission errors;

[0019] Convert carbon-related data in different measurement units from multi-source heterogeneous data into a unified standard unit based on preset conversion rules;

[0020] Based on a predefined unified carbon-related data model, structured mapping and conversion are performed on cleaned multi-source heterogeneous data;

[0021] Master data management technology is used to establish unique identifiers for core entities corresponding to multi-source heterogeneous data, and structured data from different sources are integrated based on unique identifiers and business logic associations to form a carbon footprint data stream containing complete contextual information.

[0022] Optionally, running the event-driven AI perception engine and identifying instantaneous carbon pulses associated with production and operation events based on the carbon footprint data stream and the dynamic carbon emission baseline model includes the following steps:

[0023] For each carbon emission unit in the supply chain, a dynamic carbon emission baseline model is established using a time series model combined with historical carbon emission data and periodic operating parameters. The dynamic carbon emission baseline model outputs the predicted carbon emission value and prediction confidence interval of the carbon emission unit;

[0024] Collect and process actual carbon emissions data from carbon footprint data streams;

[0025] Compare actual carbon emissions data with the predicted confidence intervals of the dynamic carbon emissions baseline model;

[0026] A complex event processing engine is used to extract predefined production and operation events from the carbon footprint data stream. Production and operation events include equipment start and stop, process parameter adjustment, and order status change.

[0027] When the actual carbon emission data statistically deviates from the predicted confidence interval of the dynamic carbon emission baseline model, and the actual carbon emission data is strongly correlated with the extracted production and operation events in the time window, it is identified as an instantaneous carbon pulse.

[0028] Optionally, the method further comprises the following steps:

[0029] Build a supply chain carbon emissions knowledge graph that includes supply chain nodes, production processes, bills of materials, equipment parameters, and potential carbon impact factors;

[0030] When an instantaneous carbon pulse is identified, in the supply chain carbon emission knowledge graph, starting from the carbon emission unit node where the instantaneous carbon pulse occurs, reverse traverse is performed along the material flow, energy flow and information flow paths;

[0031] Analyze the state changes and parameter anomalies of each supply chain node and related production and operation events on the traversal path before and after the instantaneous carbon pulse occurs;

[0032] Apply a decomposition model based on regression analysis to quantify the contribution of each potential upstream influencing factor or concurrent production and operation event to the instantaneous carbon pulse amplitude;

[0033] Combining the contribution degree and business logic rules, the main root events or key influencing factors of the associated supply chain nodes that cause instantaneous carbon pulses are determined and output.

[0034] Optionally, the analyzing the carbon shock propagation risk of the instantaneous carbon pulse in the supply chain includes the following steps:

[0035] Based on supply chain nodes, the supply chain is abstracted into a dynamic network graph containing node attributes and edge attributes, where supply chain nodes represent suppliers or production units, and node edges represent material or order dependencies;

[0036] The initial parameters of the carbon shock propagation model are set based on the carbon emission transmission coefficient, inventory buffer level, production capacity elasticity, and sensitivity to upstream supply disruptions at each supply chain node;

[0037] The intensity of the identified instantaneous carbon pulse and the associated supply chain nodes are used as the initial input of the carbon shock propagation model. The system dynamics simulation is used on a dynamic network diagram to simulate the step-by-step propagation process of the instantaneous carbon pulse along the supply chain.

[0038] Predict the affected nodes downstream of the instantaneous carbon pulse and quantify the additional carbon emissions of the affected nodes caused by the propagation process within a preset time window in the future;

[0039] The additional carbon emissions of all affected nodes are integrated to generate a risk assessment report on the chain impact of instantaneous carbon pulses on the supply chain.

[0040] Optionally, combining the carbon shock transmission risk and the carbon absorption capacity of each node in the supply chain and using a dynamic matching algorithm to obtain an optimal absorption allocation plan for instantaneous carbon pulses includes the following steps:

[0041] Dynamically evaluate the potential carbon absorption capacity limit of each supply chain node and the corresponding marginal absorption cost function;

[0042] Setting the optimization goal of coordinated absorption to minimize the overall absorption cost and maximize the emission reduction efficiency;

[0043] The optimization goal is used as the objective function, and the upper limit of the carbon absorption capacity and marginal absorption cost function of each supply chain node, as well as the requirement that the total absorption capacity of all supply chain nodes must cover the additional carbon emissions, are used as constraints to construct the absorption allocation model.

[0044] A heuristic optimization algorithm is used to solve the consumption allocation model, and the output is the optimal consumption allocation plan that includes the specific consumption volume, execution time window and expected cost that each supply chain node should bear.

[0045] Optionally, the method of utilizing smart contracts and oracles on a blockchain platform associated with the supply chain and adopting an incentive allocation mechanism to automatically execute the optimal absorption allocation plan so that each node in the supply chain completes all carbon impact absorption tasks in the optimal absorption allocation plan includes the following steps:

[0046] Based on the optimal carbon absorption allocation plan, a carbon absorption task smart contract containing task objectives, acceptance criteria and incentive rules is generated and deployed for each designated task-bearing node on the blockchain platform associated with the supply chain;

[0047] When the carbon absorption task smart contract meets the preset conditions, it sends the encrypted absorption task instruction to the registered digital identity of the task undertaking node;

[0048] The task-bearing node completes the consumption task and submits the key performance data and emission reduction evidence hash during the execution process to the oracle on the blockchain platform through the interface;

[0049] Use the oracle to obtain the actual carbon emission data of the task-undertaking node during the task cycle from the carbon footprint data stream, and feed the actual carbon emission data into the corresponding carbon absorption task smart contract after cross-validation;

[0050] Based on the carbon absorption task smart contract and according to the acceptance criteria, the task completion status is determined, the allocation of pre-stored carbon credits is executed according to the incentive rules to complete the incentive distribution, and all transactions and status changes are recorded on the blockchain platform.

[0051] In the second aspect, the present invention provides an AI-based carbon footprint and supply chain carbon emission collaborative management platform, including a memory, a processor, and a computer program stored on the memory and runnable on the processor. When the processor executes the computer program, it implements the AI-based carbon footprint and supply chain carbon emission collaborative management method as described in the first aspect.

[0052] The beneficial effects of the present invention are:

[0053] By establishing a dynamic perception mechanism, the present invention can capture changes in carbon emissions in the supply chain in real time, completely changing the traditional lagging management model and achieving a fundamental shift from passive response to active prevention. The present invention has intelligent analysis capabilities, can accurately identify and predict the propagation path and impact range of carbon emission shocks, provide a scientific basis for decision-making, and greatly improve management accuracy and effectiveness. By constructing a collaborative optimization mechanism, the present invention achieves efficient coordination and cooperation between various nodes in the supply chain, avoids the information island problem under the traditional model, and significantly improves the overall carbon emission reduction effect. The present invention adopts an automated execution method, reduces manual intervention, reduces operating costs and error rates, and improves management efficiency. The present invention establishes a transparent incentive mechanism, mobilizes the enthusiasm of all participants, and forms a benign synergistic effect. The application of the present invention will significantly reduce corporate carbon management costs, improve compliance efficiency, provide strong technical guarantees for achieving carbon neutrality goals, and has important economic value and environmental significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a flow chart of an AI-based method for collaborative management of carbon footprint and supply chain carbon emissions in one of the implementations of this application.

[0055] Figure 2 This is a platform architecture diagram of an AI-based carbon footprint and supply chain carbon emissions collaborative management platform in one of the implementation methods of this application. DETAILED DESCRIPTION

[0056] The following will be combined with the accompanying drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the embodiments described are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of this application.

[0057] The terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of this application can be implemented in an order other than that illustrated or described herein, and that the objects distinguished by "first," "second," and the like are generally of the same type, and do not limit the number of objects; for example, the first object can be one or more. In addition, the term "and / or" in the specification and claims refers to at least one of the connected objects, and the character " / " generally indicates that the objects connected are in an "or" relationship.

[0058] Figure 1 The figure is a flow chart of an AI-based method for collaboratively managing carbon footprint and supply chain carbon emissions in one embodiment. Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 1 At least part of the steps in the above process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but may be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but may be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps. Figure 1 As shown, the AI-based carbon footprint and supply chain carbon emission collaborative management method disclosed in the present invention specifically includes the following steps:

[0059] S101. Access the supply chain carbon-related multi-source heterogeneous data from each node in the supply chain in real time, and pre-process the multi-source heterogeneous data to form a carbon footprint data stream.

[0060] This involves connecting sensor devices at each node via pre-deployed edge computing gateways and API extractors using standard IoT protocols such as MQTT and OPC-UA. Data from ERP and SCM systems is also obtained through direct database connections and RESTful API calls. Scheduled pull tasks are configured to capture heterogeneous data from multiple sources, including equipment energy consumption readings, production batch information, and material flow records. This data is then buffered and distributed in real time using stream processing technologies such as Apache Kafka. Format validation and metadata tagging are performed on the incoming data. Missing values ​​are filled using linear interpolation, and outliers are identified and corrected using the 3σ criterion. Different measurement units are standardized to standard units, such as kilograms of CO2 equivalent, according to pre-set conversion rules. Structured mapping is performed based on a unified data model. Master data management technologies are used to establish unique UUID identifiers for core entities, and data from different sources is integrated and correlated according to business logic. For example, energy consumption data from production equipment is linked to production order information from the ERP system to form a carbon footprint data stream with complete context. This process achieves standardized integration of heterogeneous data, providing a high-quality, up-to-date carbon emissions data foundation for subsequent AI analysis.

[0061] The carbon footprint data stream includes product carbon footprint data and raw material carbon footprint data. Product carbon footprint data represents the total greenhouse gas emissions generated by the supply chain product throughout its entire life cycle, from raw material extraction to waste disposal. Raw material carbon footprint data represents the greenhouse gas emissions generated by the raw materials used in the supply chain product during extraction, initial processing, and transportation to the factory. In one embodiment, carbon footprint data can be calculated using the following three methods: emission factor method, mass balance method, and actual measurement method.

[0062] The emission factor method calculates total emissions by multiplying activity data (such as energy consumption and mileage) by a corresponding emission factor (average emissions per unit of activity). The calculation logic is: Carbon emissions = Activity data (AD) × Emission factor (EF). For example, if a company uses 10,000 kWh of electricity annually and the grid emission factor is 0.5 kg CO₂e / kWh, the annual carbon emissions calculation method is: 10,000 kWh × 0.5 kg CO₂e / kWh = 5,000 kg CO₂e. The emission factor method is currently the most widely used method, widely used in carbon accounting at the national, regional, and enterprise levels.

[0063] The mass balance method, based on the law of conservation of matter, calculates net emissions within a system boundary by analyzing input and output carbon flows (such as raw materials, fuels, and products). Calculation logic: Emissions = Input Carbon - Output Carbon (Products / Waste).

[0064] Here's an example calculation: In a steel industry (blast furnace ironmaking), the inputs are 1 ton of iron ore (0.1% carbon) and 0.5 tons of coke (85% carbon), resulting in 0.425 tons of carbon. Outputs are 0.9 tons of pig iron (4% carbon) (0.036 tons of carbon), 0.1 tons of slag (1% carbon) (0.001 tons of carbon), and waste gas (primarily CO2) (0.388 tons of carbon). The net emissions are 0.425 - (0.036 + 0.001 + 0.388) = 0 (theoretical equilibrium). Mass balance is applicable to industrial processes (such as chemical and steel production).

[0065] The measurement method directly measures greenhouse gas concentrations and flows at emission sources using sensors, continuous emissions monitoring systems (CEMS), or laboratory analysis. The calculation is: Emissions = Gas Concentration x Exhaust Flow x Time. Gas concentration is measured by sensors or laboratory analysis (e.g., CO2 volume percentage). Exhaust flow refers to the exhaust gas emission rate from the flue / duct (m³ / h). Time represents the monitoring period (hours, years, etc.).

[0066] Example: The CO2 concentration monitored at the chimney of a coal-fired power plant is 12% (volume ratio), the exhaust flow rate is 50 m³ / s, and the annual operating hours are 8,000 hours. The annual CO2 emissions are: 12% x 50 m³ / s x 3,600 s / h x 8,000 x 1.96 kg / m³ = 34,000 tons of CO2 (Note: 1.96 kg / m³ is the CO2 density). The field measurement method is suitable for stationary emission sources (such as chimneys and industrial pipelines).

[0067] Summary: The emission factor method is simple and rapid, suitable for indirect emissions (such as purchased electricity) and preliminary estimates, and relies on industry average data. The mass balance method is preferred for industrial processes, tracking carbon flows (such as chemicals and steel), requires detailed material data, and has higher accuracy. The field measurement method can accurately monitor stationary sources (such as chimneys), but is costly and is used for regulatory compliance (such as carbon trading).

[0068] S102. Run an event-driven AI perception engine and identify instantaneous carbon pulses associated with production operation events based on carbon footprint data streams and dynamic carbon emission baseline models.

[0069] Among them, for each carbon emission unit, the ARIMA time series model is used to combine historical carbon emission data and periodic operating parameters to establish a dynamic carbon emission baseline model. This model can output predicted carbon emission values ​​and corresponding confidence intervals. The complex event processing engine extracts predefined production and operation events from the carbon footprint data stream in real time, including key events such as equipment start and stop, process parameter adjustment, and order status change, and uses sliding time window technology to perform event sequence analysis. When the actual carbon emission data statistically deviates from the predicted confidence interval of the baseline model, the system will further check whether the anomaly is strongly correlated with the extracted production and operation events within the time window. The Pearson correlation coefficient is calculated to quantify the correlation strength. When the absolute value of the correlation coefficient is greater than 0.7, it is identified as an instantaneous carbon pulse. For example, after a production line receives a large number of orders, the equipment runs at full capacity, resulting in a sudden increase of 30% in carbon emissions, which exceeds the predicted interval and is strongly correlated with the order event.

[0070] S103. Analyze the carbon shock transmission risk of instantaneous carbon pulses in the supply chain.

[0071] The supply chain is abstracted as a dynamic network graph with node and edge attributes, where nodes represent suppliers or production units, and edges represent material or order dependencies. The initial parameters of the carbon shock propagation model are set based on each node's carbon emission transfer coefficient, inventory buffer level, production capacity elasticity, and sensitivity to upstream supply disruptions. The identified instantaneous carbon pulse intensity and associated nodes serve as the model's initial inputs, and a system dynamics simulation is used to simulate the cascading propagation of the carbon pulse along the supply chain. The simulation considers propagation delays, attenuation effects, and inter-node interactions, using a Monte Carlo method with 1,000 random simulations to account for uncertainty. The impact of the instantaneous carbon pulse on downstream nodes is predicted, and the additional carbon emissions caused by the propagation process over the next 30 days are quantified. For example, a carbon pulse from an upstream steel supplier propagates through the material flow to an automaker, resulting in a 15% increase in carbon emissions at downstream nodes. The additional carbon emissions of all affected nodes are combined to generate a chain reaction risk assessment report, enabling accurate prediction and quantitative assessment of carbon shocks.

[0072] S104. Combining the carbon shock transmission risk and the carbon absorption capacity of each node in the supply chain and using a dynamic matching algorithm to match, we can obtain the optimal absorption allocation plan for instantaneous carbon pulses.

[0073] The potential carbon absorption capacity ceiling and corresponding marginal absorption cost function of each supply chain node are dynamically assessed. Absorption capacity includes the node's emission reduction potential, room for technological improvement, and available resources. A multi-objective optimization model is constructed with the optimization objectives of minimizing overall absorption cost and maximizing emission reduction efficiency. Constraints include the upper limit of each node's carbon absorption capacity, the requirement that the total absorption capacity cover additional carbon emissions, and time window constraints. A heuristic optimization algorithm, such as a genetic algorithm, is used to solve this NP-hard problem. The optimal solution is found through evolutionary operations such as population initialization, selection, crossover, and mutation. The algorithm sets a population size of 100, iterates for 500 generations, has a crossover probability of 0.8, and a mutation probability of 0.1. For example, if a carbon shock requires the absorption of 1,000 tons of CO2, the algorithm allocates 300 tons to node A (with the lowest cost), 400 tons to node B (with the highest capacity), and 300 tons to node C (with the most flexible timeframe). The output is an optimal allocation plan that includes the specific absorption capacity of each node, the execution time window, and the expected cost, minimizing costs while ensuring emission reduction efficiency.

[0074] S105. Utilize smart contracts and oracles on the blockchain platform associated with the supply chain, and adopt an incentive allocation mechanism to automatically execute the optimal absorption allocation plan, so that each node in the supply chain can complete all carbon impact absorption tasks in the optimal absorption allocation plan.

[0075] Among them, reference Figure 2 Based on the optimal carbon footprint allocation plan, a smart contract is generated and deployed on blockchain platforms such as Ethereum or Hyperledger Fabric for each designated task node. The contract contains the task objectives, acceptance criteria, and incentive rules. When the contract meets the preset trigger conditions, such as the execution time, it automatically sends an AES-256-encrypted carbon footprint task instruction to the task node's registered digital identity. After completing the task, the node submits key performance data and a SHA-256 hash of emission reduction evidence to a blockchain oracle via a standardized API. The oracle, acting as a data bridge between on-chain and off-chain, retrieves the node's actual carbon emissions data during the task cycle from the carbon footprint data stream. After multi-source cross-validation to ensure data authenticity, it feeds the verification results into the corresponding smart contract. The smart contract automatically determines task completion based on the preset acceptance criteria and determines compliance when the actual emission reduction reaches or exceeds 90% of the target. Reward points are calculated based on the incentive rules, and incentives are automatically allocated by automatically allocating pre-deposited carbon credits. This mechanism achieves decentralized automatic execution, significantly enhancing transparency and credibility.

[0076] In one embodiment, real-time access to carbon-related multi-source heterogeneous data from various supply chain nodes includes the following steps:

[0077] Through pre-deployed edge computing gateways and developed application program interface extractors, the system can connect to IoT devices, enterprise resource planning systems, and supply chain management systems at various supply chain nodes using standard IoT protocols, direct database connections, or application program interface calls.

[0078] Configure data subscription and scheduled pull tasks to capture multi-source heterogeneous data including equipment energy consumption readings, production batch information, material flow records, transportation vehicle carbon emission factors, and external environmental impact parameters;

[0079] Use stream processing technology to buffer and distribute incoming multi-source heterogeneous data in real time;

[0080] Perform preliminary format verification and metadata tagging on multi-source heterogeneous data from different data sources, and establish a data archive for each supply chain node to record the data source type, access frequency, and data quality profile.

[0081] In this implementation, to effectively aggregate basic information related to supply chain carbon emissions, the primary task is to establish reliable connections with data sources at various supply chain nodes and extract the required data. First, edge computing gateways are pre-deployed near physical data sources (e.g., sensors and energy meters on the factory floor). These gateways act as data outposts, interacting directly with IoT devices via standard IoT protocols (e.g., MQTT, a lightweight publish / subscribe messaging protocol for inter-device communication, or OPC UA, a secure and scalable communication framework for industrial automation) to collect data in real time. Second, specialized application programming interface (API) extractors are developed for existing enterprise information systems, such as enterprise resource planning (ERP) and supply chain management (SCM) systems. These extractors retrieve information through direct database connections (e.g., securely querying relevant tables in SQL databases using JDBC or ODBC drivers) or by invoking system-provided APIs. For example, an edge gateway might read minute-by-minute power consumption from a production line's electricity meter via Modbus and send it via MQTT. Meanwhile, an API extractor might call the ERP system's API every hour to retrieve completed production orders and their bills of materials. This step effectively bridges data silos, ensuring that raw carbon-related data from disparate physical locations and information systems can be stably and continuously integrated into subsequent processing flows.

[0082] After establishing a data connection, the data capture mechanism must be precisely configured to ensure comprehensive and timely access to all heterogeneous, multi-source data relevant to carbon footprint calculation. This involves configuring data subscription tasks and scheduled pull tasks. Data subscription typically applies to data sources that support event-driven or push notifications. For example, when an IoT device reports a new energy consumption reading or a key state change in an enterprise system occurs, data is automatically pushed to the collection end. Scheduled pull tasks, on the other hand, target data sources that require active queries, periodically requesting data at fixed intervals (such as every minute, every hour, or every day). The scope of captured data is extensive, encompassing energy consumption readings directly output by devices such as smart electricity, water, and gas meters; detailed records of material flows through various links in the supply chain (including material name, quantity, source, destination, and transportation method); and the carbon emission factors of the transportation vehicles themselves (for example, the CO2 equivalent emissions per kilometer or per ton-kilometer for a particular truck model). Furthermore, external environmental parameters that influence energy consumption or carbon emissions, such as local temperature and humidity, are collected. This configuration ensures that critical information isn't missed. For example, it can promptly capture unusual energy consumption during the production of a batch of products or record differences in carbon emissions due to different transportation methods. The core effect is to build a continuously updated, context-rich raw data set, laying a solid foundation for subsequent accurate carbon emissions accounting and analysis.

[0083] Once various data types are captured through configured tasks, advanced stream processing technologies are used to efficiently buffer and distribute this real-time, multi-source, heterogeneous data. Because data may come from hundreds or thousands of sources, and the rate and volume of data generated can fluctuate dramatically, directly sending this data to subsequent processing modules can lead to system congestion or data loss. Therefore, stream processing technologies, such as Apache Kafka or similar message queue systems, are introduced to serve as a "reservoir" and "sorting center" for data before it enters subsequent processing. These technologies provide high-throughput, real-time data buffering, temporarily storing incoming data to form an orderly data stream. This not only smooths data spikes, ensuring that backend processing systems can consume data at their own pace, but also provides temporary data storage during brief backend system failures, preventing data loss.

[0084] The stream processing platform also supports distributing the same data to multiple downstream applications or processing units. For example, a stream of energy consumption data from a factory can be sent to a carbon footprint calculation engine for carbon emissions accounting, a real-time monitoring dashboard for energy consumption anomaly warnings, and a long-term storage system for archiving. This buffering and distribution mechanism ensures the elasticity and reliability of the data pipeline, making the entire data processing process more robust and able to flexibly respond to various dynamic changes. The result is a significant improvement in the stability of data access and the flexibility and efficiency of subsequent processing. After buffering and distribution, the data undergoes preliminary quality control and context enrichment before it enters deeper processing. This stage primarily involves preliminary format verification and metadata tagging of multi-source, heterogeneous data from different sources, and the establishment of dedicated data archives for each key node in the supply chain (such as a single factory, warehouse, or transportation hub).

[0085] Format validation ensures that received data conforms to the expected structure and type. For example, it checks whether numeric data is truly numeric, whether dates and times follow standard formats, and whether structured data such as JSON / XML is complete and valid. If format errors are detected, they can be flagged, isolated, or corrected according to pre-set rules. Next comes metadata tagging, which adds descriptive information to each data item or batch, such as the specific sensor, ERP module, or API source, the access timestamp, and the original unit of measurement (e.g., kWh, tons, cubic meters). Furthermore, a data archive is established for each supply chain node to systematically record the types of all data sources related to that node (e.g., IoT device data, ERP system data), the frequency of data access, and the quality of the node's historical data (e.g., data completeness ratio, accuracy assessment, etc.). This data archive helps track the health of data sources and the credibility of the data. This series of operations directly improves the standardization and usability of data entering subsequent analysis processes. Metadata gives the data clearer business meaning, and the data archive provides a basis for data quality management and problem tracing.

[0086] In one embodiment, preprocessing multi-source heterogeneous data to form a carbon footprint data stream includes the following steps:

[0087] Use algorithms including linear interpolation, mean filling, or business rule-based algorithms to fill in missing key carbon-related parameters in multi-source heterogeneous data;

[0088] Use statistical methods to identify and correct abnormal data points in multi-source heterogeneous data caused by sensor failure or data transmission errors;

[0089] Convert carbon-related data in different measurement units from multi-source heterogeneous data into a unified standard unit based on preset conversion rules;

[0090] Based on a predefined unified carbon-related data model, structured mapping and conversion are performed on cleaned multi-source heterogeneous data;

[0091] Master data management technology is used to establish unique identifiers for core entities corresponding to multi-source heterogeneous data, and structured data from different sources are integrated based on unique identifiers and business logic associations to form a carbon footprint data stream containing complete contextual information.

[0092] In this embodiment, first, in order to ensure the integrity and continuity of the carbon footprint data stream, it is necessary to effectively fill in the missing values ​​of key carbon-related parameters in multi-source heterogeneous data. These key parameters may include equipment energy consumption, material consumption or transportation mileage at a specific time point. The filling method is selected based on the data characteristics and business logic. One common method is linear interpolation, and another method is mean filling, which is to fill in the average value of the parameter in a certain historical period or the average value of the adjacent non-missing values. In more complex cases, algorithms based on business rules will be used. For example, if the device is in a "shutdown" state, its missing energy consumption value can be filled with a preset extremely low standby power consumption value. The implementation of these filling algorithms is intended to minimize the adverse effects of missing data on the accuracy of subsequent carbon emission analysis, ensure the consistency of the carbon footprint data stream, and provide a more reliable data basis for AI model training and carbon emission accounting.

[0093] Next, statistical methods are needed to identify and correct outliers in the multi-source, heterogeneous data, caused by sensor failures, data transmission errors, or other interference. These outliers, such as energy consumption readings far outside the normal range or illogical material flow records, can severely distort carbon emissions analysis results if left unaddressed. A common identification method relies on the statistical distribution of the data, such as the 3-sigma principle (for normally distributed data, approximately 99.7% of data points should fall within the range of three standard deviations from the mean) or using the interquartile range (IQR) of a boxplot to define outlier boundaries. Once an outlier is identified, correction methods can include directly removing the data point (if the data volume is sufficient and the point's impact is minimal), replacing it with a similar approach to missing value handling, such as replacing it with the mean or median of a nearby normal value, or correcting it based on predicted values ​​from a time series model. This step aims to clean the data, remove noise, and improve the overall quality and credibility of the dataset, thereby ensuring the accuracy and robustness of subsequent carbon emissions calculations and AI model analysis.

[0094] Next, based on pre-defined conversion rules, carbon-related data in various units of measurement within the multi-source, heterogeneous data is converted to internationally recognized or industry-standard units. Since raw data may originate from different systems and devices, their units of measurement may vary. For example, electricity consumption may be recorded in kilowatt-hours (kWh) or megawatt-hours (MWh), and fuel consumption may be recorded in liters (L), cubic meters (m³), kilograms (kg), or tons (t). Ultimately, carbon emissions are typically expressed in carbon dioxide equivalents (CO2e) (e.g., kg CO2e or t CO2e). The core of the conversion rules is the application of accurate conversion factors and emission factors. These emission factors are typically derived from official databases (such as IPCC guidelines and national greenhouse gas inventories) or industry-specific standards. Implementing a unified unit conversion eliminates calculation errors and comparison difficulties caused by inconsistent units of measurement, ensuring that all carbon-related data is integrated, analyzed, and calculated at the same scale, laying the foundation for a standardized carbon footprint view.

[0095] After data cleansing and unit unification, it needs to be structured and mapped and transformed based on a predefined unified carbon-related data model. This unified carbon-related data model is a standardized data structure blueprint that specifies the core data fields (such as timestamp, emission source identifier, activity type, energy type, consumption, carbon emissions, and relevant product / batch information) that should be included in the carbon footprint data stream, as well as the data type and value range of each field, and the hierarchy and relationships between fields. The mapping process accurately maps each data item from different raw data sources (after cleansing and unit conversion) to the corresponding fields in the unified data model. The transformation process may involve some logical processing, such as generating new fields in the model based on the combination of several fields in the raw data, or classifying or encoding data according to specific rules to conform to the model definition. For example, "Equipment A Energy Consumption" and "Equipment A Operating Hours" in the raw data may be mapped and transformed into "Stationary Combustion Activity Data" in the unified model and associated with the corresponding "Emission Unit." The purpose of this step is to integrate disorganized multi-source data into a standardized and consistent logical framework, forming a standardized data set with a clear structure that is easy for machines to read and program to process, thereby greatly improving the efficiency and accuracy of subsequent data analysis, AI modeling, and carbon emissions report generation.

[0096] Finally, to achieve comprehensive traceability and refined management of supply chain carbon emissions, master data management (MDM) technology is employed to establish unique and authoritative identities for core supply chain entities (such as products, materials, suppliers, production plants, warehouses, and transportation routes) corresponding to multi-source, heterogeneous data. This means that regardless of the name or code used by an entity across different information systems, it is assigned a globally unified "master data ID." For example, a specific product model may be coded P001 in the ERP system and ITEM_XYZ in the SCM system. Through MDM, these codes are linked to the same master data ID. Once these unique identifiers are established, structured data from different data sources describing the same business activity or process can be deeply integrated based on these IDs and predefined business logic associations (such as order numbers linking material procurement, production, processing, warehousing, and transportation). For example, energy consumption data (from equipment monitoring), material consumption data (from the MES system), and subsequent transportation carbon emissions data (from the logistics system) associated with a production batch (with a unique batch ID) can be precisely linked and aggregated. In this way, the final carbon footprint data stream is no longer scattered data points, but a dynamic data network that contains complete contextual information and can reflect complex business relationships. It greatly enhances the insight of the data and provides a solid foundation for refined carbon emission accounting, traceability analysis and optimized decision-making.

[0097] In one embodiment, running an event-driven AI perception engine and identifying instantaneous carbon pulses associated with production and operation events based on a carbon footprint data stream and a dynamic carbon emissions baseline model includes the following steps:

[0098] For each carbon emission unit in the supply chain, a dynamic carbon emission baseline model is established using a time series model combined with historical carbon emission data and periodic operating parameters. The dynamic carbon emission baseline model outputs the predicted carbon emission value and prediction confidence interval of the carbon emission unit;

[0099] Collect and process actual carbon emissions data from carbon footprint data streams;

[0100] Compare actual carbon emissions data with the predicted confidence intervals of the dynamic carbon emissions baseline model;

[0101] A complex event processing engine is used to extract predefined production and operation events from the carbon footprint data stream. Production and operation events include equipment start and stop, process parameter adjustment, and order status change.

[0102] When the actual carbon emission data statistically deviates from the predicted confidence interval of the dynamic carbon emission baseline model, and the actual carbon emission data is strongly correlated with the extracted production and operation events in the time window, it is identified as an instantaneous carbon pulse.

[0103] In this implementation, a dynamic carbon emissions baseline model is constructed for each distinct carbon emission unit in the supply chain, such as a specific production line or a key energy-consuming piece of equipment, using advanced time series models. This model is closely integrated with detailed historical carbon emissions data and periodic operating parameters for each unit. This historical data includes not only past carbon emissions, but also operational parameters that may include factors such as production schedule adjustments, equipment shift changes, and seasonal load fluctuations. The selected time series model, such as an ARIMA (autoregressive integrated moving average) or a more complex deep learning model like an LSTM (long short-term memory network), can learn and capture the complex dynamic relationships between these factors and carbon emissions. The core function of this dynamic carbon emissions baseline model is to output two key pieces of information: a predicted carbon emissions value for the carbon emission unit at a specific future point in time or time period; and a confidence interval associated with the predicted value, which defines a range. Simultaneously, the system continuously collects and processes the latest actual carbon emissions data from the carbon footprint data stream generated in the previous step. This carbon footprint data stream is a high-quality data set formed after a series of preprocessing operations such as multi-source heterogeneous data access, cleaning, unit conversion and structured mapping.

[0104] The data collection process ensures real-time or near-real-time data, enabling the AI ​​perception engine to obtain the most realistic carbon emissions performance for each carbon emission unit at the current moment or within the most recent statistical period. The processing process involves accurately matching this actual data with the corresponding carbon emission unit and ensuring that its data format and units are consistent with the requirements of the dynamic carbon emission baseline model to facilitate subsequent direct comparison. For example, if the baseline model predicts carbon emissions in hours (in kg CO₂e), the collected actual energy consumption data (perhaps in kWh per minute) must be aggregated, converted, and mapped to the same time granularity and emission units. The collected and processed actual carbon emissions data is then rigorously compared with the prediction confidence interval output by the dynamic carbon emission baseline model for the corresponding carbon emission unit. This comparison is performed on a data point-by-data point or statistical period-by-period basis. For each actual carbon emission value, the system checks whether it falls within the predicted confidence interval provided by the baseline model. If the actual carbon emission value exceeds the upper limit of this interval, or in some cases falls below the lower limit, this constitutes a statistically significant deviation. This deviation indicates that the current carbon emission behavior of the carbon emission unit is significantly different from its normal state predicted based on historical data and operating parameters.

[0105] While performing the carbon emissions data comparison, a dedicated Complex Event Processing (CEP) engine proactively extracts predefined production and operational events from the carbon footprint data stream. These events are determined based on experience and business knowledge and typically represent activities that could directly or indirectly lead to significant changes in carbon emissions. Specific event types include, but are not limited to, equipment startups and shutdowns (e.g., large compressors and furnaces), adjustments to key process parameters (e.g., changes to setpoints such as temperature, pressure, and flow rates on a production line), and changes in order status (e.g., urgent orders accelerating production or the start of a large, unscheduled order). By configuring specific event patterns and rules, the CEP engine identifies these events in real time from the continuous data stream and records their occurrence time, the units involved, and related attributes. For example, if the current signal of a device rapidly rises from zero and stabilizes at the operating current, the CEP engine can identify this as an "equipment startup" event. This step provides potential operational explanations and context for unusual fluctuations in carbon emissions.

[0106] When the actual carbon emission data of a carbon emission unit deviates statistically significantly from the predicted confidence interval of its dynamic carbon emission baseline model, and this deviation can be strongly correlated with one or more production and operation events extracted by the complex event processing engine in the time window, the system will identify and confirm that this is an "instantaneous carbon pulse". The judgment of "strong correlation" is usually based on two aspects: one is the close proximity in time, that is, the abnormal fluctuation of carbon emissions follows a production and operation event (or during the event); the other is the logical causal relationship, that is, the production and operation event that occurred may indeed lead to the observed carbon emission changes based on process knowledge or historical experience. The pulse intensity can be quantified as the difference between the actual emissions and the predicted baseline value (or the upper limit of the confidence interval). or The completion of this step marks the accurate identification and definition of a significant abnormal carbon emission event supported by a specific operational event, providing a clear target for subsequent risk assessment and traceability analysis.

[0107] In one embodiment, the method further comprises the steps of:

[0108] Build a supply chain carbon emissions knowledge graph that includes supply chain nodes, production processes, bills of materials, equipment parameters, and potential carbon impact factors;

[0109] When an instantaneous carbon pulse is identified, in the supply chain carbon emission knowledge graph, starting from the carbon emission unit node where the instantaneous carbon pulse occurs, reverse traverse is performed along the material flow, energy flow and information flow paths;

[0110] Analyze the state changes and parameter anomalies of each supply chain node and related production and operation events on the traversal path before and after the instantaneous carbon pulse occurs;

[0111] Apply a decomposition model based on regression analysis to quantify the contribution of each potential upstream influencing factor or concurrent production and operation event to the instantaneous carbon pulse amplitude;

[0112] Combining the contribution degree and business logic rules, the main root events or key influencing factors of the associated supply chain nodes that cause instantaneous carbon pulses are determined and output.

[0113] In this implementation, to intelligently analyze and trace the source of carbon emissions, a comprehensive supply chain carbon emissions knowledge graph is constructed. This knowledge graph is a structured knowledge base that stores and links various supply chain entities and their relationships in a graph format. Entities primarily include supply chain nodes, such as suppliers, production plants, warehouses, and distribution centers; detailed production process flows, including each process step and its sequence; the bill of materials (BOM) required for product manufacturing, which specifies the raw materials and semi-finished products that make up the final product; detailed parameters of key equipment, such as model, power rating, energy efficiency rating, and normal operating process parameter range; and other potential factors influencing carbon emissions, such as ambient temperature, energy type, operator skill level, or shift schedule. These entities serve as "nodes" in the graph, while relationships between them, such as "Factory A uses process B," "Process B consumes material C," "Material C is processed by equipment D," and "Equipment D is affected by ambient temperature," serve as "edges." This approach integrates dispersed, heterogeneous supply chain data into a semantically rich network, clearly depicting the complex pathways and interactions that generate carbon emissions.

[0114] When the system identifies a transient carbon spike, to trace its source, it begins by tracing back through the established supply chain carbon emissions knowledge graph, starting from the specific carbon emission unit node where the spike occurred (for example, the node representing a production line or piece of equipment in the knowledge graph), along predefined material flow, energy flow, and information flow paths. The material flow path traces the source of the raw materials and semi-finished products consumed by the unit and the relevant upstream processing links. The energy flow path traces the supply source and conversion process of the energy used by the unit, such as electricity, steam, and fuel. The information flow path may trace the source of production instructions, scheduling plans, or process parameter settings that affected the unit's operation. Reverse traversal means starting from the result (the carbon spike) and working in the opposite direction of the normal flow to identify possible causes. For example, if an assembly line experiences a sudden increase in carbon emissions, the reverse traversal might trace back to its component supplier along the material flow, to the power supply equipment along the energy flow, and to the production order system along the information flow. This traversal is not aimless, but is based on clearly defined entity relationships in the knowledge graph (such as "supplied by...", "driven by...", "received instructions from...").

[0115] After identifying potential upstream supply chain nodes and production operations events associated with the transient carbon pulse through reverse traversal, a detailed analysis of these "circled" elements is required, focusing on state changes and parameter anomalies within a time window before and during the transient carbon pulse. This requires accessing and examining historical and real-time data for these nodes and events during critical time periods. For example, if the traversal path leads to a batch of materials from an upstream supplier, the quality inspection report for that batch of materials will be reviewed for anomalies in key indicators (such as purity and moisture content). If it leads to a key piece of production equipment, its operating parameters (such as temperature, pressure, speed, and current) will be checked to see if they deviate from the normal set range or exhibit abnormal fluctuations. If it is associated with a production operation event, such as a process adjustment, the specific content and execution of the adjustment will be analyzed to ensure compliance with standards. By comparing the changes in these parameters before and after the pulse, and comparing them with normal baselines or preset thresholds, significant deviations, unexpected state transitions, or operations beyond the norm can be identified. The effect of this step is to concretize abstract correlations into observable data anomalies or state changes, providing direct evidence for determining whether these upstream factors actually cause carbon pulses.

[0116] In order to more accurately determine the actual impact of each identified potential upstream influencing factor or concurrent production and operation event on the observed instantaneous carbon pulse amplitude (i.e., excess carbon emissions), a decomposition model based on regression analysis will be used for quantitative evaluation. This model attempts to establish a relationship between the instantaneous carbon pulse amplitude Y and multiple potential influencing factors. (These factors are the parameter anomalies or state changes identified in the previous step and quantified, such as temperature deviation, material impurity content percentage, etc.) The mathematical relationship between them. A simplified multiple linear regression model can be expressed as ,in The coefficient indicates the effect of a factor when other factors remain unchanged. The average effect of each unit change on the carbon pulse amplitude is, Represents the random error that the model cannot explain. After training the model with historical data, the estimated coefficients to decompose the amplitude of a specific instantaneous carbon pulse, thereby quantifying each specific abnormality that occurs The contribution of factor i to this pulse is approximately .

[0117] Finally, after quantifying the contribution of each potential influencing factor to the magnitude of the instantaneous carbon pulse, the system combines these quantitative results with a pre-established business logic rule base to comprehensively identify and output the primary root cause events or key influencing factors at the relevant supply chain nodes that caused the instantaneous carbon pulse. The business logic rule base stores causal relationship judgment criteria derived from domain expert knowledge, industry best practices, or equipment operating manuals. For example, a rule might read: "If the contribution of factor A (e.g., equipment aging) exceeds threshold T1, and the contribution of factor B (e.g., improper maintenance) also exceeds threshold T2, then determine 'equipment aging and improper maintenance' as the primary root cause." The system then uses these rules to infer and match factors based on the quantified contribution of each factor. The output typically lists one or more of the most likely root causes (e.g., "quality of raw material batch XX from an upstream supplier failed") or key influencing factors (e.g., "temperature control module failure of furnace No. 1 in the production workshop"), possibly accompanied by their contribution scores and a relevant chain of evidence.

[0118] In one embodiment, analyzing the carbon shock propagation risk of an instantaneous carbon pulse in a supply chain includes the following steps:

[0119] Based on supply chain nodes, the supply chain is abstracted into a dynamic network graph containing node attributes and edge attributes, where supply chain nodes represent suppliers or production units, and node edges represent material or order dependencies;

[0120] The initial parameters of the carbon shock propagation model are set based on the carbon emission transmission coefficient, inventory buffer level, production capacity elasticity, and sensitivity to upstream supply disruptions at each supply chain node;

[0121] The intensity of the identified instantaneous carbon pulse and the associated supply chain nodes are used as the initial input of the carbon shock propagation model. The system dynamics simulation is used on a dynamic network diagram to simulate the step-by-step propagation process of the instantaneous carbon pulse along the supply chain.

[0122] Predict the affected nodes downstream of the instantaneous carbon pulse and quantify the additional carbon emissions of the affected nodes caused by the propagation process within a preset time window in the future;

[0123] The additional carbon emissions of all affected nodes are integrated to generate a risk assessment report on the chain impact of instantaneous carbon pulses on the supply chain.

[0124] In this embodiment, the entire supply chain is abstracted into a dynamic network diagram. In this diagram, each independent supplier, production plant, warehouse, or distribution center is considered a "node." These nodes have their own attributes, such as their normal carbon emission intensity, production capacity limit, current inventory level, and sensitivity to environmental regulations. The "edges" connecting these nodes represent the material flow relationship between them (such as the flow of raw materials from suppliers to factories) or order dependencies (such as the output of factory A is the input of factory B). Edge attributes can include the mode commonly used to transport this material flow, the average transportation time, and the reliability of the connection. This network diagram is "dynamic", meaning that the attributes of the nodes and edges will change based on actual operating conditions and the passage of time, thereby more realistically reflecting the real-time status and interdependencies of the supply chain.

[0125] In order for the carbon shock propagation model to accurately simulate the actual situation, a series of initial parameters need to be set according to the specific characteristics of each node in the supply chain. These parameters include the "carbon emission transmission coefficient" of each node, which is used to quantify the extent to which the carbon emissions of the node itself may be affected when the carbon emissions of its upstream node increase. For example, if the carbon footprint of the unit product of upstream supplier A increases , while node B has If the raw materials of node B come from A, the carbon emissions of node B may increase by Related. In addition, it is also necessary to consider the "inventory buffer level", that is, the amount of inventory that the node has that can be used to cope with upstream supply fluctuations or meet short-term demand surges. Higher inventory can absorb the impact of upstream carbon shocks for a certain period of time. The "capacity elasticity" of the node is also an important parameter. It reflects the flexibility of the node to adjust its own production activities (which may involve enabling production lines with different carbon efficiencies) when faced with changes in demand or supply constraints. Finally, it is also necessary to evaluate the "sensitivity of each node to upstream supply disruptions", that is, when a problem occurs in an upstream supply link (possibly caused by a carbon emission event), the degree of direct impact on the normal operation and carbon emission performance of the node. By accurately setting these parameters, subsequent simulations can be made closer to the behavior patterns of the real supply chain.

[0126] After the network diagram is constructed and node parameters are set, the carbon shock propagation model uses the previously identified instantaneous carbon pulse, including its intensity (i.e., excess carbon emissions) and the initial supply chain node where it occurred, as initial input. System dynamics simulation is then used on this dynamic network diagram to simulate how this instantaneous carbon pulse propagates downstream along the material and order flows of the supply chain. System dynamics simulation considers inter-node interactions, feedback loops, and time delays. For example, if a carbon pulse causes a temporary increase in the carbon footprint of a product or limits production at initial node A, its immediate downstream node B may experience an increase in its own product carbon footprint due to receiving high-carbon-footprint materials, or be forced to resort to an alternative, more carbon-intensive production plan due to material shortages. The simulation model then steps through time, calculating at each point in time how the carbon shock propagates from one node to the next, and how it affects the carbon emissions of the receiving node.

[0127] Through system dynamics simulation, it is possible to predict which downstream supply chain nodes will be impacted by the propagation of the initial instantaneous carbon pulse, and to quantify the additional carbon emissions expected to be generated by these affected nodes within a preset time window in the future (such as the next week or month). The simulation model tracks each downstream node that may be affected, and estimates the additional carbon emissions due to the carbon shock from the upstream (which may be high-carbon materials, supply delays or interruptions, etc.) based on its normal carbon emissions baseline. This quantification process takes into account the intensity of the impact, the carbon emission transfer characteristics of the node, and the node's response measures (such as activating backup plans). For example, for the affected downstream node k, its total additional carbon emissions generated by the propagation of this carbon pulse in the future time window T This step clarifies the scope of potential victims of the carbon shock and provides each affected party with a specific value for the additional carbon emissions burden they may bear, providing a quantitative basis for subsequent risk assessment and response strategy formulation.

[0128] After predicting and quantifying the additional carbon emissions of all potentially affected downstream nodes, this data will be combined to generate a risk assessment report on the chain impact of the instantaneous carbon pulse on the entire supply chain. This report will summarize the total additional carbon emissions expected to be generated by the entire supply chain network within a preset time window caused by the initial carbon pulse, namely , where k represents the number of nodes affected downstream. The report not only presents this overall impact numerically but also includes a list of affected nodes, a breakdown of their respective additional emissions, a diagram of the primary pathways through which the impact propagates, and an analysis of the potential short- and medium-term impacts on the overall carbon performance of the supply chain. It also identifies nodes that play a key role in the propagation of the carbon impact (e.g., those that amplify the impact or are inherently vulnerable).

[0129] In one embodiment, combining the carbon shock transmission risk and the carbon absorption capacity of each node in the supply chain and using a dynamic matching algorithm to obtain the optimal absorption allocation plan for the instantaneous carbon pulse includes the following steps:

[0130] Dynamically evaluate the potential carbon absorption capacity limit of each supply chain node and the corresponding marginal absorption cost function;

[0131] Setting the optimization goal of coordinated absorption to minimize the overall absorption cost and maximize the emission reduction efficiency;

[0132] The optimization goal is used as the objective function, and the upper limit of the carbon absorption capacity and marginal absorption cost function of each supply chain node, as well as the requirement that the total absorption capacity of all supply chain nodes must cover the additional carbon emissions, are used as constraints to construct the absorption allocation model.

[0133] A heuristic optimization algorithm is used to solve the consumption allocation model, and the output is the optimal consumption allocation plan that includes the specific consumption volume, execution time window and expected cost that each supply chain node should bear.

[0134] In this implementation, a dynamic assessment is conducted on each node in the supply chain that can potentially participate in carbon emission reduction (such as production plants, logistics centers, etc.) to determine its potential upper limit of carbon absorption capacity and the corresponding marginal absorption cost function. It refers to the maximum carbon emission reduction that node i can achieve within a certain period of time by taking various emission reduction measures (such as process improvement, equipment upgrade, energy substitution, purchase of carbon sinks, etc.). This assessment is dynamic because it takes into account the node's current operating conditions, technology level, available resources, and changes in external market conditions (such as carbon price and technology cost). It describes the amount of carbon emissions per additional unit absorbed (or reduced) by node i. Costs to pay. Typically, as the volume of consumption increases, marginal costs tend to rise, as companies prioritize lower-cost emission reduction measures over more expensive ones. For example, simple energy-saving improvements may be inexpensive initially, but achieving larger-scale emissions reductions may require investment in expensive new technologies.

[0135] After clarifying the absorption capacity and cost characteristics of each node, it is necessary to set the optimization goal of coordinated absorption. This goal is usually twofold: on the one hand, it is to minimize the overall absorption cost of the entire supply chain while completing the necessary carbon absorption tasks; on the other hand, it may also pursue maximizing the overall emission reduction efficiency or total emission reduction within a certain cost budget. The most common goal is to minimize the overall absorption cost. This means that when faced with a total additional carbon emissions generated by an instantaneous carbon pulse and its propagation, When the consumption task is reasonably distributed among the supply chain nodes with consumption capacity, the consumption task can be completed. This total absorption target has the lowest total economic cost. Maximizing emission reduction efficiency may focus on emission reduction per unit cost, or maximizing carbon benefits under a specific incentive mechanism.

[0136] Then, based on the set optimization objectives and the parameters of each node, a mathematical consumption allocation model is constructed. The core of the model is an objective function and a series of constraints. If the optimization goal is to minimize the overall consumption cost, the objective function can be expressed as ,in is the consumption of node i The total cost of unit carbon emissions can be expressed as the marginal cost function The integral is obtained, that is .

[0137] The main constraints include: the consumption capacity allocated to each node i It cannot exceed its own carbon absorption capacity limit, that is, Secondly, the total absorption capacity of all supply chain nodes involved in absorption must at least cover (or be equal to) the total additional carbon emissions caused by the instantaneous carbon pulse and its propagation, that is, In addition, other business-related constraints may also be included, such as the minimum absorption capacity that a specific node must bear, and the combination restrictions of different types of emission reduction measures.

[0138] Since the constructed absorption and allocation model usually involves nonlinear cost functions and multiple constraints, it is a complex optimization problem that is difficult to solve directly using traditional analytical methods. Therefore, heuristic optimization algorithms are used to find approximate optimal solutions. Commonly used heuristic algorithms include genetic algorithms (GA), simulated annealing (SA), particle swarm optimization (PSO) or ant colony algorithms (ACO). These algorithms can perform global searches in a huge solution space by simulating natural evolution, physical annealing processes or swarm intelligence behaviors, effectively avoiding falling into local optimality, and finding high-quality solutions within a reasonable time. The output of the algorithm is a specific optimal absorption and allocation plan, which will list in detail the specific amount of carbon absorption that each selected supply chain node (that is, those nodes that are assigned absorption tasks) should bear. , the time windows for executing these consumption tasks (taking into account the urgency of the tasks and the execution capacity of the nodes), and the expected costs calculated according to their marginal consumption cost functions .

[0139] In one embodiment, the optimal absorption allocation plan is automatically executed using smart contracts and oracles on a blockchain platform associated with the supply chain, and an incentive allocation mechanism is adopted to ensure that each node in the supply chain completes all carbon impact absorption tasks in the optimal absorption allocation plan, including the following steps:

[0140] Based on the optimal carbon absorption allocation plan, a carbon absorption task smart contract containing task objectives, acceptance criteria and incentive rules is generated and deployed for each designated task-bearing node on the blockchain platform associated with the supply chain;

[0141] When the carbon absorption task smart contract meets the preset conditions, it sends the encrypted absorption task instruction to the registered digital identity of the task undertaking node;

[0142] The task-bearing node completes the consumption task and submits the key performance data and emission reduction evidence hash during the execution process to the oracle on the blockchain platform through the interface;

[0143] Use the oracle to obtain the actual carbon emission data of the task-undertaking node during the task cycle from the carbon footprint data stream, and feed the actual carbon emission data into the corresponding carbon absorption task smart contract after cross-validation;

[0144] Based on the carbon absorption task smart contract and according to the acceptance criteria, the task completion status is determined, the allocation of pre-stored carbon credits is executed according to the incentive rules to complete the incentive distribution, and all transactions and status changes are recorded on the blockchain platform.

[0145] In this embodiment, after obtaining the optimal carbon absorption allocation plan, to ensure transparency, trustworthiness, and automated execution, a dedicated carbon absorption task smart contract is automatically generated and deployed on a blockchain platform associated with the supply chain for each supply chain node designated to undertake carbon absorption. This smart contract is a self-executing piece of computer code, written into the blockchain, making it tamper-proof and transparent. The smart contract clearly encapsulates the core elements of the node's task: first, a clear task objective, for example, node A must reduce emissions by X tons of carbon dioxide equivalent within the next month. Second, specific, quantifiable acceptance criteria detail how emissions reductions will be calculated, for example, by comparing energy consumption data adjusted for output before and after the task, or based on certified third-party emission reduction verification reports. Finally, clear incentive rules specify the rewards that the node will receive upon successful completion of the task, for example, Y carbon credits for each ton of carbon emissions absorbed, or a direct reward of a certain amount. Deploying a smart contract involves compiling and publishing its code to the blockchain network, making it a callable and executable on-chain program.

[0146] When the pre-set activation conditions in a carbon absorption task smart contract are met—for example, when the current time reaches the task start date and time set in the smart contract, or when a precondition (such as an upstream task completion signal) is confirmed on the blockchain—the smart contract automatically triggers and sends an encrypted absorption task instruction to the designated task-carrying node's blockchain-registered digital identity (typically an encrypted public key address or a digital certificate tied to a corporate identity). This instruction contains all the key information necessary to execute the task, such as the specific emission reduction target, execution time window, acceptance criteria, and reference implementation guidelines. Because it is sent in an encrypted format, the instruction's confidentiality and targeted nature are guaranteed. The task-carrying node receives and decrypts this instruction through its blockchain client or blockchain-connected application, officially receiving and commencing its absorption task. This automated instruction delivery reduces delays and errors associated with manual intervention, ensuring that tasks are initiated promptly and according to optimal planning.

[0147] After receiving the encrypted carbon reduction task instructions, the task-assigning node begins executing its assigned carbon reduction task according to the instructions. This may involve a series of specific actions, such as adjusting production and operation plans, optimizing energy usage, implementing energy-saving technology upgrades, and purchasing green electricity or carbon credits. During task execution, the node submits its key performance indicators (KPIs) and the hash values ​​of relevant emission reduction evidence to the blockchain platform through a pre-defined interface. KPIs may include real-time energy consumption readings, production parameter adjustment records, and equipment operation logs. Emission reduction evidence may include external audit reports, monitoring data screenshots, and purchase receipts. Submitting hash values ​​rather than the original data ensures data integrity and immutability while protecting data privacy. Any slight change to the original data will result in a completely different hash value. The oracle, as a trusted third-party service, is responsible for securely transmitting this off-chain information to the blockchain.

[0148] The oracle not only receives data submitted by the task-assigning node but also proactively obtains the node's actual carbon emissions data during the task execution cycle from a previously constructed and continuously updated carbon footprint data stream. This data is independent of the data submitted by the node itself and is derived from objective data sources such as sensors and information systems deployed at each node in the supply chain. The oracle cross-validates and compares the node-submitted evidence (authenticated by hash values) with objective data obtained from the carbon footprint data stream to verify the accuracy of task execution and emissions reduction results. For example, if a node claims to have reduced energy consumption through a certain measure, the oracle will search the carbon footprint data stream for the node's actual energy consumption records for the corresponding period for verification. After completing cross-validation and confirming the consistency and reliability of the data, the oracle will apply this verified carbon emissions data, which reflects the actual completion of the task, to the corresponding carbon absorption task smart contract on the blockchain.

[0149] Finally, after receiving verified actual carbon emissions data fed by an oracle, the carbon absorption task smart contract automatically determines task completion according to the contract's pre-programmed acceptance criteria. For example, the contract might determine whether the actual emissions reductions have met the task target. If the task is deemed successfully completed (or reaches a certain predetermined level of completion), the smart contract automatically triggers and executes the pre-set incentive rules. This typically involves automatically transferring a predetermined number of carbon credits or reward funds from a pre-existing account (perhaps established by a core supply chain enterprise or platform) to the registered digital identity account of the node that successfully completed the task. The entire incentive distribution process requires no manual approval and is driven entirely by code logic, resulting in rapid and transparent results. All transactions and status changes related to task execution, data verification, outcome determination, and incentive distribution are permanently and immutably recorded on the blockchain platform, accessible and auditable by all authorized parties.

[0150] The present invention also discloses an AI-based carbon footprint and supply chain carbon emission collaborative management platform, including a memory, a processor, and a computer program stored on the memory and runnable on the processor. When the processor executes the computer program, it implements the AI-based carbon footprint and supply chain carbon emission collaborative management method as described in any of the above embodiments.

[0151] Among them, the processor can adopt a central processing unit (CPU). Of course, according to actual usage, other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. can also be adopted. The general-purpose processor can adopt a microprocessor or any conventional processor, etc., and this application does not impose any restrictions on this.

[0152] Among them, the memory can be an internal storage unit of a computer device, such as a hard disk or memory of a computer device, or an external storage device of a computer device, such as a plug-in hard disk, smart memory card (SMC), secure digital card (SD) or flash memory card (FC) equipped on the computer device. In addition, the memory can also be a combination of an internal storage unit and an external storage device of a computer device. The memory is used to store computer programs and other programs and data required by the computer device. The memory can also be used to temporarily store data that has been output or is to be output. This application does not impose any restrictions on this.

[0153] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of protection of the present application is limited to these examples. In line with the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of different aspects of one or more embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.

[0154] The one or more embodiments of this application are intended to encompass all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of this application should be included in the scope of protection of this application.

Claims

1. An AI-based method for collaborative management of carbon footprint and supply chain carbon emissions, characterized by: The steps include: Real-time access to supply chain carbon-related multi-source heterogeneous data from each supply chain node, and pre-processing of multi-source heterogeneous data to form a carbon footprint data stream; Run an event-driven AI perception engine and identify instantaneous carbon pulses associated with production and operation events based on carbon footprint data streams and a dynamic carbon emissions baseline model; Analyze the risk of carbon shock transmission in the supply chain caused by instantaneous carbon pulses; Combining the carbon shock transmission risk and the carbon absorption capacity of each node in the supply chain and using a dynamic matching algorithm to match, we can obtain the optimal absorption allocation plan for instantaneous carbon pulses. Utilize smart contracts and oracles on the blockchain platform associated with the supply chain, and adopt an incentive allocation mechanism to automatically execute the optimal absorption allocation plan, so that each node in the supply chain can complete all carbon impact absorption tasks in the optimal absorption allocation plan.

2. The AI-based carbon footprint and supply chain carbon emissions collaborative management method according to claim 1 is characterized in that: The carbon footprint data stream includes product carbon footprint data and carbon footprint raw material data. The product carbon footprint data represents the total amount of greenhouse gas emissions generated by the supply chain products throughout their entire life cycle from raw material mining to waste treatment. The carbon footprint raw material data represents the greenhouse gas emissions generated by the raw materials used in the supply chain products during the extraction, primary processing and transportation to the factory.

3. The AI-based carbon footprint and supply chain carbon emissions collaborative management method according to claim 1 is characterized in that: The real-time access to carbon-related multi-source heterogeneous data from each supply chain node includes the following steps: Through pre-deployed edge computing gateways and developed application program interface extractors, the system can connect to IoT devices, enterprise resource planning systems, and supply chain management systems at various supply chain nodes using standard IoT protocols, direct database connections, or application program interface calls. Configure data subscription and scheduled pull tasks to capture multi-source heterogeneous data including equipment energy consumption readings, production batch information, material flow records, transportation vehicle carbon emission factors, and external environmental impact parameters; Use stream processing technology to buffer and distribute incoming multi-source heterogeneous data in real time; Perform preliminary format verification and metadata tagging on multi-source heterogeneous data from different data sources, and establish a data archive for each supply chain node to record the data source type, access frequency, and data quality profile.

4. The AI-based carbon footprint and supply chain carbon emission collaborative management method according to claim 1 or 3 is characterized in that: The preprocessing of multi-source heterogeneous data to form a carbon footprint data stream includes the following steps: Use algorithms including linear interpolation, mean filling, or business rule-based algorithms to fill in missing key carbon-related parameters in multi-source heterogeneous data; Use statistical methods to identify and correct abnormal data points in multi-source heterogeneous data caused by sensor failure or data transmission errors; Convert carbon-related data in different measurement units from multi-source heterogeneous data into a unified standard unit based on preset conversion rules; Based on a predefined unified carbon-related data model, structured mapping and conversion are performed on cleaned multi-source heterogeneous data; Master data management technology is used to establish unique identifiers for core entities corresponding to multi-source heterogeneous data, and structured data from different sources are integrated based on unique identifiers and business logic associations to form a carbon footprint data stream containing complete contextual information.

5. The AI-based carbon footprint and supply chain carbon emission collaborative management method according to claim 1 is characterized in that: The operation of the event-driven AI perception engine and identifying instantaneous carbon pulses associated with production and operation events based on the carbon footprint data stream and the dynamic carbon emission baseline model includes the following steps: For each carbon emission unit in the supply chain, a dynamic carbon emission baseline model is established using a time series model combined with historical carbon emission data and periodic operating parameters. The dynamic carbon emission baseline model outputs the predicted carbon emission value and prediction confidence interval of the carbon emission unit; Collect and process actual carbon emissions data from carbon footprint data streams; Compare actual carbon emissions data with the predicted confidence intervals of the dynamic carbon emissions baseline model; A complex event processing engine is used to extract predefined production and operation events from the carbon footprint data stream. Production and operation events include equipment start and stop, process parameter adjustment, and order status change. When the actual carbon emission data statistically deviates from the predicted confidence interval of the dynamic carbon emission baseline model, and the actual carbon emission data is strongly correlated with the extracted production and operation events in the time window, it is identified as an instantaneous carbon pulse.

6. The AI-based carbon footprint and supply chain carbon emissions collaborative management method according to claim 5 is characterized in that: The method further comprises the steps of: Build a supply chain carbon emissions knowledge graph that includes supply chain nodes, production processes, bills of materials, equipment parameters, and potential carbon impact factors; When an instantaneous carbon pulse is identified, in the supply chain carbon emission knowledge graph, starting from the carbon emission unit node where the instantaneous carbon pulse occurs, reverse traverse is performed along the material flow, energy flow and information flow paths; Analyze the state changes and parameter anomalies of each supply chain node and related production and operation events on the traversal path before and after the instantaneous carbon pulse occurs; Apply a decomposition model based on regression analysis to quantify the contribution of each potential upstream influencing factor or concurrent production and operation event to the instantaneous carbon pulse amplitude; Combining the contribution degree and business logic rules, the main root events or key influencing factors of the associated supply chain nodes that cause instantaneous carbon pulses are determined and output.

7. The AI-based carbon footprint and supply chain carbon emissions collaborative management method according to claim 6 is characterized in that: The analysis of the carbon shock propagation risk of an instantaneous carbon pulse in the supply chain includes the following steps: Based on supply chain nodes, the supply chain is abstracted into a dynamic network graph containing node attributes and edge attributes, where supply chain nodes represent suppliers or production units, and node edges represent material or order dependencies; The initial parameters of the carbon shock propagation model are set based on the carbon emission transmission coefficient, inventory buffer level, production capacity elasticity, and sensitivity to upstream supply disruptions at each supply chain node; The intensity of the identified instantaneous carbon pulse and the associated supply chain nodes are used as the initial input of the carbon shock propagation model. The system dynamics simulation is used on a dynamic network diagram to simulate the step-by-step propagation process of the instantaneous carbon pulse along the supply chain. Predict the affected nodes downstream of the instantaneous carbon pulse and quantify the additional carbon emissions of the affected nodes caused by the propagation process within a preset time window in the future; The additional carbon emissions of all affected nodes are integrated to generate a risk assessment report on the chain impact of instantaneous carbon pulses on the supply chain.

8. The AI-based carbon footprint and supply chain carbon emissions collaborative management method according to claim 7 is characterized in that: The method of combining the carbon shock transmission risk and the carbon absorption capacity of each node in the supply chain and using a dynamic matching algorithm to obtain the optimal absorption allocation plan for instantaneous carbon pulses includes the following steps: Dynamically evaluate the potential carbon absorption capacity limit of each supply chain node and the corresponding marginal absorption cost function; Setting the optimization goal of coordinated absorption to minimize the overall absorption cost and maximize the emission reduction efficiency; The optimization goal is used as the objective function, and the upper limit of the carbon absorption capacity and marginal absorption cost function of each supply chain node, as well as the requirement that the total absorption capacity of all supply chain nodes must cover the additional carbon emissions, are used as constraints to construct the absorption allocation model. A heuristic optimization algorithm is used to solve the consumption allocation model, and the output is the optimal consumption allocation plan that includes the specific consumption volume, execution time window and expected cost that each supply chain node should bear.

9. The AI-based carbon footprint and supply chain carbon emissions collaborative management method according to claim 8 is characterized in that: The method of utilizing smart contracts and oracles on a blockchain platform associated with the supply chain and adopting an incentive allocation mechanism to automatically execute the optimal absorption allocation plan so that each node in the supply chain completes all carbon impact absorption tasks in the optimal absorption allocation plan includes the following steps: Based on the optimal carbon absorption allocation plan, a carbon absorption task smart contract containing task objectives, acceptance criteria and incentive rules is generated and deployed for each designated task-bearing node on the blockchain platform associated with the supply chain; When the carbon absorption task smart contract meets the preset conditions, it sends the encrypted absorption task instruction to the registered digital identity of the task undertaking node; The task-bearing node completes the consumption task and submits the key performance data and emission reduction evidence hash during the execution process to the oracle on the blockchain platform through the interface; Use the oracle to obtain the actual carbon emission data of the task-undertaking node during the task cycle from the carbon footprint data stream, and feed the actual carbon emission data into the corresponding carbon absorption task smart contract after cross-validation; Based on the carbon absorption task smart contract and according to the acceptance criteria, the task completion status is determined, the allocation of pre-stored carbon credits is executed according to the incentive rules to complete the incentive distribution, and all transactions and status changes are recorded on the blockchain platform.

10. An AI-based carbon footprint and supply chain carbon emissions collaborative management platform, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements the AI-based carbon footprint and supply chain carbon emission collaborative management method as described in any one of claims 1 to 9.

Images