Carbon footprint calculation method and system based on artificial intelligence technology

Abstract

The present invention proposes a carbon footprint accounting method and system based on artificial intelligence technology, which relates to the field of environmental protection science and technology. The method includes: collecting activity data, and obtaining a battery life cycle flow chart based on the activity data; dividing each link into different areas, dividing the activity data in the area into time periods and classifying them, and selecting a carbon footprint accounting method according to the classification to obtain the carbon footprint accounting results under the classification; integrating the obtained results to obtain comprehensive accounting results, and constructing a prediction model based on the existing carbon footprint data; setting up collection equipment in key areas, and correcting and optimizing the prediction model through actual carbon emission data and predicted data in key areas; regularly updating data and prediction models, which not only improves the accuracy and comprehensiveness of carbon footprint accounting, but also enhances the adaptability and prediction ability of the prediction model, providing strong support for the sustainable development of the battery industry.

Description

Technical Field

[0001] The present application relates to the field of environmental protection science and technology, and in particular to a carbon footprint calculation method and system based on artificial intelligence technology. Background Art

[0002] As the world pays more and more attention to environmental protection and sustainable development, carbon footprint accounting has become an important means to evaluate the environmental impact of enterprises, products and even the entire supply chain. Especially in the battery industry, since the production, use and disposal of batteries involve multiple links and complex supply chains, its carbon footprint accounting is particularly complex and important.

[0003] Traditional carbon footprint accounting methods often rely on manual collection and organization of data, and then calculate through preset algorithms or models. However, this method has many limitations, such as incomplete data collection, single accounting method, and limited prediction ability. Especially for products such as batteries with long life cycles and complex supply chains, traditional accounting methods are difficult to accurately reflect their carbon footprint throughout their life cycle.

[0004] In addition, with the advancement of technology and changes in the market, battery production processes, raw material usage, energy consumption, etc. are constantly changing. Changes in these factors will directly affect the carbon footprint of batteries, making it difficult for traditional static accounting methods to adapt to the rapidly changing market environment.

[0005] Therefore, a more intelligent, flexible and accurate carbon footprint accounting method is needed to cope with the complex and changing supply chain and rapidly changing market environment of the battery industry. The carbon footprint accounting method based on artificial intelligence technology has emerged. It uses advanced technologies such as big data and machine learning to achieve comprehensive and accurate calculation and prediction of the carbon footprint of the battery throughout its life cycle.

[0006] However, the existing carbon footprint accounting methods based on artificial intelligence technology still have some shortcomings, such as insufficiently detailed data classification, insufficient flexibility in the selection of accounting methods, and insufficient timeliness in the optimization of prediction models. These problems limit the effectiveness and accuracy of the methods in practical applications. Summary of the invention

[0007] Based on the above problems, this application provides a carbon footprint accounting method and system based on artificial intelligence technology. By collecting activity data from various links in the supply chain, a battery life cycle flow chart is constructed, and appropriate accounting methods are selected according to different links and time periods for carbon footprint accounting; at the same time, machine learning technology is used to build a prediction model to predict the carbon emissions of future activities, and the prediction model is calibrated and optimized according to actual data; this not only improves the accuracy and comprehensiveness of carbon footprint accounting, but also enhances the adaptability and prediction ability of the prediction model, providing strong support for the sustainable development of the battery industry.

[0008] The purpose of this application is achieved by the following technical solutions:

[0009] This application provides a carbon footprint calculation method based on artificial intelligence technology, the method comprising:

[0010] S1. Collect activity data from all links of the supply chain and obtain a battery life cycle flow chart based on the activity data; the links include raw material mining, transportation, processing, battery production, use, recycling and disposal;

[0011] S2. Divide each link into different regions, divide the activity data in the region into time periods and classify them, select the carbon footprint accounting method according to the classification in different time periods to obtain the carbon footprint accounting results under the classification in the time period; the classification includes direct emissions, energy consumption and electricity consumption; determine the key areas according to the regional carbon emissions in the same time period;

[0012] S3. Integrate the results obtained from different accounting methods to obtain comprehensive accounting results, and generate a carbon footprint report based on the comprehensive accounting results;

[0013] S4. Use machine learning to build a prediction model based on existing carbon footprint data to predict carbon emissions from future activities;

[0014] S5. Setting up collection equipment in key areas to obtain actual carbon emission data in the key areas, and calibrating and optimizing the prediction model based on the actual carbon emission data and prediction data in the key areas;

[0015] S6. Update data and prediction models regularly, compare the actual carbon emission data with the prediction results of the latest prediction model and the prediction results of the optimal prediction model in the previous cycle, and determine the current optimal model based on the comparison results.

[0016] Preferably, the S1 includes:

[0017] Utilize multiple data sources to collect activity data from all links of the supply chain;

[0018] Through graphical tools, the various links in the battery life cycle are connected in sequence to display the flow path of materials and energy and obtain a battery life cycle flow chart.

[0019] Preferably, the S2 includes:

[0020] Determine the boundaries of each link;

[0021] Each link is further divided into different areas based on geographical location, function and / or management responsibility;

[0022] Divide time into different sections;

[0023] categorizing the collected activity data, including direct emissions, energy consumption, and electricity consumption;

[0024] In each time period, the classified activity data is calculated according to the selected accounting method to obtain the carbon footprint accounting results of the classification in the time period;

[0025] Compare carbon emissions in various regions over the same time period and identify key areas.

[0026] Preferably, the classified activity data is calculated according to a selected calculation method in each time period, including:

[0027] For direct emissions during the production process, a life cycle assessment method is used to obtain the first carbon emissions;

[0028] For carbon emissions generated by energy consumption, the emission factor method is used to obtain the second carbon emissions;

[0029] For carbon emissions from electricity consumption, the marginal carbon emission coefficient method is used to obtain the third carbon emission amount;

[0030] Among them, the second carbon emission is:

[0031]

[0032] Among them, C2 is the total second carbon emission of each region within the preset time period, Rij is the usage of the jth energy in the ith region within the preset time period; Fij is the conversion or transmission efficiency of the jth energy in the ith region within the preset time period; Yij is the emission factor of the jth energy in the ith region; Zij is the quality factor of the jth energy in the ith region; Xij is the correction factor, m is the total number of regions in each link, and n is the type of energy consumption in the ith region within the preset time period;

[0033] Among them, the third emission is:

[0034]

[0035] Among them, C2 is the total third carbon emission of each region in the preset time period, m is the total number of regions in each link, Eik is the amount of electricity consumed by the i-th region from the k-th power source in the preset time period; MCEFik is the marginal carbon emission coefficient when the i-th region obtains electricity from the k-th power source in the preset time period; v is the total number of power sources in the i-th region in the preset time period; is the total carbon emissions of the k-th power grid within the preset time period, is the load of the kth power node in the ith region within the preset time period.

[0036] Preferably, comparing the carbon emissions of various regions within the same time period to determine the key regions includes:

[0037] Standardize the carbon emissions in the region to obtain standardized carbon emissions;

[0038] The normalized emissions are:

[0039]

[0040] Among them, BCzi is the standardized carbon emissions of the ith region within the preset time period, Czi is the carbon emissions of the ith region within the preset time period, Si is the area of ​​the ith region; Hi is the production scale score of the ith region within the preset time period; DEi is the emission reduction potential score of the ith region within the preset time period; w1, w2, and w3 are all the first weight coefficients;

[0041] The standardized carbon emissions of each region are compared with the preset threshold, and the regions where the standardized carbon emissions exceed the preset threshold are identified as key regions.

[0042] Preferably, the S4 includes:

[0043] Collect carbon footprint accounting results and other relevant data, including but not limited to direct emissions, energy consumption, electricity consumption, production scale, energy efficiency, process improvements and environmental factors;

[0044] Establishing a machine learning model based on the carbon footprint calculation results and other relevant data, and training and optimizing the machine learning model;

[0045] Through cross-validation, the generalization ability of the model on unknown data is evaluated and the model parameters are further adjusted;

[0046] Through the trained machine learning model, the carbon emissions of future activities are predicted, including the carbon emissions of future activities in various regions.

[0047] Preferably, the S5 includes:

[0048] Obtaining the difference between the predicted data and the actual data of each key area as the first difference;

[0049] If the first difference exceeds the preset difference range, the corresponding area is taken as the first abnormal area;

[0050] Obtain at least one factor affecting the first difference in each first abnormal area and an influence weight of the corresponding factor;

[0051] Obtaining scores of influencing factors according to factors and weights obtained in the plurality of first abnormal areas;

[0052]

[0053] Among them, Ysα is the score of the αth influencing factor, Ysαβ is the influence weight of the αth influencing factor on the first difference of the βth first abnormal area; z is the number of first abnormal areas;

[0054] The influencing factors with scores greater than the preset score threshold are used as inputs of the prediction model to optimize the model.

[0055] Preferably, the S6 includes:

[0056] Obtaining the difference between the predicted data and the actual data of each key area through the latest prediction model as a second difference; obtaining a first comprehensive evaluation coefficient according to a plurality of second differences;

[0057] Obtaining the difference between the predicted data and the actual data of each key area through the previous optimal prediction model as a third difference; obtaining a second comprehensive evaluation coefficient according to a plurality of third differences;

[0058] The current optimal prediction model is determined according to the comprehensive evaluation coefficient, and the model corresponding to the smaller comprehensive evaluation coefficient is selected as the current optimal model.

[0059] This application provides a carbon footprint accounting system based on artificial intelligence technology, the system comprising:

[0060] A data collection module is used to collect activity data from various links of the supply chain and obtain a battery life cycle flow chart based on the activity data; the links include raw material mining, transportation, processing, battery production, use, recycling and disposal;

[0061] The accounting module is used to divide each link into different areas, divide the activity data in the area into time periods and classify them, and select the carbon footprint accounting method according to the classification in different time periods to obtain the carbon footprint accounting results under the classification in the time period; the classification includes direct emissions, energy consumption and electricity consumption; and determine the key areas according to the regional carbon emissions in the same time period;

[0062] The integration module is used to integrate the results obtained from different accounting methods to obtain comprehensive accounting results, and generate a carbon footprint report based on the comprehensive accounting results;

[0063] The prediction module is used to build a prediction model based on existing carbon footprint data using machine learning to predict carbon emissions from future activities;

[0064] An optimization module is used to set up collection equipment in key areas to obtain actual carbon emission data in key areas, and to calibrate and optimize the prediction model based on the actual carbon emission data and prediction data in key areas;

[0065] The update module is used to regularly update data and prediction models, compare the actual carbon emission data with the prediction results of the latest prediction model and the prediction results of the optimal prediction model in the previous cycle, and determine the current optimal model based on the comparison results.

[0066] The present application also provides an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any method described in the present application when executing the computer program.

[0067] The beneficial effects of the present invention include: by collecting activity data from various links of the supply chain (raw material mining, transportation, processing, battery production, use, recycling and disposal) and generating a battery life cycle flow chart, the method can comprehensively and systematically calculate the carbon footprint of the battery, providing accurate data support for reducing carbon emissions. Each link is divided into different areas, and the corresponding accounting method is selected according to the classification of activity data (direct emissions, energy consumption, electricity consumption) in each time period, so that the carbon footprint is calculated more finely and accurately. This helps to identify key areas and high emission periods, so as to carry out targeted emission reduction measures. Using machine learning technology to build a prediction model, it is possible to predict the carbon emissions of future activities based on historical data, including the carbon emissions of future activities in each region. This not only helps to plan emission reduction strategies in advance, but also can correct and optimize the prediction model through real-time or regular data feedback to improve the prediction accuracy. By comparing the predicted data and actual data of key areas, abnormal areas and influencing factors are discovered and identified in time. These factors are quantified and used as inputs to the prediction model to optimize the model, thereby ensuring that the model always maintains a high prediction accuracy and adaptability. By regularly updating data and forecasting models, and comparing actual carbon emission data with the forecast results of different forecasting models, the optimal forecasting model can be selected. This method can ensure that the model is always based on the latest data and information, provide accurate carbon emission forecasts, and provide strong support for decision-making. Based on the carbon footprint reports and forecast results generated by this method, corporate decision makers, government regulators, and the public can have a clearer understanding of the carbon emissions of the supply chain and formulate and implement effective emission reduction strategies. This will help reduce overall carbon emissions and promote the development of the industry in a low-carbon, green, and sustainable direction.

[0068] In summary, the carbon footprint accounting method based on artificial intelligence technology provides strong support for reducing carbon emissions and promoting sustainable development through comprehensive data collection, refined accounting, intelligent prediction and optimization, timely anomaly detection and model adjustment, and scientific model evaluation and updating. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 It is a schematic diagram of a carbon footprint calculation method based on artificial intelligence technology provided in an embodiment of the present application. DETAILED DESCRIPTION

[0070] Below, the present application is further described in conjunction with the accompanying drawings and specific implementation methods. It should be noted that, under the premise of no conflict, the various embodiments or technical features described below can be arbitrarily combined to form a new embodiment.

[0071] See also Figure 1 Some embodiments of the present application provide a carbon footprint calculation method based on artificial intelligence technology, the method comprising:

[0072] S1. Collect activity data from all links of the supply chain and obtain a battery life cycle flow chart based on the activity data; the links include raw material mining, transportation, processing, battery production, use, recycling and disposal;

[0073] S2. Divide each link into different regions, divide the activity data in the region into time periods and classify them, and select the carbon footprint accounting method according to the classification in different time periods to obtain the carbon footprint accounting results under the classification in the time period; the classification includes direct emissions, energy consumption and electricity consumption; determine the key areas according to the regional carbon emissions in the same time period; for example, for direct emissions in the production process, LCA (life cycle assessment) is used; for energy consumption, the emission factor method is used; for electricity consumption, the marginal carbon emission coefficient method is used;

[0074] S3. Integrate the results obtained from different accounting methods to obtain comprehensive accounting results, and generate a carbon footprint report based on the comprehensive accounting results;

[0075] S4. Use machine learning to build a prediction model based on existing carbon footprint data to predict carbon emissions from future activities;

[0076] S5. Setting up collection equipment in key areas to obtain actual carbon emission data in the key areas, and calibrating and optimizing the prediction model based on the actual carbon emission data and prediction data in the key areas;

[0077] S6. Update data and prediction models regularly, compare the actual carbon emission data with the prediction results of the latest prediction model and the prediction results of the optimal prediction model in the previous cycle, and determine the current optimal model based on the comparison results.

[0078] The activity data includes:

[0079] Data during production:

[0080] Raw material mining and processing: including energy consumption and waste generation during the raw material mining process, as well as energy consumption and emissions during the processing process.

[0081] Battery production: involves the use of production equipment, energy consumption (such as electricity, gas, etc.), waste and emissions generated during the production process, etc.

[0082] Data during transportation:

[0083] Raw material transportation: The type of transportation used, transportation distance, energy consumption and emissions during transportation of raw materials from the mining site to the processing plant.

[0084] Product transportation: Similar data generated during the transportation of batteries from the production plant to the point of sale or end user.

[0085] Data during use:

[0086] Energy consumption: For batteries, especially when they are used in devices that require continuous power supply such as electric vehicles, the amount of electricity consumed during use is an important activity data.

[0087] Usage conditions: including the frequency of battery use, load conditions, ambient temperature, etc. These factors may affect the energy efficiency and life of the battery, and thus affect its carbon emissions.

[0088] Data during recycling and disposal:

[0089] Recycling efficiency: the recycling rate of the battery during recycling, the energy consumption and emissions generated during the recycling process, etc.

[0090] Disposal method: The treatment method of discarded batteries (such as landfill, incineration, recycling, etc.) and its corresponding environmental impact data.

[0091] Indirect activity data:

[0092] Supply chain data: including energy consumption, emissions and related activities of suppliers at each link in the supply chain. Although these data are not directly generated by the accounting object, they have an important impact on its carbon footprint.

[0093] Background data: Such as regional electricity production structure, energy efficiency standards, etc., are crucial for calculating indirect emissions such as electricity consumption.

[0094] When collecting activity data, it is usually necessary to use a variety of methods and technical means, such as field measurements, monitoring equipment, supplier data, government reports and databases, industry standards and guidelines, models and software tools, etc. The accuracy and reliability of these data are crucial to the accuracy and effectiveness of carbon footprint accounting.

[0095] In some embodiments, the S1 includes:

[0096] Utilize multiple data sources to collect activity data from all links of the supply chain;

[0097] Through graphical tools, the various links in the battery life cycle are connected in sequence to display the flow path of materials and energy and obtain a battery life cycle flow chart.

[0098] The working principle of the above technical solution is: first, detailed activity data is collected from all links of the battery supply chain (including raw material mining, transportation, processing, battery production, use, recycling and disposal). These data include but are not limited to energy consumption, raw material use, transportation distance, production process parameters, etc.

[0099] Based on the collected activity data, these links are connected in sequence through graphical tools to show the flow path of materials and energy, thereby constructing a battery life cycle flow chart. This flow chart is the basis for subsequent accounting and forecasting work.

[0100] Each link of the supply chain is further subdivided into different regions to facilitate more accurate carbon footprint calculation. Regional division can be based on factors such as geographical location, function or management responsibility. Activity data in each region are classified according to time periods (such as years, quarters, months, etc.), and appropriate carbon footprint accounting methods are selected for accounting according to the nature of the data (direct emissions, energy consumption, electricity consumption) in each time period. For example, for direct emissions in the production process, the LCA (life cycle assessment) method is used; for energy consumption, the emission factor method is used; for electricity consumption, the marginal carbon emission coefficient method is used. In the same time period, the carbon emissions of each region are compared to determine which regions are the key areas of carbon footprint, which will be the focus of subsequent prediction and optimization. The results obtained by different accounting methods are integrated to obtain the comprehensive carbon footprint accounting results of the entire life cycle of the battery; based on the comprehensive accounting results, a detailed carbon footprint report is generated, which may include carbon emissions of each link and region, key area analysis, emission reduction potential assessment, etc.

[0101] Using machine learning algorithms, a prediction model is built based on existing carbon footprint data and other relevant data (such as production scale, energy efficiency, process improvement, etc.). By training and optimizing the prediction model, it can accurately predict the carbon emissions of future activities, including the carbon emissions of future activities in various regions.

[0102] Set up high-precision carbon emission collection equipment in key areas to obtain real-time carbon emission data in these areas. Compare the actual carbon emission data with the predicted data of the prediction model to analyze the reasons for the difference. Correct and optimize the prediction model based on the actual data to improve the prediction accuracy and adaptability of the model.

[0103] Regularly collect the latest actual carbon emission data and supply chain activity data to update the input data set of the forecast model. Use the updated data set to retrain the forecast model to obtain the latest forecast results. Compare the actual carbon emission data with the forecast results of the latest forecast model and the forecast results of the optimal forecast model in the previous cycle to evaluate the forecast performance of the current model. Based on the comparison results, select the current optimal forecast model as the basis for future carbon emission forecasts. This process ensures the continuous optimization and adaptability of the forecast model.

[0104] The effect of the above technical solution is: by collecting activity data in detail from each link of the supply chain and building a battery life cycle flow chart based on these data, the carbon footprint accounting is more comprehensive, covering the entire life cycle from raw material mining to final disposal. For different links and time periods, appropriate accounting methods (such as LCA, emission factor method, marginal carbon emission coefficient method, etc.) are adopted to improve the precision and accuracy of the accounting. The use of machine learning technology to build a prediction model can predict the carbon emissions of future activities based on historical data and current trends, providing forward-looking guidance for the company's carbon management. The continuous correction and optimization of the prediction model enables it to adapt to changes in the external environment such as the market, technology and policies, and maintain a high level of prediction accuracy. By generating a detailed carbon footprint report, companies can clearly understand their carbon emissions in each link, identify high-emission areas and emission reduction potential points. Based on the accounting results and forecast data, companies can formulate targeted emission reduction strategies, optimize production processes and supply chain management, and promote the green transformation and sustainable development of enterprises.

[0105] In some embodiments, the S2 includes:

[0106] Determine the boundaries of each link;

[0107] Each link is further divided into different areas based on geographical location, function and / or management responsibility factors; for example, in the battery production link, it can be divided into raw material warehouse, production workshop, quality inspection area, packaging area and other areas; the production workshop is divided into multiple areas according to the process flow;

[0108] Divide time into different periods, such as years, quarters, or months;

[0109] categorizing the collected activity data, including direct emissions, energy consumption, and electricity consumption;

[0110] In each time period, the classified activity data is calculated according to the selected accounting method to obtain the carbon footprint accounting results of the classification in the time period;

[0111] Compare carbon emissions in various regions over the same time period and identify key areas.

[0112] The working principle of the above technical solution is: first, clarify the starting and ending points of each link in the supply chain to ensure the comprehensiveness and accuracy of data collection. This helps to define the scope of carbon footprint accounting for each link to avoid duplication or omission.

[0113] Each supply chain link is further subdivided into multiple regions based on factors such as geographic location, function and / or management responsibility. This division helps to identify carbon emission sources more accurately and adopt more appropriate accounting methods based on the characteristics of different regions. For example, in the battery production link, it can be divided into raw material warehouses, production workshops, quality inspection areas, packaging areas and other areas according to the production process. The production workshop can also be further divided into multiple sub-areas according to the process flow. Divide the timeline into different sections, such as years, quarters or months, so that carbon emissions in different time periods can be independently accounted for and compared. This time division helps to capture the cyclical changes in carbon emissions and provide a basis for predicting future carbon emission trends. The collected activity data is sorted according to categories such as direct emissions, energy consumption and electricity consumption. This classification helps to distinguish different types of carbon emission sources and provide clear input data for subsequent accounting work.

[0114] In each time period, the classified activity data are accounted for according to the selected accounting method. The selection of accounting methods should be based on the availability of data, the accuracy of accounting and the applicability of the method. For example, for direct emissions in the production process, the LCA (Life Cycle Assessment) method can be used; for energy consumption, the emission factor method can be used; for electricity consumption, the marginal carbon emission coefficient method can be used. The application of these methods can calculate the carbon footprint accounting results for each category in each time period.

[0115] Compare the carbon emissions of various regions in the same time period and identify key areas. Key areas are areas with relatively high carbon emissions or significant emission reduction potential. By identifying key areas, companies can prioritize and invest resources to achieve effective emission reduction targets. This step provides an important basis for companies to formulate emission reduction strategies and set priorities.

[0116] To summarize, the working principle of the S2 stage is to achieve detailed accounting of the carbon footprint of each link in the supply chain through steps such as clarifying link boundaries, refining regional divisions, time division, data classification and organization, application of accounting methods and identification of key areas, so as to provide basic data support for subsequent forecasts, reports and decisions.

[0117] The effect of the above technical solution is: by further dividing each link into different areas and taking into account factors such as geographical location, function and / or management responsibility, it is possible to more accurately identify carbon emission sources and reduce errors in the accounting process. The activity data is sorted according to categories such as direct emissions, energy consumption and electricity consumption, making the accounting process clearer and helping to improve the accuracy of the accounting results. Dividing time into different sections (such as years, quarters or months) makes carbon emission data in different time periods comparable; it helps companies understand the cyclical changes in carbon emissions and provides strong support for the formulation of emission reduction strategies. Comparison of carbon emissions in different regions within the same time period helps to identify key areas and provide a basis for companies to formulate targeted emission reduction measures. After determining the key areas, priority attention can be given to and resources can be invested to achieve effective emission reduction goals, which helps to optimize resource allocation and improve emission reduction efficiency. Through the accounting results, companies can identify which links or regions have higher carbon emissions, thereby adjusting production plans, improving process flows or upgrading equipment and facilities to reduce carbon emissions.

[0118] In some embodiments, the calculating of the classified activity data according to the selected calculation method in each time period includes:

[0119] For direct emissions during the production process, a life cycle assessment method is used to obtain the first carbon emissions;

[0120] For carbon emissions generated by energy consumption, the emission factor method is used to obtain the second carbon emissions;

[0121] For carbon emissions from electricity consumption, the marginal carbon emission coefficient method is used to obtain the third carbon emission amount;

[0122] Among them, the second carbon emission is:

[0123]

[0124] Among them, C2 is the total second carbon emission of each region within the preset time period, Rij is the usage of the jth energy in the ith region within the preset time period; Fij is the conversion or transmission efficiency of the jth energy in the ith region within the preset time period; Yij is the emission factor of the jth energy in the ith region; Zij is the quality factor of the jth energy in the ith region; Xij is the correction factor, m is the total number of regions in each link, and n is the type of energy consumption in the ith region within the preset time period;

[0125] The value of the correction factor Xij depends on the specific factors you want to consider. Here are some possible factors and their value settings:

[0126] Technological progress: If technological progress leads to improved energy efficiency or reduced carbon emissions, Xij can be set to a value less than 1 to reflect this effect. This value can be determined by comparing historical data with current data.

[0127] Policy impact: If the government implements energy conservation and emission reduction policies, Xij can be set to a value that reflects the effect of these policies. This may need to be estimated based on data comparisons before and after the policy implementation.

[0128] Seasonal variation: If there are seasonal variations in energy usage and carbon emissions (e.g., coal use for heating increases in winter), Xij can be set to a value that varies with the season. This value can be determined through historical data analysis.

[0129] Economic factors: If there is a relationship between the level of economic activity (such as GDP growth rate) and energy use and carbon emissions, Xij can be set to a value that reflects this relationship. This needs to be determined based on economic models or statistical analysis.

[0130] Other unknown factors: If there are other unknown but potentially important influencing factors, Xij can be set to a conservative estimate and adjusted based on new data in subsequent studies.

[0131] Among them, the third emission is:

[0132]

[0133] Among them, C2 is the total third carbon emission of each region in the preset time period, m is the total number of regions in each link, Eik is the amount of electricity consumed by the i-th region from the k-th power source in the preset time period; MCEFik is the marginal carbon emission coefficient when the i-th region obtains electricity from the k-th power source in the preset time period; v is the total number of power sources in the i-th region in the preset time period; is the total carbon emissions of the k-th power grid within the preset time period, is the load of the kth power node in the ith region within the preset time period.

[0134] The working principle of the above technical solution is: Direct emission accounting in the production process (using life cycle assessment method):

[0135] How it works: The Life Cycle Assessment (LCA) method is a systematic methodology used to evaluate the environmental impact of a product, process or service throughout its life cycle. For direct emissions during the production process, the LCA method comprehensively considers the environmental load of each stage by tracking direct greenhouse gas emissions from all stages, from raw material acquisition, production, use to waste disposal. This method is based on the boundary setting of "cradle to grave" or "cradle to gate", and collects data from each stage in detail, including raw material consumption, energy consumption, waste generation, etc., and then conducts quantitative analysis to derive the direct carbon emissions of the entire production process.

[0136] Carbon emissions from energy consumption (using emission factor method):

[0137] The emission factor method is a method for calculating carbon emissions based on energy consumption and corresponding emission factors. In the above formula, C2 represents the total second carbon emissions of each region within a preset time period, that is, the carbon emissions generated by energy consumption. This method calculates the total carbon emissions by considering the energy consumption (Rij) of different regions and types, combined with their respective conversion or transmission efficiency (Fij), emission factors (Yij), quality factors (Zij) and correction factors (Xij). This method can more accurately reflect the contribution of energy consumption to carbon emissions, while taking into account the influence of multiple factors.

[0138] The marginal carbon emission factor method (MCEF) is a method for calculating the carbon emissions generated by electricity consumption based on the carbon emission intensity of the electricity source. In the above formula, C3 represents the total third carbon emissions of each region in a preset time period, that is, the carbon emissions generated by electricity consumption. This method first determines the marginal carbon emission factor (MCEFik) of different electricity sources, which is usually related to the carbon emissions of the power grid and the load of the power node. Then, according to the amount of electricity consumed by each region from each power source (Eik), combined with the corresponding marginal carbon emission coefficient, the total carbon emissions of electricity consumption are calculated. This method can reflect the direct connection between electricity consumption and carbon emissions, while taking into account the diversity of electricity sources and the differences in their carbon emission intensity.

[0139] In summary, the working principles of these accounting methods are based on scientific and systematic assessment methods, aiming to comprehensively and accurately reflect the carbon emissions from different sources. By comprehensively applying these methods, it can provide strong data support for enterprises to formulate emission reduction strategies and evaluate emission reduction effects.

[0140] The effects of the above technical solutions are as follows: All three methods are based on a scientific and systematic evaluation system, which can comprehensively and accurately reflect the carbon emissions from different sources. The life cycle assessment method takes into account the entire life cycle of the product from cradle to gate (or grave), ensuring the comprehensiveness of direct emission accounting; the emission factor rule accurately calculates the carbon emissions generated by energy consumption by considering multiple factors (such as conversion efficiency, emission factor, quality factor and correction factor); the marginal carbon emission coefficient rule accurately calculates the carbon emissions generated by electricity consumption by considering the diversity of electricity sources and the difference in carbon emission intensity; the formula provides a method for accurately calculating carbon emissions; whether it is direct emissions in the production process, carbon emissions generated by energy consumption or carbon emissions generated by electricity consumption, the corresponding formula can be used to obtain quantitative results; this precise quantification helps enterprises accurately understand their own carbon emissions and provide a reliable basis for formulating emission reduction strategies. The formula takes into account a variety of influencing factors, such as energy usage, conversion efficiency, emission factor, quality factor, correction factor, and carbon emission intensity and load of electricity sources. This comprehensive consideration makes the accounting results more comprehensive and accurate, and can reflect the contribution of different links, different regions and different electricity sources to carbon emissions.

[0141] In some embodiments, comparing the carbon emissions of various regions within the same time period to determine the key regions includes:

[0142] Standardize the carbon emissions in the region to obtain standardized carbon emissions;

[0143] The normalized emissions are:

[0144]

[0145] Among them, BCzi is the standardized carbon emissions of the ith region within the preset time period, Czi is the carbon emissions of the ith region within the preset time period, Si is the area of ​​the ith region; Hi is the production scale score of the ith region within the preset time period; DEi is the emission reduction potential score of the ith region within the preset time period; w1, w2, and w3 are all the first weight coefficients;

[0146] The standardized carbon emissions of each region are compared with the preset threshold, and the regions where the standardized carbon emissions exceed the preset threshold are identified as key regions.

[0147] The working principle and effect of the above technical solution are as follows: by standardizing the carbon emissions of each region, the influence of factors such as area, production scale and emission reduction potential is eliminated, so that the carbon emissions of each region can be fairly compared on the same scale. Then, by comparing the standardized carbon emissions with the preset threshold, key areas with relatively high carbon emissions are identified, providing a scientific basis for the subsequent formulation of emission reduction policies and implementation of measures.

[0148] First, in order to fairly compare carbon emissions between different regions, the carbon emissions of each region need to be standardized.

[0149] The normalized carbon emissions (BCzi) are calculated using the following formula:

[0150]

[0151] Among them, BCzi is the standardized carbon emissions of the ith region within the preset time period, Czi is the carbon emissions of the ith region within the preset time period, Si is the area of ​​the ith region; Hi is the production scale score of the ith region within the preset time period; DEi is the emission reduction potential score of the ith region within the preset time period; w1, w2, and w3 are all the first weight coefficients.

[0152] The normalized carbon emissions (BCzi) are compared with the preset threshold. If BCzi exceeds the preset threshold, the area is identified as a critical area. Critical areas usually mean that these areas have high carbon emissions and need to be paid special attention to and take corresponding emission reduction measures.

[0153] In some embodiments, the S4 includes:

[0154] Collect carbon footprint accounting results and other relevant data, including but not limited to direct emissions, energy consumption, electricity consumption, production scale, energy efficiency, process improvements and environmental factors;

[0155] Establishing a machine learning model based on the carbon footprint calculation results and other relevant data, and training and optimizing the machine learning model;

[0156] Through cross-validation, the generalization ability of the model on unknown data is evaluated and the model parameters are further adjusted;

[0157] Through the trained machine learning model, the carbon emissions of future activities are predicted, including the carbon emissions of future activities in various regions.

[0158] The working principle and effect of the above technical solution are as follows: First, comprehensive data needs to be collected, including direct results of carbon footprint accounting (such as direct emissions in the production process, carbon emissions generated by energy consumption, carbon emissions generated by electricity consumption, etc.) and other data related to carbon emissions. These data include but are not limited to direct emission data, energy consumption, electricity consumption records, production scale indicators, energy efficiency assessments, process improvement records, and possible impacts of environmental factors (such as climate conditions, policy changes, etc.) on carbon emissions.

[0159] After collecting enough data, the next step is to build a machine learning model; this model will learn the complex relationship between data through algorithms based on the carbon footprint accounting results and other relevant data to identify the key factors and patterns that affect carbon emissions; select appropriate machine learning algorithms (such as regression models, neural networks, random forests, etc.) and build a model framework based on the characteristics of the data and the prediction target; use the collected data to train the model and optimize the model performance by adjusting the model parameters so that it can accurately reflect the relationship between carbon emissions and various influencing factors. In order to ensure the accuracy and reliability of the model, model evaluation is required. This is usually achieved through cross-validation, that is, dividing the data set into a training set and a test set (or more subsets), training the model on the training set, and evaluating the model performance on the test set. By dividing the training set and test set multiple times and repeating the training and testing process to evaluate the performance of the model on different data subsets, it helps to reduce overfitting and improve the generalization ability of the model on unknown data. Based on the results of cross-validation, further adjust the model parameters, such as learning rate, number of iterations, regularization terms, etc., to optimize model performance. After the model has been fully trained and validated, the model can be used to predict carbon emissions from future activities. This includes the carbon emissions that may be generated in different regions over different time periods, as well as the changing trends of carbon emissions under different policies, technologies or market conditions. Input relevant data of future activities (such as production plans, energy consumption forecasts, changes in electricity sources, etc.) into the trained machine learning model. The model calculates and outputs the predicted carbon emissions of future activities based on the input data. Based on the prediction results, enterprises, governments or research institutions can formulate corresponding emission reduction strategies, optimize production plans or adjust energy structures to achieve low-carbon development goals.

[0160] In some embodiments, the S5 includes:

[0161] Obtaining the difference between the predicted data and the actual data of each key area as the first difference;

[0162] If the first difference exceeds the preset difference range, the corresponding area is taken as the first abnormal area;

[0163] Obtain at least one factor affecting the first difference in each first abnormal area and an influence weight of the corresponding factor;

[0164] Obtaining scores of influencing factors according to factors and weights obtained in the plurality of first abnormal areas;

[0165]

[0166] Among them, Ysα is the score of the αth influencing factor, Ysαβ is the influence weight of the αth influencing factor on the first difference of the βth first abnormal area; z is the number of first abnormal areas;

[0167] The influencing factors with scores greater than the preset score threshold are used as inputs of the prediction model to optimize the model.

[0168] The working principle of the above technical solution is: first, compare the predicted carbon emission data of each key area with the actual carbon emission data, and calculate the difference between them (first difference). This difference reflects the performance deviation of the prediction model in this area.

[0169] Next, the first difference of each key area is checked to see if it exceeds the preset difference range. If the first difference of a region exceeds this range, then the region is considered a first anomaly region; these anomaly regions are indicators of poor performance of the prediction model or incorrect data input. For each first anomaly region, the factors affecting the first difference are further analyzed. These factors may include changes in production scale, differences in energy efficiency, effects of process improvements, changes in environmental factors, etc. At the same time, the weight of each factor's influence on the first difference is determined, that is, the relative importance of the factor in causing the prediction deviation. Then, the system calculates the score of each influencing factor based on the factors and their weights obtained in multiple first anomaly regions. This score is obtained by summing the influence weights of the factor in all first anomaly regions, reflecting the contribution of the factor to the overall prediction deviation; finally, the influencing factors with a score greater than the preset score threshold are used as inputs to the prediction model to optimize the model. These high-scoring factors are considered to be the main causes of the model's prediction deviation, so by incorporating them into the model training process, the accuracy and generalization ability of the model can be improved.

[0170] Specifically, the optimization process may include:

[0171] Model parameters are adjusted to better capture the relationship between these high-scoring factors and carbon emissions.

[0172] Introduce new features or variables to more fully reflect the impact of these factors on carbon emissions.

[0173] Retrain the model using a dataset containing these high-scoring factors to improve the model’s accuracy in predicting future carbon emissions.

[0174] The effect of the above technical solution is: by comparing the difference between the predicted data and the actual data and identifying the abnormal areas, the deficiencies in the prediction model can be discovered in time. In-depth analysis of the abnormal areas to find the key factors affecting the accuracy of the prediction will help optimize the model and thus improve the accuracy of future predictions. Analyzing the various factors and their weights that affect the first difference enables the model to more comprehensively consider various potential influences and reduce the prediction deviation caused by changes in a single factor. Incorporating high-scoring influencing factors into the model optimization process can enhance the model's adaptability to different scenarios and changes and improve the robustness of the model. Through the analysis of influencing factors and the calculation of scores, decision makers can be provided with clearer and more quantitative information to help them better understand which factors are most critical to carbon emissions forecasts; it helps decision makers to formulate more scientific and reasonable emission reduction strategies and plans to achieve effective control and management of carbon emissions.

[0175] In some embodiments, the S6 includes:

[0176] Obtaining the difference between the predicted data and the actual data of each key area through the latest prediction model as a second difference; obtaining a first comprehensive evaluation coefficient according to a plurality of second differences;

[0177] Obtaining the difference between the predicted data and the actual data of each key area through the previous optimal prediction model as a third difference; obtaining a second comprehensive evaluation coefficient according to a plurality of third differences;

[0178] Among them, the comprehensive evaluation coefficient is:

[0179] PZ1=x1*DZ1+x2*DA1+x3*Dmax1

[0180] PZ2=x1*DZ2+x2*DA2+x3*Dmax2

[0181] Among them, PZ1 is the first comprehensive evaluation coefficient, DZ1 is the sum of the second differences of each key area, DA1 is the average value of the second differences of each key area; Dmax1 is the maximum value of the second differences of each key area; PZ2 is the second comprehensive evaluation coefficient, DZ2 is the sum of the third differences of each key area, DA2 is the average value of the third differences of each key area; Dmax2 is the maximum value of the third differences of each key area; x1, x2, x3 are all second weight coefficients;

[0182] The current optimal prediction model is determined according to the comprehensive evaluation coefficient, and the model corresponding to the smaller comprehensive evaluation coefficient is selected as the current optimal model.

[0183] The working principle and effect of the above technical solution are as follows: First, the carbon emissions of each key area are predicted using the latest prediction model, and the predicted data are compared with the actual data to calculate the difference as the second difference. These second differences reflect the prediction accuracy of the latest prediction model in each area.

[0184] Similarly, the carbon emissions of each key area are predicted using the optimal prediction model determined last time, and the predicted data are compared with the actual data to calculate the difference as the third difference. These third differences reflect the prediction accuracy of the previous optimal prediction model in each area.

[0185] Calculation of the first comprehensive evaluation coefficient (PZ1): Based on the second difference of the latest prediction model, the sum (DZ1), average value (DA1) and maximum value (Dmax1) of the second difference of each key area are calculated. Then, these statistics are weighted and summed using the preset second weight coefficient (x1, x2, x3) to obtain the first comprehensive evaluation coefficient (PZ1). This coefficient comprehensively reflects the performance of the latest prediction model in terms of overall prediction accuracy, average prediction accuracy and maximum prediction deviation.

[0186] Similarly, according to the third difference of the previous optimal prediction model, the sum (DZ2), average value (DA2) and maximum value (Dmax2) of the third difference of each key area are calculated. Then, these statistics are weighted and summed using the same second weight coefficient (x1, x2, x3) to obtain the second comprehensive evaluation coefficient (PZ2). This coefficient reflects the performance of the previous optimal prediction model in the corresponding aspect.

[0187] Finally, the first comprehensive evaluation coefficient (PZ1) is compared with the second comprehensive evaluation coefficient (PZ2). Since the smaller the comprehensive evaluation coefficient, the better the performance of the prediction model (that is, the smaller the prediction error), the model corresponding to the smaller comprehensive evaluation coefficient is selected as the current optimal prediction model.

[0188] Through this process, the system can continuously evaluate and optimize the prediction model. Whenever a new prediction model is developed or an existing model is improved, the step S6 can be repeated to re-determine the current optimal prediction model. This continuous optimization and iteration process helps to maintain the advancement and accuracy of the prediction model.

[0189] By comparing the prediction performance of different prediction models in multiple key areas (by calculating the difference and constructing a comprehensive evaluation coefficient), the current optimal prediction model is determined, and subsequent carbon emission prediction and decision support are carried out on this basis. Since the current optimal prediction model is selected for carbon emission prediction, the prediction error will be significantly reduced. This makes the prediction results closer to the actual carbon emissions and improves the accuracy of the prediction. By continuously iterating and optimizing the prediction model, the overall performance of the model will be improved. This includes improvements in prediction accuracy, robustness, adaptability, etc. Based on accurate prediction data and the current optimal model, decision makers can make decisions faster and reduce decision delays caused by inaccurate data or poor model performance.

[0190] Some embodiments of the present application provide a carbon footprint accounting system based on artificial intelligence technology, the system comprising:

[0191] A data collection module is used to collect activity data from various links of the supply chain and obtain a battery life cycle flow chart based on the activity data; the links include raw material mining, transportation, processing, battery production, use, recycling and disposal;

[0192] The accounting module is used to divide each link into different areas, divide the activity data in the area into time periods and classify them, and select the carbon footprint accounting method according to the classification in different time periods to obtain the carbon footprint accounting results under the classification in the time period; the classification includes direct emissions, energy consumption and electricity consumption; the key areas are determined according to the regional carbon emissions in the same time period; for example, for direct emissions in the production process, LCA (life cycle assessment) is used; for energy consumption, the emission factor method is used; for electricity consumption, the marginal carbon emission coefficient method is used;

[0193] The integration module is used to integrate the results obtained from different accounting methods to obtain comprehensive accounting results, and generate a carbon footprint report based on the comprehensive accounting results;

[0194] The prediction module is used to build a prediction model based on existing carbon footprint data using machine learning to predict carbon emissions from future activities;

[0195] An optimization module is used to set up collection equipment in key areas to obtain actual carbon emission data in key areas, and to calibrate and optimize the prediction model based on the actual carbon emission data and prediction data in key areas;

[0196] The update module is used to regularly update data and prediction models, compare the actual carbon emission data with the prediction results of the latest prediction model and the prediction results of the optimal prediction model in the previous cycle, and determine the current optimal model based on the comparison results.

[0197] In some embodiments, the data collection module includes:

[0198] A first data collection unit is used to obtain and collect activity data from various links of the supply chain using multiple data sources;

[0199] The flow chart acquisition unit is used to connect the various links of the battery life cycle in sequence through a graphical tool, display the flow path of materials and energy, and obtain the battery life cycle flow chart.

[0200] In some embodiments, the accounting module includes:

[0201] A boundary determination unit, used to determine the boundary of each link;

[0202] The regional division unit is used to further divide each link into different areas based on geographical location, function and / or management responsibility factors; for example, in the battery production link, it can be divided into raw material warehouse, production workshop, quality inspection area, packaging area and other areas; the production workshop can be divided into multiple areas according to the process flow;

[0203] Time division unit, used to divide time into different sections, such as year, quarter or month;

[0204] A classification unit for classifying the collected activity data, wherein the classification includes direct emissions, energy consumption, and electricity consumption;

[0205] The accounting unit is used to calculate the classified activity data according to the selected accounting method in each time period to obtain the carbon footprint accounting result under the classification in the time period;

[0206] The key area determination unit is used to compare the carbon emissions of various areas within the same time period and determine the key areas.

[0207] In some embodiments, the accounting unit includes:

[0208] A first accounting subunit is used to obtain a first carbon emission amount using a life cycle assessment method for direct emissions in a production process;

[0209] The second accounting subunit is used to obtain a second carbon emission amount by adopting an emission factor method for carbon emission generated by energy consumption;

[0210] The third accounting subunit is used to obtain a third carbon emission amount by using a marginal carbon emission coefficient method for carbon emissions generated by electricity consumption;

[0211] Among them, the second carbon emission is:

[0212]

[0213] Among them, C2 is the total second carbon emission of each region within the preset time period, Rij is the usage of the jth energy in the ith region within the preset time period; Fij is the conversion or transmission efficiency of the jth energy in the ith region within the preset time period; Yij is the emission factor of the jth energy in the ith region; Zij is the quality factor of the jth energy in the ith region; Xij is the correction factor, m is the total number of regions in each link, and n is the type of energy consumption in the ith region within the preset time period;

[0214] Among them, the third emission is:

[0215]

[0216] Among them, C2 is the total third carbon emission of each region in the preset time period, m is the total number of regions in each link, Eik is the amount of electricity consumed by the i-th region from the k-th power source in the preset time period; MCEFik is the marginal carbon emission coefficient when the i-th region obtains electricity from the k-th power source in the preset time period; v is the total number of power sources in the i-th region in the preset time period; is the total carbon emissions of the k-th power grid within the preset time period, is the load of the kth power node in the ith region within the preset time period.

[0217] In some embodiments, the key area determination unit includes:

[0218] A standardization processing subunit is used to perform standardization processing on the carbon emissions in the region to obtain the standardized carbon emissions;

[0219] The normalized emissions are:

[0220]

[0221] Among them, BCzi is the standardized carbon emissions of the ith region within the preset time period, Czi is the carbon emissions of the ith region within the preset time period, Si is the area of ​​the ith region; Hi is the production scale score of the ith region within the preset time period; DEi is the emission reduction potential score of the ith region within the preset time period; w1, w2, and w3 are all the first weight coefficients;

[0222] The comparison and determination subunit is used to compare the standardized carbon emissions of each area with a preset threshold value, and determine the area where the standardized carbon emissions exceed the preset threshold value as a key area.

[0223] In some embodiments, the prediction module includes:

[0224] A second collection unit collects carbon footprint calculation results and other relevant data, including but not limited to direct emissions, energy consumption, electricity consumption, production scale, energy efficiency, process improvement and environmental factors;

[0225] A model building unit, used to build a machine learning model according to the carbon footprint calculation results and other relevant data, and to train and optimize the machine learning model;

[0226] The validation unit is used to evaluate the generalization ability of the model on unknown data through cross-validation and further adjust the model parameters;

[0227] The prediction unit is used to predict the carbon emissions of future activities, including the carbon emissions of future activities in each region, through the trained machine learning model.

[0228] In some embodiments, the optimization module includes:

[0229] A first difference value obtaining unit, used to obtain the difference between the predicted data and the actual data of each key area as a first difference value;

[0230] A first abnormal unit, used for determining that if the first difference exceeds a preset difference range, the corresponding area is a first abnormal area;

[0231] An influence determination unit, configured to obtain at least one factor affecting the first difference in each first abnormal area and an influence weight of the corresponding factor;

[0232] A factor scoring unit, used to obtain scores of influencing factors according to factors and weights obtained in multiple first abnormal areas;

[0233]

[0234] Among them, Ysα is the score of the αth influencing factor, Ysαβ is the influence weight of the αth influencing factor on the first difference of the βth first abnormal area; z is the number of first abnormal areas;

[0235] The optimization unit is used to use the influencing factors whose scores are greater than a preset score threshold as inputs of the prediction model to optimize the model.

[0236] In some embodiments, the update module includes:

[0237] The first evaluation unit is used to obtain the difference between the predicted data and the actual data of each key area through the latest prediction model as the second difference; and obtain the first comprehensive evaluation coefficient according to the plurality of second differences;

[0238] The second evaluation unit is used to obtain the difference between the predicted data and the actual data of each key area through the previous optimal prediction model as a third difference; and obtain a second comprehensive evaluation coefficient according to a plurality of third differences;

[0239] Among them, the comprehensive evaluation coefficient is:

[0240] PZ1=x1*DZ1+x2*DA1+x3*Dmax1

[0241] PZ2=x1*DZ2+x2*DA2+x3*Dmax2

[0242] Among them, PZ1 is the first comprehensive evaluation coefficient, DZ1 is the sum of the second differences of each key area, DA1 is the average value of the second differences of each key area; Dmax1 is the maximum value of the second differences of each key area; PZ2 is the second comprehensive evaluation coefficient, DZ2 is the sum of the third differences of each key area, DA2 is the average value of the third differences of each key area; Dmax2 is the maximum value of the third differences of each key area; x1, x2, x3 are all second weight coefficients;

[0243] The model selection unit is used to determine the current optimal prediction model according to the comprehensive evaluation coefficient, and select the model corresponding to the one with the smaller comprehensive evaluation coefficient as the current optimal model.

[0244] The working principle and effect of the above technical solution are the same as those in the embodiment of the method of the present application and will not be described in detail here.

[0245] The present application also provides an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any method described in the present application are implemented.

[0246] This application is explained from the perspectives of purpose of use, effectiveness, progress and novelty, and has met the functional enhancement and usage requirements emphasized by the Patent Law. The above description and drawings of this application are only the preferred embodiments of this application, and are not intended to limit this application. Therefore, all structures, devices, features, etc. that are similar or identical to this application, that is, all equivalent replacements or modifications made in accordance with the scope of the patent application of this application, should fall within the scope of protection of the patent application of this application.

Claims

1. A carbon footprint calculation method based on artificial intelligence technology, characterized in that: The method comprises: S1. Collect activity data from all links of the supply chain and obtain a battery life cycle flow chart based on the activity data; the links include raw material mining, transportation, processing, battery production, use, recycling and disposal; S2. Divide each link into different regions, divide the activity data in the region into time periods and classify them, select the carbon footprint accounting method according to the classification in different time periods to obtain the carbon footprint accounting results under the classification in the time period; the classification includes direct emissions, energy consumption and electricity consumption; determine the key areas according to the regional carbon emissions in the same time period; S3. Integrate the results obtained from different accounting methods to obtain comprehensive accounting results, and generate a carbon footprint report based on the comprehensive accounting results; S4. Use machine learning to build a prediction model based on existing carbon footprint data to predict carbon emissions from future activities; S5. Setting up collection equipment in key areas to obtain actual carbon emission data in the key areas, and calibrating and optimizing the prediction model based on the actual carbon emission data and prediction data in the key areas; S6. Update data and prediction models regularly, compare actual carbon emission data with the prediction results of the latest prediction model and the prediction results of the best prediction model in the previous cycle, and determine the current best model based on the comparison results; The S2 includes: Determine the boundaries of each link; Each link is further divided into different areas based on geographical location, function and / or management responsibility; Divide time into different sections; categorizing the collected activity data, including direct emissions, energy consumption, and electricity consumption; In each time period, the classified activity data is calculated according to the selected accounting method to obtain the carbon footprint accounting results of the classification in the time period; Compare carbon emissions in various regions over the same time period and identify key areas; In each time period, the classified activity data are calculated according to the selected accounting method, including: For direct emissions during the production process, a life cycle assessment method is used to obtain the first carbon emissions; For carbon emissions generated by energy consumption, the emission factor method is used to obtain the second carbon emissions; For carbon emissions from electricity consumption, the marginal carbon emission coefficient method is used to obtain the third carbon emission amount; Among them, the second carbon emission is: ; Among them, C2 is the total second carbon emission of each region within the preset time period, Rij is the usage of the jth energy in the ith region within the preset time period; Fij is the conversion or transmission efficiency of the jth energy in the ith region within the preset time period; Yij is the emission factor of the jth energy in the ith region; Zij is the quality factor of the jth energy in the ith region; Xij is the correction factor, m is the total number of regions in each link, and n is the type of energy consumption in the ith region within the preset time period; Among them, the third emission is: ; ; Among them, C3 is the total third carbon emission of each region in the preset time period, m is the total number of regions in each link, Eik is the amount of electricity consumed by the i-th region from the k-th power source in the preset time period; MCEFik is the marginal carbon emission coefficient when the i-th region obtains electricity from the k-th power source in the preset time period; v is the total number of power sources in the i-th region in the preset time period; is the total carbon emissions of the k-th power grid within the preset time period, is the load of the kth power node in the ith region within the preset time period.

2. The method according to claim 1, characterized in that: The S1 includes: Utilize multiple data sources to collect activity data from all links of the supply chain; Through graphical tools, the various links in the battery life cycle are connected in sequence to display the flow path of materials and energy and obtain a battery life cycle flow chart.

3. The method according to claim 1, characterized in that The above-mentioned carbon emissions of various regions are compared within the same time period to identify key regions; include: Standardize the carbon emissions in the region to obtain standardized carbon emissions; The normalized emissions are: ; Among them, BCzi is the standardized carbon emissions of the ith region within the preset time period, Czi is the carbon emissions of the ith region within the preset time period, Si is the area of ​​the ith region; Hi is the production scale score of the ith region within the preset time period; DEi is the emission reduction potential score of the ith region within the preset time period; w1, w2, and w3 are all the first weight coefficients; The standardized carbon emissions of each region are compared with the preset threshold, and the regions where the standardized carbon emissions exceed the preset threshold are identified as key regions.

4. The method according to claim 1, characterized in that: The S4 includes: Collect carbon footprint accounting results and other relevant data, including but not limited to direct emissions, energy consumption, electricity consumption, production scale, energy efficiency, process improvements and environmental factors; Establishing a machine learning model based on the carbon footprint calculation results and other relevant data, and training and optimizing the machine learning model; Through cross-validation, the generalization ability of the model on unknown data is evaluated and the model parameters are further adjusted; Through the trained machine learning model, the carbon emissions of future activities are predicted, including the carbon emissions of future activities in various regions.

5. The method according to claim 1, characterized in that The S5 includes: Obtaining the difference between the predicted data and the actual data of each key area as the first difference; If the first difference exceeds the preset difference range, the corresponding area is taken as the first abnormal area; Obtain at least one factor affecting the first difference in each first abnormal area and an influence weight of the corresponding factor; Obtaining scores of influencing factors according to factors and weights obtained in the plurality of first abnormal areas; ; in, For the The score of the influencing factor, No. The factors affecting The influence weight of the first difference of the first abnormal area; Z is the number of the first abnormal area; The influencing factors with scores greater than the preset score threshold are used as inputs of the prediction model to optimize the model.

6. The method according to claim 1, characterized in that The S6 includes: Obtaining the difference between the predicted data and the actual data of each key area through the latest prediction model as a second difference; obtaining a first comprehensive evaluation coefficient according to a plurality of second differences; Obtaining the difference between the predicted data and the actual data of each key area through the previous optimal prediction model as a third difference; obtaining a second comprehensive evaluation coefficient according to a plurality of third differences; The current optimal prediction model is determined according to the comprehensive evaluation coefficient, and the model corresponding to the smaller comprehensive evaluation coefficient is selected as the current optimal model.

7. The carbon footprint accounting system based on artificial intelligence technology is characterized by: The system comprises: A data collection module is used to collect activity data from various links of the supply chain and obtain a battery life cycle flow chart based on the activity data; the links include raw material mining, transportation, processing, battery production, use, recycling and disposal; The accounting module is used to divide each link into different areas, divide the activity data in the area into time periods and classify them, and select the carbon footprint accounting method according to the classification in different time periods to obtain the carbon footprint accounting results under the classification in the time period; the classification includes direct emissions, energy consumption and electricity consumption; and determine the key areas according to the regional carbon emissions in the same time period; The integration module is used to integrate the results obtained from different accounting methods to obtain comprehensive accounting results, and generate a carbon footprint report based on the comprehensive accounting results; The prediction module is used to build a prediction model based on existing carbon footprint data using machine learning to predict carbon emissions from future activities; An optimization module is used to set up collection equipment in key areas to obtain actual carbon emission data in key areas, and to calibrate and optimize the prediction model based on the actual carbon emission data and prediction data in key areas; The update module is used to regularly update data and prediction models, compare the actual carbon emission data with the prediction results of the latest prediction model and the prediction results of the optimal prediction model in the previous cycle, and determine the current optimal model based on the comparison results; The accounting module includes: Determine the boundaries of each link; Each link is further divided into different areas based on geographical location, function and / or management responsibility; Divide time into different sections; categorizing the collected activity data, including direct emissions, energy consumption, and electricity consumption; In each time period, the classified activity data is calculated according to the selected accounting method to obtain the carbon footprint accounting results of the classification in the time period; Compare carbon emissions in various regions over the same time period and identify key areas; In each time period, the classified activity data are calculated according to the selected accounting method, including: For direct emissions during the production process, a life cycle assessment method is used to obtain the first carbon emissions; For carbon emissions generated by energy consumption, the emission factor method is used to obtain the second carbon emissions; For carbon emissions from electricity consumption, the marginal carbon emission coefficient method is used to obtain the third carbon emission amount; Among them, the second carbon emission is: ; Among them, C2 is the total second carbon emission of each region within the preset time period, Rij is the usage of the jth energy in the ith region within the preset time period; Fij is the conversion or transmission efficiency of the jth energy in the ith region within the preset time period; Yij is the emission factor of the jth energy in the ith region; Zij is the quality factor of the jth energy in the ith region; Xij is the correction factor, m is the total number of regions in each link, and n is the type of energy consumption in the ith region within the preset time period; Among them, the third emission is: ; ; Among them, C3 is the total third carbon emission of each region in the preset time period, m is the total number of regions in each link, Eik is the amount of electricity consumed by the i-th region from the k-th power source in the preset time period; MCEFik is the marginal carbon emission coefficient when the i-th region obtains electricity from the k-th power source in the preset time period; v is the total number of power sources in the i-th region in the preset time period; is the total carbon emissions of the k-th power grid within the preset time period, is the load of the kth power node in the ith region within the preset time period.

8. An electronic device, characterized in that: The electronic device comprises a memory and a processor, the memory stores a computer program, and the processor implements the steps of the method according to any one of claims 1 to 6 when executing the computer program.

Images