Carbon data quality diagnosis and processing method based on double-source monitoring and artificial intelligence

Abstract

The invention discloses a carbon data quality diagnosis and processing method based on double-source monitoring and artificial intelligence, and the method comprises a training stage and an application stage, and the application stage comprises the steps: obtaining to-be-diagnosed original data of a material monitoring method, original data of an online monitoring method, and production operation data; the steps of validity judgment, cleaning and accounting are executed; the current online monitoring method carbon emission is utilized, and the predicted material monitoring method carbon emission is obtained through calculation of the artificial intelligence prediction model; and calculating an accumulative error between the predicted carbon emission of the material monitoring method and the carbon emission of the current material monitoring method in a set period, comparing the accumulative error with a preset threshold value, and if the accumulative error exceeds the threshold value, determining that the carbon emission of the current material monitoring method is risk data. The whole process does not need manual intervention step by step, full-chain automatic operation from data inflow to result output is achieved, and the efficiency and consistency of data quality management and control are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of carbon data quality diagnosis and processing technology, and in particular to a carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence. Background Technology

[0002] In the current carbon emission trading market, the accounting and reporting of greenhouse gas emissions in the cement and power generation industries mainly follow the requirements of the "Guidelines for Enterprise Greenhouse Gas Emission Accounting and Reporting (Cement Industry)" and the "Guidelines for Enterprise Greenhouse Gas Emission Accounting and Reporting (Power Generation Facilities)" (hereinafter collectively referred to as the "Industry Guidelines"). The Industry Guidelines, specifically for calculating carbon dioxide emissions from cement clinker production, primarily employ the material balance method. This method relies on static data such as monthly fossil fuel consumption, laboratory analysis data, and industry default values ​​for calculation. In actual production practice, some cement companies have exceeded the "minimum monthly measurement timescale" requirement in the Guidelines, using automatic monitoring equipment such as belt scales to achieve continuous, high-frequency data acquisition of key parameters such as fuel consumption and clinker production at the second level, and calculating emissions based on this real-time monitoring data. In this patented technical solution, this method of "real-time acquisition of key parameters and emission calculation through automatic monitoring equipment for raw materials and fuels at the production end" is defined as the material monitoring method; while the power generation industry guidelines have explicitly recognized this material monitoring method as one of the benchmark methods for emission accounting.

[0003] Meanwhile, the European and American carbon markets generally adopt another type of emission monitoring technology approach, which is the online monitoring method. Its core is to use a continuous emission monitoring system (CEMS) to continuously and in real time monitor end emission parameters such as flow rate and carbon dioxide concentration in the flue gas of cement and power generation enterprises, and calculate the emission amount.

[0004] Currently, while material monitoring and online monitoring methods are being used in parallel in various carbon market practices, the final verification of corporate carbon emission data still relies on a "single method result acceptance" model. However, due to fundamental differences in monitoring objects, data sources, and calculation logic, the two methods each carry the risk of data bias: for example, material monitoring methods are susceptible to fluctuations in raw material composition, while online monitoring methods are limited by equipment calibration accuracy. These issues not only affect the accuracy of carbon emission data but may also lead to unfairness in accounting results among companies, urgently requiring innovative technological solutions to address. Summary of the Invention

[0005] The purpose of this invention is to provide a carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence. The entire process does not require manual intervention at each stage, realizing the fully automated operation of the entire chain from data inflow to result output, which greatly improves the efficiency and consistency of data quality control and lays a solid foundation for the reliable application of carbon emission data.

[0006] To achieve the above objectives, the present invention provides the following solution: A method for carbon data quality diagnosis and processing based on dual-source monitoring and artificial intelligence includes the following steps: Training phase: S1. Obtain raw data from material monitoring, online monitoring, and production operation data within a historical time period; S2. The validity of the raw data from the material monitoring method and the online monitoring method is judged and cleaned, and the carbon emissions from the material monitoring method and the online monitoring method are calculated respectively. S3. Based on production operation data, an artificial intelligence classification algorithm is used to automatically classify production conditions and obtain multiple condition categories. S4. For each operating condition category, using the carbon emissions from online monitoring as input and the carbon emissions from material monitoring as output, construct and train an artificial intelligence prediction model for that operating condition category until the model performance indicators meet the preset threshold. Application phase: S5. Obtain the raw data of the material monitoring method, the raw data of the online monitoring method, and the production operation data to be diagnosed, and execute the validity judgment, cleaning and accounting steps in S2 to obtain the current carbon emissions of the material monitoring method and the current carbon emissions of the online monitoring method. S6. Based on production operation data, determine the current operating condition category and call the corresponding trained artificial intelligence prediction model. S7. Using the current online monitoring method for carbon emissions, the predicted material monitoring method for carbon emissions is calculated using the artificial intelligence prediction model in S4. S8. Calculate the cumulative error between the predicted carbon emissions from the material monitoring method and the current carbon emissions from the material monitoring method within a set period, and compare the cumulative error with a preset threshold. If the error exceeds the threshold, the current carbon emissions from the material monitoring method are determined to be risky data.

[0007] Preferably, in S2, the validity of the raw data from the material monitoring method and the raw data from the online monitoring method is assessed and cleaned, and the carbon emissions from the material monitoring method and the online monitoring method are calculated separately, specifically including: If either the raw data from the material monitoring method or the raw data from the online monitoring method is invalid for a certain period of time, the data for that period is invalid; invalid data is identified through artificial intelligence algorithms, and invalid data and valid data are marked separately; the carbon emissions from the material monitoring method and the carbon emissions from the online monitoring method for the valid data are calculated.

[0008] Preferably, in S4, an artificial intelligence prediction model for this work condition category is constructed and trained, as shown in the following formula:

[0009] in, Training phase category Carbon emissions from material monitoring methods The training phase category k represents the carbon emissions from the online monitoring method.

[0010] Preferably, in step S4, the online monitoring emissions are input into the trained model to predict the material monitoring emissions, using the following formula:

[0011] in, For the training phase working condition category The predicted carbon emissions from the i-th material monitoring method. For the training phase working condition category The i-th online monitoring method for carbon emissions.

[0012] Preferably, in S4, the performance indicators of the artificial intelligence prediction model meeting the preset threshold specifically include: Calculate training phase categories The root mean square percentage error between the predicted carbon emissions from the material monitoring method and the calculated carbon emissions from the material monitoring method The formula is as follows:

[0013] in, For the training phase working condition category Root mean square percentage error For the training phase working condition category The number of valid data pairs; if the root mean square percentage error is greater than a preset threshold. Then collect more valid data until the root mean square percentage error is less than or equal to a preset threshold. .

[0014] Preferably, in S8, the cumulative error between the carbon emissions calculated by the material monitoring method and the current carbon emissions calculated by the material monitoring method within a set period is calculated. The formula is as follows:

[0015] in, Application stage category Cumulative error Application stage category The predicted carbon emissions from the i-th material monitoring method. Application stage category The carbon emissions of the i-th material monitoring method.

[0016] Preferably, the application phase also includes hierarchical data processing steps: Invalid data handling: When the data corresponding to the current carbon emissions from the material monitoring method is deemed invalid, while the carbon emissions from the current online monitoring method are valid, the predicted carbon emissions from the material monitoring method will be multiplied by a penalty factor to obtain the final carbon emissions. When both the current carbon emission data from the material monitoring method and the current carbon emission data from the online monitoring method are deemed invalid, the maximum carbon emission data from the material monitoring method within a preset time period will be used as the final carbon emission data. When the current material monitoring method for carbon emissions is valid, but the current online monitoring method for carbon emissions is deemed invalid, the current material monitoring method for carbon emissions is multiplied by a penalty factor to obtain the final carbon emissions.

[0017] Preferably, effective data processing is also included: When the current carbon emission data obtained through material monitoring is valid but is deemed risky, the predicted carbon emission data obtained through material monitoring will be multiplied by a penalty factor to obtain the final carbon emission data. When the current material monitoring method carbon emission data is valid and determined to be non-risk data, the current material monitoring method carbon emission data shall be directly used as the final carbon emission data.

[0018] The present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence as described above.

[0019] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects: (1) This invention addresses the inherent problems of material monitoring methods, such as default value deviations, human error loopholes, and high regulatory costs, as well as online monitoring methods, such as insufficient coverage of dispersed sources, system errors, and misjudgments of carbon emissions from alternative fuels / raw materials. It innovatively establishes a two-way fusion logic. By dynamically predicting material monitoring data using an artificial intelligence model and combining it with the classification and accounting logic of the material monitoring method, it indirectly identifies the carbon emissions of alternative fuels / raw materials from online monitoring methods, thus filling the technical blind spots of online monitoring. Furthermore, it uses predicted material monitoring data to fill in invalid data and cross-verifies high-frequency real-time data, reducing the space for falsification in manual sampling and sample preparation, thereby reducing data uncertainty from the source.

[0020] (2) The method of this invention addresses the problem of significant differences in carbon emission characteristics under different categories in cement / power generation production. It innovatively uses an algorithm to classify the data and builds an artificial intelligence model for each category. This breaks through the adaptability limitations of a one-size-fits-all model, enabling the model to accurately capture the intrinsic relationship between material consumption and flue gas emissions under different categories, and significantly improves the accuracy of data adjustment, supplementation, and diagnosis.

[0021] (3) The method of this invention breaks through the segmented and manually-led mode of traditional data management. After the parameters of the material monitoring method and the online monitoring method are input into the algorithm, the data acquisition link can use intelligent sensing and automatic transmission technology to capture multi-source monitoring data in real time; the carbon emission accounting process is automatically calculated by the algorithm according to preset rules and models, and the accounting results are quickly output; the anomaly diagnosis module can automatically identify abnormal data caused by falsification in the material method through cross-verification of data from the two methods; the intelligent completion link fills in the gaps in material data based on the prediction results of artificial intelligence models when there are gaps due to equipment failure, etc., rather than traditional interpolation or experience estimation. The data completion relies on the physical correlation of materials and online monitoring, and the credibility is much higher than that of static accounting. The whole process does not require manual intervention at each step, realizing the full-chain automated operation from data inflow to result output, greatly improving the efficiency and consistency of data quality control, and laying a solid foundation for the reliable application of carbon emission data. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence provided by the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] like Figure 1 As shown, the present invention provides a carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence, comprising the following steps: Training phase: S1. Obtain raw data from material monitoring, online monitoring, and production operation data within a historical time period; S2. The validity of the raw data from the material monitoring method and the online monitoring method is judged and cleaned, and the carbon emissions from the material monitoring method and the online monitoring method are calculated respectively. S3. Based on production operation data, an artificial intelligence classification algorithm is used to automatically classify production conditions. The classification algorithm used is one of the following: KNN, random forest, support vector machine, etc., to obtain multiple condition categories. S4. For each operating condition category, using the carbon emissions from online monitoring as input and the carbon emissions from material monitoring as output, construct and train an artificial intelligence prediction model for that operating condition category until the model performance indicators meet the preset threshold. Application phase: S5. Obtain the raw data of the material monitoring method, the raw data of the online monitoring method, and the production operation data to be diagnosed, and execute the validity judgment, cleaning and accounting steps in S2 to obtain the current carbon emissions of the material monitoring method and the current carbon emissions of the online monitoring method. S6. Based on production operation data, determine the current operating condition category and call the corresponding trained artificial intelligence prediction model. S7. Using the current online monitoring method for carbon emissions, the predicted material monitoring method for carbon emissions is calculated using the artificial intelligence prediction model in S4. S8. Calculate the cumulative error between the predicted carbon emissions from the material monitoring method and the current carbon emissions from the material monitoring method within a set period, and compare the cumulative error with a preset threshold. If the error exceeds the threshold, the current carbon emissions from the material monitoring method are determined to be risky data.

[0027] The method of this invention specifically includes: 1. Artificial intelligence model training 1.1 Data Acquisition Based on existing accounting guidelines and online monitoring standards, raw monitoring data and production operation data are collected over a certain period of time, requiring raw monitoring data to be collected at least once per minute.

[0028] 1.2 Data Cleaning and Accounting Invalid data refers to missing emission data or data from periods of calibration or maintenance. Data for a given time period is invalid if either material monitoring (MMR) or online monitoring (OLM) data is invalid. Artificial intelligence can be used to identify invalid data and mark it separately from valid data. Calculate carbon emissions for valid minutes using MMR and OLM methods. Consolidate carbon emissions into lower-frequency emission data, such as MMR and OLM emission data every 30 minutes.

[0029] 1.3 Data Classification By using artificial intelligence and production operation data, automatic operating conditions are classified, and the corresponding low-frequency emission data will be assigned to each category to form training data.

[0030] 1.4 Artificial Intelligence Model Construction Using data pairs from different categories, train an artificial intelligence model that connects the data from material monitoring methods and online monitoring methods, including machine learning and deep learning models. For example, use one of the following models: linear regression, random forest, or neural network.

[0031]

[0032] In the formula: —Training Phase Categories Carbon emissions from material monitoring methods; —Training Phase Categories Carbon emissions are monitored online. For example, using a linear model for modeling:

[0033] —Training Phase Categories Carbon emissions from material monitoring methods; —Training Phase Categories Carbon emissions are monitored online. — The coefficients of the first-order linear model of the k-th class.

[0034]

[0035] In the formula: — The coefficients of the first-order linear model of the k-th class. —Training Phase Categories Carbon emissions from the i-th material monitoring method; —Training Phase Categories The i-th online monitoring method for carbon emissions; —Training Phase Categories The number of valid data pairs in the data.

[0036] The material monitoring method uses a constructed artificial intelligence model and online emission monitoring to predict emissions.

[0037]

[0038] In the formula: —Training Phase Categories The predicted carbon emissions from the i-th material monitoring method; —Training Phase Categories The i-th online monitoring method for carbon emissions.

[0039] Calculate training phase categories The root mean square percentage error between the predicted carbon emissions from the material monitoring method and the calculated carbon emissions from the material monitoring method The error should be less than or equal to a certain threshold. The formula for calculating the error during the training phase is as follows:

[0040] In the formula: —Training Phase Categories Root mean square percentage error; —Training Phase Categories The predicted carbon emissions from the i-th material monitoring method; —Training Phase Categories The Middle Carbon emissions from material monitoring methods; —Training Phase Categories The number of valid data pairs in the data.

[0041] If the error exceeds the threshold, more valid data should be collected until the cumulative error is less than or equal to the threshold.

[0042] 2. Data quality diagnosis during the application phase 2.1 Data Preprocessing The newly collected monitoring data in the application phase should be cleaned, calculated, and classified in accordance with the contents of 1.1-1.3.

[0043] 2.2 Material Monitoring Method for Data Prediction Using effective data, we construct successful artificial intelligence models for each category and monitor emissions online to predict emissions using the material monitoring method.

[0044]

[0045] In the formula: ——Application Stage Category The predicted carbon emissions from the i-th material monitoring method; ——Application Stage Category The i-th online monitoring method for carbon emissions.

[0046] 2.3 Cumulative Error Test The cumulative error between the predicted carbon emissions from the material monitoring method and the calculated carbon emissions from the material monitoring method during the application phase is calculated. The formula for calculating the cumulative error during the application phase is as follows:

[0047] In the formula: ——Application Stage Category Cumulative error; ——Application Stage Category The predicted carbon emissions from the i-th material monitoring method; ——Application Stage Category The carbon emissions of the i-th material monitoring method.

[0048] The cumulative error should be less than the threshold. Data above the threshold is considered risky. Data below the threshold... All the data can be used as training data to update the artificial intelligence model.

[0049] 3. Data processing in the application phase 3.1 Invalid data handling methods (as shown in Table 1) (1) Data from the material monitoring method is invalid, while data from the online monitoring method is valid. Using artificial intelligence models and online monitoring data, predicted carbon emissions from the material monitoring method are calculated to fill in invalid data from the material monitoring method. To penalize invalid data from the material monitoring method, the final carbon emissions should be calculated by multiplying the predicted carbon emissions by a certain coefficient.

[0050] (2) Data from the material monitoring method is invalid, and data from the online monitoring method is invalid. When both methods are ineffective, the maximum carbon emission value of the material monitoring method within a certain period prior to this time period shall be used.

[0051] (3) Data from the material monitoring method is valid, while data from the online monitoring method is invalid. While the material monitoring data is valid, this result cannot be verified using artificial intelligence models. To penalize emissions from invalid online monitoring data, the carbon emissions from the material monitoring method should be multiplied by a certain coefficient to obtain the final result.

[0052] Table 1

[0053] 3.2 Effective data processing methods (as shown in Table 2) (1) Valid risk data The existence of risks in the material monitoring method (LMMD) data indicates that the data at this stage is unreliable. Therefore, artificial intelligence models and online monitoring data are used to calculate and predict LMMD carbon emissions to supplement the risk data. To penalize the risk data, the final carbon emissions should be the predicted carbon emissions multiplied by a coefficient.

[0054] (2) Valid non-risk data Data from material monitoring methods that are valid and have low-risk test results will not be processed in any way.

[0055] Table 2

[0056] 4. Automated algorithms and programs By building automated algorithms and programs, once the raw data from material monitoring and online monitoring methods are input into the program, the entire process, including data validity assessment, carbon emission calculation, quality diagnosis, and data processing, will be automatically performed.

[0057] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements a representation learning method based on a multimodal knowledge graph as described above.

[0058] This invention addresses the issues of numerous verification steps and high regulatory costs associated with material monitoring methods by integrating an intelligent abnormal emission identification algorithm. Through real-time comparison of data from both methods, it enables rapid location of abnormal data. Furthermore, it shifts from traditional post-event manual verification to real-time automatic early warning, significantly reducing the difficulty of carbon market regulation, improving regulatory efficiency, and providing accurate regulatory basis for government carbon data management departments.

[0059] This invention utilizes a real-time data acquisition and analysis system to upgrade carbon emission monitoring from traditional monthly static accounting to high-frequency dynamic monitoring. By combining a classification model with real-time data from both methods, it achieves continuous tracking of carbon emissions at the second / minute level. This not only captures instantaneous emission fluctuations but also reflects macroeconomic trends, providing dynamic decision-making support for enterprise carbon emission management and overcoming the limitations of the lag in static accounting.

[0060] This invention clarifies the data processing rules after quality diagnosis and data accounting, making it authoritative data that can be directly accepted by the carbon emission trading market and government management departments. Compared with the results of a single method, the combined results integrate the advantages of both methods, satisfying the fairness requirements of carbon trading while providing a high-quality data source for industry or social carbon emission accounting, thus solving the problems of insufficient credibility and limited applicability of traditional data.

[0061] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0062] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for carbon data quality diagnosis and processing based on dual-source monitoring and artificial intelligence, characterized in that, Includes the following steps: Training phase: S1. Obtain raw data from material monitoring, online monitoring, and production operation data within a historical time period; S2. The validity of the raw data from the material monitoring method and the raw data from the online monitoring method is judged and cleaned, and the carbon emissions from the material monitoring method and the carbon emissions from the online monitoring method are calculated respectively. S3. Based on the production operation data, an artificial intelligence classification algorithm is used to automatically classify the production conditions to obtain multiple condition categories. S4. For each of the above operating conditions, with the carbon emissions from the online monitoring method as input and the carbon emissions from the material monitoring method as output, construct and train an artificial intelligence prediction model for that operating condition until the model performance index meets the preset threshold. Application phase: S5. Obtain the raw data of the material monitoring method, the raw data of the online monitoring method, and the production operation data to be diagnosed, and execute the validity judgment, cleaning and accounting steps in S2 to obtain the current carbon emissions of the material monitoring method and the current carbon emissions of the online monitoring method. S6. Based on the production operation data, determine the current operating condition category and call the corresponding trained artificial intelligence prediction model. S7. Using the current online monitoring carbon emissions, the predicted material monitoring carbon emissions are calculated using the artificial intelligence prediction model in S4. S8. Calculate the cumulative error between the predicted carbon emissions from the material monitoring method and the current carbon emissions from the material monitoring method within a set period, and compare the cumulative error with a preset threshold. If the error exceeds the threshold, the current carbon emissions from the material monitoring method are determined to be risk data.

2. The carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence according to claim 1, characterized in that, In step S2, the validity of the raw data from the material monitoring method and the raw data from the online monitoring method is assessed and cleaned, and the carbon emissions from the material monitoring method and the online monitoring method are calculated respectively, specifically including: If either the raw data from the material monitoring method or the raw data from the online monitoring method is invalid for a certain period of time, the data for that period is invalid; invalid data is identified by an algorithm, and invalid data and valid data are marked separately; the carbon emissions from the material monitoring method and the carbon emissions from the online monitoring method for the valid data are calculated.

3. The carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence according to claim 1, characterized in that, In step S4, an artificial intelligence prediction model for this work condition category is constructed and trained, as shown in the following formula: ； in, Training phase category Carbon emissions from material monitoring methods The training phase category k represents the carbon emissions from the online monitoring method.

4. The carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence according to claim 1, characterized in that, In step S4, the online monitoring emissions are input into the trained model to predict the material monitoring emissions, using the following formula: ； in, For the training phase working condition category The predicted carbon emissions from the i-th material monitoring method. For the training phase working condition category The i-th online monitoring method for carbon emissions.

5. The carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence according to claim 1, characterized in that, In step S4, the performance indicators of the artificial intelligence prediction model meeting the preset threshold specifically include: Calculate training phase categories The root mean square percentage error between the predicted carbon emissions from the material monitoring method and the calculated carbon emissions from the material monitoring method The formula is as follows: ； in, For the training phase working condition category Root mean square percentage error For the training phase working condition category The number of valid data pairs; if the root mean square percentage error is greater than a preset threshold. Then collect more valid data until the root mean square percentage error is less than or equal to a preset threshold. .

6. The carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence according to claim 1, characterized in that, In step S8, the cumulative error between the carbon emissions from the material monitoring method and the current carbon emissions from the material monitoring method within a set period is calculated. The formula is as follows: ； in, Application stage category Cumulative error Application stage category The predicted carbon emissions from the i-th material monitoring method. Application stage category The carbon emissions of the i-th material monitoring method.

7. The carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence according to claim 1, characterized in that, The application phase also includes hierarchical data processing steps: Invalid data handling: When the data corresponding to the current carbon emissions from the material monitoring method is determined to be invalid, while the carbon emissions from the current online monitoring method are valid, the predicted carbon emissions from the material monitoring method are multiplied by a penalty coefficient to obtain the final carbon emissions. When both the current carbon emission amount measured by the material monitoring method and the current carbon emission amount measured by the online monitoring method are determined to be invalid, the maximum carbon emission amount measured by the material monitoring method within a preset time period is used as the final carbon emission amount. When the carbon emission amount measured by the current material monitoring method is valid, but the carbon emission amount measured by the current online monitoring method is determined to be invalid, the carbon emission amount measured by the current material monitoring method is multiplied by a penalty coefficient to obtain the final carbon emission amount.

8. A carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence according to claim 6, characterized in that, It also includes effective data processing: When the current carbon emissions from the material monitoring method are valid and are determined to be risky data, the predicted carbon emissions from the material monitoring method are multiplied by a penalty coefficient to obtain the final carbon emissions. When the carbon emission data obtained by the current material monitoring method is valid and is determined to be non-risk data, the original carbon emission data obtained by the material monitoring method is directly used as the final carbon emission data.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements a carbon data quality diagnosis and processing method based on dual-source monitoring and artificial intelligence as described in any one of claims 1 to 7.

Images