Carbon emission intelligent monitoring system and method based on Internet of Things

Abstract

The invention relates to the technical field of energy management and carbon accounting, in particular to an intelligent carbon emission monitoring system and method based on the Internet of Things, and the system comprises a distributed sensing network module which collects the energy consumption and emission data of an independent accounting unit in real time; the edge computing module is deployed at a local edge computing node to preprocess the data acquired by the distributed sensing network module; the cloud co-processing module receives the preprocessed data and carries out dynamic accounting and multi-source data fusion; the carbon accounting model module automatically matches accounting requirements according to the industry to which the enterprise belongs and generates a monitoring point layout scheme; and the intelligent visualization module is used for generating a multi-dimensional carbon emission data analysis chart and a compliance report meeting MRV requirements based on data of the edge calculation module, the cloud co-processing module and the carbon accounting model module. Full-process automatic management from data acquisition, real-time monitoring to verification is realized, and the accuracy, integrity and traceability of carbon emission data are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of energy management and carbon accounting, and specifically to an intelligent carbon emission monitoring system and method based on the Internet of Things. Background Art

[0002] As one of the main sources of carbon emissions, the accuracy and timeliness of corporate carbon emission data is not only related to the effective formulation and implementation of their own energy-saving and emission reduction measures, but also affects the government's macro-level carbon management and policy formulation.

[0003] Currently, carbon emissions data collection relies primarily on manual reporting, sampling monitoring, or estimation of industry average emission factors, which present significant flaws. Manual reporting is prone to data omissions or distortion, and the accounting cycle is long, making it difficult to respond to regulatory needs in a timely manner. While traditional sensors can collect some data, they lack unified standards, leading to prominent data silos and difficulty in cross-system integration. Furthermore, accounting methods based on static emission factors fail to reflect actual production fluctuations and are subject to significant errors. Third-party verification relies on periodic on-site audits, which are costly and inefficient.

[0004] Existing technologies generally face challenges such as data lag, insufficient accuracy, and low credibility, making it difficult to meet the increasingly stringent carbon regulation and carbon market trading needs. Summary of the Invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a carbon emission intelligent monitoring system and method based on the Internet of Things.

[0006] In a first aspect, the technical solution of the present invention provides an intelligent carbon emission monitoring system based on the Internet of Things, comprising: Distributed sensor network module for real-time collection of energy consumption and emission data of independent accounting units; The edge computing module is deployed on the local edge computing node and is used to pre-process the data collected by the distributed sensor network module, including data cleaning, anomaly detection and preliminary emission calculation; The cloud collaborative processing module is connected to the edge computing module to receive pre-processed data and perform dynamic calculations and multi-source data fusion; The carbon accounting model module has a built-in industry characteristic knowledge base and a differentiated accounting unit division algorithm, which is used to automatically match accounting requirements according to the industry to which the enterprise belongs and generate a monitoring point layout plan; The intelligent visualization module generates multi-dimensional carbon emission data analysis charts and compliance reports that meet MRV requirements based on data from the edge computing module, cloud collaborative processing module, and carbon accounting model module.

[0007] Through distributed sensor networks, energy consumption and emission data of independent accounting units are collected in real time. Edge computing and cloud collaborative processing technologies are combined to achieve data cleaning, anomaly detection and dynamic accounting. A standardized carbon accounting model is established to achieve full-process automated management from data collection, real-time monitoring to verification and validation, ensuring the accuracy, completeness and traceability of carbon emission data.

[0008] Through the collaborative work of a distributed sensor network, edge computing modules, and cloud-based collaborative processing modules, the entire process, from data collection and preprocessing to dynamic accounting and visual report generation, is fully automated, reducing manual intervention and improving data processing efficiency and accuracy. The combination of edge computing and cloud processing enables real-time collection and processing of energy consumption and emissions data, ensuring high-precision and real-time carbon emissions data, and providing reliable data support for corporate carbon accounting, carbon trading, and green certification. The carbon accounting model module's built-in industry-specific knowledge base and accounting unit division algorithm automatically match accounting requirements to the company's industry and generate a monitoring point layout plan to meet the carbon emissions monitoring needs of different industries.

[0009] As a further limitation of the technical solution of the present invention, the local edge computing node presets a lightweight emission factor library; The edge computing module includes a pre-processing unit, an anomaly detection unit, and a preliminary emission calculation unit; The preprocessing unit is used to convert the raw sensor data collected by the distributed sensor network module into structured time series data through the industrial protocol parsing engine; eliminate invalid data and fill missing values ​​based on sliding window statistics; normalize the data and unify the dimension and sampling frequency.

[0010] The anomaly detection unit uses a temporal convolutional network model to detect mutation points and outliers in the data stream and calculate the anomaly probability to input into the preliminary emission calculation unit; combines the industry process rule library to mark logical anomaly data; generates anomaly reports and triggers data re-collection or equipment calibration instructions to the distributed sensor network module; The preliminary emission calculation unit calls the lightweight emission factor library preset in the local edge computing node, calculates the process-level carbon emissions in real time according to the formula, and uploads it to the cloud collaborative processing module.

[0011] The edge computing module's preprocessing unit rapidly converts raw sensor data into structured time-series data, eliminating invalid data and filling missing values, improving data quality and availability. The anomaly detection unit promptly identifies sudden changes and outliers in the data, labels logically abnormal data based on the industry's process rule base, and triggers data re-collection or equipment calibration instructions to ensure data accuracy and reliability. The preliminary emissions calculation unit calculates process-level carbon emissions in real time and uploads them to the cloud, providing a timely and accurate data foundation for subsequent dynamic accounting.

[0012] As a further limitation of the technical solution of the present invention, the anomaly detection unit normalizes the latest time series data and inputs it into the trained time series convolutional network model to capture local patterns through causal convolution, capture long-term dependencies through dilated convolution, and output the anomaly probability of the center point of the current window.

[0013] The causal and dilated convolutions of the temporal convolutional network model capture local patterns and long-term dependencies in data streams, accurately detecting data anomalies and improving the accuracy and reliability of anomaly detection. The model can dynamically adjust detection strategies based on real-time data, adapting to data changes across different industries and production processes, ensuring the timeliness and effectiveness of anomaly detection.

[0014] As a further limitation of the technical solution of the present invention, the system also includes a dynamic verification module; The preliminary emission calculation unit calculates the confidence level of process-level carbon emissions based on the probability of abnormality, and directly uploads high-confidence results that are higher than the set threshold to the cloud-based collaborative processing module, while low-confidence data that are lower than the set threshold triggers the dynamic verification module for data review.

[0015] As a further limitation of the technical solution of the present invention, the preliminary emission calculation unit calculates the data integrity score, sensor accuracy score, process compliance score, and the abnormal probability calculated by the abnormality detection unit. Calculate the confidence level of process-level carbon emissions; the formula is as follows:

[0016]

[0017] Data integrity score ; Sensor Accuracy Score ; Process Compliance Score ; Where, is the process-level carbon emissions, Activity data collected in real time by distributed sensor network modules, is the locally preset emission factor, is the process emission factor, is the weight coefficient.

[0018] As a further limitation of the technical solution of the present invention, the cloud collaborative processing module is specifically used to receive pre-processed data uploaded by the edge computing module; associate with third-party data sources to construct a multi-dimensional feature matrix; and call an industry-specific accounting model to calculate the total enterprise-level carbon emissions according to the formula:

[0019] Where, is the process weight factor, Emissions from purchased electricity / heat.

[0020] The cloud-based collaborative processing module integrates pre-processed data uploaded by the edge computing module with third-party data sources to construct a multi-dimensional feature matrix, providing more comprehensive data support for dynamic accounting. Through industry-specific accounting models, it accurately calculates enterprise-level carbon emissions, meeting the company's overall carbon accounting and management needs and providing a scientific basis for carbon reduction decisions.

[0021] As a further limitation of the technical solution of the present invention, the carbon accounting model module includes a construction and matching unit, an accounting unit division algorithm unit, and a monitoring point layout plan generation unit; The construction and matching unit segments enterprises into several industry nodes. The deep learning OCR engine automatically parses business licenses and matches industry codes. Each industry node is associated with an emission source list, monitoring equipment rules, and accounting boundaries. The company's latitude and longitude coordinates and production scale are input to automatically adjust the density of monitoring points. The accounting unit division algorithm unit obtains the process flow chart through the enterprise DCS system, identifies the key carbon flow nodes, uses the graph segmentation algorithm to divide the production line into the smallest accounting units, and sets the unit weight according to the proportion of equipment energy consumption; and adjusts the accounting granularity in real time; The monitoring point layout plan generation unit determines the required parameters based on the industry knowledge base; screens sensitive variables through feature importance analysis; and uses integer linear programming to solve the optimal monitoring point location; The objective function is:

[0022] in, is the equipment installation cost, To cover the Type of emission source, Is a decision variable, a binary variable, indicating whether Deploy monitoring equipment:

[0023] By optimizing The number of monitoring points can be minimized.

[0024] By building and matching units, enterprises can be segmented by industry, precisely matching industry codes with accounting requirements to ensure accurate carbon accounting. The accounting unit division algorithm dynamically adjusts accounting granularity based on the enterprise's process flow diagram and equipment energy consumption, improving the level of accounting refinement. The monitoring point layout plan generation unit generates optimal monitoring point locations based on industry knowledge bases and feature importance analysis, reducing monitoring costs and improving monitoring efficiency.

[0025] As a further limitation of the technical solution of the present invention, the dynamic verification module includes a real-time verification layer, a process verification layer and a system verification layer, which is used to achieve multi-level data verification through statistical process control, material balance model and third-party data source docking.

[0026] By calculating the confidence level of process-level carbon emissions, we can distinguish between high-confidence and low-confidence data, conduct targeted review of low-confidence data, and improve the overall quality and credibility of the data. The introduction of a dynamic verification module to review low-confidence data ensures its accuracy and reliability before calculation, further enhancing the intelligence level of the system.

[0027] As a further limitation of the technical solution of the present invention, the intelligent visualization module receives process-level real-time emission data and confidence labels uploaded by the edge computing module; associates the enterprise-level accounting results output by the cloud-based collaborative processing module; calls the industry benchmark data of the carbon accounting model module; generates emission curves by hour / day / month; displays high-emission areas within the factory through a heat map; and uses a Sankey diagram to visualize the flow of carbon elements.

[0028] The intelligent visualization module automatically generates text descriptions by loading preset templates based on the industry to which the enterprise belongs. Key fields include: activity data, emission factors, and uncertainty descriptions.

[0029] The intelligent visualization module generates multi-dimensional carbon emissions data analysis charts, heat maps, Sankey diagrams, and other visualizations, providing an intuitive overview of carbon emissions and enabling companies to quickly understand their current status and trends. It also automatically generates MRV-compliant compliance reports based on pre-set templates, reducing the manual workload and improving report accuracy and compliance.

[0030] In a second aspect, the technical solution of the present invention further provides a carbon emission intelligent monitoring method based on the Internet of Things, comprising the following steps: Real-time collection of energy consumption and emission data of independent accounting units through distributed sensor networks; Pre-processing of collected data in the edge computing module, including data cleaning, anomaly detection, and preliminary emission calculations; Upload the pre-processed data to the cloud collaborative processing module for dynamic accounting and multi-source data fusion; Based on the carbon accounting model module, it automatically matches accounting requirements according to the industry to which the enterprise belongs and generates a monitoring point layout plan; Based on the data from the edge computing module, cloud collaborative processing module and carbon accounting model module, multi-dimensional carbon emission data analysis charts and compliance reports that meet MRV requirements are generated.

[0031] As a further limitation of the technical solution of the present invention, the step of preprocessing the collected data in the edge computing module includes: The industrial protocol parsing engine converts raw sensor data into structured time series data; invalid data is eliminated and missing values ​​are filled based on sliding window statistics; the data is normalized to unify the dimension and sampling frequency; Use a time series convolutional network model to detect mutation points and outliers in data streams and calculate anomaly probabilities; combine industry process rule libraries to mark logically abnormal data; generate anomaly reports and trigger data re-collection or equipment calibration instructions; Call the lightweight emission factor library preset in the local edge computing node to calculate the process-level carbon emissions in real time according to the formula; upload the calculation results to the cloud collaborative processing module.

[0032] As a further limitation of the technical solution of the present invention, the specific steps of using the temporal convolutional network model to detect mutation points and outliers in the data stream and calculate the anomaly probability include: Normalize the latest time series data and input it into the trained time series convolutional network model; Capturing local patterns through causal convolution and long-term dependencies through dilated convolution; Output the abnormal probability of the center point of the current window.

[0033] As a further limitation of the technical solution of the present invention, the step of calling the lightweight emission factor library preset in the local edge computing node and calculating the process-level carbon emissions in real time according to the formula includes: Calculate the confidence level of process-level carbon emissions in combination with the abnormal probability; Upload high-confidence results that are higher than the set threshold directly to the cloud; Low-confidence data below the set threshold will trigger the dynamic verification module to conduct data review.

[0034] As a further limitation of the technical solution of the present invention, the specific steps of performing dynamic accounting and multi-source data fusion include: Receive pre-processed data uploaded by the edge computing module; Link third-party data sources to build a multi-dimensional feature matrix; Call industry-specific accounting models to calculate total enterprise-level carbon emissions.

[0035] As a further limitation of the technical solution of the present invention, the specific steps of the carbon accounting model module include: Segment enterprises by industry and build industry nodes; automatically parse business licenses through a deep learning OCR engine to match industry codes; and associate emission source inventories, monitoring equipment rules, and accounting boundaries; Obtain process flow diagrams through the enterprise DCS system to identify key carbon flow nodes; use graph segmentation algorithms to divide the production line into minimum accounting units; set unit weights based on the proportion of equipment energy consumption; Determine the required measurement parameters based on the industry knowledge base; screen decision variables through feature importance analysis; and use integer linear programming to solve the optimal monitoring point location.

[0036] As a further limitation of the technical solution of the present invention, the specific steps of visualization and report generation include: Receive process-level real-time emission data and confidence labels uploaded by the edge computing module; Associate the enterprise-level accounting results output by the cloud collaborative processing module; Calling industry benchmark data of carbon accounting model module; Generate emission curves by hour / day / month; Use heat maps to show high-emission areas within the plant; Use Sankey diagrams to visualize the flow of carbon elements; Load preset templates based on the company's industry and automatically generate compliance reports that meet MRV requirements.

[0037] The beneficial effects of this invention lie in its ability to not only enable data collection, monitoring, anomaly detection, and automated verification, but also provide carbon emissions forecasting and analysis, as well as intelligent decision support. This system effectively addresses the pain points of traditional carbon management, such as the high level of manual intervention and poor data timeliness, providing comprehensive support for enterprises from carbon emissions monitoring to emission reduction decision-making. By continuously optimizing algorithmic models and deepening integration with the carbon trading market, it will facilitate the early realization of more precise carbon asset management and low-carbon transition. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0039] Figure 1 This is a schematic block diagram of a system according to an embodiment of the present invention.

[0040] Figure 2 The figure is a schematic flow chart of a method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, features, and advantages of the present invention more obvious and easy to understand, the technical solutions of the present invention will be clearly and completely described below in conjunction with the drawings in the specific embodiments. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0042] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0043] like Figure 1 As shown, an embodiment of the present invention provides an intelligent carbon emission monitoring system based on the Internet of Things, including: Distributed sensor network module for real-time collection of energy consumption and emission data of independent accounting units; The edge computing module is deployed on the local edge computing node and is used to pre-process the data collected by the distributed sensor network module, including data cleaning, anomaly detection and preliminary emission calculation; The cloud collaborative processing module is connected to the edge computing module to receive pre-processed data and perform dynamic calculations and multi-source data fusion; The carbon accounting model module has a built-in industry characteristic knowledge base and a differentiated accounting unit division algorithm, which is used to automatically match accounting requirements according to the industry to which the enterprise belongs and generate a monitoring point layout plan; The intelligent visualization module generates multi-dimensional carbon emission data analysis charts and compliance reports that meet MRV requirements based on data from the edge computing module, cloud collaborative processing module, and carbon accounting model module.

[0044] In some embodiments, a lightweight emission factor library is pre-installed on local edge computing nodes. This library is a database of emission factors designed specifically for IoT edge computing environments. By streamlining data structures, optimizing storage methods, and improving computational efficiency, it enables rapid access and calculation of emission factors on local nodes, thereby supporting real-time or near-real-time carbon emissions monitoring.

[0045] The edge computing module includes a pre-processing unit, an anomaly detection unit, and a preliminary emission calculation unit; The preprocessing unit is used to convert the raw sensor data collected by the distributed sensor network module into structured time series data through the industrial protocol parsing engine; eliminate invalid data and fill missing values ​​based on sliding window statistics; normalize the data and unify the dimension and sampling frequency.

[0046] The anomaly detection unit uses a temporal convolutional network model to detect mutation points and outliers in the data stream and calculate the anomaly probability to input into the preliminary emission calculation unit; combines the industry process rule library to mark logical anomaly data; generates anomaly reports and triggers data re-collection or equipment calibration instructions to the distributed sensor network module; The preliminary emission calculation unit calls the lightweight emission factor library preset in the local edge computing node, calculates the process-level carbon emissions in real time according to the formula, and uploads it to the cloud collaborative processing module.

[0047] An end-to-end data security protection system has been established: at the data transmission layer, a combination of hardware and software is employed. The hardware uses an encryption card to implement the encryption process, while the software uses the national secret SM4 algorithm for identity authentication, combined with the SM4-CBC encryption mode to ensure data transmission security. The data storage layer is based on a trusted execution environment, and key sensitive data (such as enterprise production data) is processed using homomorphic encryption, supporting accounting operations in an encrypted state. The access control module implements attribute-based dynamic authorization (ABAC), includes multiple fine-grained access policies, and supports spatial and temporal constraints (such as allowing access only from specified IP segments during business hours). The audit tracking system uses a modified Hyperledger Fabric framework to store key operation logs on-chain.

[0048] In some embodiments, the anomaly detection unit normalizes the latest time series data and inputs it into a trained time series convolutional network model to capture local patterns through causal convolution, capture long-term dependencies through dilated convolution, and output the anomaly probability of the center point of the current window.

[0049] A training set is constructed using historical normal operating data (e.g., three months of sensor data). Synthetic anomalies (e.g., impulse noise, trend drift) are injected to generate labeled samples, with a sample size of N. Causal dilation convolution stacking (e.g., dilation factors [1, 2, 4, 8]) is used to ensure temporal dependencies. The output layer consists of a single neuron with sigmoid activation, which directly predicts the anomaly probability.

[0050]

[0051] in, is the true label, is the predicted probability.

[0052] In some embodiments, the system further comprises a dynamic verification module; The preliminary emission calculation unit calculates the confidence level of process-level carbon emissions based on the probability of abnormality, and directly uploads high-confidence results that are higher than the set threshold to the cloud-based collaborative processing module, while low-confidence data that are lower than the set threshold triggers the dynamic verification module for data review.

[0053] The specific steps of the dynamic verification module reviewing the low-confidence data include: The system conducts a confidence assessment on the collected and calculated carbon emission data, generating a data confidence score based on multiple indicators such as data completeness, accuracy, timeliness, and source reliability; Filter data points or data segments with confidence levels lower than the preset threshold, mark them as low-confidence data, and trigger the review process.

[0054] Applying an improved CUSUM control chart algorithm to low-confidence data, the sliding window is used to analyze short-term fluctuations in the data stream to detect abnormal fluctuations or mutation points. If abnormal fluctuations are detected, further analysis is conducted on the causes of the fluctuations, such as equipment failure, data transmission errors, or abnormal production processes.

[0055] A material balance model was constructed based on first principles to analyze the flow of carbon elements in the production process involved in low-confidence data; Check the logical consistency of data in the production process and verify whether the carbon emission data conforms to the material balance relationship. For example, in the catalytic cracking unit of the petrochemical industry, the carbon element flow matrix is ​​used to perform hourly mass balance verification.

[0056] Connect to third-party data sources, such as power grid emission factors and meteorological data, to verify the cross-system consistency of low-confidence data; compare the differences between third-party data and data within the system, analyze the causes of the differences, and judge the credibility of the data.

[0057] Based on the review results, low-confidence data will be corrected. Correction methods include data interpolation, recalculation or adjustment of data sources; the corrected data will be reincorporated into the accounting system, and the correction process and reasons will be recorded to form a complete chain of evidence to support subsequent verification and audit work.

[0058] Regularly evaluate and adjust the review rules and thresholds of the dynamic verification module to improve the system's intelligence level and verification efficiency.

[0059] In some embodiments, the preliminary emission calculation unit calculates the data integrity score, the sensor accuracy score, the process compliance score, and the anomaly probability calculated by the anomaly detection unit. Calculate the confidence level of process-level carbon emissions; the formula is as follows:

[0060]

[0061] Data integrity score ; Sensor Accuracy Score ; Process Compliance Score ; Where, is the process-level carbon emissions, Activity data collected in real time by distributed sensor network modules, is the locally preset emission factor, is the process emission factor, is the weight coefficient.

[0062] The values ​​in the embodiment of the present invention are 0.2, 0.3, 0.3, and 0.2.

[0063] Data completeness score (based on the proportion of missing values ​​and time series continuity, normalized to 0-1); sensor accuracy score (based on the error range in the equipment calibration certificate, such as ±0.5% is recorded as 0.95 points); process compliance score (compared to the industry rule base, such as whether the cement kiln temperature data is within the preset threshold).

[0064] For example, high confidence (Confidence ≥ 0.8): directly upload to the cloud collaborative processing module; Low confidence (Confidence < 0.8): triggers the material balance review of the dynamic verification module Expected total data points: calculated based on the sampling frequency (e.g., 1-minute sampling, 60 points expected per hour). Invalid data: includes over-range values ​​(e.g., current sensor output > 20mA) and fill values ​​due to communication interruptions (e.g., 0xFFFF). Example: A process is missing 5 data points within 1 hour and has 2 over-range values ​​(total expected 60 points):

[0065] Sensor error percentage: Obtained from the equipment calibration certificate (e.g., ±0.5% is recorded as 0.005). Industry maximum allowable error: Refer to the standard (e.g., HJ 75-2017 stipulates that CEMS flow rate error ≤ 5%).

[0066] Example: The error of a flue gas flow sensor is ±2%, and the industry allows 5%:

[0067] Threshold range: From the industry knowledge base (e.g., cement kiln temperature threshold: 1200±50°C). Exceedance: The percentage deviation between the actual value and the threshold. Example: Five parameters were monitored, of which the kiln temperature exceeded the standard (measured 1300°C, threshold 1250°C), while the rest were in compliance:

[0068] =0.3 Take the cement industry as an example: Input features: raw meal feed scale data (completeness score 0.9, due to missing 5 minutes of data; accuracy score 0.8, equipment error ±1%); kiln outlet temperature (compliance score 0.6, due to exceeding the preset range by 50°C).

[0069] Model output: TCN anomaly probability =0.3; The confidence interval width of the emission calculation model is 0.15 (converted score 0.85, set threshold).

[0070] Comprehensive calculation: Confidence=0.2×0.983+0.3×0.6+0.3×0+0.2×0.7=0.5166 (low confidence), triggering the clinker production-coal consumption balance model in the dynamic verification module for cross-validation.

[0071] The system has established a scientific and complete data credibility assessment system to quantitatively evaluate data quality from multiple dimensions. The integrity assessment mainly checks the temporal and spatial coverage of the data and identifies missing data and abnormal discontinuities. The accuracy assessment comprehensively considers factors such as the accuracy of the monitoring equipment and the sampling frequency to quantify the uncertainty of the data. The timeliness assessment focuses on the timeliness of data updates and downgrades the weight of lagging data. The system generates a detailed quality file for each data point, recording all quality events throughout its entire life cycle. These assessment results are not only used to guide data use, but also support the generation of data quality reports to meet compliance requirements. In terms of credibility management, the system takes special account of the data characteristics of different industries. For example, the power industry focuses on the regular calibration records of monitoring equipment, while the chemical industry is more concerned with the degree of closure of material balances.

[0072] In some embodiments, the cloud collaborative processing module is specifically configured to receive pre-processed data uploaded by the edge computing module; associate with third-party data sources to construct a multi-dimensional feature matrix; and call an industry-specific accounting model to calculate the total enterprise-level carbon emissions according to the formula:

[0073] Where, is the process weight factor, Emissions from purchased electricity / heat.

[0074] In some embodiments, the carbon accounting model module includes a construction and matching unit, an accounting unit division algorithm unit, and a monitoring point layout plan generation unit; The construction and matching unit segments enterprises into several industry nodes. The deep learning OCR engine automatically parses business licenses and matches industry codes. Each industry node is associated with an emission source list, monitoring equipment rules, and accounting boundaries. The company's latitude and longitude coordinates and production scale are input to automatically adjust the density of monitoring points. The accounting unit division algorithm unit obtains the process flow chart through the enterprise DCS system, identifies the key carbon flow nodes, uses the graph segmentation algorithm to divide the production line into the smallest accounting units, and sets the unit weight according to the proportion of equipment energy consumption; and adjusts the accounting granularity in real time; The monitoring point layout plan generation unit determines the required parameters based on the industry knowledge base; screens sensitive variables through feature importance analysis; and uses integer linear programming to solve the optimal monitoring point location; The objective function is:

[0075] in, is the equipment installation cost, To cover the Type of emission source, is the decision variable.

[0076]

[0077] By optimizing The number of monitoring points can be minimized.

[0078] When the constraints cannot be met (e.g., insufficient budget), the following steps are performed: According to the contribution of emission sources (e.g. clinker calcination accounts for 70% of the total plant emissions), Sort in descending order; Priority will be given to monitoring high-contribution emission sources, and secondary sources (such as heating in office areas) may be temporarily uncovered; mobile monitoring equipment will be used to conduct periodic retests on uncovered sources.

[0079] In some embodiments, the dynamic verification module includes a real-time verification layer, a process verification layer, and a system verification layer, which is used to implement multi-level data verification through statistical process control, material balance models, and third-party data source docking.

[0080] The system has established a multi-level dynamic verification system to ensure the accuracy and reliability of carbon emission data. The real-time verification layer uses statistical process control methods to continuously monitor abnormal fluctuations in data flows. The process verification layer constructs a material balance model based on first principles and implements cross-process cross-validation through carbon element flow analysis. The system verification layer performs data consistency checks between systems by connecting to third-party data sources. Targeted at the characteristics of different industries, the system has designed specialized verification schemes, such as a blast furnace carbon flow tracking model for the steel industry and a clinker output and coal consumption correlation model for the cement industry. Abnormal data discovered during the verification process will automatically trigger a review process, and the system will correct the data or issue an early warning based on preset rules. The entire verification process forms a complete chain of evidence to support subsequent verification and audit work.

[0081] In some embodiments, the intelligent visualization module receives process-level real-time emission data and confidence labels uploaded by the edge computing module; associates the enterprise-level accounting results output by the cloud collaborative processing module; calls the industry benchmark data of the carbon accounting model module; generates emission curves by hour / day / month; displays high-emission areas within the factory through heat maps; and uses Sankey diagrams to visualize the flow of carbon elements.

[0082] The intelligent visualization module automatically generates text descriptions by loading preset templates based on the industry to which the enterprise belongs. Key fields include: activity data, emission factors, and uncertainty descriptions.

[0083] The system provides powerful data visualization capabilities and supports multi-dimensional carbon emission data analysis and display. Based on modern Web visualization technology, the system can dynamically render various complex charts, such as carbon flow Sankey diagrams, emission heat maps, etc. The report generation module has a built-in template library that meets MRV requirements and supports the automatic generation of various compliance reports. In response to the management needs of different industries, the system provides special display solutions, such as designing unit benchmarking analysis dashboards for power companies and developing route carbon emission visualization tools for aviation companies. The interactive analysis function allows users to obtain the required information through natural language queries. The system will automatically identify the query intent and generate corresponding analysis results. All visualization components support drill-down analysis, and users can drill down from a macro overview to specific device-level data.

[0084] The knowledge base is stored using graph database technology, efficiently representing complex industry knowledge and relationships. When the system detects anomalies or optimization opportunities, it generates targeted recommendations based on the knowledge base content. Decision support functions combine a rules engine and machine learning models to provide comprehensive support from problem diagnosis to resolution. The system also supports the continuous accumulation of knowledge, enriching the knowledge base through analysis of historical data and user feedback. Based on the management characteristics of different industries, the system provides differentiated decision support solutions, such as raw material ratio optimization recommendations for cement companies and cooling system tuning strategies for data centers. These features make the system not only a monitoring tool but also an intelligent carbon management assistant.

[0085] like Figure 2 As shown, an embodiment of the present invention provides a carbon emission intelligent monitoring method based on the Internet of Things, comprising the following steps: S1. Real-time collection of energy consumption and emission data of independent accounting units through distributed sensor networks; S2. Preprocess the collected data in the edge computing module, including data cleaning, anomaly detection, and preliminary emission calculation; S3, upload the pre-processed data to the cloud collaborative processing module for dynamic accounting and multi-source data fusion; S4. Based on the carbon accounting model module, automatically match accounting requirements according to the industry to which the enterprise belongs and generate a monitoring point layout plan; S5. Based on the data from the edge computing module, cloud collaborative processing module and carbon accounting model module, generate multi-dimensional carbon emission data analysis charts and compliance reports that meet MRV requirements.

[0086] In some embodiments, the step of preprocessing the collected data in the edge computing module includes: The industrial protocol parsing engine converts raw sensor data into structured time series data; invalid data is eliminated and missing values ​​are filled based on sliding window statistics; the data is normalized to unify the dimension and sampling frequency; Use a time series convolutional network model to detect mutation points and outliers in data streams and calculate anomaly probabilities; combine industry process rule libraries to mark logically abnormal data; generate anomaly reports and trigger data re-collection or equipment calibration instructions; Call the lightweight emission factor library preset in the local edge computing node to calculate the process-level carbon emissions in real time according to the formula; upload the calculation results to the cloud collaborative processing module.

[0087] In some embodiments, the specific steps of using a temporal convolutional network model to detect mutation points and outliers in a data stream and calculate anomaly probabilities include: Normalize the latest time series data and input it into the trained time series convolutional network model; Capturing local patterns through causal convolution and long-term dependencies through dilated convolution; Output the abnormal probability of the center point of the current window.

[0088] The steps of calling the lightweight emission factor library preset in the local edge computing node and calculating the process-level carbon emissions in real time according to the formula include: Calculate the confidence level of process-level carbon emissions in combination with the abnormal probability; Upload high-confidence results that are higher than the set threshold directly to the cloud; Low-confidence data below the set threshold will trigger the dynamic verification module to conduct data review.

[0089] In some embodiments, the specific steps of performing dynamic accounting and multi-source data fusion include: Receive pre-processed data uploaded by the edge computing module; Link third-party data sources to build a multi-dimensional feature matrix; Call industry-specific accounting models to calculate total enterprise-level carbon emissions.

[0090] In some embodiments, the specific steps of the carbon accounting model module include: Segment enterprises by industry and build industry nodes; automatically parse business licenses through a deep learning OCR engine to match industry codes; and associate emission source inventories, monitoring equipment rules, and accounting boundaries; Obtain process flow diagrams through the enterprise DCS system to identify key carbon flow nodes; use graph segmentation algorithms to divide the production line into minimum accounting units; set unit weights based on the proportion of equipment energy consumption; Determine the required measurement parameters based on the industry knowledge base; screen decision variables through feature importance analysis; and use integer linear programming to solve the optimal monitoring point location.

[0091] In some embodiments, the specific steps of visualization and report generation include: Receive process-level real-time emission data and confidence labels uploaded by the edge computing module; Associate the enterprise-level accounting results output by the cloud collaborative processing module; Calling industry benchmark data of carbon accounting model module; Generate emission curves by hour / day / month; Use heat maps to show high-emission areas within the plant; Use Sankey diagrams to visualize the flow of carbon elements; Load preset templates based on the company's industry and automatically generate compliance reports that meet MRV requirements.

[0092] Although the present invention has been described in detail with reference to the accompanying drawings and in conjunction with preferred embodiments, the present invention is not limited thereto. Without departing from the spirit and essence of the present invention, persons of ordinary skill in the art may make various equivalent modifications or substitutions to the embodiments of the present invention, and such modifications or substitutions shall be within the scope of the present invention. Any changes or substitutions that can be easily conceived by persons skilled in the art within the technical scope disclosed in the present invention shall be within the scope of protection of the present invention.

Claims

1. An intelligent carbon emission monitoring system based on the Internet of Things, characterized in that: include: Distributed sensor network module for real-time collection of energy consumption and emission data of independent accounting units; The edge computing module is deployed on the local edge computing node and is used to pre-process the data collected by the distributed sensor network module, including data cleaning, anomaly detection and preliminary emission calculation; The cloud collaborative processing module is connected to the edge computing module to receive pre-processed data and perform dynamic calculations and multi-source data fusion; The carbon accounting model module has a built-in industry characteristic knowledge base and a differentiated accounting unit division algorithm, which is used to automatically match accounting requirements according to the industry to which the enterprise belongs and generate a monitoring point layout plan; The intelligent visualization module generates multi-dimensional carbon emission data analysis charts and compliance reports that meet MRV requirements based on data from the edge computing module, cloud collaborative processing module, and carbon accounting model module.

2. The carbon emission intelligent monitoring system based on the Internet of Things according to claim 1 is characterized in that: The local edge computing node is pre-installed with a lightweight emission factor library; The edge computing module includes a pre-processing unit, an anomaly detection unit, and a preliminary emission calculation unit; The preprocessing unit is used to convert the raw sensor data collected by the distributed sensor network module into structured time series data through the industrial protocol parsing engine; eliminate invalid data and fill missing values ​​based on sliding window statistics; normalize the data and unify the dimension and sampling frequency. The anomaly detection unit uses a temporal convolutional network model to detect mutation points and outliers in the data stream and calculate the anomaly probability to input into the preliminary emission calculation unit; combines the industry process rule library to mark logical anomaly data; generates anomaly reports and triggers data re-collection or equipment calibration instructions to the distributed sensor network module; The preliminary emission calculation unit calls the lightweight emission factor library preset in the local edge computing node, calculates the process-level carbon emissions in real time according to the formula, and uploads it to the cloud collaborative processing module.

3. The carbon emission intelligent monitoring system based on the Internet of Things according to claim 2 is characterized in that: The anomaly detection unit normalizes the latest time series data and inputs it into the trained time series convolutional network model. It captures local patterns through causal convolution and long-term dependencies through dilated convolution, and outputs the anomaly probability of the center point of the current window.

4. The carbon emission intelligent monitoring system based on the Internet of Things according to claim 3 is characterized in that: The system also includes a dynamic verification module; The preliminary emission calculation unit calculates the confidence level of process-level carbon emissions based on the probability of abnormality, and directly uploads high-confidence results that are higher than the set threshold to the cloud-based collaborative processing module, while low-confidence data that are lower than the set threshold triggers the dynamic verification module for data review.

5. The carbon emission intelligent monitoring system based on the Internet of Things according to claim 4 is characterized in that: The preliminary emission calculation unit calculates the data integrity score, sensor accuracy score, and process compliance score, combined with the anomaly probability calculated by the anomaly detection unit Calculate the confidence level of process-level carbon emissions; the formula is as follows: Data integrity score ; Sensor Accuracy Score ; Process Compliance Score ; Where, is the process-level carbon emissions, Activity data collected in real time by distributed sensor network modules, is the locally preset emission factor, is the process emission factor, is the weight coefficient.

6. The carbon emission intelligent monitoring system based on the Internet of Things according to claim 5 is characterized in that: The cloud-based collaborative processing module is specifically used to receive pre-processed data uploaded by the edge computing module; associate third-party data sources to build a multi-dimensional feature matrix; and call industry-specific accounting models to calculate the total enterprise-level carbon emissions according to the formula: Where, is the process weight factor, Emissions from purchased electricity / heat.

7. The carbon emission intelligent monitoring system based on the Internet of Things according to claim 6 is characterized in that: The carbon accounting model module includes a construction and matching unit, an accounting unit division algorithm unit, and a monitoring point layout plan generation unit; The construction and matching unit segments enterprises into several industry nodes. The deep learning OCR engine automatically parses business licenses and matches industry codes. Each industry node is associated with an emission source list, monitoring equipment rules, and accounting boundaries. The company's latitude and longitude coordinates and production scale are input to automatically adjust the density of monitoring points. The accounting unit division algorithm unit obtains the process flow chart through the enterprise DCS system, identifies the key carbon flow nodes, uses the graph segmentation algorithm to divide the production line into the smallest accounting units, and sets the unit weight according to the proportion of equipment energy consumption; Adjust accounting granularity in real time; The monitoring point layout plan generation unit determines the required measurement parameters based on the industry knowledge base; selects decision variables through feature importance analysis; and uses integer linear programming to solve the optimal monitoring point location; The objective function is: in, is the equipment installation cost, To cover the Type of emission source, is the decision variable.

8. The carbon emission intelligent monitoring system based on the Internet of Things according to claim 6 is characterized in that: The intelligent visualization module receives the process-level real-time emission data and confidence labels uploaded by the edge computing module; associates the enterprise-level accounting results output by the cloud collaborative processing module; calls the industry benchmark data of the carbon accounting model module; generates emission curves by hour / day / month; displays high-emission areas within the factory through heat maps; uses Sankey diagrams to visualize the flow of carbon elements; and automatically generates text descriptions by loading preset templates based on the industry to which the enterprise belongs.

9. A carbon emission intelligent monitoring method based on the Internet of Things, characterized in that: The following steps are involved: Real-time collection of energy consumption and emission data of independent accounting units through distributed sensor networks; Pre-processing of collected data in the edge computing module, including data cleaning, anomaly detection, and preliminary emission calculations; Upload the pre-processed data to the cloud collaborative processing module for dynamic accounting and multi-source data fusion; Based on the carbon accounting model module, it automatically matches accounting requirements according to the industry to which the enterprise belongs and generates a monitoring point layout plan; Based on the data from the edge computing module, cloud collaborative processing module and carbon accounting model module, multi-dimensional carbon emission data analysis charts and compliance reports that meet MRV requirements are generated.

10. The method for intelligent carbon emission monitoring based on the Internet of Things according to claim 9, characterized in that: The steps for preprocessing the collected data in the edge computing module include: The industrial protocol parsing engine converts raw sensor data into structured time series data; invalid data is eliminated and missing values ​​are filled based on sliding window statistics; the data is normalized to unify the dimension and sampling frequency; Use a time series convolutional network model to detect mutation points and outliers in data streams and calculate anomaly probabilities; combine industry process rule libraries to mark logically abnormal data; generate anomaly reports and trigger data re-collection or equipment calibration instructions; Call the lightweight emission factor library preset in the local edge computing node to calculate the process-level carbon emissions in real time according to the formula; upload the calculation results to the cloud collaborative processing module.

Images