Industrial park carbon source tracking detection method and system

Abstract

The invention belongs to the technical field of carbon emission monitoring, and particularly relates to an industrial park carbon source tracking detection method and system.A data acquisition layer is used for acquiring multi-source data related to carbon emission in an industrial park through various heterogeneous devices; the edge calculation layer is in communication connection with the data acquisition layer and is used for carrying out preprocessing and feature extraction on the multi-source data; the block chain layer is in communication connection with the edge computing layer and is used for carrying out credible storage and verification on the preprocessed data to form credible data which cannot be tampered; the central analysis layer is in communication connection with the block chain layer and is used for carrying out fusion, inversion and trend prediction analysis on the credible data to generate a carbon tracking result; and the application layer is in communication connection with the central analysis layer and is used for visually displaying the carbon tracking result and providing decision support. Through a multi-technology fusion mode, comprehensive, real-time and credible monitoring of carbon emission of the industrial park is realized, and a powerful technical support is provided for carbon management of the park.

Description

Technical Field

[0001] This invention belongs to the field of carbon emission monitoring technology, specifically a method and system for tracking and detecting carbon sources in industrial parks. Background Technology

[0002] The statements in this section merely refer to the background art related to this invention and do not necessarily constitute prior art.

[0003] Industrial parks are concentrated areas of industrial production, and their carbon emissions account for a large proportion of total carbon emissions. Current carbon source tracking and monitoring technologies for industrial parks have several shortcomings: Traditional methods heavily rely on energy consumption data (such as electricity and gas meter readings) reported by companies within industrial parks for indirect estimation, resulting in an extremely limited data dimension. This approach completely ignores the direct and indirect carbon emissions from multiple key aspects of the production process, including material consumption, chemical reactions, waste disposal, and transportation and logistics within the park. This leads to serious biases and prejudices in the final carbon assessment results. More critically, existing technologies lack effective mechanisms for verifying the authenticity of the data. Throughout the entire chain from collection and transmission to storage, data may be distorted or lost due to equipment failure, network interference, or human tampering, failing to provide a reliable data foundation for carbon accounting and supervision, thus constituting a fundamental obstacle to accurate carbon management.

[0004] Most existing carbon accounting systems aggregate and analyze data offline on a monthly or quarterly basis, which is a lagging "post-event accounting" model. This model cannot reflect the dynamic changes in carbon emissions in the industrial park, let alone cope with sudden carbon emission exceedances caused by production equipment failures, process anomalies, etc. By the time managers discover abnormal emissions, pollution has already occurred, missing the best opportunity for control and potentially causing unnecessary environmental consequences and economic losses. Summary of the Invention

[0005] To address the technical problems mentioned above, this invention provides a method and system for tracking and detecting carbon sources in industrial parks. By integrating multiple technologies, it achieves comprehensive, real-time, and reliable monitoring of carbon emissions in industrial parks, providing strong technical support for carbon management in these parks.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: The first aspect of the present invention provides a carbon source tracking and detection system for industrial parks, comprising: The data acquisition layer is used to collect multi-source data related to carbon emissions within the industrial park through various heterogeneous devices. The heterogeneous devices include at least multiple drones and ground sensor networks for collecting environmental data of the park, an IoT gateway for collecting energy consumption and emission data of enterprises in the park, and a transportation management unit for collecting vehicle activity data within the park. The edge computing layer communicates with the data acquisition layer and is used for preprocessing and feature extraction of multi-source data; The blockchain layer communicates and connects with the edge computing layer to perform trusted storage and verification of pre-processed data, forming tamper-proof and trustworthy data. The central analysis layer, which communicates and connects with the blockchain layer, is used to fuse, invert, and predict trends in trusted data to generate carbon tracking results. The application layer communicates with the central analysis layer and is used to visualize carbon tracking results and provide decision support.

[0007] As a further technical solution, the data acquisition layer includes: The drone monitoring unit consists of multiple drones, each equipped with at least a carbon dioxide sensor, a temperature and humidity sensor, a wind speed and direction sensor, and an image acquisition device, used to collect environmental data within a certain spatial range at a certain height above the ground. A ground-based sensor network, consisting of multiple sensors arranged in a grid, is used to collect environmental data at the ground level. The IoT gateway is configured to connect to the production management system of enterprises within the park and collect data on the enterprises' energy consumption, raw material consumption, and waste emissions. The transportation management unit is configured to collect activity data of transport vehicles within the park through positioning terminals and RFID devices. The power data interface unit is configured to acquire electricity purchase data and carbon emission factor data of enterprises within the park.

[0008] As a further technical solution, the drone monitoring unit is configured to periodically fly and scan along a preset route to collect environmental data of the park.

[0009] As a further technical solution, in the ground sensor network, the sensors in each grid include, but are not limited to, carbon dioxide sensors, temperature and humidity sensors, wind speed and direction sensors, atmospheric pressure sensors, particulate matter sensors, and volatile organic compound sensors.

[0010] As a further technical solution, the edge computing layer includes edge computing nodes deployed near the data source, and the edge computing nodes are configured as follows: The collected data is denoised based on wavelet transform and median filtering algorithms. Based on the convolutional neural network (CNN) model, deep features related to carbon emissions are extracted from the raw data to reduce the data dimensionality. Based on the lightweight image recognition model MobileNet, images captured by drones are analyzed to identify emission source types, locations, and abnormal states.

[0011] As a further technical solution, the blockchain layer specifically refers to a consortium blockchain network that adopts the PBFT consensus mechanism. The nodes in the consortium blockchain network include park management agencies, enterprises, and regulators.

[0012] As a further technical solution, the blockchain layer also includes: The smart contract module is configured to: verify the format legality, source authenticity and transmission integrity of the data to be stored, and only allow data that passes verification to be timestamped and broadcast for storage; The distributed storage module is configured to encrypt and store data on each node based on a hash algorithm, ensuring the immutability of the data.

[0013] As a further technical solution, the central analysis layer includes: The data fusion engine is configured to: assign weights to credible data from different sources and perform probability synthesis based on the DS evidence theory algorithm to construct a carbon flow model for the park; The emissions inversion module is configured to: dynamically adjust the model parameters based on the improved Gaussian plume model with the introduction of atmospheric turbulence correction factors and topography correction factors, so as to realize the inversion of the location and intensity of carbon emission sources. The trend prediction module is configured to use a Long Short-Term Memory (LSTM) network model trained on historical data to predict future carbon emission trends and identify key influencing factors.

[0014] As a further technical solution, the carbon tracking results are visualized, specifically by using a 3D visualization platform to dynamically display the distribution and changing trends of carbon sources on the 3D terrain of the park in the form of heat maps and / or bar charts.

[0015] A second aspect of the present invention provides a method for carbon source tracing and detection in industrial parks, comprising the following steps: Acquire environmental data, industrial process data, transportation activity data, and electricity trading data within the park to form a comprehensive carbon-related data pool; The collected data is preprocessed, features are extracted, and preliminary image analysis is performed, and the corresponding hash values ​​are calculated. The preprocessed data and its corresponding hash value are uploaded to the blockchain layer, and after verification, trusted data is obtained. The analysis results are obtained by fusing and analyzing reliable data, inverting emission sources, and predicting trends. The analysis results are visualized and can trigger alerts and provide decision support.

[0016] Compared with existing technologies, one or more of the above technical solutions have the following beneficial effects: 1. By deploying drone swarms, ground sensor networks, and IoT gateways, an integrated air-space-ground multi-source data acquisition system was constructed, simultaneously collecting environmental, production, transportation, and energy data. This changes the traditional approach of relying solely on energy consumption data reported by enterprises. Furthermore, blockchain technology is used to perform tamper-proof and reliable verification of the entire data chain, ensuring the authenticity and reliability of carbon data and providing a solid data foundation for precise carbon management.

[0017] 2. The collected data is processed in real time at the edge computing end, enabling the system to instantly acquire dynamic changes in carbon emissions within the park. Upon detecting sudden emissions exceeding standards due to equipment malfunctions or process anomalies, the system can immediately trigger a multi-level early warning mechanism, allowing managers to respond quickly and transforming "post-event accounting" into "in-process intervention," effectively preventing environmental accidents and economic losses.

[0018] 3. An improved Gaussian plume model that integrates atmospheric turbulence and terrain correction factors is adopted, and the model parameters are dynamically adjusted based on accurate meteorological data (such as wind speed and wind direction) collected in real time by the ground sensor network. This significantly improves the inversion accuracy of emission sources in complex industrial environments and enables rapid and accurate identification of abnormal emission sources.

[0019] 4. Based on artificial intelligence algorithms such as LSTM, the system performs in-depth mining and fusion analysis of multi-source reliable data, which can not only clearly show the carbon flow process in the park, but also predict the short-term carbon emission trend and automatically identify key influencing factors. The system can also generate quantifiable and actionable personalized emission reduction plans based on an expert knowledge base, providing managers with scientific decision support and driving the park to shift from passive control to proactive governance.

[0020] 5. The system adopts a modular design, and the IoT gateway supports multiple industrial protocols, enabling seamless integration with heterogeneous management systems within the park and breaking down data silos. Through a collaborative model of "drone patrol + ground sensor fixed-point monitoring" and edge computing data processing, it effectively reduces overall deployment and maintenance costs while ensuring high accuracy and wide coverage, making large-scale application and promotion possible. Attached Figure Description

[0021] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0022] Figure 1 This is a schematic diagram of a carbon source tracing and detection process provided in one or more embodiments of the present invention; Figure 2 A schematic diagram of the data flow between the data acquisition layer and the edge computing layer provided in one or more embodiments of the present invention; Figure 3 A schematic diagram of the data flow in the central analysis layer provided for one or more embodiments of the present invention; Figure 4 This is a schematic diagram of the application layer provided for one or more embodiments of the present invention. Detailed Implementation

[0023] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0024] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0025] The relevant terms are explained below: Carbon source tracking and monitoring refers to the process of using technological means to monitor, identify, quantify, and trace the location, intensity, timing, and responsible parties of carbon emission sources within an industrial park.

[0026] Multi-source data fusion refers to the technology of collaboratively processing heterogeneous data from different sources such as environmental sensors, industrial control systems, and logistics and transportation systems to form a more comprehensive and accurate understanding of carbon emissions.

[0027] Edge computing refers to providing near-end computing, storage, and application services at the network edge, close to the data source, in order to reduce data transmission latency and alleviate pressure on the cloud.

[0028] Trusted blockchain storage: This refers to using the decentralized, immutable, and traceable characteristics of blockchain technology to encrypt, store, and verify collected carbon emission data, ensuring its authenticity and reliability.

[0029] As described in the background section, current carbon source tracing and detection technologies face many problems that urgently need to be addressed: 1. Limited Data Dimensions: Traditional carbon source tracking methods often rely on only a single type of data, such as simply relying on energy consumption data reported by enterprises to estimate carbon emissions. This approach ignores carbon emissions generated in multiple stages of the production process, such as material consumption, waste disposal, and transportation within the industrial park. This results in a serious bias in the overall assessment of the park's carbon emissions and fails to fully reflect the actual carbon emission situation of the park.

[0030] 2. Poor real-time performance: Most existing carbon accounting systems collect and analyze data on a monthly or quarterly basis. This lagging approach makes it impossible for park managers to keep abreast of dynamic changes in carbon emissions. When sudden abnormal carbon emissions occur, such as excessive emissions due to equipment failure, the inability to take effective countermeasures quickly may result in serious environmental impacts and economic losses.

[0031] 3. Low data reliability: There is a lack of effective supervision and verification mechanisms during data collection, transmission, and storage. Some companies may falsify or conceal carbon emission data for their own benefit; sensor equipment is prone to failure in harsh environments, leading to data distortion; data loss or tampering may also occur during data transmission due to network problems, all of which seriously affect the reliability of carbon source analysis results.

[0032] 4. Cost-benefit imbalance: Some high-precision carbon monitoring technologies, such as satellite remote sensing-based methods, can acquire carbon emission data over a large area, but the equipment purchase cost is high, and data processing is complex and expensive, making it unaffordable for most small and medium-sized industrial parks. On the other hand, low-cost monitoring methods often suffer from insufficient accuracy and limited coverage, making it difficult to meet actual monitoring needs.

[0033] 5. Insufficient technological compatibility: Significant differences exist between the production equipment and management systems of various enterprises, making it difficult for existing carbon monitoring systems to effectively interact and integrate with these heterogeneous systems. This results in severe data silos, hindering unified management and analysis of carbon emission data within the industrial park and impacting the overall efficiency of carbon management.

[0034] 6. Lack of intelligent decision support: Most current carbon monitoring systems can only collect and display data, lacking in-depth data analysis and intelligent decision support capabilities. When faced with massive amounts of carbon emission data, park managers struggle to quickly identify key issues and potential risks, making it difficult to formulate scientific and reasonable emission reduction strategies.

[0035] This embodiment presents a method and system for carbon source tracking and detection in industrial parks, constructing a collaborative carbon tracking system integrating "edge-chain-cloud". Data is collected from multiple sources, including drones and sensors, preprocessed using edge computing, and its reliability is ensured by blockchain technology. Finally, the data is integrated, analyzed, and intelligently predicted on a central platform. Through technological integration and cross-validation of data, the system addresses the problems of incomplete, outdated, and unreliable data in traditional methods, achieving real-time, accurate, and reliable traceability and intelligent management of carbon emissions.

[0036] Example 1: like Figure 1 As shown, the carbon source tracking and detection method for industrial parks includes the following steps: S1: Multi-source data collaborative acquisition. By deploying drone swarms, ground sensor networks, IoT gateways, and transportation management systems, environmental data (CO2 concentration, meteorological parameters), industrial process data (energy consumption, raw materials, waste discharge), and transportation activity data (location, fuel consumption) within the park are collected in real time. External electricity trading data is also obtained through API interfaces to form a comprehensive carbon-related data pool. S2: Edge-side data preprocessing and feature extraction. Near the data source, edge computing nodes are used to denoise and filter the collected raw data, and deep learning models are applied for image recognition and feature extraction to compress data volume, preliminarily identify emission sources and abnormal states, and provide high-quality input for upper-level analysis. S3: Blockchain-based trusted data storage and verification. Preprocessed data and its hash value are uploaded to the consortium blockchain network. Smart contracts automatically verify the data format, source, and integrity. After verification, a timestamp is added and the data is distributed for storage, ensuring the data's immutability and traceability. S4: Central System Multi-Source Data Fusion and In-Depth Analysis. In the central system, methods such as DS evidence theory are used to fuse reliable data from multiple sources to construct a carbon flow model for the industrial park; an improved Gaussian plume model is used for emission source inversion and tracing; and an LSTM model is used to learn historical data and predict future carbon emission trends. S5: Visualization and Intelligent Decision Support. The analysis results are dynamically displayed through a 3D visualization platform, showing the carbon source distribution in real time. The system triggers early warnings based on multi-level thresholds and automatically generates and recommends precise emission reduction strategies and implementation plans based on expert systems and case reasoning.

[0037] S1. Multi-source data collaborative acquisition.

[0038] (1) Obtain environmental data.

[0039] A swarm of 10-20 drones is used, each equipped with a high-precision carbon dioxide sensor with a measurement accuracy of ±2ppm, a temperature and humidity sensor with an accuracy of ±0.5℃ and ±2%RH, and a wind speed and direction sensor with a measurement range of 0-30m / s and 0-360°. The drones' flight paths are planned using 3D modeling software to ensure a monitoring point every 50 meters. The flight altitude is dynamically adjusted based on the height of buildings in the park, generally between 50 and 100 meters. Each drone completes a full park scan every 10 minutes, and the collected data is transmitted in real-time to a ground receiving station via 4G / 5G networks with a transmission delay of no more than 1 second.

[0040] The ground sensor network is distributed in a grid pattern, with one monitoring point set up every 200 square meters within the park, totaling 200-500 ground sensors. The sensors adopt an industrial-grade design, possessing waterproof, dustproof, and anti-interference capabilities, and an operating temperature range of -30℃ to 70℃. Data collected by the sensors is transmitted to the IoT gateway via wired Ethernet or LoRa wireless communication technology, with a data transmission interval of 1 minute to ensure real-time data transmission. The collected environmental data is aggregated and preliminarily processed by the IoT gateway before being uploaded to the edge computing nodes via wired communication.

[0041] (2) Obtain industrial and transportation data.

[0042] The IoT gateway collects real-time energy consumption data (such as electricity, steam, and natural gas) from the enterprise's production process once per second; raw material usage data (such as various chemical raw materials and metal materials) is collected once per hour; and emission data (such as the amount and composition of waste gas and wastewater) from each production stage is collected once every 30 minutes. The IoT gateway uses industrial-grade embedded devices and supports multiple industrial communication protocols, such as Modbus, OPCUA, and Profinet, enabling seamless integration with MES and EMS systems from different manufacturers.

[0043] The system integrates BeiDou positioning with vehicle-mounted terminals to access the transportation vehicle management system, acquiring real-time data such as vehicle location (accuracy of 1 meter), speed, engine RPM, and fuel consumption every 10 seconds. Simultaneously, RFID technology is used to identify vehicles entering and leaving the park, recording information such as cargo load and type of goods.

[0044] (3) Obtain power data.

[0045] Companies in the park typically use the power trading platform to purchase electricity. They use API interfaces to obtain real-time power trading contract data, including purchase time, purchase volume, power source (such as thermal power, hydropower, wind power, etc.), carbon emission factors, etc. The data is updated once a day.

[0046] S2, edge-side data preprocessing and feature extraction.

[0047] (1) Data denoising and feature extraction.

[0048] Data denoising employs a combination of algorithms. For environmental data acquired by sensors, wavelet transform denoising is used to effectively remove high-frequency and random noise. For image data acquired by UAVs, median filtering is used to remove salt-and-pepper noise. Through these algorithms, the signal-to-noise ratio of the data can be improved by more than 30%.

[0049] Feature extraction employs an automatic feature extraction method based on deep learning, utilizing convolutional neural networks (CNNs) to extract deep features related to carbon emissions from the raw data. For example, features such as equipment operating status and production load are extracted from energy consumption data in the production process; vehicle driving mode and load characteristics are extracted from transportation data. The extracted feature dimensions are reduced by more than 60% compared to the original data, significantly reducing the amount of data transmission.

[0050] (2) Preliminary image analysis.

[0051] The MobileNet model deployed on edge computing nodes has undergone specialized training and optimization, and has been trained on a large number of samples targeting common emission sources in industrial parks (such as chimneys, cooling towers, and transport vehicles). The model's recognition accuracy can reach over 95%. For images captured by drones, the model can complete the analysis of a single image within 0.5 seconds, identifying the type, location (accuracy of 5 meters), and emission status (normal or abnormal) of the emission source. For abnormal emission statuses, such as abnormal smoke concentration from chimneys, the model can preliminarily determine the severity and mark them as targets requiring priority attention.

[0052] S3. Trusted data storage and verification based on blockchain.

[0053] (1) Distributed storage.

[0054] The blockchain network adopts a consortium blockchain architecture, consisting of 10-20 nodes including the park management center, enterprise nodes, third-party testing agency nodes, and government regulatory nodes. Each node is equipped with a high-performance server with a storage capacity of no less than 10TB, ensuring the storage of a large amount of historical data. It employs the PBFT consensus mechanism, with a consensus node count of 3f+1 (where f is the number of fault-tolerant nodes), and a consensus time of no more than 5 seconds, enabling rapid data uploading and storage. When data is written to the blockchain, a unique hash value is generated using the SHA-256 hash algorithm and stored along with the data content, timestamp, and node information, ensuring data immutability.

[0055] (2) Smart contract verification.

[0056] The smart contract is written in Solidity and includes multiple modules such as data format verification, data source verification, and data integrity verification. The data format verification module validates the data according to a preset JSON format to ensure the integrity and correctness of data fields. The data source verification module verifies the legitimacy of the data sending node through a digital certificate. The data integrity verification module verifies whether the data has been tampered with during transmission through hash value comparison. Only data that passes all verification steps is allowed to be written to the blockchain, automatically triggering timestamp addition and network broadcast operations. The smart contract's execution time is less than one second, ensuring efficient data processing.

[0057] S4. Multi-source data fusion and in-depth analysis.

[0058] (1) Multi-source data fusion.

[0059] In applying the DS evidence theory fusion algorithm, the credibility of data from different sources is first assessed, and different weights are assigned based on factors such as data collection accuracy and reliability. For example, environmental monitoring data has a weight of 0.3, production energy consumption data has a weight of 0.4, transportation data has a weight of 0.2, and electricity data has a weight of 0.1. Then, the probability allocation functions of each data point are synthesized using evidence combination rules to obtain a comprehensive carbon emission assessment result. This fusion algorithm improves the data fusion accuracy by more than 25% compared to single data sources, and the constructed carbon flow model clearly demonstrates the processes of carbon generation (e.g., fuel combustion in production processes), transfer (e.g., carbon brought in by raw material transportation), and consumption (e.g., carbon carried away by products) within the industrial park.

[0060] (2) Emissions inversion optimization.

[0061] The improved Gaussian plume model incorporates atmospheric turbulence and topographic correction factors, enabling it to better adapt to the complex terrain and meteorological conditions of industrial parks. The atmospheric stability grading response mechanism classifies atmospheric stability into six levels (A, B, C) based on real-time monitored meteorological parameters such as wind speed, wind direction, and temperature gradient, with each level corresponding to different model parameter settings. For example, under stable atmospheric conditions (levels E and F), diffusion parameters are increased; under unstable atmospheric conditions (levels A and B), diffusion parameters are decreased. Through these optimizations, the model's inversion error for carbon emission source location is controlled within 10 meters, and the inversion error for emission intensity is controlled within 8%, effectively reducing the impact of interference factors such as aerosols and topographic obstruction.

[0062] (3) Trend prediction.

[0063] The LSTM machine learning model employs a multi-layer neural network structure, including an input layer, hidden layers (3-5 layers), and an output layer, with 128-256 nodes in the hidden layers. The model is trained using historical carbon emission data, production data, and meteorological data from the past 5 years as training samples. Parameters are optimized using the Adam optimization algorithm, and the training iterations exceed 1000 times to ensure the model's prediction accuracy. The model can predict carbon emission trends for the next 1-7 days with a prediction error of no more than 10%. During the prediction process, the model automatically identifies key factors affecting carbon emissions, such as adjustments to production plans leading to abrupt changes in carbon emissions, and changes in energy policies leading to trend changes in carbon emissions, and annotates these factors in the prediction results.

[0064] S5, Visualization and Intelligent Decision Support.

[0065] (1) Visual presentation.

[0066] The visualization platform is built with WebGL technology to construct a three-dimensional visualization interface, which can display the three-dimensional terrain model and carbon source distribution of the industrial park in real time. The distribution of carbon sources is presented in the form of a heat map, where the darker the color, the greater the carbon emission intensity; the carbon emission intensities of each enterprise and production link are dynamically updated in the form of bar charts, line charts, etc., with an update frequency of once per minute. Users can perform mouse interaction operations to zoom in, zoom out, and rotate the three-dimensional model to view the carbon emission detail information at any location. At the same time, the platform supports the data export function, and can export the analysis results in formats such as Excel and PDF, facilitating users for further analysis and report compilation.

[0067] (2) Intelligent warning and decision support.

[0068] The system sets multiple levels of carbon emission thresholds, including warning thresholds, alarm thresholds, and emergency thresholds, corresponding to different response measures respectively. When the carbon emission data exceeds the warning threshold, the system gives a prompt through a platform pop-up window; when it exceeds the alarm threshold, it sends text messages and email warnings to relevant personnel; when it exceeds the emergency threshold, in addition to text messages and email warnings, it can also make a call to the responsible person for reminder according to the pre-saved contact information. The intelligent decision-making module, based on expert system and case-based reasoning technologies, provides specific emission reduction suggestions for enterprises according to the analysis results and historical emission reduction cases. For example, for carbon emission over-standard caused by production processes, it is recommended to adjust process parameters, replace high-efficiency equipment, etc.; for carbon emission over-standard in the transportation link, it is recommended to optimize transportation routes, adopt new energy vehicles, etc.

[0069] This solution integrates multi-source data collection methods such as drone swarms, ground sensor networks, Internet of Things gateways, and transportation management systems, breaks through the limitation of traditional methods that only rely on single energy consumption data, covers multi-dimensional carbon-related data in environment, production, transportation, electricity, etc., and constructs a complete carbon flow model for the park, significantly improving the comprehensiveness and accuracy of carbon source assessment.

[0070] This solution realizes near-real-time monitoring of the park's carbon emission status and rapid preliminary identification of abnormal states through high-frequency data collection (from seconds to minutes) and edge computing proximal processing, enabling managers to discover and respond to sudden emission events in a timely manner, fundamentally changing the lagging mode of traditional monthly / quarterly accounting.

[0071] This solution introduces alliance blockchain and smart contract technologies, establishes a full-chain credible deposit and verification mechanism for data from collection, transmission to storage, effectively prevents data tampering and loss, and provides an immutable and traceable reliable data basis for carbon accounting, supervision, and trading.

[0072] This scheme employs an improved Gaussian plume model that integrates atmospheric turbulence and topographic correction factors, and dynamically adjusts parameters in conjunction with real-time meteorological parameters. This significantly reduces interference in complex environments and achieves high-precision inversion and source tracing of emission source location (error <10 meters) and intensity (error <8%).

[0073] This solution uses deep learning models such as LSTM to mine and analyze massive amounts of historical and real-time data, achieving high-precision prediction of future short-term (1-7 days) carbon emission trends (error <10%), and can automatically identify key influencing factors, providing a scientific basis for managers to formulate forward-looking emission reduction strategies.

[0074] This solution uses WebGL technology to dynamically and intuitively display the carbon emissions of the park in three-dimensional space, and sets up multi-level thresholds to trigger differentiated early warnings (platform prompts, SMS emails, and telephone notifications), which greatly improves the intuitiveness, interactivity and emergency response efficiency of carbon management.

[0075] This solution, through a collaborative deployment model of "drones + ground sensors" and edge computing data preprocessing, effectively reduces the overall deployment and operation costs of the system while ensuring monitoring accuracy and coverage, making it particularly suitable for large-scale, complex park applications.

[0076] This solution supports multiple industrial protocols such as Modbus and OPC UA, enabling seamless integration with heterogeneous enterprise management systems (MES / EMS) within the park, breaking down data silos. The system adopts a modular design, with each layer (data acquisition, edge computing, blockchain, and analytics) capable of independent expansion and upgrades as needed, and allowing for easy access to new data sources or analytical algorithms in the future.

[0077] Based on expert systems and case-based reasoning, this solution not only provides emission reduction recommendations but also quantifies and predicts the effectiveness of these recommendations (with an error of <15%), offering enterprises a practical and cost-effective low-carbon transformation path and enhancing the feasibility of emission reduction measures.

[0078] This solution provides park managers with a comprehensive carbon management platform that integrates monitoring, analysis, early warning, and decision-making. It not only significantly improves management efficiency but also helps promote overall energy conservation and emission reduction in the park, achieving green, low-carbon, and sustainable development.

[0079] This solution will be further explained using carbon tracking in a chemical industrial park as an example.

[0080] 1. Data Acquisition: (1) Unmanned Aerial Vehicle (UAV) Monitoring: Five advanced UAVs were deployed, each equipped with a high-precision carbon dioxide sensor, temperature and humidity sensor, wind speed and direction sensor, and a high-definition camera. The flight routes of the UAVs were carefully planned according to the terrain and production layout of the chemical industrial park to ensure comprehensive coverage of all areas of the park. The UAVs scanned the park every 15 minutes according to the preset route, collected atmospheric environmental data and captured image data in real time, and transmitted the data to the ground receiving station through wireless communication technology.

[0081] (2) Ground sensor network: Twenty ground sensors, including carbon dioxide sensors, temperature and humidity sensors, wind speed and direction sensors, and air pressure sensors, are evenly installed in the chemical industrial park. The environmental data collected by these sensors is aggregated and preliminarily processed through an IoT gateway, and then uploaded to the edge computing node via wired communication.

[0082] (3) Enterprise Data Access: Using IoT technology, the MES system, EMS system and data acquisition system of the three chemical companies in the park are seamlessly connected. Real-time data collection is performed on the raw material usage, production process parameters, energy consumption data and emissions data of each production stage. At the same time, the data is accessed through the enterprise transportation management system to obtain data such as the driving routes of transport vehicles, the types and weights of transported goods, and vehicle energy consumption.

[0083] (4) Power data access: Cooperate with the power grid company to access the power consumption data of each workshop in the automobile manufacturing park. At the same time, obtain data on the use of green electricity.

[0084] 2. Edge computing processing: (1) Data Preprocessing: The edge computing nodes use NVIDIA Jetson Nano devices and deploy advanced data preprocessing algorithms. The image data transmitted from the UAV is analyzed in real time using the lightweight deep learning model MobileNet to quickly identify emission sources such as chimneys and reactors in the images and to preliminarily determine whether the emission status is normal. At the same time, the environmental data collected by the sensors is subjected to Kalman filtering noise reduction to remove abnormal data caused by factors such as equipment noise and electromagnetic interference, thereby improving the accuracy of the data.

[0085] (2) Feature extraction: Using feature extraction algorithms, key features closely related to carbon emissions are extracted from massive amounts of raw data, such as the energy consumption characteristics of specific production processes and the emission characteristics of different transportation methods. The data after noise reduction and feature extraction are uploaded to the blockchain layer and the central analysis layer.

[0086] 3. Blockchain-based evidence storage: (1)Blockchain network construction: The Hyperledger Fabric consortium chain technology is adopted, and the chemical industrial park management party, 3 chemical enterprises, and a third-party certification agency participate in the blockchain network as nodes. The consensus mechanism is used among the nodes to ensure the consistency and immutability of data.

[0087] (2)Smart contract execution: Write smart contracts to automatically verify the data uploaded to the blockchain. The smart contracts quickly verify the data according to preset rules such as data format and data source legality. Only the verified data will be timestamped and broadcast in the blockchain network to ensure the authenticity and reliability of the data entering the subsequent analysis link.

[0088] 4. Central system analysis: (1)Multi-source data fusion: In the central system, the D-S evidence theory multi-source data fusion algorithm is used to deeply fuse the environmental data collected by drones, ground sensor data, enterprise production data, and transportation data. By establishing an association model between the data, a complete carbon flow model of the chemical industrial park is constructed, clearly showing the generation, transfer, and consumption processes of carbon in the park.

[0089] (2)Emission inversion optimization: Based on the improved Gaussian plume model and combined with the real-time atmospheric stability classification response mechanism, accurate inversion of carbon dioxide emission sources is carried out. By real-time monitoring of atmospheric stability parameters, the relevant parameters of the Gaussian plume model are dynamically adjusted to effectively reduce the influence of interference factors such as aerosols on the inversion results. For example, at a certain moment, through the inversion algorithm, an abnormal emission caused by a valve leakage in a reactor of a certain enterprise is accurately identified, and the position and intensity of the emission source are accurately calculated.

[0090] (3)Trend prediction: The LSTM machine learning model is used to deeply mine and analyze the fused data. The model takes into account various factors such as the production plan adjustment of chemical enterprises, changes in raw material supply, and changes in energy policies, and predicts the carbon emission trend in the chemical industrial park within the next 2 hours. The prediction results show that within the next 2 hours, due to a certain enterprise's plan to increase production load, the overall carbon emissions in the park will increase, providing a basis for the park managers to formulate response measures in advance.

[0091] 5. Feedback and decision-making: (1)Visual display and early warning: The central system pushes the analysis results to the chemical industrial park management platform in real time through a visual interface. On the visual interface, the distribution of carbon sources in the park is displayed in an intuitive map form, and information such as the carbon emission intensity and carbon emission trend changes of each enterprise and production link is presented in a chart form. When it is detected that the carbon emissions increase abnormally due to a valve leakage in a reactor of a certain enterprise, the system automatically triggers the early warning mechanism and sends alarms to the park managers and enterprise responsible persons via text messages, emails, etc.

[0092] (2) Decision Support: Based on the analysis results and preset emission reduction strategies, the intelligent decision-making module provides enterprises with targeted suggestions for adjusting their production plans. For example, it suggests that enterprises immediately stop production of the leaking reactor and arrange maintenance personnel for emergency repairs. At the same time, it provides enterprises with long-term emission reduction suggestions such as optimizing production processes and adjusting energy structures to help them reduce carbon emissions and achieve green production. Based on the analysis results, the park management platform adjusts its energy dispatch strategy, increases the proportion of renewable energy use, reduces dependence on fossil fuels, and further reduces the overall carbon emissions of the park.

[0093] Example 2: The industrial park carbon source tracking and monitoring system includes: The historical data acquisition module is configured as follows: 1. Data Acquisition Layer.

[0094] (1) Unmanned Aerial Vehicle (UAV) Monitoring Unit: The UAV is a quadcopter industrial-grade UAV with a flight time of no less than 60 minutes and a maximum flight radius of 5 kilometers. In addition to high-precision carbon dioxide sensors, temperature and humidity sensors, and wind speed and direction sensors, it also includes a high-definition camera (resolution of no less than 20 million pixels) and an infrared thermal imager, which can perform monitoring at night or in low visibility conditions. The UAV ground control station adopts an industrial-grade tablet computer, which supports functions such as flight path planning, real-time data viewing, and remote control, and can simultaneously control multiple UAVs to work together.

[0095] (2) Ground sensor network: The sensors adopt a modular design, which can flexibly combine different types of sensors according to actual needs. The IoT gateway adopts a redundant design, with dual network ports and dual wireless modules to ensure the reliability of data transmission. The gateway supports edge computing functions, which can perform preliminary aggregation and preprocessing of sensor data, such as data format conversion and simple outlier removal, to reduce the pressure on subsequent processing links.

[0096] (3) Enterprise Data Access Gateway: The gateway uses a high-performance industrial computer equipped with multiple Ethernet interfaces, serial ports, and wireless communication modules, supporting simultaneous access to multiple enterprise systems. The gateway has a built-in data encryption module to encrypt the collected data, ensuring data security during transmission. Simultaneously, the gateway has a resume capability, storing the collected data after a network interruption and automatically uploading it once the network is restored.

[0097] 2. Edge computing layer.

[0098] Edge computing nodes: In addition to NVIDIA Jetson Nano, other edge computing devices such as Intel Neural Compute Stick 2 can be selected, configured according to actual computing power requirements. The operating system deployed on the edge computing nodes is Linux, supporting Docker container technology for convenient deployment and updates of algorithm models. The nodes have local storage capabilities, capable of storing preprocessed data for the past 7 days, ensuring no data loss in the event of communication interruption with the upper-layer system. Simultaneously, the nodes support remote management and maintenance, allowing administrators to remotely view node operating status and update algorithm models via the network.

[0099] In this embodiment, the data flow between the data acquisition layer and the edge computing layer is as follows: Figure 2 As shown.

[0100] 3. Blockchain layer.

[0101] (1) Blockchain Nodes: The node servers adopt high-performance x86 architecture servers, equipped with multi-core processors, large-capacity memory and hard disks, and the operating system is Ubuntu Server. Nodes communicate with each other through a dedicated network with a network bandwidth of no less than 100Mbps to ensure efficient data transmission. Each node is equipped with a Hardware Security Module (HSM) to store the node's private key and digital certificate, thereby improving the node's security.

[0102] (2) Smart Contract Engine: The Ethereum Virtual Machine (EVM) is used as the smart contract engine, supporting smart contracts written in Solidity. The engine has high-performance contract execution capabilities, capable of processing more than 100 contract calls per second. At the same time, the engine supports contract upgrade and pause functions. When contract vulnerabilities are discovered, they can be repaired and paused in a timely manner to ensure the secure and stable operation of the blockchain system.

[0103] 4. Central Analysis Layer.

[0104] (1) Data Fusion Engine: Adopting a distributed computing architecture, built on the Hadoop and Spark platforms, it can process massive amounts of multi-source data in parallel. The engine has a built-in multi-source data fusion algorithm library, which includes Bayesian estimation, Kalman filtering, and other algorithms in addition to DS evidence theory. Users can choose the appropriate algorithm according to their actual needs. The data fusion engine has a processing capacity of no less than 100GB / h, ensuring that it can process a large amount of real-time data in a timely manner.

[0105] (2) Emissions Inversion Module: The module is developed using C++ and has high-efficiency computing power, capable of completing the carbon emission source inversion calculation for a medium-sized industrial park (10 square kilometers) within 1 minute. The module has a built-in meteorological data interface, which can acquire meteorological data of the park and its surroundings in real time and dynamically adjust the parameters of the inversion model based on the meteorological data. At the same time, the module supports custom inversion accuracy and range to meet the needs of different scenarios.

[0106] (3) Machine Learning Model Platform: Built on the TensorFlow and PyTorch deep learning frameworks, it supports the training, deployment, and inference of various machine learning models. The platform has model management functions, which can perform version control, performance evaluation, and updates of models. At the same time, the platform supports distributed training, which can utilize the computing power of multiple computing nodes to accelerate the model training process, reducing the training time by more than 50% compared to single-machine training.

[0107] (4) Visualization Platform: Adopting a B / S architecture, users can access it through a browser without installing dedicated software. The platform supports multi-terminal adaptation, including computers, tablets, and mobile phones, making it convenient for users to view the park's carbon emission information anytime, anywhere. The platform has a rich built-in visualization component library, including not only 3D models and common charts, but also dynamic data flow diagrams, geographic information maps, etc., which can intuitively display the analysis results in various forms.

[0108] In this embodiment, the data flow of the central analysis layer is as follows: Figure 3 As shown.

[0109] 5. Application layer.

[0110] (1) Park Management Application: In addition to basic carbon emission monitoring and early warning functions, it also has functions such as park carbon quota management, emission reduction task allocation, and emission reduction effect evaluation. It can decompose emission reduction tasks to each enterprise according to the overall carbon emission reduction target of the park and track the completion status of emission reduction tasks in real time. By comparing carbon emission data of different periods, it can evaluate the implementation effect of emission reduction measures and provide park managers with a scientific basis for decision-making.

[0111] (2) Enterprise Application: Provides enterprises with self-service query and analysis functions for carbon emission data. Enterprises can view the details of their carbon emissions in each production stage and compare them with the industry average. At the same time, the application also provides personalized emission reduction suggestions and cost-benefit analysis to help enterprises choose the optimal emission reduction plan. For example, for a certain high-energy-consuming equipment, the application will calculate the investment cost of replacing the equipment, the annual carbon emission reduction, and the investment payback period, etc., for enterprises to refer to in their decision-making.

[0112] In this embodiment, the application layer is as follows: Figure 4 As shown.

[0113] This solution will be further explained using the carbon management optimization of a certain automobile manufacturing park as an example.

[0114] 1. Data Acquisition: (1) Drone exhaust monitoring: The drone swarm is equipped with sensors specifically designed to monitor vehicle exhaust emissions, and monitors the exhaust emissions of transport vehicles in the park in real time. Through image recognition technology, the license plate numbers of high-emission vehicles are identified, and the relevant data is transmitted to the edge computing node.

[0115] (2) Power data access: Obtain power consumption data of each workshop in the automobile manufacturing park. At the same time, obtain data on the use of green electricity.

[0116] (3) Enterprise Data Access: Utilizing IoT technology, the MES and EMS systems of multiple automobile manufacturing enterprises within the park are seamlessly connected to the data acquisition system. Real-time data collection is performed on raw material usage, production process parameters, energy consumption, and emissions data from each production stage. Simultaneously, the system is integrated with the enterprise transportation management system to obtain data such as vehicle routes, types and weights of transported goods, and vehicle energy consumption.

[0117] (4) Supply chain data access: By connecting with the supplier's information system, information such as the source of raw materials, transportation method, and carbon emission factors can be obtained to track carbon emissions in the supply chain.

[0118] 2. Edge computing processing: (1) Data preprocessing: The edge computing nodes use Intel NeuralComputeStick2 devices and deploy advanced data preprocessing algorithms. Feature extraction and classification are performed on the exhaust emission data transmitted from the UAV to identify different types of pollutants and their concentrations. At the same time, data cleaning and standardization are performed on the power and environmental data collected by the sensors to improve data quality.

[0119] (2) Feature extraction: Using deep learning algorithms, key features related to carbon emissions are extracted from massive amounts of raw data, such as vehicle mileage, engine operating conditions, and energy consumption characteristics during raw material processing. The preprocessed and feature-extracted data is then uploaded to the blockchain layer and the central analysis layer.

[0120] 3. Blockchain-based evidence storage: (1) Blockchain network construction: The Ethereum consortium blockchain technology is adopted, with the management of the automobile manufacturing park, multiple automobile manufacturing companies, suppliers and third-party testing agencies participating as nodes in the blockchain network. The consensus mechanism among the nodes ensures the consistency and immutability of the data.

[0121] (2) Smart Contract Execution: Write a smart contract to automatically verify the data uploaded to the blockchain. The smart contract quickly verifies the data according to preset rules such as data format and data source legality. Only data that passes verification will be timestamped and broadcast in the blockchain network to ensure that the data entering the subsequent analysis stage is authentic and reliable.

[0122] 4. Central System Analysis: (1) Multi-source data fusion: In the central system, a Bayesian estimation multi-source data fusion algorithm is used to deeply integrate exhaust emission data, electricity consumption data, enterprise production data, supply chain data, etc. collected by UAVs. By establishing a correlation model between the data, a complete carbon flow model of the automobile manufacturing park is constructed, clearly showing the generation, transfer and consumption process of carbon in the park.

[0123] (2) Emission Inversion Optimization: Based on the improved Gaussian plume model and combined with the real-time atmospheric stability classification response mechanism, the carbon dioxide emission sources are accurately inverted. By monitoring atmospheric stability parameters in real time and dynamically adjusting the relevant parameters of the Gaussian plume model, the influence of interference factors such as aerosols on the inversion results is effectively reduced. For example, at a certain moment, the inversion algorithm accurately identifies the abnormal emissions caused by the failure of the waste gas treatment equipment in the painting workshop of a certain enterprise, and accurately calculates the location and intensity of the emission source.

[0124] (3) Trend Prediction: A deep neural network model was used to conduct in-depth mining and analysis of the fused data. The model considered various factors such as production plan adjustments by automobile manufacturing enterprises, changes in raw material supply, changes in energy policies, and market demand to predict the carbon emission trend of the automobile manufacturing park in the next three days. The prediction results show that in the next three days, due to increased market demand and enterprises' plans to expand production scale, the overall carbon emissions of the park will increase significantly, providing a basis for park managers to formulate countermeasures in advance.

[0125] 5. Feedback and Decision-Making: (1) Visualization and Early Warning: The central system pushes the analysis results to the automotive manufacturing park management platform in real time through a visual interface. The visual interface displays the distribution of carbon sources in the park in an intuitive map form, and presents information such as the carbon emission intensity and carbon emission trend changes of each enterprise and production process in chart form. When a malfunction of the exhaust gas treatment equipment in a certain enterprise's painting workshop is detected, which leads to an abnormal increase in carbon emissions, the system automatically triggers the early warning mechanism and sends alarms to park managers and enterprise leaders through SMS, email and other means.

[0126] (2) Decision Support: Based on the analysis results and preset emission reduction strategies, the intelligent decision-making module provides enterprises with targeted suggestions for adjusting their production plans. For example, it suggests that enterprises suspend production in some painting workshops and prioritize the maintenance of waste gas treatment equipment. At the same time, it provides enterprises with long-term emission reduction suggestions such as optimizing production processes and strengthening supply chain management to help them reduce carbon emissions and achieve green production. Based on the analysis results, the park management platform adjusts its energy dispatch strategy, increases the proportion of renewable energy use, reduces dependence on fossil fuels, and further reduces the overall carbon emissions of the park.

[0127] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An industrial park carbon source tracking and detection system, characterized in that, include: The data acquisition layer is used to collect multi-source data related to carbon emissions within the industrial park through various heterogeneous devices. The heterogeneous equipment includes at least multiple drones and a ground sensor network for collecting park environmental data, an IoT gateway for collecting energy consumption and emission data of enterprises in the park, and a transportation management unit for collecting vehicle activity data within the park. An edge computing layer, which is communicatively connected to the data acquisition layer, is used for preprocessing and feature extraction of multi-source data; The blockchain layer, which communicates with the edge computing layer, is used to perform trusted storage and verification of preprocessed data, forming tamper-proof and trustworthy data. The central analysis layer, which communicates with the blockchain layer, is used to fuse, invert, and predict trends in trusted data to generate carbon tracking results. The application layer communicates with the central analysis layer and is used to visualize carbon tracing results and provide decision support.

2. The industrial park carbon source tracking and detection system as described in claim 1, characterized in that, The data acquisition layer includes: The drone monitoring unit consists of multiple drones, each equipped with at least a carbon dioxide sensor, a temperature and humidity sensor, a wind speed and direction sensor, and an image acquisition device, used to collect environmental data within a certain spatial range at a certain height above the ground. A ground-based sensor network, consisting of multiple sensors arranged in a grid, is used to collect environmental data at the ground level. The IoT gateway is configured to connect to the production management system of enterprises within the park and collect data on the enterprises' energy consumption, raw material consumption, and waste emissions. The transportation management unit is configured to collect activity data of transport vehicles within the park through positioning terminals and RFID devices. The power data interface unit is configured to acquire electricity purchase data and carbon emission factor data of enterprises within the park.

3. The industrial park carbon source tracking and detection system as described in claim 2, characterized in that, The drone monitoring unit is configured to perform periodic flight scans along a preset route to collect environmental data of the park.

4. The industrial park carbon source tracking and detection system as described in claim 2, characterized in that, In the ground sensor network, the sensors in each grid include, but are not limited to, carbon dioxide sensors, temperature and humidity sensors, wind speed and direction sensors, atmospheric pressure sensors, particulate matter sensors, and volatile organic compound sensors.

5. The industrial park carbon source tracking and detection system as described in claim 1, characterized in that, The edge computing layer includes edge computing nodes deployed near the data source, and the edge computing nodes are configured as follows: The collected data is denoised based on wavelet transform and median filtering algorithms. Based on the convolutional neural network (CNN) model, deep features related to carbon emissions are extracted from the raw data to reduce the data dimensionality. Based on the lightweight image recognition model MobileNet, images captured by drones are analyzed to identify emission source types, locations, and abnormal states.

6. The industrial park carbon source tracking and detection system as described in claim 1, characterized in that, The blockchain layer specifically refers to a consortium blockchain network that adopts the PBFT consensus mechanism. The nodes in the consortium blockchain network include park management agencies, enterprises, and regulators.

7. The industrial park carbon source tracking and detection system as described in claim 6, characterized in that, The blockchain layer also includes: The smart contract module is configured to: verify the format legality, source authenticity and transmission integrity of the data to be stored, and only allow data that passes verification to be timestamped and broadcast for storage; The distributed storage module is configured to encrypt and store data on each node based on a hash algorithm, ensuring the immutability of the data.

8. The industrial park carbon source tracking and detection system as described in claim 1, characterized in that, The central analysis layer includes: The data fusion engine is configured to: assign weights to credible data from different sources and perform probability synthesis based on the DS evidence theory algorithm to construct a carbon flow model for the park; The emissions inversion module is configured to: dynamically adjust the model parameters based on the improved Gaussian plume model with the introduction of atmospheric turbulence correction factors and topography correction factors, so as to realize the inversion of the location and intensity of carbon emission sources. The trend prediction module is configured to use a Long Short-Term Memory (LSTM) network model trained on historical data to predict future carbon emission trends and identify key influencing factors.

9. The industrial park carbon source tracking and detection system as described in claim 1, characterized in that, The carbon tracking results are visualized by using a 3D visualization platform to dynamically display the distribution and changing trends of carbon sources on the 3D terrain of the park in the form of heat maps and / or bar charts.

10. A method for tracing and detecting carbon sources in industrial parks, implemented using the industrial park carbon source tracing and detection system described in any one of claims 1-9, characterized in that, Includes the following steps: Acquire environmental data, industrial process data, transportation activity data, and electricity trading data within the park to form a comprehensive carbon-related data pool; The collected data is preprocessed, features are extracted, and preliminary image analysis is performed, and the corresponding hash values ​​are calculated. The preprocessed data and its corresponding hash value are uploaded to the blockchain layer, and after verification, trusted data is obtained. The analysis results are obtained by fusing and analyzing reliable data, inverting emission sources, and predicting trends. The analysis results are visualized and can trigger alerts and provide decision support.

Images