A multi-source carbon data intelligent management and control method and system for a zero-carbon park
By constructing a multi-dimensional sensing network and dynamic correction coefficients to improve the accuracy of carbon data collection and establishing a multi-dimensional evaluation system, the problems of single data collection and insufficient calculation accuracy in the existing carbon emission control of industrial parks have been solved, and efficient zero-carbon transformation management of industrial parks has been achieved.
Patent Information
- Application Number
- CN202610222371.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-02-25
- Publication Date
- 2026-07-14
- Estimated Expiration
- 2046-02-25
Smart Images

Figure CN121724287B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of zero-carbon park management technology, specifically to a method and system for intelligent management and control of multi-source carbon data for zero-carbon parks. Background Technology
[0002] Carbon emissions refer to the total amount of greenhouse gases emitted into the atmosphere by human activities or natural processes. These greenhouse gases are primarily carbon dioxide, but also include methane, nitrous oxide, and others. In calculations, other greenhouse gases are usually converted into carbon dioxide equivalents using the "global warming potential" for unified measurement. The significance of studying carbon emissions extends far beyond addressing climate change itself; it profoundly impacts economic development, technological innovation, international competition, and individual lives. Addressing the climate crisis: Carbon emissions are the primary cause of global warming, and research on them is the scientific cornerstone for understanding and mitigating disasters such as sea-level rise and extreme weather events. Reshaping the global development paradigm: This drives humanity from "black growth" reliant on fossil fuels to a sustainable, green growth model—a fundamental transformation of civilization. Reducing dependence on imported fossil fuels and developing domestic renewable energy sources is crucial for long-term national strategic security. Carbon costs are becoming internalized as core costs for businesses. Research can help companies mitigate policy and market risks and identify business opportunities in new technologies and markets. Building brands and attracting investment: Effective carbon emission management has become a core indicator of corporate social responsibility and sustainable development capabilities, significantly enhancing brand value and attracting green investment. Guiding the direction of technological innovation: From carbon capture to green hydrogen energy, carbon emission research has pointed out the most urgent needs and directions for cutting-edge technology research and development.
[0003] The existing carbon emission control technologies in industrial parks have the following shortcomings: First, carbon data collection methods are limited, relying heavily on traditional manual statistics or single-sensor monitoring, making it difficult to achieve comprehensive integration of multi-source carbon data and easily leading to data omissions or duplicate statistics; Second, the accuracy of carbon emission calculations is insufficient, with traditional methods often using fixed emission factors for calculation, failing to fully consider the impact of dynamic factors such as the production processes of different enterprises within the park, equipment operating status, and meteorological conditions on carbon emissions, resulting in significant deviations in calculation results; Third, the evaluation system is incomplete, with existing evaluations focusing primarily on total carbon emissions, neglecting multi-dimensional indicators such as carbon emission intensity, emission reduction potential, and carbon sink contribution, making it difficult to comprehensively reflect the low-carbon development level of the industrial park. Summary of the Invention
[0004] To address the aforementioned technical problems, this invention provides a multi-source carbon data intelligent management and control method for zero-carbon parks. This method achieves comprehensive and accurate collection of multi-source carbon data, dynamic high-precision calculation, multi-dimensional comprehensive evaluation, and intelligent emission reduction decision support, providing core technical support for the construction and operation of zero-carbon parks. The multi-source carbon data intelligent management and control method for zero-carbon parks includes the following steps:
[0005] S1: Based on the classification of carbon emission sources, a multi-dimensional sensing network is constructed to collect multi-source carbon-related data in real time; and the data is preprocessed to obtain a carbon dataset.
[0006] S2. Based on the carbon data set and combined with the emission source type, the carbon emission assessment data of the park is calculated.
[0007] S3. Construct a multi-dimensional carbon emission evaluation index system for zero-carbon parks using carbon emission evaluation data from the parks, calculate the comprehensive weight of each evaluation index, and combine the values of each evaluation index to calculate the comprehensive evaluation score.
[0008] S4. Determine whether the comprehensive evaluation score meets the qualification standard. If yes, no emission reduction plan will be generated; otherwise, proceed to S5.
[0009] S5. Combining the park's zero-carbon development goals and carbon emission assessment data, construct an intelligent emission reduction decision-making model and generate emission reduction plans; monitor the implementation effect of emission reduction plans in real time, forming a closed-loop management and control system of "collection-calculation-evaluation-decision-feedback".
[0010] Preferably, the types of carbon emission sources include: stationary emission sources, mobile emission sources, building energy sources, waste treatment sources, and carbon sink sources, etc.
[0011] Preferred stationary emission sources include: industrial kilns, boilers, internal combustion engines, flare systems, and generator sets, etc.
[0012] Preferred mobile emission sources include: freight vehicles, commuter vehicles, and engineering vehicles within the park.
[0013] Preferred energy sources for buildings include: electricity, gas, and heat consumption in office buildings, factories, and dormitories.
[0014] Preferred sources of waste disposal include landfills and wastewater treatment plants.
[0015] Preferred carbon sink sources include: green spaces, wetlands, and carbon sink forests within the park.
[0016] Preferred methods for multi-source carbon-related data preprocessing include: data cleaning, data standardization, and data fusion.
[0017] Preferred methods include: data cleaning using outlier detection algorithms to remove outlier data and linear interpolation to fill in missing data; data standardization using Z-Score standardization to convert data of different dimensions into a unified standard; and data fusion using a weighted average fusion algorithm to fuse data collected from multiple devices from the same emission source to improve data reliability.
[0018] Preferred carbon emission assessment data includes: total carbon emissions, carbon emissions from each emission source, carbon flow distribution data, total carbon emission control, carbon emission intensity, emission reduction potential, carbon sink contribution, and low-carbon operation, etc.
[0019] Preferred emission sources include: stationary emission sources, mobile emission sources, building energy, waste treatment sources, and carbon sink sources.
[0020] Preferred method for obtaining carbon emissions from each emission source includes: calculating basic carbon emissions based on activity data of different emission sources and corresponding baseline emission factors; introducing a dynamic correction coefficient, which is obtained by training a gradient boosting tree algorithm using production process parameters, equipment operating efficiency, meteorological conditions, and park operating conditions; and correcting the basic carbon emissions using the dynamic correction coefficient to obtain high-precision carbon emissions from each emission source.
[0021] The preferred method for calculating dynamic correction coefficients is as follows: 1. Data acquisition and feature selection. 2. Data preprocessing. 3. Model training and optimization. 4. Correction coefficient generation.
[0022] Preferred method: Data collection and feature selection includes: collecting at least 3 years of historical data from the park, including core characteristic variables affecting carbon emissions: production process parameters, equipment operation data, meteorological data, and park operation conditions; and collecting baseline carbon emissions and measured carbon emissions for the corresponding time periods as label data for model training.
[0023] Preferred data preprocessing includes: cleaning the collected feature variable data and removing outliers; filling missing data with K-nearest neighbor interpolation; screening feature variables that are significantly correlated with carbon emissions through Pearson correlation analysis, and finally determining the model input feature set.
[0024] Preferred model training and optimization include: using the selected feature variables as input and the measured deviation ratio as output, where the measured deviation ratio is (measured carbon emissions - baseline carbon emissions) / baseline carbon emissions, constructing a gradient boosting tree model; using a grid search method to optimize the model hyperparameters; dividing the preprocessed dataset into a training set and a test set in a 7:3 ratio, training the model with the training set, and verifying the model accuracy with the test set, ensuring that the model prediction error RMSE ≤ 0.02.
[0025] Preferred method for generating correction coefficients includes: inputting real-time collected feature variables into a trained gradient boosting tree model to obtain a correction coefficient deviation value Δk; dynamic correction coefficient k = 1 + Δk, where the value range of k is constrained to 0.95~1.05.
[0026] Preferably, the multi-dimensional zero-carbon park carbon emission evaluation index system includes a target layer, a criterion layer, and an index layer.
[0027] Preferred criteria include: total carbon emission control indicators, carbon emission intensity indicators, emission reduction potential indicators, carbon sink contribution indicators, and low-carbon operation indicators.
[0028] The preferred indicators include specific indicators such as the park's annual carbon emissions, carbon emissions per unit of GDP, carbon emissions per unit of industrial output, emission reduction potential of key industries, annual carbon absorption of carbon sink forests, proportion of renewable energy, and green transportation modal share.
[0029] Preferred methods for obtaining comprehensive weights include: 1. Defining the hierarchy of the evaluation indicator system; each criterion level corresponds to one core indicator level. 2. Determining subjective weights using the analytic hierarchy process (AHP). 3. Determining objective weights using the entropy weight method. 4. Using a weighted average method to integrate subjective and objective weights to obtain combined weights, which are then normalized to obtain the comprehensive weights of each evaluation indicator.
[0030] Preferred method for obtaining comprehensive evaluation score includes: standardizing using different formulas to map the index values to the [0, 1] interval, thereby obtaining standardized annual index X'. jk The formula for calculating the comprehensive carbon emission evaluation score of the industrial park is: S=Σ(X' jk ×W j ), where S is the comprehensive evaluation score, X' jk Let W be the standardized annual value of the j-th indicator in the k-th year. j is the comprehensive weight of the j-th indicator.
[0031] Preferred methods for obtaining the comprehensive evaluation score include: comprehensive evaluation score Where k is the year number, K is the total number of years for which data was collected, k = 1, 2, ..., K; j is the number of indicator layers, J is the total number of indicator layers, j = 1, 2, ..., J; X' jk x is the standardized annual value of the j-th indicator in the k-th year; jk ' is the target value for the j-th indicator layer in year k, x k W is the unit reference value for the deviation of the k-th indicator layer. j is the comprehensive weight of the j-th indicator.
[0032] Preferred: Deviation unit reference quantity , where X m It is the maximum value of historical data or the data with the maximum allowable deviation, and β is the number of stratification levels. This method of calculation can quickly narrow down the deviation of indicator data, avoiding data distortion caused by its interference with other data, and making the calculation more accurate.
[0033] The preferred criteria layer consists of five indicators: M1 - total carbon emission control, M2 - carbon emission intensity, M3 - emission reduction potential, M4 - carbon sink contribution, and M5 - low-carbon operation.
[0034] The preferred method for determining subjective weights using the analytic hierarchy process includes: 1. constructing a judgment matrix; 2. calculating subjective weights; and 3. performing a consistency check.
[0035] Preferred method for constructing the judgment matrix includes: inviting 5 experts in the field of low carbon to compare the importance of the 5 indicators of the criterion layer in pairs according to the 1-9 scale method, obtaining the expert scoring matrix, and taking the average value as the final judgment matrix M.
[0036] The preferred method for obtaining subjective weights includes: Step 1: Normalizing the elements of each column of the judgment matrix to obtain a normalized matrix N. Step 2: Summing the elements of each row of the normalized matrix N to obtain a row sum vector S. Step 3: Normalizing the row sum vector to obtain the subjective weight W = [S1 / ΣS, S2 / ΣS, ..., S5 / ΣS].
[0037] The preferred method for consistency verification includes: Step 1: Calculate the product MW of the judgment matrix M and the subjective weight vector W; Step 2: Calculate the largest eigenvalue λ = (1 / 5) × Σ[(MW) / W]; Step 3: Calculate the consistency index CI = (λ-n) / (n-1) (n=5); Find the average random consistency index RI, calculate the consistency ratio CR = CI / RI, and determine whether the consistency ratio CR is less than a preset threshold. If it is, it indicates that the judgment matrix has good consistency and the subjective weights are effective. If not, the subjective weights are invalid.
[0038] Preferably, the intelligent emission reduction decision-making model is constructed based on a reinforcement learning algorithm, aiming to minimize the park's carbon emissions and optimize emission reduction costs. Input parameters include park carbon emission assessment data, assessment indicator scores, emission reduction technology costs, renewable energy supply capacity, and carbon sink potential. The model outputs emission reduction solutions including source reduction measures, process control measures, carbon sink enhancement measures, and carbon offsetting measures. It tracks carbon data changes in real time during the implementation of the emission reduction plan, and regularly updates carbon emission calculation results and comprehensive evaluation results. If the evaluation results do not meet the expected targets, the emission reduction plan is adjusted through the intelligent emission reduction decision-making model to achieve dynamic closed-loop management.
[0039] Preferred input parameters include carbon emission data from each emission source, evaluation scores, photovoltaic power plant installed capacity, CCUS technology costs, etc.
[0040] A multi-source carbon data intelligent management and control system for zero-carbon industrial parks includes a multi-source carbon data acquisition module, a data preprocessing module, a carbon emission calculation module, a comprehensive evaluation module, an intelligent decision-making and management module, and a data storage and visualization module. These modules are connected sequentially to form a closed-loop management and control architecture.
[0041] Multi-source carbon data acquisition module: used to build a multi-dimensional sensing network and collect multi-source carbon-related data from different types of carbon emission sources in the zero-carbon park.
[0042] Data preprocessing module: Used to preprocess multi-source carbon-related data to obtain a standardized carbon dataset.
[0043] Carbon emission calculation module: Used to calculate the total carbon emissions of the park, the carbon emissions of each emission source, and the carbon flow distribution data based on the carbon data set and the emission source type.
[0044] Comprehensive evaluation module: Used to construct a multi-dimensional carbon emission evaluation index system for zero-carbon parks, calculate the comprehensive weight of each evaluation index, and combine the numerical values of each evaluation index to obtain the comprehensive evaluation score.
[0045] Intelligent decision-making and control module: It is used to determine the comprehensive evaluation score, and based on the carbon emission calculation results and comprehensive evaluation results, it generates emission reduction plans through intelligent emission reduction decision model, and realizes real-time tracking of emission reduction effects and dynamic adjustment of plans.
[0046] Data storage and visualization module: Used to store multi-source carbon data, calculation results, evaluation results and emission reduction scheme data, and display them through a visualization interface.
[0047] The technical effects and advantages of this invention are as follows:
[0048] 1. Achieve comprehensive and accurate collection of multi-source carbon data: By constructing a multi-dimensional sensing network, covering all types of emission sources and carbon sinks in the park, including stationary sources, mobile sources, and building energy, and combining multi-device data collection and multi-system data docking, the comprehensiveness of carbon data collection is ensured; data preprocessing algorithms are used to remove abnormal data and integrate multi-source data to improve data reliability.
[0049] 2. Improve the accuracy of carbon emission calculation: Construct a dynamic carbon emission calculation model and introduce dynamic correction coefficients based on production processes, equipment operation, and meteorological conditions to overcome the limitations of the traditional fixed emission factor method; adopt differentiated calculation strategies for different emission sources, and compare and calibrate with real-time monitoring data and statistical data to achieve high-precision carbon emission calculation, with the calculation error controllable within ±2%.
[0050] 3. Construct a multi-dimensional comprehensive evaluation system: Break through the limitations of traditional single total quantity evaluation, construct an evaluation index system from multiple dimensions such as total quantity control, intensity, emission reduction potential, and carbon sink contribution, and adopt a weight model combining the analytic hierarchy process and entropy weight method to take into account both subjective experience and objective data, thereby improving the scientificity and comprehensiveness of the evaluation results.
[0051] 4. Achieve intelligent closed-loop management and control throughout the entire process: Construct a closed-loop management architecture of "collection-computation-evaluation-decision-feedback", generate targeted emission reduction plans based on reinforcement learning algorithms, track the implementation effect in real time and make dynamic adjustments, provide accurate decision support for the operation and management of zero-carbon parks, and effectively improve the efficiency of zero-carbon transformation of parks. Attached Figure Description
[0052] Figure 1 This is a flowchart illustrating a multi-source carbon data intelligent management and control method for zero-carbon industrial parks proposed in this invention.
[0053] Figure 2 This is a flowchart illustrating the method for obtaining the carbon emission amounts of each emission source in a multi-source carbon data intelligent management and control method for zero-carbon parks proposed in this invention.
[0054] Figure 3 This is a flowchart illustrating the dynamic correction coefficient calculation method in a multi-source carbon data intelligent management and control method for zero-carbon parks proposed in this invention.
[0055] Figure 4 This is a flowchart illustrating the data preprocessing method in a multi-source carbon data intelligent management and control method for zero-carbon parks proposed in this invention.
[0056] Figure 5 This is a flowchart illustrating the model training and optimization method in a multi-source carbon data intelligent management and control method for zero-carbon parks proposed in this invention.
[0057] Figure 6 This is a flowchart illustrating the comprehensive weight acquisition method in the multi-source carbon data intelligent management and control method for zero-carbon parks proposed in this invention.
[0058] Figure 7 This is a flowchart illustrating the subjective weight acquisition method in a multi-source carbon data intelligent management and control method for zero-carbon parks proposed in this invention.
[0059] Figure 8 This is a flowchart illustrating the consistency verification method in a multi-source carbon data intelligent management and control method for zero-carbon parks proposed in this invention.
[0060] Figure 9 This is a structural block diagram of a multi-source carbon data intelligent management and control system for zero-carbon industrial parks proposed in this invention. Detailed Implementation
[0061] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the invention, and should not be construed as limiting the invention. Rather, embodiments of the invention include all variations, modifications, and equivalents falling within the spirit and scope of the appended claims.
[0062] Example 1
[0063] refer to Figure 1 This embodiment proposes a multi-source carbon data intelligent management and control method for zero-carbon parks, realizing comprehensive and accurate collection of multi-source carbon data, dynamic high-precision calculation, multi-dimensional comprehensive evaluation, and intelligent emission reduction decision support, providing core technical support for the construction and operation of zero-carbon parks. The multi-source carbon data intelligent management and control method for zero-carbon parks includes the following steps:
[0064] S1: Based on the classification of carbon emission sources, a multi-dimensional sensing network is constructed to collect multi-source carbon-related data in real time; and the data is preprocessed to obtain a carbon dataset. Traditional classification methods generally include: energy activities, industrial processes, agricultural activities, forestry, and waste treatment. This classification is based on industry sectors, facilitating data reading and classification for decision-makers. However, for data processing, a more computationally efficient classification method is used. The acquired data on energy activities, industrial processes, agricultural activities, forestry, and waste treatment can be reclassified in the background before being read. Specifically, the carbon emission source types can include: stationary emission sources, mobile emission sources, building energy consumption, waste treatment sources, and carbon sink sources. Stationary emission sources can include: industrial kilns, boilers, generator sets, etc.; mobile emission sources can include: freight vehicles, commuter vehicles, engineering vehicles, etc. within the park; building energy consumption can include: electricity, gas, and heat consumption in office buildings, factories, and dormitories; waste treatment sources can include: landfills, sewage treatment plants, etc.; carbon sink sources can include: green spaces, wetlands, and carbon sink forests within the park. The multi-dimensional sensing network includes online monitoring equipment deployed at fixed emission sources, vehicle-mounted monitoring terminals deployed at mobile emission sources, smart metering equipment deployed at buildings and public facilities, gas sensors and environmental monitoring equipment deployed at waste treatment sources, and carbon sink monitoring equipment deployed at carbon sink sources. Simultaneously, it acquires production process parameters, energy consumption statistics, and meteorological data in real time through the park's enterprise production management system, energy management platform, and meteorological monitoring station. Online monitoring equipment may include: a Continuous Emission Monitoring System (CEMS) and flow meters; vehicle-mounted monitoring terminals may include: OBD interface data acquisition modules, GPS positioning modules, etc.; smart metering equipment may include: smart meters, gas meters, heat meters, etc.; carbon sink monitoring equipment may include: vegetation cover sensors, soil carbon content monitors, etc.; and meteorological data may include: temperature, humidity, wind speed, and wind direction, etc. For example, CEMS and flow meters are installed at stationary emission sources such as kilns / boilers of three coal-fired power plants and five chemical enterprises in a park to collect real-time concentrations and emission flow rates of gases such as CO2 and CH4; vehicle-mounted monitoring terminals are installed on 80 freight vehicles and 20 commuter vehicles in the park to collect mileage, fuel consumption, speed and location data; smart meters, gas meters and heat meters are installed in 15 office buildings and 10 factory buildings in the park to collect energy consumption data; methane sensors are installed at the park's landfill and sewage treatment plant to collect greenhouse gas data during waste treatment; vegetation cover sensors and soil carbon content monitors are deployed in the park's 200-acre carbon sink forest and 50-acre wetland to collect carbon sink data; and the park's enterprise ERP system, energy management platform and local weather station are connected through MPI interface to obtain production process parameters, energy statistics and meteorological data such as temperature, humidity, and wind speed.Through the deployed sensing network and data interfaces, data is collected every minute for stationary sources, every five minutes for mobile sources, every 15 minutes for building energy consumption, and every hour for carbon sequestration and meteorological data. This is just a simple example and may not be universally applicable; other scenarios will not be elaborated upon here. Preprocessing of multi-source carbon-related data can include data cleaning, data standardization, and data fusion. Data cleaning uses outlier detection algorithms (isolation forest algorithm) to remove abnormal data and linear interpolation to fill in missing data. Data standardization uses the Z-Score standardization method to convert data of different dimensions into a unified standard. Data fusion uses a weighted average fusion algorithm to merge data collected from multiple devices from the same emission source, improving data reliability. For example, the isolated forest algorithm is used to identify and remove outliers (such as concentration spikes caused by equipment malfunctions) from CEMS monitoring data; linear interpolation is used to fill in missing energy consumption data within 10%; the Z-Score standardization method is used to convert data of different dimensions such as concentration (mg / m³), flow rate (m³ / h), and energy consumption (kWh) into standardized data; a weighted average fusion algorithm (weights determined based on equipment calibration accuracy) is used to fuse the monitoring data of two CEMS systems from the same coal-fired power plant to obtain fused concentration data. These data are then aggregated to form a carbon dataset.
[0065] S2. Based on the carbon dataset and combined with emission source types, carbon emission assessment data for the industrial park is calculated. This data may include: total carbon emissions, carbon emissions from each emission source, carbon flow distribution data, total carbon emission control, carbon emission intensity, emission reduction potential, carbon sink contribution, and low-carbon operation, etc. The core quantitative indicator for total carbon emission control is the park's annual carbon emissions; the core quantitative indicator for carbon emission intensity is carbon emissions per unit of GDP; the core quantitative indicator for emission reduction potential is the emission reduction space of key industries; the core quantitative indicator for carbon sink contribution is the annual carbon absorption of carbon sink forests; and the core quantitative indicator for low-carbon operation is the proportion of renewable energy. Emission source sub-items may include: stationary emission sources, mobile emission sources, building energy consumption, waste treatment sources, and carbon sink sources. (Reference) Figure 2Specific methods for obtaining this information may include: calculating basic carbon emissions based on activity data from different emission sources (such as fuel consumption, product output, mileage, and energy consumption) and corresponding baseline emission factors (referencing emission factors in the IPCC inventory guidelines and national carbon emission accounting standards). A dynamic correction coefficient is introduced, which is obtained by training production process parameters (such as kiln temperature and pressure), equipment operating efficiency, meteorological conditions, and park operating conditions using a gradient boosting tree algorithm. This dynamic correction coefficient is used to correct the basic carbon emissions, resulting in highly accurate carbon emissions for each emission source. Differentiated calculation strategies are employed for different emission sources: For stationary emission sources, the actual emissions are calculated using the "concentration-flow-time" method, combining real-time gas concentration and flow data monitored by CEMS, and then compared and calibrated with the results from the basic calculation module. For mobile emission sources, carbon emissions are calculated based on mileage, fuel consumption, and speed data collected by vehicle terminals, combined with emission factors corresponding to the vehicle type and dynamic correction coefficients. For building energy consumption, carbon emissions are calculated based on energy consumption data collected by smart metering devices, combined with emission factors corresponding to the energy type. For waste treatment sources, greenhouse gas emissions such as methane and nitrous oxide are calculated based on waste treatment volume, sewage treatment volume, and corresponding emission factors, and then converted into carbon dioxide equivalents. For carbon sink sources, carbon absorption is calculated based on carbon sink monitoring data and included as negative carbon emission data in the total carbon emission accounting of the park. The core of the dynamic correction coefficient is to construct a prediction model by integrating multi-dimensional influencing factors through the gradient boosting tree algorithm, thereby achieving dynamic calibration of the baseline emission factor. (Reference) Figure 3 The specific steps are as follows: 1. Data Collection and Feature Screening. Collect at least three years of historical data from the park, including core characteristic variables affecting carbon emissions: production process parameters (such as kiln temperature T, furnace pressure P, reaction time t), equipment operation data (such as equipment load rate η, unit product energy consumption E), meteorological data (such as ambient temperature T, humidity RH, wind speed v), and park operating conditions (such as production shifts, capacity utilization rate μ); simultaneously collect baseline carbon emissions (calculated based on the "activity data-emission factor" method) and measured carbon emissions (calculated based on data collected by online monitoring equipment such as CEMS) for the corresponding period as label data for model training. (Reference) Figure 4 2. Data Preprocessing. The collected feature variable data were cleaned, and outliers were removed (using the 3σ criterion); missing data were imputed using K-nearest neighbor interpolation; Pearson correlation analysis was used to screen feature variables significantly correlated with carbon emissions (screening threshold |r|≥0.3), ultimately determining the model input feature set. (Reference) Figure 53. Model Training and Optimization. Using the selected feature variables from the input feature set as input, and "(measured carbon emissions - baseline carbon emissions) / baseline carbon emissions" as output (i.e., the correction coefficient deviation value), a gradient boosting tree model is constructed. The model hyperparameters are optimized using a grid search method (learning rate 0.1, tree depth 5, number of decision trees 100). The preprocessed dataset is divided into training and testing sets in a 7:3 ratio. The model is trained using the training set and validated using the testing set, ensuring that the model prediction error RMSE ≤ 0.02. 4. Correction Coefficient Generation. The real-time collected feature variables are input into the trained gradient boosting tree model to obtain the correction coefficient deviation value Δk. The dynamic correction coefficient k = 1 + Δk, where the value of k is constrained to a range of 0.95~1.05 (if the model output exceeds this range, the boundary value is used), ensuring a reasonable correction magnitude and avoiding abnormal corrections. For example, for a fixed emission source (taking a coal-fired boiler as an example): Q1 = fuel consumption × fuel emission factor. A coal-fired boiler consumes 500 tons of raw coal per month, with a raw coal emission factor of 2.62 tCO2 / t coal and a dynamic correction coefficient k = 1.02. Therefore, the monthly carbon emissions from this boiler are Q1 = 500 × 2.62 × 1.02 = 1336.2 tCO2. Mobile emission sources (taking freight vehicles as an example): Q2 = mileage × emission factor per unit mileage. Example: 10 diesel freight vehicles in the park, each with an average monthly mileage of 3000 km, a diesel vehicle emission factor of 0.2 kg CO2 / km, and a dynamic correction coefficient k = 0.98. Therefore, the monthly carbon emissions from mobile emission sources are Q2 = 10 × 3000 × 0.2 × 10⁻³ × 0.98 = 5.88 tCO2. Building energy consumption (taking electricity consumption as an example): Q3 = electricity consumption × electricity emission factor. The park's 15 office buildings consume an average of 80,000 kWh of electricity per month. The regional power grid's emission factor is 0.58 tCO2 / MWh, and the dynamic correction coefficient k=1.00. Therefore, the monthly carbon emissions from building energy consumption are Q3=80000×10⁻³×0.58×1.00=46.4 tCO2. Waste treatment sources (taking landfill as an example): Q4=waste treatment volume×landfill emission factor. The park's landfill processes 300 tons of municipal solid waste per month. The municipal solid waste landfill emission factor is 0.3 tCO2 / t waste, and the dynamic correction coefficient k=1.01. Therefore, the monthly carbon emissions from waste treatment are Q4=300×0.3×1.01=90.9 tCO2. Carbon sink sources (taking carbon sink forests as an example): Carbon sink volume is a negative emission, Q5=carbon sink forest area×carbon sink volume per unit area. Example: A 200-mu carbon sink forest in the park has an annual carbon sink of 2 tCO2 / mu·year. The average monthly carbon sink Q5 = 200 × 2 ÷ 12 ≈ 33.33 tCO2 / month (i.e., -33.33 tCO2 / month). The formula for calculating the total carbon emissions of the park is: Total carbon emissions of the park. Where i is the emission source sub-item number, I is the total number of emission source sub-items, and its value can be 5, Qi The carbon emissions from emission source i are calculated as follows: Based on the emissions from each emission source, the total monthly carbon emissions of the park are Q = (1336.2 + 5.88 + 46.4 + 90.9) + (-33.33) = 1445.05 tCO2 / month. Carbon flow distribution is expressed as the proportion of carbon emissions from each emission source to the total carbon emissions (excluding carbon sinks), used to identify the core sources of carbon emissions. Calculation formula: Carbon flow proportion of a certain emission source R i =Q i / Q'×100%, where Q' is the sum of emissions from all emission sources excluding carbon sinks. The total emissions from the above-mentioned park excluding carbon sinks are Q'=1336.2+5.88+46.4+90.9=1479.38tCO2 / month. Therefore, the carbon flow share of stationary emission sources is R1=(1336.2 / 1479.38)×100%≈90.3%, the share of mobile emission sources is R2=(5.88 / 1479.38)×100%≈0.4%, the share of building energy consumption is R3=(46.4 / 1479.38)×100%≈3.1%, and the share of waste treatment is R4=(90.9 / 1479.38)×100%≈6.2%. The core of the carbon flow distribution is stationary emission sources.
[0066] S3. Construct a multi-dimensional carbon emission evaluation index system for zero-carbon parks using carbon emission assessment data, calculate the comprehensive weight of each evaluation index, and obtain a comprehensive evaluation score by combining the numerical values of each evaluation index. Specifically, the multi-dimensional carbon emission evaluation index system for zero-carbon parks includes a target layer, a criterion layer, and an indicator layer. The target layer represents the comprehensive evaluation level of carbon emissions in the zero-carbon park; the criterion layer includes total carbon emission control indicators, carbon emission intensity indicators, emission reduction potential indicators, carbon sink contribution indicators, and low-carbon operation indicators; the indicator layer includes specific indicators such as the park's annual carbon emissions, carbon emissions per unit of GDP, carbon emissions per unit of industrial output, emission reduction space in key industries, annual carbon absorption of carbon sink forests, the proportion of renewable energy, and the green transportation modal share. For example, refer to... Figure 6 The comprehensive weighting method can include: 1. Defining the hierarchy of the evaluation indicator system (taking the criteria layer and corresponding indicator layer as an example). Five indicators are selected for the criteria layer: M1 - Total carbon emission control, M2 - Carbon emission intensity, M3 - Emission reduction potential, M4 - Carbon sink contribution, and M5 - Low-carbon operation; each criteria layer corresponds to one core indicator layer indicator (simplified calculation example). 2. Determining subjective weights using the analytic hierarchy process (AHP), specifically including: 1. Constructing a judgment matrix: Inviting five experts in the low-carbon field to compare the importance of the five indicators in the criteria layer pairwise using a 1-9 scale (1 indicates that two indicators are equally important, and 9 indicates that one indicator is extremely important than the other), obtaining an expert scoring matrix, and taking the average as the final judgment matrix M (rows / columns are M1-M5): The matrix dimension has five criteria, therefore the judgment matrix is a 5×5 matrix, with row and column headings M1, M2, M3, M4, and M5. Element meaning: Matrix element M... mn This indicates the importance of the m-th indicator relative to the n-th indicator. (Reference) Figure 7 2. Calculate subjective weights: First step: Normalize the elements of each column of the judgment matrix to obtain a normalized matrix N, with elements N mn =M mn / ΣM m (Taking the first column as an example): The sum of the first column is: 1 + 1 + 0.5 + 0.333 + 0.5 = 3.333. N 11 =1 / 3.333≈0.3, N 21 =1 / 3.333≈0.3, N 31 =0.5 / 3.333≈0.15, N 41 =0.333 / 3.333≈0.1, N 51 =0.5 / 3.333≈0.15. Second step: Summate the elements of each row of the normalized matrix N to obtain the row sum vector S=[S1, S2, S3, S4, S5]. T The calculated S = [1.428, 1.428, 0.714, 0.428, 0.714] T Step 3: Normalize the rows and vectors to obtain the subjective weights W = [S1 / ΣS, S2 / ΣS, ..., S5 / ΣS], where ΣS = 1.428 + 1.428 + 0.714 + 0.428 + 0.714 = 4.712. Therefore, W = [0.303, 0.303, 0.152, 0.091, 0.152]. T (Corresponding to the subjective weights of M1-M5) 3. Consistency check: Ensure the logical consistency of the judgment matrix and avoid contradictions. (Reference) Figure 8 The first step is to calculate the product MW of the judgment matrix M and the subjective weight vector W, resulting in MW = [1.487, 1.487, 0.6525, 0.6728, 0.6525]. T Step 2: Calculate the largest eigenvalue λ = (1 / 5) × Σ[(MW) / W], and λ ≈ 5.158. Step 3: Calculate the consistency index CI = (λ-n) / (n-1) (n=5), CI = (5.15-5) / (5-1) = 0.04; find the average random consistency index RI (RI = 1.12 when n=5), and calculate the consistency ratio CR = CI / RI = 0.035 < 0.1, indicating that the judgment matrix has good consistency and the subjective weights are effective. III. Determine the objective weights using the entropy weight method, which specifically includes: 1. Collecting indicator data: Select the data of 5 indicator layers of the park in the past 5 years (before standardization), and construct a data matrix X (5 rows: years 1-5; 5 columns: M). 11-M 51 ), 5 criterion layers and corresponding indicator layers (M) 11 -M 51 The indicator layer data collected in the past 5 years is as follows: M 11 Annual carbon emissions for the park: The total net carbon emissions of the park each year (unit: 10,000 tons of CO2) are calculated by accounting for emissions from stationary sources, mobile sources, etc., plus carbon sink offsetting. This is a derived calculation (based on direct monitoring data). 21 Carbon emissions per unit of GDP: The annual carbon emissions per unit of GDP in the park (unit: tCO2 / 10,000 yuan), calculated by "total annual carbon emissions ÷ annual GDP of the park", is a derived calculation. M 31 Emission reduction potential for key industries: The theoretical emission reduction potential (unit: 10,000 tons of CO2) for key high-emission industries (such as chemical and steel) in the park each year is calculated by "industry benchmark emission level - current emission level" and is considered assessment data. M 41 Annual carbon absorption by carbon sink forests: The annual carbon dioxide absorption by the park's carbon sink forests (unit: 10,000 tons CO2), calculated from direct monitoring data such as the area of the carbon sink forests and the carbon sink volume per unit area, is derived from calculations. M 51 Renewable energy percentage: The percentage of renewable energy (photovoltaic, wind power, etc.) generation in the park's total electricity consumption each year (unit: %), calculated directly from the "renewable energy generation" and "total electricity consumption" measured by smart meters. This is derived data. After Z-score standardization, a standardized matrix X' is obtained (eliminating the influence of dimensions). Some standardized data are as follows: Where j represents the number of indicator layers and k represents the year number, it can be simplified to 5 rows and 5 columns, for example... .
[0067] 2. Calculate the index weight matrix P: P j =X' jk / ΣX j '(The proportion of the j-th column indicator in the k-th year), taking the first column as an example, ΣX'=0.8+0.7+0.6+0.5+0.9=3.5, P=0.8 / 3.5≈0.229, and so on to obtain the complete P matrix. 3. Calculate the information entropy E of each indicator: E=-H×ΣP×ln(P), where H=1 / ln(n)=1 / ln5≈0.4605; if P=0, define P×ln(P)=0. After calculating the information entropy of each indicator: E1=0.7316 (M 11 E2 = 0.7316 (M) 21 E3 = 0.7276 (M) 31 E4 = 0.7276 (M) 41 E5 = 0.7316 (M) 514. Calculate the objective weight W: the degree of difference of the indicators D. j =1-E j Objective weight W j =D j / ΣD j The calculated value is D = [0.2684, 0.2684, 0.2724, 0.2724, 0.2684]. T W = [0.1989, 0.1989, 0.2016, 0.2016, 0.1989] (corresponding to the objective weights of M1-M5, the objective weights reflect the degree of difference in the data itself, such as M5 (the proportion of renewable energy) having the largest data difference and the highest weight). IV. Combining weights to obtain the comprehensive weights of each evaluation indicator. A weighted average method is used to combine subjective and objective weights. The subjective weight coefficient α = 0.5 (considering both expert experience and data objectivity, it can be adjusted according to actual needs), the objective weight coefficient β = 1 - α = 0.5, and the combined weight W = α × W + β × W. The calculated combined weights for each criterion layer are: W1(M1) = 0.2510; W2(M2) = 0.2510; W3(M3) = 0.1768; W4(M4) = 0.1463; W5(M5) = 0.1755. The combined weights are finally normalized to obtain the comprehensive weights of each evaluation indicator. Ensure the sum is 1: W = [0.2510, 0.2510, 0.1768, 0.1463, 0.1755]. Evaluation indicators are divided into positive indicators (the higher the value, the better the low-carbon level, such as M). 41 -Annual carbon uptake by carbon sink forests, M 51 - The proportion of renewable energy) and negative indicators (the smaller the value, the better the low-carbon level, such as M) 11 -Annual carbon emissions of the park, M 21 - Carbon emissions per unit of GDP, M 31 -Emission reduction space in key industries), each is standardized using different formulas, mapping the indicator values to the [0, 1] interval, thereby obtaining the standardized annual indicator X'. jk The formula for calculating the comprehensive carbon emission assessment score of the industrial park is: S=Σ(X' jk ×W j ), where S is the comprehensive evaluation score, X' jk W is the standardized value of the j-th indicator in the k-th year. j Let be the comprehensive weight of the j-th indicator. The comprehensive evaluation score can also be obtained through methods such as: comprehensive evaluation score... Where k is the year number, K is the total number of years for which data was collected, k = 1, 2, ..., K; j is the number of indicator layers, J is the total number of indicator layers, j = 1, 2, ..., J; x jk' is the target value for the j-th indicator level in year k. It can be set manually based on actual conditions or according to policy requirements, which will not be elaborated here. x k This is the deviation unit reference value for the k-th indicator level. Its specific value can be determined empirically. Alternatively, its value can be obtained through simple calculation, where X is the deviation unit reference value. m This refers to the maximum value of historical data or the maximum allowable deviation data, which will not be elaborated upon here. β is the stratification level number, which can be 5-10, and will not be elaborated upon here either. This method of calculation can quickly narrow down the deviation of indicator data, avoiding data distortion caused by its interference with other data, and making the calculation more accurate.
[0068] S4. Determine whether the comprehensive evaluation score meets the qualification standard. If yes, no emission reduction plan will be generated; otherwise, proceed to S5. For example, the evaluation level is divided according to the score (the threshold can be adjusted according to the actual needs of the park): (1) Excellent: S≥0.85; (2) Good: 0.7≤S<0.85; (3) Qualified: 0.6≤S<0.7; (4) Unqualified: S<0.6. In the example, 0.71 points belong to the "Good" level, indicating that the current carbon emission control level of the park needs to be improved, and targeted improvement measures need to be formulated in conjunction with the intelligent emission reduction decision model. Of course, this is just a simple example. Other situations need to be set according to the actual situation, which will not be elaborated here.
[0069] S5. Combining the park's zero-carbon development goals and carbon emission assessment data, construct an intelligent emission reduction decision-making model and generate emission reduction plans; monitor the implementation effect of the emission reduction plans in real time, forming a closed-loop management system of "collection-calculation-evaluation-decision-feedback". Specifically, the intelligent emission reduction decision-making model is built based on reinforcement learning algorithms, aiming to minimize the park's carbon emissions and optimize emission reduction costs. Input parameters include the park's carbon emission assessment data, assessment index scores, emission reduction technology costs, renewable energy supply capacity, and carbon sink potential. The model outputs emission reduction plans including source emission reduction measures (such as energy structure optimization and production process upgrades), process control measures (such as equipment energy efficiency improvement and traffic control optimization), carbon sink enhancement measures (such as carbon sink forest expansion and wetland restoration), and carbon offsetting measures (such as CCUS technology application and carbon credit purchase). Real-time tracking of carbon data changes during the implementation of emission reduction plans, regular updates to carbon emission calculation results and comprehensive evaluation results; if the evaluation results do not meet the expected goals, the emission reduction plan is adjusted through the intelligent emission reduction decision-making model to achieve dynamic closed-loop management. For example, an intelligent emission reduction decision-making model is constructed based on reinforcement learning algorithms. Input parameters include carbon emission data from various emission sources, evaluation scores, photovoltaic power station installed capacity, and CCUS technology costs. The output emission reduction plan includes: ① Source reduction: replacing coal-fired boilers in chemical plants with gas-fired boilers, and expanding the photovoltaic power station capacity in the industrial park by 10MW; ② Process control: optimizing freight vehicle routes and promoting electric commuter vehicles; ③ Carbon sink enhancement: expanding carbon sink forests by 50 acres; ④ Carbon offsetting: purchasing 10,000 tons of forestry carbon credits. Emission reduction tasks are assigned through the industrial park's energy management platform, and the implementation effect of the plan is tracked in real time. Carbon emission calculation and evaluation results are updated quarterly. If the evaluation score does not reach the target of 85 points, the scale of photovoltaic capacity expansion and the area of carbon sink forest expansion are adjusted.
[0070] Example 2
[0071] refer to Figure 9 This invention also proposes a multi-source carbon data intelligent management and control system for zero-carbon industrial parks. The system includes a multi-source carbon data acquisition module, a data preprocessing module, a carbon emission calculation module, a comprehensive evaluation module, an intelligent decision-making and management module, and a data storage and visualization module. These modules are connected sequentially to form a closed-loop management and control architecture.
[0072] Multi-source carbon data acquisition module: This module is used to construct a multi-dimensional sensing network and collect multi-source carbon-related data from different types of carbon emission sources within the zero-carbon park. The multi-source carbon data acquisition module may include a stationary source monitoring unit, a mobile source monitoring unit, a building energy consumption monitoring unit, a waste monitoring unit, a carbon sink monitoring unit, and a data interface unit. The stationary source monitoring unit deploys equipment such as CEMS and flow meters to collect gas concentration and flow data from stationary emission sources. The mobile source monitoring unit deploys vehicle-mounted terminals to collect mileage, fuel consumption, and location data from mobile emission sources. The building energy consumption monitoring unit deploys smart metering equipment to collect building energy consumption data. The waste monitoring unit deploys gas sensors to collect greenhouse gas concentration data during waste treatment. The carbon sink monitoring unit deploys vegetation and soil monitoring equipment to collect carbon sink source-related data. The data interface unit is used to connect to the park's enterprise production management system, energy management platform, and meteorological monitoring station to obtain production process, energy statistics, and meteorological data.
[0073] The data preprocessing module is used to preprocess multi-source carbon-related data to obtain a standardized carbon dataset. This module may include a data cleaning unit, a data standardization unit, and a data fusion unit. The data cleaning unit uses the Isolation Forest algorithm to remove outliers and linear interpolation to fill in missing data. The data standardization unit uses the Z-Score standardization method to unify data dimensions. The data fusion unit uses a weighted average fusion algorithm to fuse data from the same source collected by multiple devices.
[0074] Carbon emission calculation module: This module calculates the total carbon emissions of the park, the individual carbon emissions from each emission source, and the carbon flow distribution data based on a carbon dataset and the type of emission source. The module includes a basic calculation unit, a dynamic correction unit, and a sub-item calculation unit. The basic calculation unit calculates the basic carbon emissions using the "activity data-emission factor" method. The dynamic correction unit introduces a dynamic correction coefficient to correct the basic emissions. The sub-item calculation unit performs carbon emission calculations for stationary sources, mobile sources, building energy consumption, waste treatment sources, and carbon sink sources, outputting the total carbon emissions, individual emission amounts, and carbon flow distribution data.
[0075] The comprehensive evaluation module is used to construct a multi-dimensional carbon emission evaluation index system for zero-carbon industrial parks, calculate the comprehensive weight of each evaluation index, and combine the numerical values of each evaluation index to obtain a comprehensive evaluation score. This module includes an index system construction unit, a weight calculation unit, and an evaluation implementation unit. The index system construction unit establishes a multi-dimensional index system covering total emission control, intensity, emission reduction potential, carbon sink contribution, and low-carbon operation. The weight calculation unit integrates subjective and objective weights to obtain a comprehensive weight. The evaluation implementation unit calculates the evaluation score based on the comprehensive weight and classifies the evaluation levels.
[0076] The intelligent decision-making and control module is used to determine the comprehensive evaluation score and, based on the carbon emission calculation results and the comprehensive evaluation results, generate emission reduction plans through an intelligent emission reduction decision model, and realize real-time tracking of emission reduction effects and dynamic adjustment of plans. This module includes a decision model unit, a plan generation unit, and an effect feedback unit. The decision model unit constructs a decision model based on reinforcement learning algorithms with the goal of optimizing emission reduction costs and minimizing emissions. The plan generation unit outputs targeted emission reduction measures. The effect feedback unit monitors changes in carbon data during the emission reduction process in real time, updates the evaluation results, and adjusts the emission reduction plan.
[0077] Data storage and visualization module: This module stores multi-source carbon data, calculation results, evaluation results, and emission reduction scheme data, and displays them through a visual interface. The module includes a distributed storage unit and a visualization unit. The distributed storage unit uses a time-series database to store real-time monitoring data and a relational database to store configuration data and evaluation results. The visualization unit displays data through carbon emission heat maps, trend curves, and indicator radar charts, and supports data querying, report generation, and anomaly warnings.
[0078] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.
[0079] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.
Claims
1. A method for intelligent management and control of multi-source carbon data in zero-carbon industrial parks, characterized in that, The intelligent management and control method for multi-source carbon data in zero-carbon industrial parks includes the following steps: S1: Based on the classification of carbon emission sources, a multi-dimensional sensing network is constructed to collect multi-source carbon-related data in real time; and the data is preprocessed to obtain a carbon dataset. S2. Based on the carbon dataset and combined with the emission source type, the carbon emission assessment data of the park is calculated. S3. Construct a multi-dimensional carbon emission evaluation index system for zero-carbon parks using carbon emission evaluation data from the parks, calculate the comprehensive weight of each evaluation index, and combine the values of each evaluation index to calculate the comprehensive evaluation score. S4. Determine whether the comprehensive evaluation score meets the qualification standard. If yes, no emission reduction plan will be generated; otherwise, proceed to S5. S5. Combining the park's zero-carbon development goals and carbon emission assessment data, construct an intelligent emission reduction decision-making model and generate emission reduction plans; The multi-dimensional zero-carbon park carbon emission evaluation index system includes an objective layer, a criterion layer, and an index layer. The methods for obtaining the overall evaluation score include: Overall Evaluation Score Where k is the year number, K is the total number of years for which data was collected, k = 1, 2, ..., K; j is the number of indicator layers, J is the total number of indicator layers, j = 1, 2, ..., J; X' jk x is the standardized annual value of the j-th indicator in the k-th year; jk ' is the target value for the j-th indicator layer in year k, x k W is the unit reference value for the deviation of the k-th indicator layer. j The overall weight of the j-th indicator; the deviation unit is the reference value. , where X m It is the maximum value in historical data or the maximum allowable deviation data; β is the number of stratification levels, with a value of 5-10.
2. The intelligent management and control method for multi-source carbon data in zero-carbon industrial parks according to claim 1, characterized in that, The types of carbon emission sources include: stationary emission sources, mobile emission sources, building energy sources, waste treatment sources, and carbon sink sources.
3. The intelligent management and control method for multi-source carbon data in zero-carbon industrial parks according to claim 1, characterized in that, Preprocessing of multi-source carbon-related data includes: data cleaning, data standardization, and data fusion.
4. The intelligent management and control method for multi-source carbon data in zero-carbon industrial parks according to claim 1, characterized in that, Carbon emission assessment data includes: total carbon emissions, carbon emissions from each emission source, carbon flow distribution data, total carbon emission control, carbon emission intensity, emission reduction potential, carbon sink contribution, and low-carbon operations.
5. The intelligent management and control method for multi-source carbon data in zero-carbon industrial parks according to claim 1, characterized in that, The method for obtaining the carbon emissions of each emission source includes: calculating the basic carbon emissions based on the activity data of different emission sources and the corresponding baseline emission factors; introducing a dynamic correction coefficient, which is obtained by training the production process parameters, equipment operating efficiency, meteorological conditions and park operating conditions through a gradient boosting tree algorithm, and correcting the basic carbon emissions through the dynamic correction coefficient to obtain high-precision carbon emissions of each emission source.
6. The intelligent management and control method for multi-source carbon data in zero-carbon industrial parks according to claim 5, characterized in that, The steps for calculating the dynamic correction coefficient are as follows:
1. Data collection and feature selection; 2. Data preprocessing; 3. Model training and optimization; 4. Generation of correction coefficients.
7. The intelligent management and control method for multi-source carbon data in zero-carbon industrial parks according to claim 6, characterized in that, Data preprocessing includes: cleaning the collected feature variable data and removing outliers; filling missing data using K-nearest neighbor interpolation; screening feature variables that are significantly correlated with carbon emissions through Pearson correlation analysis, and finally determining the model input feature set.
8. The intelligent management and control method for multi-source carbon data in zero-carbon industrial parks according to claim 5, characterized in that, Model training and optimization include: constructing a gradient boosting tree model using the selected feature variables as input and the measured deviation ratio as output; optimizing the model hyperparameters using the grid search method; dividing the preprocessed dataset into training and test sets in a 7:3 ratio, training the model with the training set, and verifying the model accuracy with the test set, ensuring that the model prediction error RMSE ≤ 0.
02.
9. A method for intelligent management and control of multi-source carbon data in zero-carbon industrial parks according to claim 5, characterized in that, The method for generating correction coefficients includes: inputting real-time collected feature variables into a trained gradient boosting tree model to obtain the correction coefficient deviation value Δk; the dynamic correction coefficient k = 1 + Δk, where the value range of k is constrained to 0.95~1.
05.
10. A multi-source carbon data intelligent management and control system for zero-carbon industrial parks, characterized in that, The system includes: Multi-source carbon data acquisition module: It is used to build a multi-dimensional sensing network and collect multi-source carbon-related data of different types of carbon emission sources in the zero-carbon park; Data preprocessing module: It is used to preprocess multi-source carbon-related data to obtain a standardized carbon dataset; Carbon emission calculation module: It is used to calculate the total carbon emissions of the park, the carbon emissions of each emission source, and the carbon flow distribution data based on the carbon data set and the emission source type. Comprehensive evaluation module: It is used to construct a multi-dimensional carbon emission evaluation index system for zero-carbon parks, calculate the comprehensive weight of each evaluation index, and calculate the comprehensive evaluation score by combining the values of each evaluation index. The multi-dimensional zero-carbon park carbon emission evaluation index system includes an objective layer, a criterion layer, and an index layer. The methods for obtaining the overall evaluation score include: Overall Evaluation Score Where k is the year number, K is the total number of years for which data was collected, k = 1, 2, ..., K; j is the number of indicator layers, J is the total number of indicator layers, j = 1, 2, ..., J; X' jk x is the standardized annual value of the j-th indicator in the k-th year; jk ' is the target value for the j-th indicator layer in year k, x k W is the unit reference value for the deviation of the k-th indicator layer. j The overall weight of the j-th indicator; the deviation unit is the reference value. , where X m It is the maximum value in historical data or the data with the maximum allowable deviation; β is the number of stratification levels, with a value of 5-10. Intelligent decision-making and control module: It is used to determine the comprehensive evaluation score, and based on the carbon emission calculation results and comprehensive evaluation results, it generates emission reduction plans through intelligent emission reduction decision model, and realizes real-time tracking of emission reduction effect and dynamic adjustment of plan; Data storage and visualization module: It is used to store multi-source carbon data, calculation results, evaluation results and emission reduction scheme data, and display them through a visualization interface.
Citation Information
Patent Citations
Near-zero carbon emission park evaluation index system construction method
CN110472842A
Park carbon emission accounting method considering comprehensive energy system and uncertain factors
CN119179841A
Park zero-carbon detection management planning method
CN120298003A
Park carbon emission management method, equipment and medium
CN121526084A
