Carbon dioxide emission monitoring system and control method thereof

Abstract

The invention discloses a carbon dioxide emission monitoring system and a control method thereof, belongs to the technical field of digital monitoring control, and aims to solve the problems of localization, large abnormal recognition deviation and regulation and control lag of traditional monitoring. The method comprises the following steps of: acquiring production parameters and carbon flow characteristic variables of each production link through a multi-dimensional sensing network constructed by a digital sensor, and constructing a time-space association data set according to a time-space association mapping system of the production parameters and the carbon flow characteristic variables; constructing a carbon flow association rule base based on historical data mining association rules and identification common association rules, embedding a multi-level carbon flow model to construct a hierarchical digital twinborn model, and correcting deviation in combination with current data; the model is utilized to periodically predict carbon flow, acquire carbon flow four-element distribution data to judge emission abnormity, construct a reference variable sequence to identify abnormal production links and production equipment, generate target production parameters and execute feedback, and complete-flow carbon flow accurate control is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital monitoring control, more particularly to a carbon dioxide emission monitoring system and a control method thereof. BACKGROUND

[0002] In the field of green methanol production, carbon emission monitoring and control is a core link to achieve low-carbon production, which is directly related to the environmental protection and economy of the production process. At present, the monitoring of carbon dioxide emission in the industry relies on local sensor data collection, which is difficult to cover the whole production process such as raw material pretreatment, catalytic reaction and product separation, resulting in blind spots in the perception of the overall distribution of carbon flow, and the correlation influence of carbon emission in each link cannot be fully grasped. In terms of anomaly identification, the traditional method is based on fixed threshold, which can only identify the explicit anomalies that have exceeded the set range, and lacks effective capture ability for implicit anomalies that slowly deviate from the normal trend of carbon flow characteristic variables, often resulting in delayed or false abnormal warning, affecting the timeliness and accuracy of subsequent regulation. The parameter regulation link generally relies on manual experience or static model, and the formulation of regulation strategy cannot fully combine the dynamic adjustment of real-time carbon flow data, resulting in insufficient adaptability of production parameters to carbon flow optimization requirements, and it is difficult to achieve fine control of carbon emission. In addition, due to the lack of systematic understanding of the carbon flow transmission law in the whole production process, the carbon emission control of each production link is in an isolated state, and it is difficult to form a global coordinated regulation mechanism, the overall carbon flow control efficiency is low, which restricts the further optimization of carbon emission level in the green methanol production process. Therefore, in order to overcome these limitations, the present application provides a carbon dioxide emission monitoring system and a control method thereof. SUMMARY

[0003] In view of the deficiencies in the prior art, the present application aims to provide a carbon dioxide emission monitoring system and a control method thereof, which solves the problems of one-sided carbon dioxide emission monitoring, large deviation in anomaly identification, insufficient dynamicity of parameter regulation, and weak whole-process carbon flow control capability in green methanol production, realizes real-time perception, dynamic modeling, health diagnosis and intelligent regulation closed-loop management of whole-process carbon emission, and achieves the goal of accurate detection, early identification of abnormal emission and efficient regulation to reduce carbon emission.

[0004] To achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0005] The carbon dioxide emission monitoring system comprises:

[0006] A multi-dimensional sensor network constructed by digital sensors collects production parameter combinations of production equipment in each production link and corresponding carbon flow characteristic variables; a spatio-temporal correlation dataset is constructed according to a spatio-temporal correlation mapping system of production parameter combinations and carbon flow characteristic variables;

[0007] Based on the historical spatio-temporal correlation data set, the production parameter combination and the carbon flow characteristic variable are associated with the rule mining and dynamic correlation analysis, the association rules of the production parameter combination and the carbon flow characteristic variable are extracted, the common association rules are identified, the carbon flow association rule base is constructed, the carbon flow association rule base is embedded in the multi-level carbon flow model to construct the hierarchical digital twin model, and the hierarchical digital twin model is corrected based on the spatio-temporal correlation data set of the current production cycle.

[0008] According to the hierarchical digital twin model, the carbon flow characteristic variable in the current production cycle is periodically predicted based on the production parameter combination, the distribution data of the four elements of carbon flow are obtained, which are used to judge whether the carbon dioxide emission is abnormal; based on the carbon flow association rule base, the reference variable sequence is constructed, the abnormal production link and the abnormal production equipment are identified, the target production parameter combination of the abnormal production equipment is generated and executed.

[0009] Specifically, the multi-level carbon flow model includes: a micro mapping model, a production link integrated model and a whole process model.

[0010] The micro mapping model is used to quantify the micro dynamics of the carbon flow characteristic variable changing with the production parameter combination in the production equipment, the common association rules of the production parameter combination and the carbon flow characteristic variable corresponding to the production equipment are extracted from the carbon flow association rule base, the basic physical equation of carbon flow transmission is established based on the physical structure parameters of the production equipment according to the common association rules, and the mapping relationship between the production parameter combination and the carbon flow characteristic variable in the common association rules is converted into the parameter coefficient in the basic physical equation, which is constructed by physical mechanism modeling and association rule parameterization.

[0011] The production link integrated model is used to quantify the carbon flow comprehensive state in the production link, the micro mapping models in the same production link are integrated according to the physical connection relationship between the production equipment, the carbon flow transmission parameters between the production equipment are defined based on the carbon flow characteristic variables output by the production equipment, and the comprehensive carbon flow characteristic variables of each production link are constructed by the carbon flow weighted aggregation algorithm.

[0012] The whole process model is used to quantify the carbon flow comprehensive state of the whole production link, the production link integrated models of each production link are integrated according to the physical connection relationship of the production link based on the physical topological structure of the production link, the interaction parameters of the comprehensive carbon flow characteristic variables between the production links are defined based on the carbon flow transmission path between the production links, and the whole process carbon flow aggregation is constructed based on the comprehensive carbon flow characteristic variables output by the production link integrated model.

[0013] Specifically, the specific steps of constructing the carbon flow association rule base include:

[0014] obtaining a historical spatiotemporal correlation dataset and dividing the historical spatiotemporal correlation dataset according to production cycles to construct historical spatiotemporal correlation subsets;

[0015] establishing a production parameter combination-carbon flow characteristic variable mapping table for each production link according to the historical spatiotemporal correlation subsets of the respective production cycles;

[0016] applying a correlation rule mining algorithm to the mapping table of each production link of the historical spatiotemporal correlation subsets to identify explicit correlation rules between the production parameter combinations and the carbon flow characteristic variables and to construct an explicit correlation rule set;

[0017] performing dynamic correlation analysis on the production parameter combination time series and the carbon flow characteristic variables to capture implicit correlation rules between the time sequence characteristics of the production parameter combinations and the dynamic response characteristics of the carbon flow characteristic variables and to construct an implicit correlation rule set;

[0018] summarizing the explicit correlation rule set and the implicit correlation rule set of each historical spatiotemporal correlation subset to construct a production link-level correlation rule total set corresponding to the historical spatiotemporal correlation subset;

[0019] performing comparative analysis on the production link-level correlation rule total sets of different historical spatiotemporal correlation subsets, counting the occurrence frequency proportions of each correlation rule in different historical spatiotemporal correlation subsets, the correlation rules including explicit correlation rules and implicit correlation rules;

[0020] if the occurrence frequency proportion of a correlation rule is greater than a preset frequency proportion threshold, marking the correlation rule as a common correlation rule, performing standardized arrangement, and constructing a carbon flow correlation rule library.

[0021] Specifically, constructing the explicit correlation rule set includes:

[0022] performing interval discretization processing on continuous production parameter combinations and carbon flow characteristic variables and encoding processing on enumerated production parameter combinations and carbon flow characteristic variables to generate a structure variable mapping table;

[0023] applying a correlation rule mining algorithm to the structure variable mapping table to identify explicit correlation rules between the production parameter combinations and the carbon flow characteristic variables and selecting production parameter combinations and carbon flow characteristic variable combinations having statistical significance as frequent item sets by setting a support threshold;

[0024] extracting strong correlation rules greater than a preset confidence threshold from the frequent item sets as explicit correlation rules to construct the explicit correlation rule set;

[0025] Specifically, constructing the implicit correlation rule set includes:

[0026] The time sequence features of the production parameter combination are extracted by time sequence decomposition, including trend change features, periodic fluctuation features and sudden fluctuation features of the production parameter combination, and the dynamic response features of the carbon flow characteristic variables with the production parameter combination are counted, including response time features and response amplitude features;

[0027] The time association rules of the time sequence features of the production parameter combination and the dynamic response features of the carbon flow characteristic variables are identified by dynamic time warping, as the implicit association rules, and an implicit association rule set is constructed.

[0028] Specifically, the step of correcting the hierarchical digital twin model deviation comprises:

[0029] The current production cycle spatio-temporal association data set is collected, the real-time carbon flow characteristic variables and the corresponding production parameter combination are extracted, and the multi-level carbon flow model in the digital twin model associated with the carbon flow characteristic variables and the corresponding production parameter combination is located, and a data model association mapping table is established;

[0030] Based on the data model association mapping table, for the located multi-level carbon flow model, the corresponding carbon flow characteristic variable model prediction value is obtained according to the current production parameter combination, the model prediction value is compared with the corresponding real-time carbon flow characteristic variable, and the prediction deviation value is calculated;

[0031] The prediction deviation values are accumulated according to the multi-level carbon flow model respectively, the multi-level cumulative deviations are generated respectively, the multi-level cumulative deviation thresholds are configured, when the cumulative prediction deviation value is greater than the corresponding multi-level cumulative deviation threshold, the multi-level carbon flow model correction is triggered, including microscopic mapping model correction, production link integrated model correction and whole process model correction;

[0032] After the multi-level carbon flow model correction is completed, the cumulative deviation of the corrected multi-level carbon flow model is cleared, and the cumulative deviation values of each multi-level carbon flow model are continuously updated; the correction times of each multi-level carbon flow model are recorded synchronously, if the correction times of the multi-level carbon flow model exceed the preset correction threshold within the preset monitoring time window, the digital twin model abnormality early warning is triggered.

[0033] Specifically, the microscopic mapping model correction is used to eliminate the production equipment level cumulative deviation, so that the production equipment level carbon flow characteristic variable prediction value output by the microscopic mapping model is consistent with the corresponding real-time carbon flow characteristic variable, by calling the common association rules in the carbon flow association rule library, combining the production equipment level cumulative deviation value, using the parameter self-adaptive optimization algorithm to adjust the parameter coefficient of the basic physical equation in the microscopic mapping model, and correcting the microscopic mapping model;

[0034] The production process integrated model correction is used to eliminate cumulative deviations at the production process level, so that the production process integrated model output comprehensive carbon flow characteristic variables at the production process level are consistent with the corresponding real-time comprehensive carbon flow characteristic variables. By combining the output results corrected based on the micro-mapping model with the cumulative deviation value at the production process level, the carbon flow weighted aggregation algorithm is re-run to dynamically calibrate the carbon flow transfer parameters between production equipment, adjust the calculation weights of the comprehensive carbon flow characteristic variables at the production process level, and correct the production process integrated model.

[0035] The full-process model correction is used to eliminate the cumulative deviation at the full-process level, so that the full-process carbon flow characteristic variables output by the full-process model are consistent with the corresponding real-time full-process carbon flow characteristic variables. By combining the output results of the production link integrated model correction with the full-process cumulative deviation value, the full-process carbon flow aggregation algorithm is re-executed to correct the carbon flow interaction parameters between production links and correct the full-process model.

[0036] Specifically, the steps for identifying abnormal production processes and abnormal production equipment include:

[0037] Configure monitoring cycles, and at the beginning of each monitoring cycle, based on the configuration of production parameter combinations, predict the carbon flow characteristic variables corresponding to the current production parameter combinations within that monitoring cycle through a hierarchical digital twin model;

[0038] Based on the carbon flow characteristic variables of the carbon flow, and according to the comprehensive carbon flow characteristic variables output by the whole process model, the distribution data of the four elements of carbon flow are obtained respectively.

[0039] Based on the distribution data of the four elements of carbon flow throughout the entire process, the proportion of carbon emissions lost throughout the entire process to the total carbon input of raw materials is calculated and compared with the preset carbon loss ratio threshold. If it is greater than the carbon loss ratio threshold, it is determined that there is an abnormality in carbon dioxide emissions and the abnormal identification process of the production process is initiated; otherwise, the current combination of production parameters continues to be executed.

[0040] Based on the carbon flow association rule library, according to the configuration of production parameter combinations, the common association rules involving the current production parameter combinations are retrieved to obtain the corresponding carbon flow feature variables. Based on the current hierarchical digital twin model, the comprehensive carbon flow feature variable sequence of each production link in the current production cycle is constructed, and the benchmark distribution data sequence of the four elements of carbon flow in each production link is generated.

[0041] Specifically, the steps for identifying abnormal production processes and abnormal production equipment also include:

[0042] Extract the actual distribution data sequence of the four carbon flow elements in each production stage within the current production cycle, compare the actual distribution data sequence with the benchmark distribution data sequence of the corresponding production stage, calculate the absolute deviation of carbon flow, and construct the absolute deviation sequence of the four carbon flow elements in each production stage of the current production cycle.

[0043] Configure a carbon flow deviation threshold. If there is a carbon flow absolute deviation in the carbon flow absolute deviation sequence of the four carbon flow elements that is greater than the carbon flow deviation threshold, the corresponding production link is marked as an abnormal production link. Otherwise, the trend of change is extracted from the carbon flow absolute deviation sequence of each production link through a trend fitting algorithm to determine whether the production link is an abnormal production link.

[0044] Extract the actual sequence of carbon flow characteristic variables of each production equipment in the abnormal production process of the current production cycle, and generate the benchmark sequence of carbon flow characteristic variables of production equipment based on the common association rules of the combination of production parameters of each production equipment in the carbon flow association rule base.

[0045] Configure the deviation threshold for production equipment, calculate the relative deviation between the actual sequence and the baseline sequence of carbon flow characteristic variables for each production equipment, and construct a relative deviation sequence at the production equipment level to mark abnormal production equipment.

[0046] Specifically, the steps for generating the target production parameter combination for abnormal production equipment include:

[0047] Based on the labeled abnormal production equipment, the real-time production parameter combination of the abnormal production equipment in the current production cycle is retrieved, and the carbon flow feature variable corresponding to the current production parameter combination is matched from the carbon flow association rule library as the target benchmark value.

[0048] Replace the actual carbon flow characteristic variables of the abnormal production equipment in the current production cycle with the corresponding target benchmark value, obtain the comprehensive carbon flow characteristic variables output by the whole process model based on the hierarchical digital twin model, and determine whether there is an abnormality in carbon dioxide emissions; if so, trigger an early warning for abnormal production equipment.

[0049] If not, trigger the inverse solution of production parameter combinations, call the micro-mapping model of the production equipment in the hierarchical digital twin model, construct the inverse mapping relationship between the production parameter combinations and the target benchmark value; input the target benchmark value, perform the inverse solution of production parameter combinations through Newton's iteration method, set the initial parameter production adjustment step size, iterate and calculate with the target benchmark value as the constraint condition, obtain the production parameter combinations of abnormal production equipment, and determine the target production parameter combinations through boundary verification and simulation verification.

[0050] Methods for monitoring and controlling carbon dioxide emissions include:

[0051] Step S1: Collect the production parameter combinations and corresponding carbon flow characteristic variables of the production equipment in each production stage through a multi-dimensional sensor network constructed by digital sensors; construct a spatiotemporal correlation dataset based on the spatiotemporal correlation mapping system between the production parameter combinations and carbon flow characteristic variables.

[0052] Step S2: Based on the historical spatiotemporal correlation dataset, perform association rule mining and dynamic correlation analysis on the combination of production parameters and carbon flow characteristic variables, extract the association rules between the combination of production parameters and carbon flow characteristic variables, identify common association rules, construct a carbon flow association rule library, embed the carbon flow association rule library into a multi-level carbon flow model to construct a hierarchical digital twin model, and correct the bias of the hierarchical digital twin model based on the spatiotemporal correlation dataset of the current production cycle.

[0053] Step S3: Based on the hierarchical digital twin model, periodically predict the carbon flow characteristic variables in the current production cycle based on the combination of production parameters, obtain the distribution data of the four elements of carbon flow, and use them to determine whether there are any abnormalities in carbon dioxide emissions; construct a benchmark variable sequence based on the carbon flow association rule base, identify abnormal production links and abnormal production equipment, generate the target production parameter combination for abnormal production equipment, and provide execution feedback.

[0054] The beneficial effects of this invention are:

[0055] This invention effectively addresses the problems of one-sided carbon dioxide emission monitoring, large anomaly identification bias, insufficient dynamic parameter control, and weak overall carbon flow management capabilities in green methanol production by constructing a multi-dimensional sensor network, a carbon flow association rule base, and a hierarchical digital twin model. It enables real-time perception of carbon emissions throughout the entire production process, ensuring comprehensive monitoring coverage. Through a precise anomaly identification mechanism, abnormal emissions can be detected early, and specific links and equipment can be located, improving the accuracy of anomaly identification. By leveraging the synergy of dynamic modeling and intelligent control, dynamic optimization and adjustment of production parameters are achieved, forming a closed-loop management system from perception to control. This significantly enhances the overall carbon flow management capabilities, ultimately achieving the goal of accurate carbon emission detection and efficient control to reduce carbon emissions, ensuring the low-carbon nature and stability of the green methanol production process. Attached Figure Description

[0056] Figure 1 This is a schematic diagram of the carbon dioxide emission monitoring system of the present invention;

[0057] Figure 2 A flowchart illustrating the specific steps involved in constructing the carbon flow association rule base for this invention;

[0058] Figure 3 This is a flowchart illustrating the specific steps involved in correcting the bias in the digital twin model according to the present invention.

[0059] Figure 4 A flowchart illustrating the specific steps involved in identifying abnormal production processes and equipment according to the present invention;

[0060] Figure 5 This is a flowchart illustrating the specific steps involved in creating the target production parameter combination for the abnormal production equipment according to the present invention. Detailed Implementation

[0061] Please see Figure 1 This embodiment introduces a carbon dioxide emission monitoring system, including an element perception module, a twin fusion module, and an intelligent decision-making module;

[0062] The element perception module constructs a multi-dimensional sensor network covering the entire green methanol production process by deploying digital sensors. This network encompasses each production stage, including raw material pretreatment, synthesis reaction, separation and purification, and tail gas treatment. Based on the spatiotemporal correlation mapping system between the production parameter combinations and corresponding carbon flow characteristic variables for each stage, a spatiotemporal correlation dataset is built. Its core function is to collect in real-time production parameter combinations from each production stage, including operating temperature, pressure, and flow rate, as well as carbon flow characteristic variables related to the four elements: raw material carbon, conversion carbon, product carbon, and loss carbon. Through timestamp synchronization technology and data fusion algorithms, it achieves real-time monitoring of production parameter combinations and dynamic derivation calculation of carbon flow characteristic variables, providing standardized digital data support for the digital mirror construction of the carbon flow digital twin fusion module, overcoming the limitations of traditional local monitoring that only targets emission outlets.

[0063] In this embodiment, a laser gas sensor is embedded in the carbon dioxide input pipeline in the raw material pretreatment stage to collect parameters such as raw material gas flow rate and carbon dioxide purity. The cumulative input amount of raw material carbon is calculated by combining the carbon source input time series data. In the methanol synthesis reactor of the synthesis reaction stage, an in-situ fiber optic sensor is deployed to collect parameters such as bed temperature and catalyst activity response. The carbon conversion reaction efficiency and the amount of unconverted carbon are deduced based on the reaction kinetic model. Microwave sensors are installed in the product storage tanks of the separation, purification and storage stages to collect parameters such as methanol storage and density. The real-time solidification amount of product carbon is calculated by applying the law of conservation of mass. A continuous emission detection system is configured at the exhaust gas outlet of the tail gas treatment stage to collect parameters such as tail gas flow rate and carbon dioxide concentration. The cumulative carbon emission amount is calculated by combining the emission time integral. Infrared imaging sensors are set at the leakage points of production equipment to collect parameters such as leakage rate and quantify non-process carbon loss. A smart meter and carbon factor tag reader are integrated at the green electricity access point to collect parameters such as green electricity consumption. The carbon emission reduction is calculated by linking the grid carbon factor database, forming a multi-dimensional sensing network covering the entire process.

[0064] All sensors acquire raw digital signals at a set digital sampling frequency. These raw sensor signals undergo preprocessing steps including digital noise reduction algorithms, digital timestamp-based synchronization calibration, and outlier removal to construct a spatiotemporal correlated dataset containing real-time production parameter combinations and dynamic carbon flow characteristic variables. This dataset uses a timestamp matching mechanism to bind instantaneous production parameter combinations with carbon flow characteristic variables within corresponding time windows. Finally, a standardized digital format is used to construct the spatiotemporal correlated dataset, including digital identifiers such as parameter names associated with specific production equipment and carbon flow elements, parameter values, acquisition time, and sensor IDs. This provides high-quality digital input reflecting the dynamic rules of carbon changes in the production process for the subsequent twin fusion module.

[0065] The digital twin fusion module is based on the spatiotemporal correlation dataset generated by the element perception module. Through the data layer, it integrates historical spatiotemporal correlation datasets, extracts correlation rules between production parameter combinations and carbon flow characteristic variables, and constructs a carbon flow correlation rule library. The model layer embeds this rule library into a multi-level carbon flow model, building a hierarchical digital twin model. Finally, the calibration layer corrects deviations in the digital twin model based on real-time data from the current production cycle's spatiotemporal correlation dataset to maintain dynamic consistency between the virtual and real worlds. The progressive processing of the data layer, model layer, and calibration layer achieves the construction and dynamic optimization of the carbon flow digital twin.

[0066] In this embodiment, at the data layer, a multi-dimensional data integration algorithm is used to structurally fuse the time series of production parameter combinations and carbon flow characteristic variables of historical production cycles. Association rule mining and time-series pattern recognition technologies are employed to extract dynamic mapping rules between production parameter combinations and carbon flow characteristic variables in each production stage, forming a carbon flow association rule library covering the entire process from raw material pretreatment, synthesis reaction, separation and purification, to tail gas treatment. At the model layer, based on the physical topology of the production system and the carbon flow conversion mechanism, the carbon flow association rule library is embedded into a multi-level carbon flow model. This multi-level model includes a microscopic mapping model, a production stage integration model, and a full-process model, constructing a hierarchical digital twin framework containing parameter carbon flow dynamic association operators. At the calibration layer, a real-time deviation diagnosis mechanism is introduced. Using the real-time monitoring data of the current production cycle as a benchmark, least squares fitting and parameter adaptive optimization algorithms are used to correct the deviation between the model's predicted values ​​and the actual carbon flow data, dynamically updating the carbon flow association rule library to ensure consistency between the virtual state of the digital twin and the real-time carbon flow dynamics of the physical production system. The final output is a digital twin model that covers the carbon flow conversion efficiency at each level, parameter sensitivity analysis, and the carbon balance state of the entire process, providing dynamic simulation and prediction support for carbon flow regulation decisions through virtual and real interaction.

[0067] Please see Figure 2 Preferably, the specific steps for constructing a carbon flow association rule base include:

[0068] Based on the preset evaluation cycle, historical spatiotemporal correlation datasets are obtained and divided according to the production cycle to construct historical spatiotemporal correlation subsets. Each historical spatiotemporal correlation subset corresponds to the monitoring data of a complete production cycle, including the time series of production parameter combinations and carbon flow characteristic variables of each production link in the production cycle, and includes the collection timestamp and production link identifier, providing basic data units for the analysis of correlation rules by cycle.

[0069] Standardized preprocessing is performed on each historical spatiotemporal correlated subset, including time synchronization, missing value imputation, and outlier handling, to construct a structured historical spatiotemporal correlated subset. Time synchronization of production parameter combinations and carbon flow characteristic variables within the same production cycle is achieved through timestamp alignment. Interpolation is used to supplement missing data in the historical spatiotemporal correlated subset. Outliers are identified and removed based on data distribution characteristics. The data format and units of each subset are unified to form a structured historical spatiotemporal correlated subset, ensuring the comparability of data from different production cycles.

[0070] Based on the process logic of each production stage, and according to the historical spatiotemporal correlation subsets of each production cycle, a mapping table of production parameter combinations and carbon flow characteristic variables for each production stage is established, clarifying the corresponding correlation range of production parameter combinations and carbon flow characteristic variables in different production stages.

[0071] Based on the variable mapping tables of each production stage, an association rule mining algorithm is applied to the mapping tables of each production stage in each historical spatiotemporal correlation subset to identify explicit association rules between production parameter combinations and carbon flow characteristic variables. Continuous production parameter combinations and carbon flow characteristic variables are discretized into intervals, while enumerated production parameter combinations and carbon flow characteristic variables are encoded to generate a structural variable mapping table suitable for association rule mining algorithm processing. The association rule mining algorithm is then applied to the structural variable mapping table to identify explicit association rules between production parameter combinations and carbon flow characteristic variables. A support threshold is set to filter statistically significant production parameter combinations and carbon flow characteristic variable combinations as frequent itemsets. Strong association rules exceeding a pre-set confidence threshold are extracted from the frequent itemsets as explicit association rules, thus constructing an explicit association rule set.

[0072] Dynamic correlation analysis is performed on the time series of production parameter combinations and carbon flow characteristic variables to capture the implicit correlation rules between the time series characteristics of production parameter combinations and the dynamic response characteristics of carbon flow characteristic variables. The time series characteristics of production parameter combinations are extracted through time series decomposition, including trend change characteristics, periodic fluctuation characteristics, and sudden fluctuation characteristics of production parameter combinations. The dynamic response characteristics of carbon flow characteristic variables with production parameter combinations are statistically analyzed, including response time characteristics and response amplitude characteristics. The time correlation rules between the time series characteristics of production parameter combinations and the dynamic response characteristics of carbon flow characteristic variables are identified through dynamic time warping, and these rules are used as implicit correlation rules to construct a set of implicit correlation rules.

[0073] The explicit and implicit association rule sets of each historical spatiotemporal associated subset are summarized to construct a total set of production-stage association rules corresponding to the historical spatiotemporal associated subset. The total sets of production-stage association rules of different historical spatiotemporal associated subsets are compared and analyzed, and the frequency ratio of each association rule in different historical spatiotemporal associated subsets is statistically analyzed. The association rules include explicit association rules and implicit association rules. If the frequency ratio of an association rule is greater than a preset frequency ratio threshold, it is marked as a common association rule. The rules are standardized and organized according to the format of production stage identifier, production parameter combination, carbon flow characteristic variable, association rule and association strength index to construct a carbon flow association rule library that can be accurately called according to production stage.

[0074] Preferably, the specific steps for constructing a hierarchical digital twin model include:

[0075] Based on the physical topology of production equipment in the green methanol production process, and considering the spatial distribution and connectivity of the equipment, a microscopic mapping model is constructed by extracting common association rules between production parameter combinations and carbon flow characteristic variables for each production stage from a carbon flow association rule base. This microscopic mapping model is used to quantify the microscopic dynamics of carbon flow characteristic variables in production equipment as production parameter combinations change. The construction method is as follows: based on the extracted common association rules, a fundamental physical equation for carbon flow transfer is established based on the physical structural parameters of the production equipment. Then, the mapping relationship between production parameter combinations and carbon flow characteristic variables in the common association rules is transformed into parameter coefficients in the fundamental physical equation, employing a combination of physical mechanism modeling and association rule parameterization.

[0076] For example, a basic physical equation for carbon flow transfer is established based on the physical structural parameters of the production equipment. The physical structural parameters include the reactor volume and the number of trays in the distillation column. The basic physical equation includes the material balance equation and the reaction kinetic equation. The mapping relationship between the combination of production parameters and the carbon flow characteristic variables in the common association rules is transformed into the parameter coefficients in the equation. For example, for every 10°C increase in reactor temperature, the carbon conversion rate increases by 2%, which corresponds to the coefficient correction value of the temperature term in the equation.

[0077] Based on the physical connections of production equipment within the production process, a production process integrated model is constructed by integrating the micro-mapping models within the same production process according to the physical connections between the production equipment. This production process integrated model is used to quantify the overall state of carbon flow within the production process. The construction method is as follows: carbon flow transfer parameters between production equipment are defined based on the carbon flow characteristic variables output by the production equipment, and the comprehensive carbon flow characteristic variables of each production process are constructed through a carbon flow weighted aggregation algorithm.

[0078] Specifically, based on the carbon flow characteristic variables output by production equipment, such as instantaneous carbon flow rate, carbon conversion rate, and carbon loss rate, carbon flow transfer parameters between production equipment are defined to quantify the transfer efficiency and loss coefficient of the carbon flow characteristic variables from the outlet of the preceding production equipment to the inlet of the subsequent production equipment. A comprehensive carbon flow characteristic variable for each production link is constructed through a carbon flow weighted aggregation algorithm. This algorithm calculates the carbon flow characteristic variables of the production equipment according to the primary and secondary relationships of the physical connections of the production equipment, while incorporating the carbon flow transfer parameters between production equipment to correct for carbon flow loss during the transfer process. By quantifying the overall conversion efficiency, total loss, and dynamic distribution characteristics of the carbon flow within the production link, an integrated production link model for the transfer of carbon flow characteristic variables between production equipment within the production link is constructed to achieve accurate mapping of the comprehensive carbon flow state of the production link.

[0079] Based on the physical topology of the entire green methanol production process, the integrated models of each production stage are combined according to their physical connections to construct a full-process model. This full-process model is used to quantify the overall carbon flow status of the entire production process. The construction method is as follows: based on the carbon flow transport paths between production stages, interaction parameters of the comprehensive carbon flow characteristic variables between production stages are defined; then, the comprehensive carbon flow characteristic variables output by the integrated production stage model are used to aggregate the carbon flow throughout the entire process.

[0080] Specifically, the carbon flow transmission path between production stages is analyzed, and interaction parameters of comprehensive carbon flow characteristic variables between production stages are defined to quantify the transmission efficiency of comprehensive carbon flow characteristic variables from the outlet of the preceding production stage to the inlet of the subsequent production stage and the loss coefficient across production stages. Based on the comprehensive carbon flow characteristic variables output by the integrated production stage model, the carbon flow is aggregated throughout the entire process in combination with the carbon flow transmission logic corresponding to the physical connection of the production stages. The comprehensive carbon flow characteristic variables at the production stage level are weighted and summarized according to the functional weight of the production stage in the entire process, while incorporating the carbon flow interaction parameters between production stages to correct the carbon flow transmission deviation across production stages. By constructing a full-process model covering the entire production process, a global mapping of the overall carbon flow status of the entire green methanol production process is achieved.

[0081] The common association rules in the carbon flow association rule base are transformed into dynamic carbon flow association operators for production parameter combinations through a rule parameterization transformation mechanism. This operator contains a quantitative mapping relationship between the change in the production parameter combination and the response of the carbon flow characteristic variable. This relationship is then embedded into the calculation modules of the micro-mapping model, the production link integration model, and the full-process model. The dynamic carbon flow association operator for production parameter combinations is dynamically updated by real-time collection of production parameter combinations, driving the output of each level of the hierarchical digital twin model to synchronize with the physical production system's carbon flow characteristic variables, thus enabling the model to have dynamic response capabilities. The rule parameterization transformation mechanism refers to transforming the association relationship between the production parameter combination and the carbon flow characteristic variable in the common association rules into a computable function expression. For example, the association rule that the carbon flow characteristic variable changes by a certain value for every change in the production parameter combination is transformed into a mathematical formula containing a proportional coefficient and a benchmark value, forming the core calculation logic of the dynamic carbon flow association operator for production parameter combinations.

[0082] Define the carbon flow data upload protocol from the micro-mapping model to the production link level integrated model, and the carbon flow data aggregation specification from the production link integrated model to the full-process model, and construct a standardized data interaction mechanism between levels. Through this mechanism, realize the real-time interaction and consistency verification of carbon flow data at the production equipment level, production link level and full-process level, form a hierarchical digital twin model closed loop, and ensure that the carbon flow status output by the model is kept in spatiotemporal synchronization with the carbon flow dynamics of the physical production system.

[0083] Please see Figure 3 Preferably, the specific steps for correcting biases in digital twin models include:

[0084] Based on the element perception module, the spatiotemporal correlation dataset of the current production cycle is collected in real time. Real-time carbon flow feature variables and corresponding production parameter combinations are extracted. The multi-level carbon flow models in the digital twin model that associates carbon flow feature variables and corresponding production parameter combinations are located, including micro-mapping models, production link integration models, and full-process models. A data model association mapping table is established.

[0085] Based on the data model association mapping table, for the located multi-level carbon flow model, the model prediction values ​​of the corresponding carbon flow characteristic variables are obtained according to the current combination of production parameters. The model prediction values ​​are compared with the corresponding real-time carbon flow characteristic variables, and the prediction deviation value is calculated by the root mean square error formula. At the same time, the prediction deviation values ​​are accumulated for each multi-level carbon flow model, and multi-level cumulative deviations are generated, including cumulative deviations at the production equipment level, production link level, and whole process level. The time of occurrence and duration of the deviation are recorded synchronously.

[0086] Configure multi-level cumulative deviation thresholds, including production equipment level cumulative deviation thresholds, production process level cumulative deviation thresholds, and end-process level cumulative deviation thresholds; when the cumulative predicted deviation value exceeds the corresponding multi-level cumulative deviation threshold, trigger multi-level carbon flow model correction, including micro-mapping model correction, production process integrated model correction, and end-process model correction.

[0087] The micro-mapping model correction is used to eliminate cumulative bias at the production equipment level, ensuring that the predicted values ​​of the production equipment-level carbon flow characteristic variables output by the micro-mapping model are consistent with the corresponding real-time carbon flow characteristic variables. The correction method involves calling the common association rules for that production equipment in the carbon flow association rule library, combining them with the cumulative bias value at the production equipment level, and using a parameter adaptive optimization algorithm to adjust the parameter coefficients of the fundamental physical equations in the micro-mapping model. The adjustment magnitude is positively correlated with the cumulative bias at the production equipment level, thereby correcting the micro-mapping model and achieving convergence of the predicted values ​​to the real-time values. Specifically, after calling the common association rules for the corresponding production equipment, the parameter adaptive optimization algorithm is initiated based on the magnitude of the cumulative bias at the production equipment level. This algorithm adjusts the parameter coefficients of the fundamental physical equations according to the association logic between the production parameter combinations and carbon flow characteristic variables in the common association rules, with the adjustment magnitude increasing accordingly as the cumulative bias at the production equipment level increases. Through this dynamic adjustment, the predicted values ​​of the production equipment-level carbon flow characteristic variables gradually approach the real-time values, thereby eliminating the cumulative bias at the production equipment level and ensuring consistency between the two.

[0088] The production process integrated model correction is used to eliminate cumulative bias at the production process level, ensuring that the comprehensive carbon flow characteristic variables output by the production process integrated model are consistent with the corresponding real-time comprehensive carbon flow characteristic variables. The correction method is as follows: based on the output results of the micro-mapping model correction, combined with the cumulative bias values ​​at the production process level, the carbon flow weighted aggregation algorithm is re-run. This dynamically calibrates the carbon flow transfer parameters between production equipment and adjusts the calculation weights of the comprehensive carbon flow characteristic variables at the production process level. The weight correction amount is correlated with the cumulative bias at the production process level, thereby correcting the production process integrated model. Specifically, using the output results of the micro-mapping model correction as the base data, the carbon flow weighted aggregation algorithm is re-run with reference to the cumulative bias at the production process level. During the run, the carbon flow transfer parameters between production equipment are dynamically calibrated according to the magnitude of the cumulative bias at the production process level, and the calculation weights of the comprehensive carbon flow characteristic variables at the production process level are adjusted accordingly. The degree of weight correction is correlated with the cumulative bias at the production process level. Through this operation, the predicted values ​​of the comprehensive carbon flow characteristic variables at the production process level tend to be consistent with the real-time values, eliminating the cumulative bias at the production process level.

[0089] The end-to-end model correction is used to eliminate end-to-end cumulative bias, ensuring that the end-to-end carbon flow characteristic variables output by the end-to-end model are consistent with the corresponding real-time end-to-end carbon flow characteristic variables. The correction method is as follows: based on the output of the production link integrated model after correction, combined with the end-to-end cumulative bias value, the end-to-end carbon flow aggregation algorithm is re-executed to correct the carbon flow interaction parameters between production links. The parameter compensation amount matches the end-to-end cumulative bias, thereby correcting the end-to-end model and compensating for the carbon flow transfer bias across production links. Specifically, based on the output of the production link integrated model after correction, combined with the end-to-end cumulative bias, the end-to-end carbon flow aggregation algorithm is re-executed. During this process, the carbon flow interaction parameters between production links are corrected according to the magnitude of the end-to-end cumulative bias, and the parameter compensation degree matches the end-to-end cumulative bias. This method compensates for the carbon flow transfer bias across production links, ensuring that the predicted values ​​of the end-to-end carbon flow characteristic variables are consistent with the real-time values, thus eliminating end-to-end cumulative bias.

[0090] After the multi-level carbon flow model is corrected, the cumulative deviation of the corrected multi-level carbon flow model is cleared, and the cumulative deviation value of each multi-level carbon flow model is updated. The number of corrections for each multi-level carbon flow model is recorded synchronously. If the number of corrections for the same multi-level carbon flow model exceeds the preset correction threshold within the preset monitoring time window, a digital twin model anomaly warning is triggered to indicate that there may be abnormalities in the model structure or data acquisition.

[0091] During the deviation correction process, the quantitative mapping relationship of the carbon flow dynamic correlation operator of the production parameter combination is updated synchronously to ensure that the operator can accurately reflect the correlation law between the corrected model parameters and the carbon flow characteristic variables. After the correction is completed, the deviation between the output of each level model and the benchmark dataset is verified through the consistency verification mechanism to see if it is lower than the threshold. If it still exceeds the threshold, the correction process is repeated until the cumulative deviation of all multi-level carbon flow models meets the requirements, so as to achieve dynamic consistency between the digital twin model and the physical production system.

[0092] The intelligent decision-making module, supported by a hierarchical digital twin model output from the twin fusion module, aims to generate optimized production parameter combinations based on the prediction results and anomaly identification results of carbon flow characteristic variables, thereby achieving precise control of carbon flow throughout the entire green methanol production process. Its core functions are: periodically predicting carbon flow characteristic variables by configuring the monitoring cycle and based on the production parameter combinations configured within the current production cycle; constructing a benchmark variable sequence based on a carbon flow association rule base, including: the absolute deviation sequence of the four carbon flow elements in the production process and the relative deviation sequence at the production equipment level, to identify abnormal production processes and equipment; using the hierarchical digital twin model to back-calculate and generate target production parameter combinations for abnormal equipment, driving the actuators to adjust parameters, and dynamically optimizing through a feedback mechanism, ultimately controlling carbon dioxide emissions within a preset range, overcoming the limitations of traditional decision-making that relies on experience and suffers from lagging control.

[0093] In this embodiment, the hierarchical digital twin model is constructed based on the association rules between the production parameter combinations and carbon flow characteristic variables of historical processing cycles in green methanol production. After being corrected for actual data deviations, it is adapted to the current processing cycle data, providing a reliable basis for predicting the four elements of carbon flow and identifying anomalies. The monitoring cycle setting must meet the following requirements: it should be able to capture dynamic fluctuations in carbon flow in a timely manner, while avoiding increasing the data processing burden due to frequent monitoring. When identifying anomalies, the proportion of carbon loss in the entire process is first calculated and compared with the threshold to determine whether there is a global emission anomaly; then, the absolute deviation between the actual distribution of the four elements of carbon flow in each production link and the baseline distribution is compared, or the deviation trend is analyzed to mark the abnormal production link; finally, the abnormal production equipment is located by the relative deviation at the production equipment level, which greatly reduces the amount of engineering work required to directly locate the production equipment. When an anomaly is identified, it indicates that the association rules between the production parameter combinations and carbon flow characteristic variables have changed. When generating the target production parameter combination, the carbon flow characteristic variables corresponding to the current parameters of abnormal production equipment are first matched from the carbon flow association rule library as the target benchmark value. After replacing the actual value, the emissions of the entire process are checked to see if they meet the standards. If they meet the standards, the production parameter combination is reversed using the Newton-Raphson iteration method. After boundary verification and simulation, the target production parameter combination is determined. If they do not meet the standards, an early warning is triggered, indicating that the emission anomaly originates from non-production equipment. After the target parameters are transmitted to the actuator, data is collected in real time through the element perception module to form feedback, and the twin model prediction benchmark is updated synchronously and the parameter effectiveness is evaluated. If the standards are not met, a new round of regulation is initiated until the carbon flow meets the preset requirements, ensuring the low-carbon nature and stability of the production process and improving the carbon flow control capability of the entire process.

[0094] Please see Figure 4 Preferably, the specific steps for identifying abnormal production processes and equipment include:

[0095] Because the production process is dynamic, carbon flow characteristic variables fluctuate over time. Excessive monitoring can increase the data processing burden. Therefore, a monitoring cycle is configured to determine the prediction period for carbon flow characteristic variables and the frequency of anomaly identification, based on the stability requirements of the production process and the rate of change of the carbon flow characteristic variables. At the beginning of each monitoring cycle, based on the current combination of production parameters, a hierarchical digital twin model predicts the carbon flow characteristic variables corresponding to the current combination of production parameters within that monitoring cycle. Furthermore, based on the carbon flow elements of the carbon flow characteristic variables and the comprehensive carbon flow characteristic variables output by the full-process model, the distribution data of the four carbon flow elements—raw material carbon, conversion carbon, product carbon, and loss carbon—are obtained throughout the entire process.

[0096] The carbon loss ratio throughout the entire process is a core indicator reflecting whether overall emissions are abnormal. If this ratio exceeds the standard, it indicates an abnormal carbon loss in the production system. Based on the acquired distribution data of the four elements of carbon flow throughout the entire process, the proportion of carbon emissions lost throughout the process to the total carbon input of raw materials is calculated and compared with a preset carbon loss ratio threshold. If it is greater than the carbon loss ratio threshold, it is determined that carbon dioxide emissions are abnormal, and the anomaly identification process of the production process is initiated. Otherwise, the current combination of production parameters continues to be executed. This method quickly locates whether there are problems in the entire process, avoids the waste of resources caused by directly delving into the production process, and overcomes the limitations of traditional local monitoring in judging the overall emission status.

[0097] Under different combinations of production parameters, the normal range of carbon flow characteristic variables varies, and relying solely on a fixed benchmark can lead to biased anomaly detection. Based on a carbon flow association rule library, common association rules related to the current production parameter combination are retrieved according to the configuration of the current production cycle, and the corresponding carbon flow characteristic variables are obtained. Based on the current hierarchical digital twin model, a comprehensive carbon flow characteristic variable sequence for each production stage in the current production cycle is constructed to generate a benchmark distribution data sequence of the four elements of carbon flow for each production stage.

[0098] It is normal for carbon flow characteristic variables to deviate from the benchmark in actual production, but excessive deviation may indicate anomalies. We extract the actual distribution data sequences of the four carbon flow elements for each production stage within the current production cycle, compare these actual distribution data sequences with the benchmark distribution data sequences for the corresponding production stages, and calculate the absolute carbon flow deviation. For example, we calculate the proportion of the difference between the actual distribution data and the benchmark distribution data for each of the four carbon flow elements to the benchmark distribution data, and construct the absolute carbon flow deviation sequence for each production stage in the current production cycle.

[0099] Configure a carbon flow deviation threshold. If there is a carbon flow absolute deviation greater than the carbon flow deviation threshold in the carbon flow absolute deviation sequence of the four elements of carbon flow, it indicates that the carbon flow in the production process has become significantly abnormal. Then, the corresponding production process will be marked as an abnormal production process.

[0100] Otherwise, the trend of change is extracted from the absolute deviation sequence of carbon flow in each production link through a trend fitting algorithm, including the trend slope and the trend duration. If the trend slope is greater than the preset trend slope threshold and the trend duration is greater than the preset duration threshold, then the production link is determined to have an abnormal trend and is marked as an abnormal production link. Some abnormalities do not exceed the threshold instantly, but gradually deteriorate. Judging by the threshold alone will miss the opportunity for early intervention.

[0101] Extract the actual sequence of carbon flow characteristic variables of each production equipment in the abnormal production process of the current production cycle. At the same time, based on the common association rules corresponding to each production equipment in the carbon flow association rule base and the current production parameter combination, generate the benchmark sequence of carbon flow characteristic variables of the production equipment.

[0102] Configure the deviation threshold for production equipment, calculate the relative deviation between the actual sequence and the benchmark sequence of carbon flow characteristic variables for each production equipment, construct the relative deviation sequence at the production equipment level, and mark the production equipment as abnormal if there is a relative deviation in the sequence that is greater than the deviation threshold for production equipment.

[0103] Please see Figure 5 Preferably, the specific steps for generating the target production parameter combination for abnormal production equipment include:

[0104] Based on the labeled abnormal production equipment, the real-time production parameter combination of the abnormal production equipment in the current production cycle is retrieved, and the carbon flow feature variable corresponding to the current production parameter combination is matched from the carbon flow association rule library as the target benchmark value. The target benchmark value reflects the carbon flow status corresponding to the current parameters under normal operating conditions.

[0105] The actual carbon flow characteristic variables of the abnormal production equipment in the current production cycle are replaced with the corresponding target baseline values. Based on the hierarchical digital twin model, the comprehensive carbon flow characteristic variables output by the whole process model are obtained. The proportion of carbon emissions lost in the whole process to the total carbon input of raw materials is calculated and compared with the preset carbon loss ratio threshold to determine whether there is an anomaly in carbon dioxide emissions. If yes, it indicates that the target baseline value does not meet the carbon emission optimization requirements of the whole process, and the overall emissions cannot be made up by adjusting the parameters of a single production equipment. In this case, an abnormal production equipment warning is triggered to indicate that the cause of the current carbon dioxide emission anomaly is not a production equipment. If no, it indicates that the target baseline value can meet the carbon emission requirements of the whole process, and carbon flow optimization can be achieved by adjusting the parameters of a single production equipment. In this case, a reverse solution of the production parameter combination is triggered.

[0106] When the production parameter combination inverse solution is triggered, the micro-mapping model of the production equipment in the hierarchical digital twin model is called to construct the inverse mapping relationship between the production parameter combination and the target benchmark value; the target benchmark value is input, and the production parameter combination inverse solution is performed through Newton's iteration method. The initial parameter production adjustment step size is set, and the production parameter combination of abnormal production equipment is obtained by iterative calculation with the target benchmark value as the constraint condition.

[0107] Boundary verification is performed on the production parameter combinations obtained by reverse calculation to ensure that the values ​​of each parameter are within the safe operating range specified in the production equipment operating procedures, and to eliminate parameter combinations that exceed the carrying capacity of the production equipment.

[0108] The verified production parameter combination is input into the hierarchical digital twin model for simulation verification. The model is used to deduce whether the carbon flow characteristic variables of the production equipment under the parameter combination stably reach the corrected target value. At the same time, the impact on the carbon flow transfer of upstream and downstream production equipment is checked to ensure that there is no risk of chain anomalies.

[0109] Using the production parameter combination that has passed simulation verification as the target production parameter combination for abnormal production equipment can not only bring the carbon flow characteristic variables of the production equipment back to the optimized state, but also adapt to the carbon emission control requirements of the whole process, thus achieving the synergy between local production equipment parameter adjustment and global carbon flow optimization.

[0110] The target production parameter combination of the abnormal production equipment is transmitted to the corresponding actuator, including valve controller, power regulator, flow control valve, etc., to drive the abnormal production equipment to adjust the combination of production parameters such as operating temperature, pressure, and flow according to the target production parameter combination;

[0111] Simultaneously, the element perception module collects production data and carbon flow characteristic variables in real time after parameter adjustments, with the sampling frequency consistent with the monitoring cycle. This monitors the actual impact of the target production parameter combination on the carbon flow characteristic variables, forming a control execution feedback dataset. This dataset is then synchronously fed back to the twin fusion module and the intelligent decision-making module. Feedback to the twin fusion module updates the prediction benchmark of the hierarchical digital twin model, while feedback to the intelligent decision-making module assesses the effectiveness of the target production parameter combination. If the carbon flow characteristic variables still do not meet the carbon dioxide emission requirements after adjustment, a new round of control instruction generation and execution is initiated in conjunction with the intelligent decision-making module until the carbon flow characteristic variables meet the preset emission requirements, ensuring that carbon dioxide emissions throughout the green methanol production process are within compliance limits.

[0112] Example 2

[0113] This embodiment describes a method for monitoring and controlling carbon dioxide emissions, including:

[0114] Step S1: Collect the production parameter combinations and corresponding carbon flow characteristic variables of the production equipment in each production stage through a multi-dimensional sensor network constructed by digital sensors; construct a spatiotemporal correlation dataset based on the spatiotemporal correlation mapping system between the production parameter combinations and carbon flow characteristic variables.

[0115] Step S2: Based on the historical spatiotemporal correlation dataset, perform association rule mining and dynamic correlation analysis on the combination of production parameters and carbon flow characteristic variables, extract the association rules between the combination of production parameters and carbon flow characteristic variables, identify common association rules, construct a carbon flow association rule library, embed the carbon flow association rule library into a multi-level carbon flow model to construct a hierarchical digital twin model, and correct the bias of the hierarchical digital twin model based on the spatiotemporal correlation dataset of the current production cycle.

[0116] Step S3: Based on the hierarchical digital twin model, periodically predict the carbon flow characteristic variables in the current production cycle based on the combination of production parameters, obtain the distribution data of the four elements of carbon flow, and use them to determine whether there are any abnormalities in carbon dioxide emissions; construct a benchmark variable sequence based on the carbon flow association rule base, identify abnormal production links and abnormal production equipment, generate the target production parameter combination for abnormal production equipment, and provide execution feedback.

[0117] Preferably, the specific steps for constructing a carbon flow association rule base include:

[0118] Obtain the historical spatiotemporal correlation dataset, divide it according to the production cycle, and construct the historical spatiotemporal correlation subset;

[0119] Based on the historical spatiotemporal correlation subsets of each production cycle, establish a mapping table between the combination of production parameters and carbon flow characteristic variables for each production link;

[0120] An association rule mining algorithm is applied to the mapping tables of each production link in the historical spatiotemporal correlation subset to identify explicit association rules between the combination of production parameters and carbon flow characteristic variables, and to construct an explicit association rule set.

[0121] Dynamic correlation analysis was performed on the time series of production parameter combinations and carbon flow characteristic variables to capture the implicit correlation rules between the time series characteristics of production parameter combinations and the dynamic response characteristics of carbon flow characteristic variables, and a set of implicit correlation rules was constructed.

[0122] Summarize the explicit and implicit association rule sets of each historical spatiotemporal associated subset to construct the total set of production-level association rules corresponding to the historical spatiotemporal associated subset;

[0123] A comparative analysis was conducted on the set of production-level association rules for different historical spatiotemporal association subsets, and the frequency ratio of each association rule in different historical spatiotemporal association subsets was statistically analyzed. The association rules include explicit association rules and implicit association rules.

[0124] If the frequency of an association rule exceeds a preset frequency threshold, it is marked as a common association rule, standardized, and used to build a carbon flow association rule library.

[0125] Preferably, the specific steps for identifying abnormal production processes and abnormal production equipment include:

[0126] Configure monitoring cycles, and at the beginning of each monitoring cycle, based on the configuration of production parameter combinations, predict the carbon flow characteristic variables corresponding to the current production parameter combinations within that monitoring cycle through a hierarchical digital twin model;

[0127] Based on the carbon flow characteristic variables of the carbon flow, and according to the comprehensive carbon flow characteristic variables output by the whole process model, the distribution data of the four elements of carbon flow are obtained respectively.

[0128] Based on the distribution data of the four elements of carbon flow throughout the entire process, the proportion of carbon emissions lost throughout the entire process to the total carbon input of raw materials is calculated and compared with the preset carbon loss ratio threshold. If it is greater than the carbon loss ratio threshold, it is determined that there is an abnormality in carbon dioxide emissions and the abnormal identification process of the production process is initiated; otherwise, the current combination of production parameters continues to be executed.

[0129] Based on the carbon flow association rule library, according to the configuration of production parameter combinations, the common association rules involving the current production parameter combinations are retrieved to obtain the corresponding carbon flow feature variables. Based on the current hierarchical digital twin model, the comprehensive carbon flow feature variable sequence of each production link in the current production cycle is constructed, and the benchmark distribution data sequence of the four elements of carbon flow in each production link is generated.

[0130] Extract the actual distribution data sequence of the four carbon flow elements in each production stage within the current production cycle, compare the actual distribution data sequence with the benchmark distribution data sequence of the corresponding production stage, calculate the absolute deviation of carbon flow, and construct the absolute deviation sequence of the four carbon flow elements in each production stage of the current production cycle.

[0131] Configure a carbon flow deviation threshold. If there is a carbon flow absolute deviation in the carbon flow absolute deviation sequence of the four carbon flow elements that is greater than the carbon flow deviation threshold, the corresponding production link is marked as an abnormal production link. Otherwise, the trend of change is extracted from the carbon flow absolute deviation sequence of each production link through a trend fitting algorithm to determine whether the production link is an abnormal production link.

[0132] Extract the actual sequence of carbon flow characteristic variables of each production equipment in the abnormal production process of the current production cycle, and generate the benchmark sequence of carbon flow characteristic variables of production equipment based on the common association rules of the combination of production parameters of each production equipment in the carbon flow association rule base.

[0133] Configure the deviation threshold for production equipment, calculate the relative deviation between the actual sequence and the baseline sequence of carbon flow characteristic variables for each production equipment, and construct a relative deviation sequence at the production equipment level to mark abnormal production equipment.

[0134] Working principle and its effects:

[0135] The carbon dioxide emission monitoring system and control method of this invention first collects the combination of production parameters and carbon flow characteristic variables of each production stage through a multi-dimensional sensor network to construct a spatiotemporal correlation dataset. Based on historical data, association rules are mined, common rules are identified to construct a carbon flow association rule library, and a multi-level carbon flow model is embedded to form a hierarchical digital twin model, which is then combined with current data to correct deviations. This model is used to periodically predict carbon flow characteristic variables, obtain the distribution data of the four elements of carbon flow, and determine whether emissions are abnormal. A benchmark sequence is constructed based on the rule library to first locate abnormal production stages and then identify abnormal equipment, avoiding the large workload of direct location. After anomaly identification, the equipment micro-mapping model is invoked, and the target production parameter combination is solved using the Newton-Raphson iteration method. After verification, the result is transmitted to the actuator for adjustment, and the adjusted data is collected for feedback, forming a closed-loop management system. This process solves the problems of localized monitoring, large anomaly identification deviations, and lagging regulation in traditional methods, achieving precise carbon flow control throughout the entire process. It enables real-time carbon emission perception, dynamic modeling, accurate anomaly diagnosis, and intelligent regulation, reducing carbon emissions, ensuring low-carbon and stable production, and improving carbon flow control capabilities.

[0136] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A carbon dioxide emission monitoring system, characterized in that, include: A multi-dimensional sensor network constructed using digital sensors is used to collect the combination of production parameters and corresponding carbon flow characteristic variables of production equipment in each production stage. Based on the spatiotemporal correlation mapping system between the production parameter combination and the carbon flow characteristic variables, a spatiotemporal correlation dataset is constructed; Based on historical spatiotemporal correlation datasets, association rule mining and dynamic correlation analysis are performed on the production parameter combinations and carbon flow characteristic variables. The association rules between the production parameter combinations and carbon flow characteristic variables are extracted, common association rules are identified, a carbon flow association rule library is constructed, and the carbon flow association rule library is embedded into a multi-level carbon flow model to construct a hierarchical digital twin model. The bias of the hierarchical digital twin model is corrected based on the spatiotemporal correlation dataset of the current production cycle. Based on the hierarchical digital twin model, the carbon flow characteristic variables in the current production cycle are periodically predicted based on the combination of production parameters, and the distribution data of the four elements of carbon flow are obtained to determine whether there are any abnormalities in carbon dioxide emissions. Based on the carbon flow association rule base, a baseline variable sequence is constructed to identify abnormal production processes and abnormal production equipment, generate target production parameter combinations for abnormal production equipment, and provide execution feedback.

2. The carbon dioxide emission monitoring system as described in claim 1, characterized in that, The multi-level carbon flow model includes: a micro-mapping model, a production process integration model, and a full-process model; The micro-mapping model is used to quantify the micro-dynamics of carbon flow characteristic variables in production equipment as production parameter combinations change. It extracts common association rules between production parameter combinations and carbon flow characteristic variables of corresponding production equipment from the carbon flow association rule library. Based on the common association rules, it establishes the basic physical equation of carbon flow transfer based on the physical structure parameters of the production equipment. It also transforms the mapping relationship between production parameter combinations and carbon flow characteristic variables in the common association rules into parameter coefficients in the basic physical equation. The model is constructed using physical mechanism modeling and association rule parameterization. The production process integration model is used to quantify the overall carbon flow status within the production process. Based on the physical connection relationship of the production equipment within the production process, the micro-mapping model within the same production process is integrated according to the physical connection relationship between the production equipment. The carbon flow transfer parameters between the production equipment are defined based on the carbon flow characteristic variables output by the production equipment. The model is constructed by using a carbon flow weighted aggregation algorithm to build the comprehensive carbon flow characteristic variables of each production process. The full-process model is used to quantify the overall carbon flow status of the entire production process. Based on the physical topology of the production process, the production process integration model of each production process is integrated according to the physical connection relationship of the production process. Based on the carbon flow transmission path between the production processes, the interaction parameters of the comprehensive carbon flow characteristic variables between the production processes are defined. Based on the comprehensive carbon flow characteristic variables output by the production process integration model, the model is constructed by aggregating the carbon flow throughout the entire process.

3. The carbon dioxide emission monitoring system as described in claim 1, characterized in that, The specific steps for constructing the carbon stream association rule base include: Obtain the historical spatiotemporal correlation dataset, divide it according to the production cycle, and construct the historical spatiotemporal correlation subset; Based on the historical spatiotemporal correlation subsets of each production cycle, establish a mapping table between the combination of production parameters and carbon flow characteristic variables for each production link; An association rule mining algorithm is applied to the mapping tables of each production link in the historical spatiotemporal correlation subset to identify explicit association rules between the combination of production parameters and carbon flow characteristic variables, and to construct an explicit association rule set. Dynamic correlation analysis was performed on the time series of production parameter combinations and carbon flow characteristic variables to capture the implicit correlation rules between the time series characteristics of production parameter combinations and the dynamic response characteristics of carbon flow characteristic variables, and a set of implicit correlation rules was constructed. Summarize the explicit and implicit association rule sets of each historical spatiotemporal associated subset to construct the total set of production-level association rules corresponding to the historical spatiotemporal associated subset; A comparative analysis was conducted on the set of production-level association rules for different historical spatiotemporal association subsets, and the frequency ratio of each association rule in different historical spatiotemporal association subsets was statistically analyzed. The association rules include explicit association rules and implicit association rules. If the frequency of an association rule exceeds a preset frequency threshold, it is marked as a common association rule, standardized, and used to build a carbon flow association rule library.

4. The carbon dioxide emission monitoring system as described in claim 3, characterized in that, The construction of the explicit association rule set includes: The continuous production parameter combinations and carbon flow characteristic variables are discretized into intervals, and the enumerated production parameter combinations and carbon flow characteristic variables are encoded to generate a structural variable mapping table. An association rule mining algorithm is applied to the structural variable mapping table to identify explicit association rules between combinations of production parameters and carbon flow characteristic variables. By setting a support threshold, combinations of production parameters and carbon flow characteristic variables with statistical significance are selected as frequent itemsets. Strong association rules with a confidence level greater than a preset threshold are extracted from the frequent itemset and used as explicit association rules to construct an explicit association rule set. The construction of the implicit association rule set includes: The time-series features of the production parameter combination are extracted by time-series decomposition. The time-series features include the trend change features, periodic fluctuation features and sudden fluctuation features of the production parameter combination. The dynamic response features of the carbon flow feature variables with the production parameter combination are statistically analyzed. The dynamic response features include response time features and response amplitude features. By using dynamic time warping to identify the time-series characteristics of production parameter combinations and the dynamic response characteristics of carbon flow characteristic variables, a set of implicit association rules is constructed.

5. The carbon dioxide emission monitoring system as described in claim 1, characterized in that, The steps for correcting biases in the hierarchical digital twin model include: Collect the spatiotemporal correlation dataset of the current production cycle, extract real-time carbon flow feature variables and corresponding production parameter combinations, locate the multi-level carbon flow model in the digital twin model that associates carbon flow feature variables and corresponding production parameter combinations, and establish a data model association mapping table. Based on the data model association mapping table, for the multi-level carbon flow model of the location, the model prediction value of the corresponding carbon flow characteristic variable is obtained according to the current production parameter combination. The model prediction value is compared with the corresponding real-time carbon flow characteristic variable, and the prediction deviation value is calculated. The prediction deviation values ​​are accumulated separately according to the multi-level carbon flow model, and the multi-level cumulative deviation is generated separately. The multi-level cumulative deviation threshold is configured. When the cumulative prediction deviation value is greater than the corresponding multi-level cumulative deviation threshold, the multi-level carbon flow model is corrected, including micro-mapping model correction, production process integrated model correction and full-process model correction. After the multi-level carbon flow model is corrected, the cumulative deviation of the corrected multi-level carbon flow model is cleared, and the cumulative deviation value of each multi-level carbon flow model is updated. The number of corrections for each multi-level carbon flow model is recorded synchronously. If the number of corrections for the multi-level carbon flow model exceeds the preset correction threshold within the preset monitoring time window, an abnormal warning for the digital twin model is triggered.

6. The carbon dioxide emission monitoring system as described in claim 5, characterized in that, The micro-mapping model correction is used to eliminate the cumulative deviation at the production equipment level, so that the predicted value of the carbon flow characteristic variable at the production equipment level output by the micro-mapping model is consistent with the corresponding real-time carbon flow characteristic variable. By calling the common association rules in the carbon flow association rule library and combining the cumulative deviation value at the production equipment level, the parameter adaptive optimization algorithm is used to adjust the parameter coefficients of the basic physical equation in the micro-mapping model, thereby correcting the micro-mapping model. The production process integrated model correction is used to eliminate cumulative deviations at the production process level, so that the production process integrated model outputs the production process level comprehensive carbon flow characteristic variables consistent with the corresponding real-time comprehensive carbon flow characteristic variables. By combining the output results after correction based on the micro-mapping model with the production process level cumulative deviation value, the carbon flow weighted aggregation algorithm is re-run to dynamically calibrate the carbon flow transfer parameters between production equipment, adjust the calculation weights of the production process level comprehensive carbon flow characteristic variables, and correct the production process integrated model. The full-process model correction is used to eliminate the cumulative deviation at the full-process level, so that the full-process carbon flow characteristic variables output by the full-process model are consistent with the corresponding real-time full-process carbon flow characteristic variables. By combining the output results of the production link integrated model correction with the full-process cumulative deviation value, the full-process carbon flow aggregation algorithm is re-executed to correct the carbon flow interaction parameters between production links and correct the full-process model.

7. The carbon dioxide emission monitoring system as described in claim 2, characterized in that, The specific steps for identifying abnormal production processes and abnormal production equipment include: Configure monitoring cycles, and at the beginning of each monitoring cycle, based on the configuration of production parameter combinations, predict the carbon flow characteristic variables corresponding to the current production parameter combinations within that monitoring cycle through a hierarchical digital twin model; Based on the carbon flow characteristic variables of the carbon flow, and according to the comprehensive carbon flow characteristic variables output by the whole process model, the distribution data of the four elements of carbon flow are obtained respectively. Based on the distribution data of the four elements of carbon flow throughout the entire process, the proportion of carbon emissions lost throughout the entire process to the total carbon input of raw materials is calculated and compared with the preset carbon loss ratio threshold. If it is greater than the carbon loss ratio threshold, it is determined that there is an abnormality in carbon dioxide emissions and the abnormal identification process of the production process is initiated; otherwise, the current combination of production parameters continues to be executed. Based on the carbon flow association rule library, according to the configuration of production parameter combinations, the common association rules involving the current production parameter combinations are retrieved to obtain the corresponding carbon flow feature variables. Based on the current hierarchical digital twin model, the comprehensive carbon flow feature variable sequence of each production link in the current production cycle is constructed, and the benchmark distribution data sequence of the four elements of carbon flow in each production link is generated.

8. The carbon dioxide emission monitoring system as described in claim 7, characterized in that, The specific steps for identifying abnormal production processes and abnormal production equipment also include: Extract the actual distribution data sequence of the four carbon flow elements in each production stage within the current production cycle, compare the actual distribution data sequence with the benchmark distribution data sequence of the corresponding production stage, calculate the absolute deviation of carbon flow, and construct the absolute deviation sequence of the four carbon flow elements in each production stage of the current production cycle. Configure a carbon flow deviation threshold. If there is a carbon flow absolute deviation in the carbon flow absolute deviation sequence of the four carbon flow elements that is greater than the carbon flow deviation threshold, the corresponding production link is marked as an abnormal production link. Otherwise, the trend of change is extracted from the carbon flow absolute deviation sequence of each production link through a trend fitting algorithm to determine whether the production link is an abnormal production link. Extract the actual sequence of carbon flow characteristic variables of each production equipment in the abnormal production process of the current production cycle, and generate the benchmark sequence of carbon flow characteristic variables of production equipment based on the common association rules of the combination of production parameters of each production equipment in the carbon flow association rule base. Configure the deviation threshold for production equipment, calculate the relative deviation between the actual sequence and the baseline sequence of carbon flow characteristic variables for each production equipment, and construct a relative deviation sequence at the production equipment level to mark abnormal production equipment.

9. The carbon dioxide emission monitoring system as described in claim 1, characterized in that, The specific steps for generating the target production parameter combination for the abnormal production equipment include: Based on the labeled abnormal production equipment, the real-time production parameter combination of the abnormal production equipment in the current production cycle is retrieved, and the carbon flow feature variable corresponding to the current production parameter combination is matched from the carbon flow association rule library as the target benchmark value. Replace the actual carbon flow characteristic variables of the abnormal production equipment in the current production cycle with the corresponding target benchmark value, obtain the comprehensive carbon flow characteristic variables output by the whole process model based on the hierarchical digital twin model, and determine whether there is an abnormality in carbon dioxide emissions; if so, trigger an early warning for abnormal production equipment. If not, trigger the inverse solution of production parameter combinations, call the micro-mapping model of the production equipment in the hierarchical digital twin model, construct the inverse mapping relationship between the production parameter combinations and the target benchmark value; input the target benchmark value, perform the inverse solution of production parameter combinations through Newton's iteration method, set the initial parameter production adjustment step size, iterate and calculate with the target benchmark value as the constraint condition, obtain the production parameter combinations of abnormal production equipment, and determine the target production parameter combinations through boundary verification and simulation verification.

10. A method for monitoring and controlling carbon dioxide emissions, implemented based on the carbon dioxide emission monitoring system according to any one of claims 1-9, characterized in that, include: Step S1: Collect the production parameter combinations and corresponding carbon flow characteristic variables of the production equipment in each production stage through a multi-dimensional sensor network constructed by digital sensors; Based on the spatiotemporal correlation mapping system between the production parameter combination and the carbon flow characteristic variables, a spatiotemporal correlation dataset is constructed; Step S2: Based on the historical spatiotemporal correlation dataset, perform association rule mining and dynamic correlation analysis on the combination of production parameters and the carbon flow characteristic variables, extract the association rules between the combination of production parameters and the carbon flow characteristic variables, identify common association rules, construct a carbon flow association rule library, embed the carbon flow association rule library into a multi-level carbon flow model to construct a hierarchical digital twin model, and correct the bias of the hierarchical digital twin model based on the spatiotemporal correlation dataset of the current production cycle. Step S3: Based on the hierarchical digital twin model, periodically predict the carbon flow characteristic variables in the current production cycle based on the combination of production parameters, and obtain the distribution data of the four elements of carbon flow to determine whether there are any abnormalities in carbon dioxide emissions. Based on the carbon flow association rule base, a baseline variable sequence is constructed to identify abnormal production processes and abnormal production equipment, generate target production parameter combinations for abnormal production equipment, and provide execution feedback.

Images