Method and system for reducing carbon emission based on dynamic economic behaviors

Abstract

The invention relates to the technical field of data management, discloses a method and a system for reducing carbon emission based on dynamic economic behaviors, and realizes built-in monitoring, evaluation and feedback links of a low-carbon measure conduction correlation model based on an industry correlation objective function, a power enterprise decision value function and an energy consumer utility correlation type. According to the parameter and tool combination of the measures of continuously and adaptively optimizing and reducing the carbon emission according to the implementation feedback, energy consumers which cannot be covered by pure subsidy are effectively activated by improving the linkage characteristics of the utility association of the energy consumers according to the energy consumer collaborative strategy, and the energy consumer collaborative strategy is optimized for power enterprises. Model linkage of a power enterprise decision value function can change a value function decision result, reduce technical uncertainty, equivalently adjust a loss aversion coefficient and current prejudice parameters, and successfully excite core low-carbon technology investment, so that the degree of connection between a measure for reducing carbon emission and upstream and downstream industries of the power industry is improved, and the power enterprise decision value function model linkage method is suitable for popularization and application. And the low-carbon measure conduction path and effect efficiency are enhanced.

Description

Technical Field

[0001] This invention relates to a method and system for reducing carbon emissions based on dynamic economic behavior, belonging to the field of data management technology. Background Technology

[0002] With increasing global attention on climate change, reducing carbon emissions has become a crucial goal for economic development and industrial transformation in various countries. Traditional carbon emission management methods typically rely on static industry emission standards, fixed energy consumption patterns, and pre-set emission reduction pathways. These methods have limitations in data collection, processing, and analysis, making it difficult to adapt to dynamically changing economic behaviors and market environments.

[0003] Existing methods for classifying industry characteristic data and energy consumer characteristic data are mostly based on fixed rules or historical experience, lacking the ability to identify and adaptively classify dynamic economic behavior characteristics in real time. This makes it difficult to accurately reflect the structural changes and consumption pattern shifts in current economic activities when generating classification feature parameters, affecting the accuracy of subsequent matching and comparative analysis. Currently, most systems rely on pre-stored first and second benchmark feature libraries for difference comparison. However, these benchmark libraries are often updated infrequently, failing to incorporate the latest industry technology developments, policy adjustments, and market supply and demand changes in a timely manner. The disconnect between static benchmark data and dynamic reality makes the matching difference results unable to truly reflect the current potential space and urgent direction for emission reduction. Existing path feature databases often use fixed or limited sets of optimized paths, lacking the ability to dynamically adjust for real-time difference data. When comparing the path feature parameter set with the difference data, the system often only performs one-way matching, failing to form a closed-loop feedback mechanism that updates path optimization labels in real time based on the comparison results. This results in a low degree of matching between path optimization strategies and actual emission reduction needs, making it difficult to achieve precise and flexible dynamic control. Summary of the Invention

[0004] This invention provides a method and system for reducing carbon emissions based on dynamic economic behavior, which can accurately pinpoint the difference between the current situation and the baseline target from both the industry and consumer sides.

[0005] This invention provides a method for reducing carbon emissions based on dynamic economic behavior, comprising: Step 1: Obtain and store the first source dataset; identify and classify the industry feature data subset according to the first preset classification rule to generate the first classification feature parameter set; Step 2: Identify and classify the subset of energy consumer characteristic data according to the second preset classification rule to generate a second classification feature parameter set; compare the first classification feature parameter set with the pre-stored first benchmark feature library item by item, and output the first matching difference result data. Step 3: Compare the second set of classification feature parameters with the pre-stored second benchmark feature library item by item, and output the second matching difference result data; generate a comprehensive feature difference data set based on the first matching difference result data and the second matching difference result data; Step 4: Extract a current path feature parameter set sequentially from the pre-stored path feature database; compare and analyze the corresponding data items of the current path feature parameter set with the comprehensive feature difference data set to generate the current path comparison and analysis result; Step 5: Based on the current path comparison analysis results, update the path optimization identifier data in the path feature database that corresponds to the current path feature parameter set; Step 6: Determine whether all path feature parameter sets in the path feature database have been compared and analyzed. If not, return to step 4. If yes, integrate all updated path optimization identification data and generate the final optimization configuration instruction.

[0006] This invention provides a carbon emission reduction system based on dynamic economic behavior, comprising: A server is used to execute the aforementioned carbon emission reduction method based on dynamic economic behavior; The memory is connected in communication with the server.

[0007] This invention provides a method and system for reducing carbon emissions based on dynamic economic behavior. By performing real-time identification and classification of industry characteristic data and energy consumer characteristic data under preset rules, a set of first and second classification characteristic parameters reflecting the current state of economic activity is generated. This mechanism enables the system to capture and quantify the dynamic changes in economic structure and consumption patterns, providing an accurate and up-to-date data foundation for subsequent analysis and overcoming the analytical lag problem caused by the reliance on static data in traditional methods.

[0008] By matching and comparing the set of classification feature parameters with the pre-stored benchmark feature library item by item, the differences between the current situation and the benchmark target can be accurately located from the industry side and the consumer side respectively. Furthermore, by merging the two types of differences to generate a comprehensive feature difference data set, a leap from single-dimensional analysis to multi-source information collaborative diagnosis has been achieved.

[0009] This invention creatively introduces a path feature database and its cyclical comparative analysis process. By matching and analyzing each pre-stored emission reduction path with comprehensive difference data one by one, and dynamically updating the path optimization identifier based on the matching results, the system realizes the self-evaluation and optimization of the strategy library, thereby improving the targeting and effectiveness of the strategy.

[0010] By traversing all paths in the path feature database and updating their optimized identifiers, the method ultimately integrates and generates optimized configuration instructions. This approach avoids the one-sidedness of selecting a single or local path. It systematically evaluates the matching degree between all available paths and the current comprehensive differences, thereby enabling the generation of a globally coordinated emission reduction configuration scheme, which helps to maximize the overall emission reduction efficiency under multi-objective constraints. Attached Figure Description

[0011] Figure 1 This is a schematic diagram of a method for reducing carbon emissions based on dynamic economic behavior in one embodiment of the present invention. Figure 2 This is a structural connection diagram of a carbon emission reduction system based on dynamic economic behavior in one embodiment of the present invention.

[0012] Figure label: 100 - Server; 200 - Storage. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] like Figure 1 As shown, the present invention provides a method for reducing carbon emissions based on dynamic economic behavior, comprising: Step 1: Obtain and store the first source dataset. Identify and classify the industry feature data subset according to the first preset classification rule to generate a first set of classification feature parameters.

[0015] Step 2: Identify and classify the subset of energy consumer characteristic data according to the second preset classification rule to generate a second set of classification feature parameters. Then, perform a step-by-step matching and comparison between the first set of classification feature parameters and a pre-stored first benchmark feature library, outputting the first matching difference result data.

[0016] Step 3: Perform item-by-item matching and comparison between the second set of classification feature parameters and the pre-stored second benchmark feature library, and output the second matching difference result data. Based on the first matching difference result data and the second matching difference result data, generate a comprehensive feature difference data set.

[0017] Step 4: Sequentially extract a set of current path feature parameters from the pre-stored path feature database. Compare and analyze the corresponding data items of the current path feature parameter set with the comprehensive feature difference data set to generate the current path comparison analysis result.

[0018] Step 5: Based on the current path comparison analysis results, update the path optimization identifier data in the path feature database that corresponds to the current path feature parameter set.

[0019] Step 6: Determine whether all path feature parameter sets in the path feature database have undergone comparative analysis. If not, return to Step 4. If yes, integrate all updated path optimization identifier data to generate the final optimization configuration instruction.

[0020] Understandably, the system receives heterogeneous parameters from industries, power companies, and energy consumers. Specifically, these heterogeneous parameters include industry signal transmission path parameters, power company decision-making impact parameters, and energy consumer energy-saving impact parameters. By performing real-time identification and classification of industry characteristic data and energy consumer characteristic data under preset rules, first and second category characteristic parameter sets reflecting the current state of economic activity are generated. This mechanism enables the system to capture and quantify dynamic changes in economic structure and consumption patterns, providing an accurate and up-to-date data foundation for subsequent analysis and overcoming the analytical lag problem caused by traditional methods relying on static data. Using these heterogeneous parameters, an industry-related objective function, a power company decision-making value function, and an energy consumer utility correlation equation are constructed. Based on these parameters, a low-carbon measure transmission correlation model is established. A dynamic economic behavior transmission path is established. Specifically, this path includes a price signal transmission path, a publicity information transmission path, and a technological innovation transmission path. A dynamic economic behavior transmission path is selected. Using the selected path, a model simulation of carbon emission reduction measures is performed on the low-carbon measure transmission correlation model. Obtain simulation results to optimize the selected dynamic economic behavior transmission path. Return to the previous step and select another dynamic economic behavior transmission path until all dynamic economic behavior transmission paths have been selected.

[0021] Specifically, the logic of the low-carbon measures transmission correlation model is that industry signals systematically change the behavioral patterns of market participants by influencing their psychological decision-making parameters and constraints, ultimately driving the achievement of emission reduction targets. Simultaneously, emission reduction results influence industry adjustments and the updating of participant beliefs through feedback mechanisms, forming a dynamic cyclical system. By matching and comparing the set of categorical feature parameters with a pre-stored benchmark feature library item by item, the differences between the current situation and benchmark targets can be accurately located from both the industry and consumer sides. Furthermore, by integrating these two types of differences to generate a comprehensive feature difference data set, a leap from single-dimensional analysis to multi-source information collaborative diagnosis is achieved.

[0022] This application relates to a carbon emission reduction method based on dynamic economic behavior. By receiving heterogeneous parameters from industries, power companies, and energy consumers, it obtains the behavior of micro-entities and uses their behavior as a key mediating variable in the energy policy transmission path, thereby improving the systematic nature of carbon emission reduction measures in dynamic economic behavior. The heterogeneous parameters include industry signal transmission path parameters, power company decision-making impact parameters, and energy consumer energy-saving impact parameters. Using these heterogeneous parameters, an industry-related objective function, a power company decision-making value function, and an energy consumer utility correlation function are constructed. Adjusting carbon emission reduction measures based on heterogeneous behavior can significantly improve emission reduction efficiency.

[0023] Based on industry-related objective functions, power enterprise decision-making value functions, and energy consumer utility correlations, a low-carbon measure transmission correlation model is established. This model establishes price signal transmission paths, publicity information transmission paths, and technological innovation transmission paths for dynamic economic behavior. Using the selected dynamic economic behavior transmission paths, the model simulates the implementation of carbon emission reduction measures, obtaining simulation results to optimize the chosen transmission paths. This system incorporates built-in monitoring, evaluation, and feedback mechanisms, forming a closed-loop learning system. By matching each pre-stored emission reduction path with comprehensive difference data and dynamically updating the path optimization identifier based on the matching results, the system achieves self-evaluation and optimization of the strategy library, improving the targeting and effectiveness of the strategies.

[0024] Based on implementation feedback, the parameters and tool combinations of carbon emission reduction measures are continuously adaptively optimized. The energy consumer-oriented collaborative strategy effectively activates energy consumers that cannot be covered by simple subsidies by enhancing the linkage characteristics of energy consumer utility correlation. For power companies, the model linkage of the power company's decision-making value function can change the decision-making results of the value function, reduce technological uncertainty, and effectively adjust the loss aversion coefficient and current bias parameters, successfully incentivizing investment in core low-carbon technologies. This improves the connection between carbon emission reduction measures and upstream and downstream industries in the power sector, enhancing the transmission path and effectiveness of low-carbon measures. By traversing all paths in the path feature database and updating the optimized identifiers, the optimized configuration instructions are finally integrated and generated. This method avoids the one-sidedness of single or local path selection. It systematically evaluates the matching degree of all optional paths with the current comprehensive differences, thereby enabling the overall coordinated emission reduction configuration scheme to be generated in a coordinated manner, which helps to maximize the overall emission reduction effectiveness under multi-objective constraints.

[0025] In one embodiment of this application, the first source dataset includes a predefined subset of industry characteristic data and a subset of energy consumer characteristic data. Step 1: Obtain and store the first source dataset. Identify and classify the industry characteristic data subset according to a first preset classification rule to generate a first set of classification feature parameters, including: Step 11 involves performing structured parsing on each piece of original industry data in the industry feature data subset, extracting the industry code, energy consumption type identifier, and production period distribution data. Specifically, this includes: formatting and filtering noise for each piece of original industry data in the industry feature data subset, removing invalid data (such as null values ​​missing key information or abnormal data with incorrect formatting), and uniformly converting unstructured / semi-structured data (such as text descriptions, table fragments, and string records) from different sources (such as enterprise declaration reports, industry statistical logs, and data uploaded from energy monitoring systems) into standardized text or field formats to ensure data parsingability. Based on pre-defined field identification rules (such as field name keyword matching, data type feature identification, and industry data specification mapping), three core information items are located and extracted from each piece of preprocessed original data: identifying industry classification codes that conform to national / industry standards (such as the National Economic Industry Classification Code) to ensure that data from different enterprises within the same industry can be uniformly identified; extracting information representing energy consumption categories from the enterprise's original records (which may be enterprise-defined identifiers, non-standard abbreviations, or old classification identifiers, such as electricity, raw coal, natural gas, and diesel, or internal enterprise codes). Extract information related to the company's production activities, including start and end times, continuous production duration, intervals between intermittent production periods, and the proportion of production load in different time periods, to fully reconstruct the company's time distribution characteristics. Validate the three core data items extracted. If any fields are missing (e.g., lack of explicit energy consumption type identifiers), supplement them appropriately by linking to historical data from the same company, industry default rules, or manual input prompts, ensuring that the core fields of each data entry are complete and usable.

[0026] Step 12: Match the energy consumption type identifier with the preset energy consumption type mapping table to determine and mark the standard energy consumption type label corresponding to each piece of original industry data. Specifically, this includes: "Pre-constructing and maintaining an energy consumption type mapping table," which contains the correspondence between all non-standard identifiers and standard energy consumption types. This table covers various non-standard forms such as enterprise-defined identifiers, historical classification identifiers, industry slang, and abbreviations, as well as corresponding national / industry standard energy consumption type labels (e.g., electricity, coal, natural gas, refined oil, and renewable energy). The energy consumption type identifiers of each piece of original industry data extracted in step 11 are compared one by one with the non-standard identifiers in the mapping table. A strategy of prioritizing exact matching and using fuzzy matching as a fallback is adopted: First, a completely consistent exact match is performed to quickly locate the corresponding standard label. If no exact match result is found, fuzzy matching is performed based on keyword similarity and semantic relevance (e.g., matching both electricity consumption and electricity usage to the electricity label), ensuring that each non-standard identifier can be mapped to a unique standard label. An independent standard energy consumption type label field is added to each piece of original industry data. The matched standard label is written into this field, binding the original data with standardized classification identifiers. This ensures that all subsequent classification operations based on energy consumption types are performed based on a unified standard, avoiding classification deviations caused by inconsistent identifiers.

[0027] Step 13: Based on industry codes and standard energy consumption type labels, group raw industry data with the same codes and labels into the same data group. Specifically, this includes: clearly defining industry codes and standard energy consumption type labels as the core dimension combination for grouping. Industry codes ensure that data within a group belongs to the same industry category, while standard energy consumption type labels ensure that the energy consumption type of data within a group is consistent. This dual-dimensional combination enables precise segmentation within the same industry and energy type, avoiding data mixing between different industries and energy types. Establish a grouping index system, using each combination of industry code and standard energy consumption type label as a unique grouping key, and assigning an independent group identifier (such as a group ID) to each grouping key, forming a grouping index table for quickly locating data. Traverse all raw industry data tagged with standard labels, query the grouping index table based on the industry code and standard energy consumption type label of each data entry to determine its corresponding group identifier, and then group the data entry into the dataset corresponding to that group identifier. Within the same group, aggregate all raw data with the same industry code and standard energy consumption type label. Data between different groups does not overlap, ensuring high homogeneity of data within each group and providing a clean data foundation for subsequent targeted feature calculations.

[0028] Step 14: For the production time distribution data within each data group, calculate the time period coverage and intensity peak to generate corresponding time period characteristic data. Specifically, this includes: extracting the production time period distribution data obtained in Step 11 from all the original data of each group; uniformly converting time period information in different data formats (such as time intervals, load percentage records, continuous production duration, etc.) into standardized time period sequence data (such as production status sequences divided by hours, load distribution data statistically analyzed by day) to ensure data computability. Based on the standardized time period sequence data, calculate the proportion of the time range covered by the production activities of this group to the preset full cycle time (such as 1 day, 1 week, 1 month, set according to the actual application scenario). For example, if all enterprises in a group produce from 8:00 to 20:00 daily, with a full cycle of 24 hours, then the time period coverage = 12 hours / 24 hours = 50%. If there are intermittent production situations, accumulate the total duration of all production time periods and compare it with the full cycle time to obtain the time period coverage of this group, reflecting the production time coverage range of this industry-energy type combination. The energy consumption intensity data (such as energy consumption per unit time and production load intensity) of all raw data within a group during production periods are statistically analyzed. The maximum value is selected as the intensity peak, and the specific time period corresponding to the peak is recorded (e.g., 10:00-11:00 daily is the peak energy consumption intensity period). If multiple data points within a group have the same intensity peak, the corresponding peak time periods are merged to form a complete peak feature record. The calculated time period coverage (including the full-cycle dimension on which the calculation is based), intensity peak values, and peak corresponding time periods are integrated into structured time period feature data, serving as the core time dimension feature of the group and providing quantitative support for the generation of subsequent classification feature parameters.

[0029] Step 15: Merge the common fields of all original industry data within the same data group and integrate them with the calculated time-period feature data to generate the classification feature parameters for that data group. Specifically, this includes: traversing all original industry data within the same group, identifying and extracting common basic fields (i.e., common fields), including industry code, standard energy consumption type label, industry name, region, enterprise size level, and energy consumption unit of measurement. These common fields are deduplicated and standardized. If a small number of data have inconsistent values ​​for common fields (e.g., incorrect region labeling for individual data), corrections are made based on the values ​​of the majority of data within the group or related industry statistics, ultimately forming a unique set of common fields for that group. The merged set of common fields is then structurally combined with the time-period feature data (time-period coverage, intensity peak, peak time period) calculated in Step 14 to form a complete feature set containing basic attributes, energy type, and time-period features. For example, the integrated result for a certain group might include: industry code C30 (non-metallic mineral products industry), standard energy consumption type label (electricity), region (East China), enterprise size (large), time period coverage (60%), peak intensity (1200 kW / h), and peak time period (9:00-11:00). The integrated feature set undergoes format standardization processing. Each feature item is defined according to preset parameter naming rules and data type specifications (such as text, numerical, and time types), forming a unique classification feature parameter for that group. This parameter can comprehensively and accurately characterize the core features of the combination of industry code and standard energy consumption type, providing a standardized data source for subsequent comparative analysis with the benchmark feature library.

[0030] Step 16: Summarize the classification feature parameters generated from all data groupings to form the first classification feature parameter set. Specifically, this includes: traversing all groups where classification feature parameters have been generated, extracting the classification feature parameters for each group one by one, ensuring that no feature parameters corresponding to any industry code or standard energy consumption type combination are omitted. Following the preset collection storage specifications, organize all collected classification feature parameters in an orderly manner to construct a structured first classification feature parameter set. Each element in the set represents a classification feature parameter for a group, and each element is assigned a unique parameter identifier to facilitate subsequent item-by-item matching and index queries. Simultaneously, establish an index system for the set to support rapid retrieval by key dimensions such as industry code and standard energy consumption type tags, improving the efficiency of subsequent matching and comparison. Perform integrity and consistency checks on the constructed first classification feature parameter set, checking for duplicate feature parameters, omissions of key industry or energy type grouping data, and whether the format of the feature parameters conforms to preset standards. If any problems are found, correct and supplement them promptly. The resulting set of first-classification feature parameters can comprehensively cover the main energy consumption scenarios of all industries, and each feature parameter can reflect the dynamic production characteristics of the corresponding industry-energy type combination in real time, providing an accurate and up-to-date data foundation for benchmark matching and difference analysis in the subsequent step 2.

[0031] Specifically, it receives industry signal parameters. It analyzes the strategy space within the industry signal transmission path parameters. It obtains industry heterogeneity parameters for carbon tax rates, emission reduction subsidies, energy efficiency standards, and the intensity of information dissemination. It receives parameters influencing power company decisions. It analyzes the strategy space within these parameters. It obtains heterogeneity parameters for power company emission reduction investment scale, production plans, and technology choices. It receives parameters influencing energy conservation by energy consumers. It analyzes the strategy space within these parameters. It obtains heterogeneity parameters for energy consumer decisions regarding energy consumption patterns and the adoption of energy-saving technologies.

[0032] Understandably, the low-carbon measures transmission and correlation model includes three heterogeneous entities: industries, power companies, and energy consumers. The industry's strategy space includes carbon tax rates, emission reduction subsidies, energy efficiency standards, and the intensity of information dissemination. Power companies' strategy space includes the scale of emission reduction investment, production plans, and technology choices. Energy consumers' strategy space includes energy consumption patterns and decisions regarding the adoption of energy-saving technologies.

[0033] Data on industry signal parameters, power company decision-making impact parameters, and energy consumer energy conservation impact parameters were obtained through questionnaire surveys. The energy consumer questionnaire covered core dimensions such as household energy consumption structure, willingness to pay for low-carbon products, perception of social norms, intertemporal decision-making preferences, and psychological reference points. The power company questionnaire included content such as energy usage status, barriers to low-carbon technology adoption, policy response mechanisms, industry benchmarking behavior, and risk preferences.

[0034] Data collection employed stratified random sampling, comprehensively considering economic development level, industrial structure, and energy consumption characteristics. Enterprise questionnaires targeted different industry sectors and varying sizes (large, medium, and small), and the basic characteristics of the sample distribution were largely consistent with macroeconomic statistical data, demonstrating good representativeness.

[0035] In this embodiment, the industry heterogeneity parameters of carbon tax rates, emission reduction subsidies, energy efficiency standards, and information dissemination intensity; the power company heterogeneity parameters of emission reduction investment scale, production plans, and technology choices; and the energy consumer heterogeneity parameters of energy consumption patterns and energy-saving technology adoption decisions aim to provide computational parameters for simulating the transmission and evolution of carbon emission reduction measures among agents with bounded rationality. The core logic of the model is that the transmission and correlation model of low-carbon measures systematically changes the behavioral patterns of market participants by influencing their psychological decision-making parameters and constraints, ultimately driving the achievement of emission reduction targets. Simultaneously, the emission reduction results influence industry adjustments and the updating of participant beliefs through feedback mechanisms, forming a dynamic cyclical system.

[0036] In one embodiment of this application, step 2 includes: Step 21 involves standardizing and cleaning each piece of raw consumer data in the energy consumer characteristic data subset, extracting the consumer account identifier, energy-consuming equipment list, and historical energy consumption curve data. Specifically, this includes format standardization and noise filtering for each piece of raw consumer data, removing invalid data lacking key information such as missing consumer account identifiers or valid energy consumption records, as well as abnormal data with anomalous values ​​(e.g., daily energy consumption exceeding the average of similar users by more than 10 times) or incorrect formats (e.g., inconsistent timestamp formats, mixed text and numbers in the equipment model field). For unstructured / semi-structured data from different sources (e.g., smart meter records, user APP reported data, power company customer service registration information), such as handwritten equipment list text and string-formatted energy consumption records, a standardized format corresponding to the field and value is uniformly converted (e.g., converting 150 kWh of electricity consumption in May 2024 to a statistical period of 2024-05, with an electricity consumption of 150 kWh), ensuring data consistency and parsability. Based on pre-defined field recognition rules (such as keyword matching of core field names like account ID, device model, and electricity consumption curve, and distinguishing between identifier (text), device (text + number), and energy consumption (numerical + time) data according to data type characteristics), three core information items are accurately extracted from each pre-processed data entry: First, a unique identifier conforming to the unified standards of power companies / energy management platforms (such as user ID and meter number) is extracted to ensure that all data for each consumer can be uniquely associated, avoiding confusion between different account data. Second, complete information on various energy-consuming devices used by the user is extracted, including device name (such as air conditioner, refrigerator, and industrial machine tool) and model specifications. Key parameters such as (e.g., KFR-35GW / BP3N1Y-IF), purchase time, and rated power are collected. If the equipment model in the original data is incomplete (e.g., only marked "air conditioner" without specifying the model), it is supplemented by associating with the user's historical repair records, default equipment brand models, or sending supplementary information prompts to the user to ensure the accuracy of the equipment information. Energy consumption records of users within a preset historical period (e.g., the past 3 months or 1 year) are extracted, including daily / hourly energy consumption values, specific timestamps of energy consumption occurrences, continuous energy consumption duration, and intervals of intermittent energy consumption, to fully restore the temporal distribution characteristics of user energy consumption and provide data support for subsequent load pattern analysis. The three core information items are cross-validated. For example, the power company database is queried through the consumer account identifier to verify whether the list of energy-consuming equipment is consistent with the equipment information declared when the account was registered. If there are discrepancies, the data is marked as data to be verified. The continuity of historical energy consumption curve data is verified. If there is a long period of missing data (e.g., no energy consumption records for 7 consecutive days), the data for that period is filled or removed by the average energy consumption of users in the same group to ensure that the core information items are complete and usable.

[0037] Step 22: Based on the equipment model information in the energy-consuming equipment list, query the pre-built equipment energy efficiency benchmark library to associate each piece of original consumer data with a corresponding set of equipment energy efficiency levels. Specifically, this includes: pre-building and dynamically maintaining the "equipment energy efficiency benchmark library." The data sources for this library include national energy efficiency standards (such as GB12021.3-2022 "Minimum Allowable Values ​​of Energy Efficiency and Energy Efficiency Grades for Room Air Conditioners"), equipment energy efficiency catalogs published by industry associations, and technical parameter manuals provided by equipment manufacturers. The core content stored in the library includes: the correspondence between equipment models and energy efficiency grades, rated energy consumption indicators for different energy efficiency grades (such as an APF value ≥ 4.2 for a Level 1 energy efficiency air conditioner), and the classification standards for energy efficiency grades (such as grades 1-5, with Level 1 being the highest energy efficiency). The library is updated according to the latest energy efficiency standards (such as the newly added grading standards after the energy efficiency label reform) to ensure the timeliness of the benchmark data. Each equipment model in the energy-consuming equipment list extracted in Step 21 is compared one by one with the model data in the equipment energy efficiency benchmark library. A combined strategy of precise matching and fuzzy matching is employed: First, precise matching is performed using the complete equipment model to quickly locate the corresponding energy efficiency level (e.g., model KFR-35GW / BP3N1Y-IF directly matches to Level 1 energy efficiency). If no precise match is found (e.g., the equipment model is a customized version or an older model), fuzzy matching is performed based on key parameters such as equipment brand, product series, and rated power (e.g., if equipment of the same brand, series, and power is all Level 2 energy efficiency, then the equipment is assumed to be Level 2 energy efficiency). The fuzzy matching result is marked in the matching result for subsequent manual verification. For all energy-consuming equipment in each consumer account, the matched energy efficiency levels are aggregated to form a set of equipment energy efficiency levels for that account. For example, a household user's set of equipment energy efficiency levels might be "Air conditioner: Level 1, Refrigerator: Level 2, Water heater: Level 3," while an industrial user's set might be "Machine tool: Level 2, Air compressor: Level 1, Lighting equipment: Level 3." Each set must clearly indicate the name, model, energy efficiency level, and matching method (precise / fuzzy) of each piece of equipment to ensure the traceability of energy efficiency data.

[0038] Step 23: Based on historical energy consumption curve data, calculate the energy consumption volatility and typical load pattern for each consumer account within a preset period. Specifically, this includes: standardizing the historical energy consumption curve data extracted in Step 21 according to a uniform time granularity (e.g., hourly, daily). For example, converting data with different recording frequencies (e.g., some devices record every 15 minutes, some every hour) into hourly energy consumption data; filling in short-term data gaps using interpolation (e.g., if no record is found in a certain hour, fill it with the average energy consumption of the preceding and following hours), ensuring the continuity and comparability of the energy consumption curves. Energy consumption volatility reflects the stability of a user's energy consumption. The calculation logic is as follows: within a preset period (e.g., one month), first calculate the difference between the daily energy consumption and the average daily energy consumption within that period; then calculate the average of the absolute values ​​of all differences; finally, divide by the average daily energy consumption to obtain the volatility (e.g., if a user's average monthly energy consumption is 100 kWh, and the average absolute difference between the daily energy consumption and the average is 10 kWh, then the volatility is 10%). Higher volatility indicates more unstable energy consumption (e.g., large fluctuations in energy consumption due to production plan adjustments for industrial users); lower volatility indicates more stable energy consumption (e.g., the daily energy consumption patterns of residential users). After calculation, each consumption account is labeled with its corresponding energy consumption volatility value and stability level (e.g., low volatility ≤5%, medium volatility 5%-15%, high volatility >15%). Typical load patterns are used to characterize the temporal distribution characteristics of user energy consumption; the operation procedure is as follows: Time Period Division: Divide the day into several standard time periods (e.g., residential users are divided into 0-6 AM, 6-12 AM, 12-6 PM, and 6-12 AM; industrial users are divided into production periods of 8-8 PM and non-production periods of 8-11 PM), or customize time periods according to industry characteristics (e.g., shopping mall users are divided into business periods of 9-10 PM and closing periods of 10-9 PM). Load characteristic statistics: Energy consumption data for each time period is statistically analyzed, calculating the average energy consumption, energy consumption percentage (the proportion of energy consumption in that time period to the total daily energy consumption), and the time of peak energy consumption for each time period. Based on the energy consumption percentage and peak characteristics of each time period, user energy consumption patterns are categorized into typical types, such as peak load patterns for residential users (energy consumption percentage exceeding 40% between 6 PM and 10 PM) and stable load patterns (energy consumption percentage difference between time periods ≤ 10%); and continuous high load patterns for industrial users (energy consumption percentage exceeding 80% during production periods with no significant fluctuations) and intermittent load patterns (large energy consumption fluctuations during production periods with multiple peaks). The typical load pattern for each consumer account must clearly indicate the time period division criteria, energy consumption characteristics of each time period, and pattern type to ensure the distinguishability of load patterns.

[0039] Step 24 involves binding the set of equipment energy efficiency levels, energy consumption volatility, and typical load patterns as a set of associated feature data to the corresponding consumer account identifier. Specifically, this includes integrating the set of equipment energy efficiency levels generated in Step 22, the energy consumption volatility (including numerical values ​​and stability levels) calculated in Step 23, and the typical load patterns (including time period characteristics and pattern types) into a set of structured associated feature data. During the integration process, it is necessary to ensure that the field names for each feature item are consistent (e.g., set of energy efficiency levels, energy consumption volatility (%), typical load pattern type) and that the data types are standardized (clearly distinguishing between text and numerical types) to avoid field confusion. Using the consumer account identifier extracted in Step 21 as the unique key, the integrated associated feature data is bound to this account identifier, forming a one-to-one correspondence between account identifier and associated feature data. For example, the data bound to account identifier user A001 includes: set of equipment energy efficiency levels: air conditioner level 1, refrigerator level 2; energy consumption volatility: 8% (medium volatility); typical load pattern: residential peak load pattern (45% of load occurs between 6 PM and 10 PM). All bound data is stored in a structured database, and a query system indexed by consumer account identifiers is established to support quick retrieval of corresponding associated feature data by account identifier. Simultaneously, secondary indexes (such as indexes by typical load pattern type and energy consumption volatility level) are created for each associated feature item to facilitate rapid data filtering during subsequent cluster analysis. All bound data is traversed to check for issues such as duplicate account identifiers or missing associated feature data (e.g., an account not being bound to an energy efficiency level set). If any such issues are found, the previous steps are returned to correct them, ensuring that the associated feature data for each account is complete and unique.

[0040] Step 25: Based on the similarity of typical load patterns, a clustering algorithm is used to divide multiple sets of associated feature data into several consumer groups. Specifically, this includes using the similarity of typical load patterns as the core clustering dimension, while also incorporating energy consumption volatility to assist clustering. This is because typical load patterns directly reflect the temporal distribution of user energy consumption (such as peak hours and energy intensity), and are a core indicator for distinguishing user energy consumption behavior. Energy consumption volatility can further refine the differences in energy consumption stability among users within a group, ensuring the homogeneity of the group. For the typical load patterns of any two consumer accounts, the similarity is calculated using the "cosine similarity of time period energy consumption ratio": the energy consumption ratio of each time period for the two accounts is used as a vector, and the cosine value between the vectors is calculated (the value range is 0-1). The closer the cosine value is to 1, the more similar the load patterns of the two are (e.g., if two residential users both have peak energy consumption from 6 PM to 10 PM, and the difference in the ratio of each time period is ≤5%, then the similarity is ≥0.95); the closer the cosine value is to 0, the greater the difference in patterns (e.g., if an industrial user has high load during production hours and a residential user has high load at night, the similarity is ≤0.3). Simultaneously, the cosine similarity is corrected by considering the difference in energy consumption volatility (e.g., a volatility difference of ≤3% between two accounts is considered high similarity) to obtain the final comprehensive similarity. The K-means algorithm, suitable for behavioral feature clustering (automatically determining the optimal number of groups based on data distribution), is employed. The operation process is as follows: randomly select several representative consumer accounts (e.g., accounts with different load patterns and volatility) as initial cluster centers; calculate the comprehensive similarity between each consumer account and each cluster center, and assign the account to the group containing the cluster center with the highest similarity; for all accounts within each group, recalculate the average characteristics of typical load patterns (e.g., average energy consumption percentage in each time period) and average energy consumption volatility as new cluster centers; repeat the above assignment and update steps until the change in cluster centers is ≤ a preset threshold (e.g., change in energy consumption percentage in each time period ≤ 1%), or the number of iterations reaches a preset upper limit (e.g., 50 times), ensuring the stability of the clustering results. After clustering is completed, the validity of each group is checked: ensure that the comprehensive similarity of typical load patterns of all accounts in the group is ≥0.8 (homogeneity), and the average similarity between groups is ≤0.4 (heterogeneity); if a group is too small (e.g., less than 10 accounts) or has insufficient homogeneity (similarity <0.6), the group is split or merged to form a reasonable number of consumer groups with clear boundaries, and a unique group identifier (e.g., G001, G002) is assigned to each group.

[0041] Step 26: For each consumer group, statistically analyze the distribution ratio of energy efficiency levels of various devices within that group, calculate the average energy consumption volatility of that group, and generate classification characteristic parameters for that consumer group. Specifically, this includes summarizing the set of device energy efficiency levels for all accounts within each consumer group and statistically analyzing the percentage of different energy efficiency levels for each type of device (e.g., air conditioners, machine tools). For example, in group G001 (high-energy-consuming residential group), air conditioners have a 10% energy efficiency level 1 rating, a 30% rating level 2 rating, and a 60% rating level 3 or below; refrigerators have a 15% energy efficiency level 1 rating, a 40% rating level 2 rating, and a 45% rating level 3 or below. The statistics must be calculated by device type to ensure a clear distribution ratio for each device type, and the sum of the energy efficiency distribution ratios for all devices within the same group is 100%. Calculation of average energy consumption volatility for the group: Calculate the arithmetic mean of the energy consumption volatility of all consumer accounts within each group, which will be used as the average energy consumption volatility for that group. For example, group G002 (Industrial Stable Energy Consumption Group) contains 100 accounts, each with a volatility between 3% and 7%, averaging 5.2%. Therefore, the average energy consumption volatility of this group is 5.2%, and the corresponding stability level (e.g., low volatility) is indicated. Before calculation, outliers within the group (e.g., an account's volatility exceeding three times the group average) must be removed to ensure the accuracy of the average. Classification Feature Parameter Integration: The core information of each group is integrated into structured classification feature parameters, including the following key items: group identifier (e.g., G001); description of core group characteristics (e.g., high-energy-consuming residential group, peak nighttime load pattern); distribution ratio of energy efficiency levels for each equipment type (e.g., air conditioners: Level 1 10%, Level 2 30%, Level 3 and below 60%; refrigerators: Level 1 15%, Level 2 40%, Level 3 and below 45%); average energy consumption volatility (e.g., 8.5%) and stability level (e.g., medium volatility); typical load pattern (e.g., 45% energy consumption between 6 PM and 10 PM, evenly distributed energy consumption during the day); group size (e.g., containing 200 consumer accounts). The integrated classification feature parameters must be in a unified format, clearly expressed, and able to comprehensively reflect the energy consumption behavior and equipment characteristics of the group, ensuring item-by-item matching during subsequent comparisons with the benchmark database.

[0042] Step 27: Summarize the classification feature parameters generated by all consumer groups to form a second classification feature parameter set. This involves iterating through all consumer groups whose classification feature parameters have been generated, extracting the classification feature parameters for each group one by one, ensuring no group is missed (including smaller groups with unique characteristics). During collection, the parameters are arranged in order of group identifier for easy indexing and querying later. Following a preset storage standard, all collected group classification feature parameters are organized into a second classification feature parameter set. The set uses a structured data format (such as JSON or a database table), where each element represents a group's classification feature parameter, and each element contains a unique group identifier as a primary key. It supports fast retrieval by key dimensions such as group identifier, typical load pattern, and average energy consumption volatility (e.g., filtering out high-volatility groups with "average energy consumption volatility > 15%").

[0043] Completeness and consistency verification: Perform a comprehensive verification of the constructed set of second-classification feature parameters. The process involves several steps: First, checking if all consumer accounts are covered (i.e., all accounts are grouped into a single group without omissions) and whether any groups lack categorization feature parameters. Second, verifying the consistency of the categorization feature parameter format across groups (e.g., the way energy efficiency distribution ratios are expressed and the precision of volatility values) and identifying duplicate group identifiers or contradictory feature descriptions (e.g., labeling the same group as both "low volatility" and "volatility 12%)." Third, checking the reasonableness of feature parameter values ​​(e.g., whether the sum of energy efficiency distribution ratios equals 100% and volatility falls within the 0%-100% range). If any anomalies are found, the process is corrected by returning to the previous steps. After successful verification, the second categorization feature parameter set is stored in a database sharing the same source as the first categorization feature parameter set. A cross-set association index is then established (e.g., associating industry-side and consumer-side parameters by energy consumption type, region, etc.) to improve the efficiency of generating comprehensive difference data in subsequent steps. The resulting second categorization feature parameter set comprehensively and accurately reflects the heterogeneity characteristics of all current energy consumers.

[0044] Specifically, the total industry revenue is obtained by multiplying the weights of low-carbon measures by the industry's overall revenue. The loss cost is obtained by multiplying the industry's funding pool pressure weight by the measure implementation cost. The industry volatility cost is obtained by multiplying the industry growth weight by the industry's economic growth fluctuation function. The practitioner income cost is obtained by multiplying the industry feedback weight by the practitioner income function. The industry revenue formula is obtained by successively subtracting the total industry revenue from the loss cost, industry volatility cost, and practitioner income cost. The industry maximum revenue function is obtained by multiplying the industry discount factor by the industry revenue formula. The maximum value of the maximum revenue function is then calculated over the time interval of the measure implementation. Finally, the maximum value of the summed revenue function is obtained, yielding the industry-related objective function.

[0045] Understandably, the objective function for industry-related relationships is: Formula 1.

[0046] In formula 1, For the period t, due to the reduction in carbon emissions The resulting industry benefits. The implementation costs of the measures include subsidy expenditures, publicity costs, and carbon tax collection and management costs. A function that measures the fluctuation of industry economic growth, i.e., short-term industry economic growth fluctuations. A function to measure the income of practitioners. to The weighting parameter reflects the trade-off between four dimensions: the weight of low-carbon measures, the weight of industry funding pressure, the weight of industry growth, and the weight of industry feedback. This is the industry discount factor.

[0047] In this embodiment, the industry-related objective function employs adaptive learning to maximize the benefits of carbon emission reduction across different periods. Each period, based on the previous period's performance (such as the degree of achievement of the objective function), marginal adjustments are made to the industry heterogeneity parameters to optimize the objective function in the next period. This multi-period dynamic structure enables the model to capture the feedback and co-evolution process of industry carbon emission reduction measures, providing a powerful analytical tool for long-term analysis of these measures.

[0048] In one embodiment of this application, step 2 further includes: Step 28: From the first set of classification feature parameters, sequentially extract a current set of classification feature parameters to be matched. This set includes industry code, standard energy consumption type label, common field data, and time period feature data. Using the industry code and standard energy consumption type label as the combined query key, search the first benchmark feature library to locate and obtain the corresponding benchmark feature parameter set. Compare the common field data in the current set of classification feature parameters to be matched with the corresponding benchmark common field data in the benchmark feature parameter set, calculating the absolute difference of each value to generate a first set of numerical differences.

[0049] Step 29: Compare the time-period feature data in the current classification feature parameter group to be matched with the corresponding benchmark time-period feature data in the benchmark feature parameter group to generate a first pattern difference identifier. Based on the first numerical difference set and the first pattern difference identifier, generate a difference feature vector for the current classification feature parameter group to be matched. Bind the industry code, standard energy consumption type label, and difference feature vector to the difference feature vector and save it as a difference result unit. Determine whether all classification feature parameter groups in the first classification feature parameter set have been matched and compared. If not, return to step 28. If yes, summarize all saved difference result units and output them as the first matching difference result data.

[0050] For each piece of original industry data in the industry feature data subset, structured parsing is performed to extract the industry code, energy consumption type identifier, and production time distribution data. The energy consumption type identifier is matched with a pre-defined energy consumption type mapping table to determine and label the standard energy consumption type tag corresponding to each piece of original industry data. Based on the industry code and standard energy consumption type tag, original industry data with the same code and tag are grouped into the same data group. For the production time distribution data within each data group, time period coverage and intensity peaks are calculated to generate corresponding time period feature data. Common fields of all original industry data within the same data group are merged and integrated with the calculated time period feature data to generate classification feature parameters for that data group. The classification feature parameters generated from all data groups are summarized to form the first set of classification feature parameters.

[0051] Specifically, the production revenue function is obtained based on the difference between corporate revenue and production costs. The carbon emission tax cost is obtained by multiplying carbon emissions by the carbon tax rate. The low-carbon input function is obtained based on the difference between the production revenue function and the carbon emission tax cost. The emission reduction revenue function is obtained by summing the low-carbon input function and emission reduction subsidies. The emission reduction investment cost is generated based on energy efficiency standards and emission reduction infrastructure investment. Finally, the decision value function for power companies is obtained by using the difference between the emission reduction revenue function and the emission reduction investment cost.

[0052] Understandably, the decision-making value function for power companies is: Formula 2.

[0053] In formula 2, For corporate revenue, For production costs. The cost of paying carbon emissions due to carbon tax rates. Carbon tax rate, where E represents carbon emissions. The emission reduction revenue function resulting from emission reduction subsidies. To meet energy efficiency standards The required investment cost for emissions reduction typically increases with the higher the standard. Companies assess profits within the framework of prospect theory. Their decisions are influenced by the industry's average profit or expected profit.

[0054] In this embodiment, the power company's decision-making is based on the power company decision-making value function of prospect theory, where the company's profits are directly affected by all carbon emission reduction measures.

[0055] In one embodiment of this application, step 3 includes: Step 31: From the second set of classification feature parameters, sequentially extract the classification feature parameters of a consumer group to be matched. This includes: confirming the completeness of the second set of classification feature parameters (e.g., whether it contains all consumer groups generated by clustering, and whether the parameter fields of each group are complete), and establishing a sequential traversal index for the set (e.g., sorted in ascending order by group identifier alphabetical / numerical order) to avoid duplicate extraction or omissions. Following the preset traversal order, extract the first consumer group classification feature parameter that has not yet been matched from the set. The extracted content consists of complete parameter items for the group, including core information such as group identifier, energy efficiency level distribution ratio of each equipment type, average energy consumption fluctuation rate, typical load pattern, and group size. Ensure that the extracted parameters are all the features of the group, with no missing fields. Mark the extracted group parameters as pending comparison, and synchronously record the extraction timestamp and benchmark library version (to ensure the basis for subsequent traceability comparison). If missing parameter fields are found during extraction (e.g., a group without energy efficiency distribution data for air conditioning equipment), mark the parameters as incomplete and pause the comparison, return to the previous steps to supplement the data, and then re-extract.

[0056] Step 32: Index the second benchmark feature library based on the group identifier to obtain the benchmark consumer characteristic parameters corresponding to the group identifier. Specifically, the benchmark library stores benchmark parameters categorized by consumer group type (e.g., high-energy-consuming residential group, stable-energy-consuming industrial group, peak-consuming commercial group). Each group type corresponds to a unique benchmark identifier, and the benchmark parameters completely match the dimensions of the consumer group classification feature parameters, including: benchmark energy efficiency level distribution ratio (by equipment type), benchmark average energy consumption volatility, benchmark typical load pattern, etc. Use the identifier of the current group to be matched (e.g., G001, high-energy-consuming residential group) as the search key to perform precise indexing in the second benchmark feature library. If a perfectly matching group identifier exists in the benchmark library (e.g., G001 corresponds to the benchmark parameters for the high-energy-consuming residential group), all benchmark consumer characteristic parameters under that identifier are directly extracted. If no perfectly matching identifier exists (e.g., a newly added niche group type), the closest benchmark type is fuzzily matched based on typical load patterns and group attributes (e.g., the high-fluctuation residential niche group matches the benchmark for the high-energy-consuming residential group), and the fuzzy match is marked. If no matching benchmark type exists, general benchmark parameters (covering basic energy-saving standards for all consumption types) are called, and it is recorded that there is no specific benchmark for subsequent manual verification. After extracting the benchmark parameters, the completeness of their fields is verified (e.g., whether they contain the benchmark energy efficiency distribution for all equipment types in the current group). If the benchmark parameters are missing data for a certain type of equipment (e.g., the benchmark library does not have an energy efficiency benchmark for new energy vehicle charging piles), it is marked that there is no benchmark for that equipment type, and subsequent comparisons will only be performed on equipment types with benchmarks.

[0057] Step 33: Compare the distribution ratio of device energy efficiency levels in the classification feature parameters of the current consumer group to be matched with the distribution ratio of benchmark energy efficiency levels in the benchmark consumer feature parameters, and calculate the ratio difference of each level to generate a second set of numerical differences. Specifically, this includes: first, splitting the distribution ratio of device energy efficiency levels of the current consumer group to be matched with the benchmark parameters according to device type and energy efficiency level to ensure that the comparison dimensions correspond one-to-one (e.g., first splitting the device types such as air conditioners, refrigerators, and machine tools, and then comparing the level 1, level 2, level 3 and below energy efficiency levels under each type separately), avoiding cross-device type and cross-level mixed comparisons.

[0058] Comparison of scale values ​​at each level: Compare each of the split dimensions one by one: Taking the proportion of Level 1 energy efficiency in air conditioning equipment as an example, if the proportion in the current group is 10% and the benchmark parameter is 25%, then the direction and magnitude of the difference between the two proportions are clearly defined. If the proportion of Level 3 and below energy efficiency in the current group is 60% and the benchmark parameter is 40%, then this level's proportion is marked as higher than the benchmark. This comparison is performed for each energy efficiency level of each equipment type, and the direction of the difference (higher than, lower than, or equal to the benchmark) is marked to ensure the intuitiveness of the difference. The comparison results of all dimensions are integrated into a structured second set of numerical differences. Within the set, equipment types are categorized, and each equipment type includes the proportion difference and direction of the difference for each energy efficiency level. For example: Air conditioning equipment: Level 1 energy efficiency proportion is 15 percentage points lower than the benchmark, Level 2 energy efficiency proportion is equal to the benchmark, and Level 3 and below energy efficiency proportion is 15 percentage points higher than the benchmark; Refrigerator equipment: Level 1 energy efficiency proportion is 10 percentage points lower than the benchmark, Level 2 energy efficiency proportion is 5 percentage points higher than the benchmark, and Level 3 and below energy efficiency proportion is 5 percentage points higher than the benchmark. The set also marks equipment types without a benchmark to avoid invalid difference data interfering with the analysis.

[0059] Step 34: Compare the average energy consumption volatility in the classification feature parameters of the current consumer group to be matched with the benchmark average energy consumption volatility in the benchmark consumer feature parameters, calculate their relative change rate, and generate a second volatility difference identifier. Specifically, this includes: clarifying the core values ​​of the average energy consumption volatility (e.g., 8.5%) and the benchmark average energy consumption volatility (e.g., 5%) of the current consumer group to be matched, while taking into account the type characteristics of the group (e.g., the volatility threshold allowed for industrial groups is higher than that for residential groups) to ensure that the comparison meets the requirements of the scenario.

[0060] Relative rate of change determination and identifier generation: First, determine the relative direction of volatility change (higher than / lower than / equal to the benchmark); then, based on preset severity grading rules (e.g., relative change rate ≤10% indicates no significant difference, 10%-30% indicates slight deviation, 30%-50% indicates moderate deviation, and >50% indicates severe deviation), determine the degree of deviation; finally, generate a second volatility difference label, which includes numerical comparison, direction of change, severity level, and description of behavioral impact. For example: the current group's average energy consumption volatility is 8.5%, the benchmark value is 5%, which is 70% higher than the benchmark, and the level is severely high; this difference reflects the group's unstable energy consumption behavior, which is likely to increase the pressure on the grid's carbon emissions. Special case annotation: if the volatility data for the current group or the benchmark is missing, the label will indicate that the data is missing and cannot be determined; if the volatility is equal to the benchmark, it will be labeled as the average energy consumption volatility being consistent with the benchmark, and the energy consumption stability meeting emission reduction requirements.

[0061] Step 35: Based on the second set of numerical differences and the second volatility difference identifier, generate characteristic description data of differences for the current consumer group. Specifically, this includes: merging the second set of numerical differences (differences in energy efficiency distribution) and the second volatility difference identifier (differences in energy use stability). During the integration, the data is sorted into a hierarchy of core differences, secondary differences, and no differences, prioritizing differences that have a significant impact on carbon emissions (such as a high proportion of inefficient equipment and a high degree of volatility). The differential characteristic description data is divided into three core modules. Each module supplements the impact analysis with dynamic economic behavior background: It details the key situations where the energy efficiency level of each equipment type deviates from the benchmark, and the direct impact of this difference on carbon emissions. For example, the proportion of air conditioning equipment with level 1 energy efficiency is 15 percentage points lower than the benchmark, while the proportion of level 3 and below is 20 percentage points higher than the benchmark. This difference leads to an average annual carbon emission of air conditioners in this group being about 12% higher than the benchmark level. The high proportion of inefficient refrigerators further exacerbates carbon emissions. Based on volatility difference indicators, it explains the degree of deviation of energy consumption stability from the benchmark and the indirect carbon emission impact. For example, the average energy consumption volatility is significantly higher than the benchmark (70%). High-frequency energy consumption fluctuations lead to a widening of the peak-valley difference in the power grid, increasing the carbon emissions of supporting peak-shaving power sources, and indirectly pushing up the group's total life-cycle carbon emissions. It extracts the core emission reduction shortcomings of this group. For example, the core emission reduction potential of this group lies in replacing inefficient air conditioning and refrigerator equipment, while guiding users to use energy steadily through price signals to reduce volatility. Ensure that the descriptive data is consistent with the preceding difference data (e.g., there should be no situation where the energy efficiency ratio is lower than the benchmark but the description is that the energy efficiency level meets the standard), and that the language is concise and clear, so as to facilitate subsequent path matching analysis.

[0062] Step 36: Associate and bind the group identifier with the difference feature description data, saving it as a consumer difference result unit. Specifically, this includes binding the group identifier with the difference feature description data to form a consumer difference result unit containing the following core fields: a unique unit ID (e.g., DC001); a group identifier (e.g., G001); difference feature description data (complete energy efficiency, volatility differences, and impact analysis); a comparison benchmark version (e.g., the second benchmark feature library V2.0, ensuring traceability of the benchmark basis); a comparison completion timestamp (for recording dynamically updated difference data); and a data integrity marker (e.g., including energy efficiency differences for air conditioners / refrigerators, but without charging pile benchmark data). Store the constructed difference result unit in a temporary database table. During storage, verify the uniqueness of the group identifier (to avoid generating multiple difference units from the same group). If duplicates are found, overwrite the old unit and record the update log. Simultaneously, verify the integrity of the unit fields. Units lacking key information (e.g., benchmark version) need to be supplemented by returning to previous steps. Establish an index centered on the group identifier for all stored difference result units to facilitate subsequent rapid retrieval and aggregation.

[0063] Step 37: Determine whether all consumer group classification feature parameters in the second classification feature parameter set have been matched and compared. If not, return to step 31. If yes, summarize all saved consumer difference result units and output as the second matching difference result data. Specifically, this includes: querying the traversal progress of the second classification feature parameter set in real time, and counting the number of compared groups and the total number of groups: if the number of compared groups is less than the total number, it is determined that not all have been completed, the current traversal position is recorded (e.g., processed to G005), and the process returns to step 31 to continue extracting the next unmatched group parameter. If the number of comparisons equals the total number, then it is considered complete. At the same time, check for groups with incomplete labeling parameters or no baseline, and separately label the information of these special groups during the summary.

[0064] Organize all stored consumer difference result units in ascending order of group identifier to form a structured summary list; supplement the summary dimension information, including: the total number of groups compared, the distribution of core difference types (e.g., 18 groups have low energy efficiency differences, 10 groups have high volatility differences), and special group descriptions (e.g., 2 groups have no specific benchmark, and a general benchmark is used for comparison); perform consistency verification on the summary data to ensure no duplicate units and no missing key information. Integrate the summarized difference result units with the dimension information to generate standardized second-match difference result data, with the output format adapted to subsequent data fusion needs (e.g., JSON, database view, Excel report, etc.). The data includes: detailed difference descriptions for each group, difference type statistics, and special case explanations, which can directly support the linking and merging with industry-side difference data in step 38.

[0065] Specifically, a current utility function for energy consumers is generated based on the sum of their consumption utility and the industry's discounted consumption utility. A social utility correlation is generated using the intensity of information dissemination. An industry information dissemination utility function is generated based on the sum of the current utility function and the social utility correlation function. A total expenditure function for energy consumers is generated using the carbon cost of consumption and the cost of energy-saving equipment. Finally, an energy consumer utility correlation is generated based on the difference between the industry information dissemination utility function and the total expenditure function.

[0066] Understandably, the energy consumer utility relationship is as follows: Formula 3.

[0067] In formula 3, For the consumption utility of energy consumers. The consumer utility discounted for the industry. The social utility of information dissemination intensity is directly determined by the intensity of information dissemination. adjust. This is a function of total energy consumer expenditure, which includes the implicit carbon costs of energy consumption and the actual expenditure on energy-efficient equipment. Implicit carbon costs are subject to carbon taxes. Indirectly, the actual cost of consumer energy-saving equipment is affected by subsidies. Direct impact.

[0068] Understandably, the carbon cost of energy consumption for consumers includes the implicit carbon cost arising from energy consumption, and this implicit carbon cost is subject to carbon tax. Indirect impact.

[0069] In this embodiment, a significant "willingness-behavior" gap exists among energy consumers. Although 65.2% of respondents indicated they were willing to accept a 10% green premium, only 18.1% chose to purchase energy-efficient products immediately, while 47.3% indicated they would postpone their decision. In the matching experiment targeting intertemporal preferences, the consumption utility of energy consumers discounted by industry was far higher than the index-discounted assumption.

[0070] The intensity of information dissemination has proven to be a powerful driving force. Over 50% of energy consumers indicated that they are influenced by information dissemination, and 69.5% acknowledged that social norms have a significant impact on their decisions. By constructing energy consumer utility correlations and measuring the path coefficients of information dissemination intensity, carbon consumption costs, and energy-saving equipment costs on purchase intentions, we calibrated the social norm sensitivity of energy consumers. These parameters exhibit a normal distribution within the population.

[0071] In one embodiment of this application, the consumer group classification characteristic parameters include group identifier, distribution ratio of device energy efficiency levels, and average energy consumption volatility. The benchmark consumer characteristic parameters include the benchmark energy efficiency level distribution ratio and the benchmark average energy consumption volatility. Step 3 further includes: Step 38: Link and merge the data of each difference result unit in the first matching difference result data and each consumer difference result unit in the second matching difference result data according to the predefined association mapping relationship to generate a comprehensive feature difference data set containing difference information of industry dimension and consumer dimension.

[0072] Specifically, the process utilizes industry-specific heterogeneity parameters such as carbon tax rates and emission reduction subsidies. These parameters are used as moderating factors. The process also incorporates the decision-making value function of power companies and the utility relationship between energy consumers. Based on these moderating factors, the price signal transmission path is obtained within the power company decision-making value function and the energy consumer utility relationship.

[0073] Understandably, the price signaling path describes the transmission process of the effects of economic tools on carbon emission reduction measures. This path alters the relationship between the decision-making value function of power companies and the utility of energy consumers by changing two regulatory factors: carbon tax rates and emission reduction subsidies. This, in turn, changes corporate profits and energy consumer spending, thereby regulating emission reduction investment or green consumption and achieving emission reduction effects. For example, carbon tax rate measures induce emission reduction behavior by altering corporate cost structures.

[0074] In one embodiment of this application, step 4 includes: Step 41: Extract a current path feature parameter set from the pre-stored path feature database. This set includes the path identifier, applicable industry scope, target consumer scope, and expected impact parameters. Specifically, the path feature database stores all preset emission reduction path parameter sets. Each path corresponds to a unique path identifier, and the parameter dimensions cover the applicable scope and the core dimension of expected impact. For example, P001 (publicity information transmission path - residential energy efficiency improvement) and P002 (price signal transmission path - industrial energy consumption regulation). Clearly define the path type and core objectives; accurately label the industry code and standard energy consumption of the path. Consumption type labels (e.g., C30 Non-metallic mineral products industry + electricity consumption) support single-industry, multi-industry, or full-industry adaptation; consumer group identifiers for path adaptation (e.g., G001 Residential high-energy-consumption group, G003 Industrial high-fluctuation group) support single-group, multi-group, or full-group coverage; categorized by industry weight + consumer weight, while also linking core parameters of dynamic economic behavior transmission (e.g., energy efficiency standard adjustment weight and information publicity intensity impact weight for the publicity transmission path; carbon tax rate weight and emission reduction subsidy weight for the price signal transmission path), with weights reflecting the priority of the path's impact on industry / consumption-side emission reduction.

[0075] Extract the first set of path feature parameters that has not yet completed matching analysis from the database in ascending order of path identifier. During extraction, simultaneously record the path type (e.g., promotional information transmission path) and database version (to ensure traceability of the preset basis for the path). If any parameters are found to be missing during extraction (e.g., no consumer weight in the expected impact parameters), mark the parameters as incomplete and pause matching until the parameters are supplemented and extracted again. After extraction, verify the completeness of core fields (path identifier, applicable industry scope, target consumer scope, expected impact parameters). For example, for promotional information transmission paths, additional verification is needed to check the existence of energy efficiency standard adjustment weights and information promotion intensity weights. If they are missing, mark the key transmission parameters as missing to facilitate adjustment of matching rules in subsequent analysis.

[0076] Step 42: Match the applicable industry scope with the industry codes and type labels in the comprehensive feature difference data set, filter out the matching difference result units, and form an industry matching set. Specifically, this includes: breaking down the applicable industry scope of the current path into a combination of industry code + standard energy consumption type label (such as C30 + electricity, C25 + coal), and clarifying the adaptation rules (such as precise matching, fuzzy matching, and fuzzy matching allows adaptation to upstream and downstream related industries).

[0077] Comprehensive Difference Data Screening: From the comprehensive feature difference data set, select difference result units that match the industry code, standard energy consumption type label, and path applicability scope. If the path applicability scope is C30 non-metallic mineral products industry + electricity consumption, only industry difference result units with industry code C30 and energy type label "electricity" in the comprehensive difference set are selected. If the path label is adapted to high-energy-consuming industries and electricity consumption, then difference units with industry codes C25 (non-metallic mineral products industry), C31 (ferrous metal smelting), and other high-energy-consuming industries and energy type "electricity" are selected. If the path has no specific applicable industry scope, then all industry difference result units in the comprehensive difference set are selected. All selected industry difference result units are categorized and organized according to industry code + energy type label to form a structured industry matching set. Each unit in the set is labeled with the matching type (precise / fuzzy) with the path, while industry units with no difference (such as industry difference feature vector of 0, no emission reduction potential) are excluded to ensure that the matching set only includes industry units with emission reduction potential.

[0078] Step 43: Match the target consumer range with the group identifiers in the comprehensive feature difference data set, and filter out matching consumer difference result units to form a consumer matching set. Specifically, this includes: breaking down the target consumer range of the current path into consumer group identifiers (e.g., G001, G003), and defining the adaptation rules (e.g., full match, feature match; feature match allows adaptation to groups with similar load patterns / energy efficiency characteristics). From the comprehensive feature difference data set, filter out consumer difference result units whose group identifiers match the target range of the path. Full match: If the target range of the path is the high-energy-consuming residential group G001, only consumer difference units with group identifier G001 will be selected. Feature match: If the path is a publicity information transmission path - inefficient equipment replacement, then consumer difference units with a proportion of level 3 and below in the distribution ratio of energy efficiency levels of all equipment (regardless of group identifier) ​​will be selected. If the path has no specific target consumer range, all consumer difference result units will be selected. The selected consumer difference result units will be classified and organized according to group identifier + core difference type (such as low energy efficiency, high volatility) to form a consumer matching set. The core emission reduction shortcomings of each unit will be marked in the set (such as high proportion of inefficient air conditioners, severely high energy consumption volatility) to facilitate accurate matching of the expected impact parameters of the path during subsequent weighted calculation.

[0079] Step 44: For each unit in the industry matching set, perform a weighted calculation based on its difference feature vector and the industry weight in the expected impact parameter to obtain the unit matching score, and calculate the industry dimension score based on all unit scores. Specifically, this includes: first, parsing the "industry weight" in the expected impact parameter of the current path, splitting the industry weight according to the difference type (such as energy efficiency difference weight, time period feature difference weight, numerical difference weight), and different weights for different transmission paths (such as the energy efficiency standard weight accounting for 60% and the time period feature weight accounting for 20% in the publicity information transmission path; and the time period feature weight accounting for 50% in the price signal transmission path).

[0080] Unit matching score calculation: For each difference unit in the industry matching set, calculate the score one by one: Extract the differential feature vector of the unit (such as differences in energy efficiency values ​​and time-period pattern differences). Differential characteristics are assigned weights based on industry (e.g., energy efficiency difference weight is 0.6; if the energy efficiency of this unit is 15% lower than the benchmark, then the score for this dimension = (1-15%) × 0.6; time period pattern difference weight is 0.2; if the pattern similarity is 80%, then the score for this dimension = 80% × 0.2). The scores of each dimension are summarized to obtain the matching score of this unit. The higher the score, the stronger the adaptability of the path to the industry unit (e.g., a score of 90 indicates that the path can effectively solve the emission reduction shortcomings of this industry). The total score for the industry dimension is calculated by taking the weighted average of the matching scores of all units (weighted according to the emission reduction potential of the industry unit, such as higher weight for high-energy-consuming industry units) to obtain the industry dimension score of the path, with a score range of 0-100 points. At the same time, the core basis for the score is marked (e.g., because the energy efficiency difference matching score of the C30 industry is high, the score of this dimension is raised to 85 points).

[0081] Step 45: For each unit in the consumer matching set, perform a weighted calculation based on its difference feature description data and the consumer weight in the expected impact parameters to obtain the unit matching score, and calculate the consumer dimension score based on all unit scores. Specifically, this includes: parsing the "consumer weight" in the expected impact parameters of the current path, splitting the consumer weight according to the difference type (such as equipment energy efficiency distribution weight, energy consumption volatility weight, load mode weight), and different weights for different transmission paths (such as the information promotion intensity correlation weight accounting for 50% and the energy efficiency distribution weight accounting for 30% in the promotion information transmission path). Unit Matching Score Calculation: For each difference unit in the consumer matching set, calculate the score one by one: Extract the difference feature description data of the unit (e.g., 60% of air conditioners are inefficient, volatility is severely high); assign values ​​to different types of difference features according to consumer weight (e.g., energy efficiency distribution weight 0.3, if the unit's inefficiency rate is higher than the benchmark 20%, then the score for this dimension = (1-20%) × 0.3; volatility weight 0.2, if volatility is severely high, then the score for this dimension = 50% × 0.2); summarize the scores of each dimension to obtain the matching score of the unit. The higher the score, the stronger the adaptability of the path to the consumer unit (e.g., a score of 88 indicates that the path can effectively guide the group to save energy). Consumer Dimension Total Score Calculation: Take the weighted average of all unit matching scores (weighted according to the size of the consumer group, such as larger groups with higher weights) to obtain the consumer dimension score of the path, with a score range of 0-100 points, and mark the core basis for the score (e.g., due to the high adaptability of the information promotion of group G001, the score of this dimension is raised to 82 points).

[0082] Step 46: Merge industry dimension scores and consumer dimension scores according to preset rules to generate a current path comparison analysis result including a comprehensive score and a grade identifier. Specifically, the merging rules are defined in advance and the weights need to be adjusted according to the path type (e.g., for the promotional information transmission path, which emphasizes the consumer dimension, the industry dimension weight is 0.4 and the consumer dimension weight is 0.6; for the technological innovation transmission path, which emphasizes the industry dimension, the industry weight is 0.7 and the consumer weight is 0.3). The rules also specify the scoring grade classification standards (e.g., a comprehensive score ≥90 is grade S, 80-89 is grade A, 70-79 is grade B, and <70 is grade C). Comprehensive score calculation: Merge industry dimension scores and consumer dimension scores according to preset weights. For example, for the promotional information transmission path P001, the industry dimension score is 85 (weight 0.4) and the consumer dimension score is 82 (weight 0.6), so the comprehensive score = 85 × 0.4 + 82 × 0.6 = 83.2. After calculation, the comprehensive score is rounded (e.g., 83) to ensure the score is concise and easy to read. Rating and Result Generation: Based on the comprehensive score, a rating label is matched (e.g., 83 points corresponds to Grade A). The final result of the current path comparison analysis includes the following core contents: path label, path type (e.g., the path of disseminating promotional information); industry dimension score, consumer dimension score, comprehensive score; rating label (e.g., Grade A); core matching conclusion (e.g., "This path has good adaptability to the C30 industry and the G001 group, with a comprehensive rating of Grade A, and can effectively promote energy efficiency improvement by increasing the intensity of information dissemination").

[0083] Specifically, the process involves incorporating industry-specific heterogeneity parameters related to energy efficiency standards and the intensity of information dissemination. These parameters are used as moderating factors. The process also incorporates the power company's decision-making value function and the energy consumer's utility correlation equation. Based on these moderating factors, the transmission path of the promotional information is determined within the power company's decision-making value function and the energy consumer's utility correlation equation.

[0084] Understandably, the information dissemination path describes the transmission process of the effects of non-economic tools for reducing carbon emissions, such as information dissemination measures. This information dissemination path achieves the goal of meeting energy efficiency standards by changing energy efficiency standards and the intensity of information dissemination. The required investment costs for emissions reduction establish a technological anchor. The dissemination of information also enhances social norm sensitivity and activates the social utility of information dissemination by altering its intensity, thereby driving changes in consumption habits and technology choices and achieving emissions reduction effects. For example, energy benchmarking information promotes the adoption of energy-saving technologies by activating social comparison mechanisms.

[0085] In one embodiment of this application, step 5 includes: Step 51: Analyze the current path comparison analysis results to obtain its comprehensive score and grade label. Specifically, this includes: accurately extracting two types of core information from the current path comparison analysis results: quantitative indicators: comprehensive score (e.g., 83 points), grade label (e.g., Grade A), and sub-item scores for industry / consumer dimensions; qualitative conclusions: core matching conclusions (e.g., suitable for C30 industry and G001 group, significant effect of information promotion intensity adjustment). Key information annotation: Annotate the parsed information, highlighting: advantages (e.g., consumer dimension score of 82 points, strong adaptability); weaknesses (e.g., industry dimension score of 85 points, insufficient time period feature matching); path type association features (e.g., the energy efficiency standard adjustment weight adaptability needs to be highlighted for the promotional information transmission path). Analysis result verification: Ensure that the extracted comprehensive score and grade label are consistent with the original comparison analysis results, and that there are no analysis errors (e.g., mistakenly extracting 88 points instead of 83 points). If an error is found, return to step 46 to regenerate the comparison analysis results.

[0086] Step 52: Based on the pre-defined mapping relationship, determine the path optimization suggestion code and priority marker corresponding to the result. Specifically, this includes: the mapping relationship is pre-built and dynamically updated, with core dimensions being the comprehensive score, level identifier, and path type, leading to the optimization suggestion code and priority marker. For example: For a publicity information transmission path with a comprehensive score of 80-89 (Level A), the optimization suggestion code is X002 (strengthen publicity intensity + refine energy efficiency standards) + high priority marker; for a price signal transmission path with a comprehensive score <70 (Level C), the optimization suggestion code is Y003 (adjust carbon tax rate + increase emission reduction subsidies) + medium priority marker; for all paths with a comprehensive score ≥90 (Level S), the optimization suggestion code is Z001 (retain current parameters + expand applicable scope) + highest priority marker. The mapping relationship also associates with path weakness dimensions. For example, for Level A paths with low industry-specific scores, the optimization suggestion code supplements and adjusts the applicable industry scope accordingly. Matching and Optimization Information: Based on the parsed comprehensive score, level identifier, and path type, precise matching is performed within the pre-defined mapping relationships: First, a subset of mapping relationships is filtered by path type (e.g., only viewing the mapping rules for the transmission path of promotional information); then, the corresponding optimization suggestion code (e.g., X002) and priority marker (e.g., high) are matched by the comprehensive score / level identifier; if the path has obvious shortcomings (e.g., the consumer dimension score is only 60 points), a note is added after the optimization suggestion code (e.g., X002 - supplementing consumer energy guidance); ensure that the optimization suggestion code matches the path type and adaptability results (e.g., the promotional information path does not match the optimization code related to price signals), and the priority marker conforms to the score level (e.g., the priority of a level A path is not lower than medium).

[0087] Step 53: Using the path identifier in the current path feature parameter set as an index, locate the corresponding storage record in the path feature database. Specifically, this includes using the path identifier (such as P001) in the current path feature parameter set as a unique index key. This index key is unique in the path feature database and can directly locate a single path record.

[0088] Database retrieval and location: A retrieval operation is performed in the path feature database. If the database is relational (e.g., MySQL), a query is used to retrieve the complete stored record of the path. If the database is non-relational (e.g., MongoDB), the path identifier is used as the document primary key to locate the corresponding document record. If no corresponding record is found (e.g., due to an incorrect path identifier), the record is marked as missing and an alarm is triggered. Manual verification is then performed before re-locating the record. After locating the record, its current status is verified (e.g., whether it has been analyzed but not updated) to ensure that the update operation targets the path record in this matching analysis, avoiding updates to historical versions or erroneous records.

[0089] Step 54: Write the path optimization suggestion code and priority flag as the updated path optimization identification data into the specified field of the stored record to complete the update. Specifically, the path feature database has a pre-defined path optimization identification data field for each path record, which contains two subfields: optimization suggestion code and priority flag. The field format is standardized text (e.g., the code is 6 characters and the priority is highest / high / medium / low).

[0090] Data writing operation: Write the optimization suggestion code (such as X002) and priority flag (such as high) determined in step 52 into the dedicated field of the corresponding record: If the field has existing data (such as the old optimization code), overwrite the existing data, and retain the update log (such as 2025-12-19 update: the original code X001 is adjusted to the new code X002, and the original priority is adjusted to the new high priority). If a field is empty, write new data directly and mark the first update timestamp. During writing, ensure the data format conforms to database specifications (e.g., no special characters, consistent code length) to avoid field errors. After writing, re-retrieve the path record, verifying that the optimization suggestion code and priority marker fields match expectations. Simultaneously, update the last update time and update status of the path record (e.g., from pending update to updated) to ensure the update operation takes effect. If database locks, insufficient field permissions, or other anomalies occur during the writing process, pause the update and trigger an alarm. Re-execute the write operation after the anomaly is resolved to ensure that the optimization marker data for each path is successfully updated.

[0091] Specifically, this involves incorporating industry-specific heterogeneity parameters for emission reduction subsidies, power company heterogeneity parameters for technology selection, and energy consumer heterogeneity parameters for energy-saving technology adoption decisions. Emission reduction subsidies, technology selection, and energy-saving technologies are used as moderating factors. The power company decision value function is then invoked. Based on these moderating factors, the technology innovation transmission path is obtained within the power company decision value function.

[0092] Understandably, the technological innovation transmission path describes the impact of low-carbon measures on low-carbon technology research and development. This path involves altering R&D subsidies or the carbon market to change the expected returns on innovation in the decision-making value function of power companies, thereby enabling them to innovate in green technologies and create long-term emission reduction potential. R&D subsidies are inherent in subsidy expenditures. Policies aimed at reducing carbon emissions influence companies' technology choices by changing the expected return stream of innovation investment.

[0093] In one embodiment of this application, step 6 includes: Step 61: After completing a path update, query the status records in the path feature database.

[0094] Step 62: Determine if there is a set of path feature parameters in the state of "unanalyzed". If it exists, return to step 4 to process the next set. If it does not exist, read all updated path optimization identifier data from the database.

[0095] Step 63: Sort and group all the read data according to their priority tags to form a structured configuration data package.

[0096] Step 64: Fill the configuration data packet according to the preset instruction template to generate the final optimized configuration instruction.

[0097] Understandably, the moderating factors of the selected dynamic economic behavior transmission path are adjusted using an industry-related objective function. These moderating factors are then substituted into the power company's decision-making value function and the energy consumer utility correlation equation, respectively, to obtain the total carbon emission output. The total carbon emission output is then used to update the power company's decision-making value function and the energy consumer utility correlation equation. The process of adjusting the moderating factors of the selected dynamic economic behavior transmission path using the industry-related objective function is repeated until the total carbon emission output reaches the preset target. The moderating factors of the selected dynamic economic behavior transmission path are then obtained.

[0098] Specifically, the model's dynamic process employs discrete-time steps, with each period t comprising four stages. The industry, based on its industry-related objective function and the previous simulation results, releases current carbon emission reduction measures, including carbon tax rates, emission reduction subsidies, energy efficiency standards, and the intensity of information dissemination. Power companies and energy consumers use these carbon emission reduction measure variables as inputs, substituting them into their respective decision functions to make simultaneous decisions. The low-carbon measure transmission correlation model simulates energy market supply and demand, calculates total carbon emissions, updates the information dissemination intensity among stakeholders, assesses the effectiveness of carbon emission reduction measures, and optimizes policy parameters for the next period based on adaptive learning rules. This multi-period dynamic structure enables the low-carbon measure transmission correlation model to capture the two-way feedback and co-evolution process between measures and behaviors, providing a powerful analytical tool for analyzing the long-term implementation path of carbon emission reduction measures.

[0099] It is understandable to use price signal transmission paths to simulate the implementation of carbon emission reduction measures through a low-carbon measure transmission correlation model.

[0100] In the baseline rational scenario, energy consumers are traditional rational agents. Scenario 1, which involves simple subsidies, assumes that actual expenditures are subsidized. =10%, directly affecting the total energy consumer expenditure function. Scenario 2, concerning subsidies and social norms, builds upon Scenario 1 by setting the intensity of information dissemination. It is 0.8.

[0101] Simulation results reveal the differences brought about by behavior. Under the baseline rational scenario, subsidies alone could achieve a 58.2% penetration rate for energy-saving products in the fifth year. However, under the more realistic behavioral model, the final penetration rate in Scenario 1 was only 48.3%, nearly 10 percentage points lower than the rational scenario, clearly demonstrating how behavioral barriers erode the effectiveness of the policy.

[0102] Scenario 2 outperformed Scenario 1, achieving a final penetration rate of 65.7%, 17.4 percentage points higher than Scenario 1. From the perspective of the industry-related objective function, Scenario 2 yielded significantly higher environmental benefits than Scenario 1. Although information dissemination increased the industry's advertising costs, the economies of scale resulting from the increased penetration rate reduced the subsidy cost per unit of emission reduction, ultimately leading to higher industry benefits in Scenario 2.

[0103] Intensive information dissemination, by updating energy consumers' beliefs about reducing carbon emissions, provides a reliable quality signal to uncertainty-averse groups, effectively increasing their expected base utility for adoption decisions. Simple subsidies cannot cover the middle group—those insensitive to prices but influenced by social norms—while combined measures achieve precise coverage.

[0104] Simply put, the simulation model demonstrates the correlation between technological innovation transmission paths and the implementation of low-carbon measures to reduce carbon emissions.

[0105] Collaboration between energy consumers and power companies can generate system-level optimization. When energy consumers become more energy-conscious and power companies increase their technological supply, it triggers a positive feedback loop where demand drives supply and supply creates demand.

[0106] Simulations show that the cost of industry low-carbon transition under the scenario of coordinated carbon emission reduction measures is 23.8% lower than that of implementing carbon emission reduction measures independently. In the industry-related objective function, this is reflected in higher industry returns and a better industry economic growth fluctuation function under the same implementation cost.

[0107] The optimized sequence for implementing carbon emission reduction measures is essentially a dynamic adjustment process of the weight parameters in the industry-related objective function. Initially, a high weight is set for industry feedback and low-carbon measures to establish the direction of emission reduction. Carbon emission reduction measures focus on enhancing information dissemination and social norms to cultivate the market. In the medium term, the weight for industry growth is moderately increased, and the focus of carbon emission reduction measures shifts to the promotional costs of economic incentives to expand market size. In the later stage, after the market matures, the weight of low-carbon measures is strengthened, and carbon tax rates and energy efficiency standards are gradually increased.

[0108] like Figure 2 As shown, the present invention provides a carbon emission reduction system based on dynamic economic behavior, comprising: Server 100 is used to execute the carbon emission reduction method based on dynamic economic behavior.

[0109] The memory 200 is communicatively connected to the server 100.

[0110] This embodiment relates to a carbon emission reduction system based on dynamic economic behavior. The server 100 performs model simulation of carbon emission reduction measures by executing a low-carbon measure transmission correlation model, obtains simulation results, and optimizes the selected dynamic economic behavior transmission path. It realizes a closed-loop learning system formed by the built-in monitoring, evaluation and feedback links of the low-carbon measure transmission correlation model based on industry correlation objective function, power enterprise decision value function and energy consumer utility correlation. According to the data stored in the memory 200, the system continuously and adaptively optimizes the parameters and tool combinations of carbon emission reduction measures through feedback. For energy consumer collaborative strategies, it effectively activates energy consumers that cannot be covered by simple subsidies by improving the linkage characteristics of energy consumer utility correlation. For power enterprises, the model linkage of power enterprise decision value function can change the decision result of value function, reduce technological uncertainty, and equivalently adjust the loss aversion coefficient and current bias parameter, successfully incentivizing investment in core low-carbon technologies. This improves the linkage between carbon emission reduction measures and upstream and downstream industries of the power industry.

[0111] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0112] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for reducing carbon emissions based on dynamic economic behavior, characterized in that, include: Step 1: Obtain and store the first source dataset; The industry feature data subset is identified and classified according to the first preset classification rule to generate a first classification feature parameter set. Step 2: Identify and classify the subset of energy consumer characteristic data according to the second preset classification rule to generate a second classification feature parameter set; The first set of classification feature parameters is matched and compared with the pre-stored first benchmark feature library item by item, and the first matching difference result data is output. Step 3: Compare the second set of classification feature parameters with the pre-stored second benchmark feature library item by item, and output the second matching difference result data; generate a comprehensive feature difference data set based on the first matching difference result data and the second matching difference result data; Step 4: Extract a set of current path feature parameters sequentially from the pre-stored path feature database; The current path feature parameter set and the comprehensive feature difference data set are compared and analyzed for corresponding data items to generate the current path comparison and analysis results; Step 5: Based on the current path comparison analysis results, update the path optimization identifier data in the path feature database that corresponds to the current path feature parameter set; Step 6: Determine whether all path feature parameter sets in the path feature database have been compared and analyzed. If not, return to step 4. If yes, integrate all updated path optimization identification data and generate the final optimization configuration instruction.

2. The carbon emission reduction method based on dynamic economic behavior according to claim 1, characterized in that, The first source dataset includes a predefined subset of industry characteristic data and a subset of energy consumer characteristic data; Step 1: Obtain and store the first source dataset; The industry feature data subset is identified and classified according to a first preset classification rule to generate a first classification feature parameter set, including: Step 11: Perform structured parsing on each piece of original industry data in the industry feature data subset to extract the industry code, energy consumption type identifier, and production period distribution data; Step 12: Match the energy consumption type identifier with the preset energy consumption type mapping table to determine and mark the standard energy consumption type label corresponding to each piece of original industry data; Step 13: Based on industry codes and standard energy consumption type labels, group the original industry data with the same codes and labels into the same data group; Step 14: For the production time period distribution data within each data group, calculate the time period coverage and intensity peak to generate corresponding time period feature data; Step 15: Merge the common fields of all original industry data within the same data group and integrate them with the calculated time period feature data to generate the classification feature parameters of the data group. Step 16: Summarize all the classification feature parameters generated by grouping the data to form the first classification feature parameter set.

3. The carbon emission reduction method based on dynamic economic behavior according to claim 2, characterized in that, Step 2 includes: Step 21: Standardize and clean each piece of original consumer data in the subset of energy consumer characteristic data to extract the consumer account identifier, energy-consuming equipment list and historical energy consumption curve data. Step 22: Based on the equipment model information in the energy-consuming equipment list, query the pre-set equipment energy efficiency benchmark library and associate the corresponding set of equipment energy efficiency levels for each piece of original consumer data; Step 23: Based on historical energy consumption curve data, calculate the energy consumption volatility and typical load pattern of each consumer account within a preset period; Step 24: The set of equipment energy efficiency levels, energy consumption fluctuation rate and typical load mode are treated as a set of associated feature data and bound to the corresponding consumer account identifier. Step 25: Based on the similarity of typical load patterns, the multiple sets of associated feature data are divided into several consumer groups using a clustering algorithm; Step 26: For each consumer group, calculate the distribution ratio of energy efficiency levels of each device within it, and calculate the average energy consumption fluctuation rate of the group to generate the classification feature parameters of the consumer group. Step 27: Summarize the classification feature parameters generated by all consumer groups to form the second classification feature parameter set.

4. The carbon emission reduction method based on dynamic economic behavior according to claim 3, characterized in that, Step 2 also includes: Step 28: From the first set of classification feature parameters, sequentially extract a current set of classification feature parameters to be matched. The set of classification feature parameters includes industry code, standard energy consumption type label, common field data, and time period feature data. Using the industry code and standard energy consumption type label as the combined query key, search in the first benchmark feature library to locate and obtain the corresponding benchmark feature parameter set. Compare the common field data in the current set of classification feature parameters to be matched with the corresponding benchmark common field data in the benchmark feature parameter set item by item, calculate the absolute difference of each value, and generate the first set of numerical differences. Step 29: Compare the time period feature data in the current classification feature parameter group to be matched with the corresponding benchmark time period feature data in the benchmark feature parameter group to generate a first pattern difference identifier; generate a difference feature vector for the current classification feature parameter group to be matched based on the first numerical difference set and the first pattern difference identifier; bind the industry code, standard energy consumption type label and difference feature vector and save them as a difference result unit; determine whether all classification feature parameter groups in the first classification feature parameter set have been matched and compared; if not, return to step 28; if yes, summarize all saved difference result units and output as the first matching difference result data.

5. The carbon emission reduction method based on dynamic economic behavior according to claim 4, characterized in that, Step 3 includes: Step 31: Extract a consumer group classification feature parameter to be matched from the second classification feature parameter set in sequence; Step 32: Index the second benchmark feature library according to the group identifier to obtain the benchmark consumer feature parameters corresponding to the group identifier; Step 33: Compare the distribution ratio of device energy efficiency levels in the current consumer group classification feature parameters to be matched with the distribution ratio of benchmark energy efficiency levels in the benchmark consumer feature parameters, calculate the ratio difference of each level, and generate a second set of numerical differences. Step 34: Compare the average energy consumption volatility in the current consumer group classification feature parameters to be matched with the benchmark average energy consumption volatility in the benchmark consumer feature parameters, calculate their relative change rate, and generate a second volatility difference identifier. Step 35: Generate difference characteristic description data for the current consumer group based on the second numerical difference set and the second volatility difference identifier; Step 36: Associate and bind the group identifier with the difference feature description data, and save it as a consumer difference result unit; Step 37: Determine whether all consumer group classification feature parameters in the second classification feature parameter set have been matched and compared; if not, return to step 31; if yes, summarize all saved consumer difference result units and output as the second matching difference result data.

6. The carbon emission reduction method based on dynamic economic behavior according to claim 5, characterized in that, Consumer group classification characteristic parameters include group identifier, distribution ratio of equipment energy efficiency level, and average energy consumption fluctuation rate; benchmark Consumer characteristic parameters include the distribution proportion of benchmark energy efficiency levels and the benchmark average energy consumption volatility; step 3 also includes: Step 38: Link and merge the data of each difference result unit in the first matching difference result data and each consumer difference result unit in the second matching difference result data according to the predefined association mapping relationship to generate a comprehensive feature difference data set containing difference information of industry dimension and consumer dimension.

7. The carbon emission reduction method based on dynamic economic behavior according to claim 6, characterized in that, Step 4 includes: Step 41: Extract a set of current path feature parameters from the pre-stored path feature database, which includes path identifier, applicable industry scope, target consumer scope, and expected impact parameters; Step 42: Match the applicable industry scope with the industry codes and type tags in the comprehensive feature difference data set, filter out the matching difference result units, and form an industry matching set; Step 43: Match the target consumer range with the group identifiers in the comprehensive feature difference data set, filter out the matching consumer difference result units, and form a consumer matching set; Step 44: For each unit in the industry matching set, perform a weighted calculation based on its difference feature vector and the industry weight in the expected impact parameter to obtain the unit matching score, and calculate the industry dimension score based on all unit scores. Step 45: For each unit in the consumer matching set, perform a weighted calculation based on its difference feature description data and the consumer weight in the expected impact parameters to obtain the unit matching score, and calculate the consumer dimension score based on all unit scores. Step 46: Merge industry dimension scores and consumer dimension scores according to preset rules to generate a current path comparison analysis result that includes a comprehensive score and a rating identifier.

8. The carbon emission reduction method based on dynamic economic behavior according to claim 7, characterized in that, Step 5 includes: Step 51: Analyze the current path comparison analysis results to obtain its comprehensive score and level label; Step 52: Based on the preset mapping relationship, determine the path optimization suggestion code and priority tag corresponding to the result; Step 53: Using the path identifier in the current path feature parameter set as an index, locate the corresponding storage record in the path feature database; Step 54: Write the path optimization suggestion code and priority flag as the updated path optimization identifier data into the specified field of the stored record to complete the update.

9. The carbon emission reduction method based on dynamic economic behavior according to claim 8, characterized in that, Step 6 includes: Step 61: After completing a path update, query the status records in the path feature database; Step 62: Determine if there is a set of path feature parameters in the state of "unanalyzed". If it exists, return to step 4 to process the next set. If it does not exist, read all updated path optimization identifier data from the database. Step 63: Sort and group all the read data according to their priority tags to form a structured configuration data package; Step 64: Fill the configuration data packet according to the preset instruction template to generate the final optimized configuration instruction.

10. A carbon emission reduction system based on dynamic economic behavior, characterized in that, include: A server for executing the carbon emission reduction method based on dynamic economic behavior as described in any one of claims 1 to 9; The memory is connected in communication with the server.

Images