A method and system for allocating carbon emission budgets across all sectors based on two-flow bayesian hierarchical
By combining Bayesian hierarchical models with industry-specific indicators and historical data, the uncertainty and localization issues of carbon budget decomposition models have been resolved, enabling accurate carbon emission prediction and risk quantification, and improving the scientific nature and operability of the budget.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CEEC HUNAN ELECTRIC POWER DESIGN INST
- Filing Date
- 2026-03-27
- Publication Date
- 2026-05-29
Smart Images

Figure CN121920796B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of carbon emission prediction technology, and in particular to a method and system for allocating carbon emission budgets across the entire domain based on dual-stream Bayesian hierarchical structure. Background Technology
[0002] Carbon emission budgeting is a "cap-and-control" approach used to plan and constrain emissions. As carbon budgeting theory has improved, its application has expanded from the regional level down to the enterprise level, and from the field of financial management to the field of corporate management. The definitions and model tools used differ at different levels.
[0003] In the scientific and policy practice of carbon budget decomposition, the mainstream model system supporting carbon budgeting mainly includes the following three categories: comprehensive assessment models for long-term global scenario simulation, medium- and long-term energy system optimization models focusing on the optimization of regional energy technology portfolios and investment paths, and computable general equilibrium models for assessing the macroeconomic impact of policies such as carbon pricing.
[0004] However, current carbon budget allocation practices still face significant challenges. On the one hand, existing models generally suffer from high uncertainty and insufficient localization and refinement; on the other hand, the allocation results are often out of sync with local planning systems, industrial structures, and development needs, making it difficult to implement the budget. These problems urgently need to be addressed. Summary of the Invention
[0005] This invention provides a method and system for allocating carbon emission budgets across the entire domain based on dual-stream Bayesian hierarchical structure, in order to solve the technical problems mentioned in the background art.
[0006] To achieve the above objectives, the technical solution of the present invention is implemented as follows:
[0007] This invention provides a method for allocating carbon emission budgets across the entire domain based on dual-stream Bayesian hierarchical structure, comprising the following steps:
[0008] S1. Extract sub-indicators for each industry from different industries;
[0009] S2. Based on the economic development forecast and cumulative carbon emission intensity reduction target for the study area in the next T years, calculate the maximum permissible total carbon emissions for the study area in the next T years. ;
[0010] S3. Collect historical data for various industry sub-indicators, and combine this data with the planning and actual conditions of the study area to provide a priori values for the historical average annual growth rate of carbon emissions for each industry. ;
[0011] S4. Combine the historical data of each industry's sub-indicators with the prior values of each industry's historical average annual carbon emission growth rate. The data is input into a Bayesian hierarchical model to calculate the predicted values and 95% confidence intervals for each industry's sub-indicators. The predicted values for each industry's sub-indicators include the final, balanced predicted value for the next T years, the autonomously extrapolated values of all activity levels for the next T years, the endogenously derived carbon emission intensity for each industry in the next T years, and the newly added carbon emissions. .
[0012] Furthermore, step S1 specifically includes the following steps:
[0013] S11. Extract sub-indicators for each industry from the processes of industry, construction, agriculture, forestry, animal husbandry and fishery, transportation, service industry, residents' life, processing and conversion and industrial production.
[0014] S12. Carbon emissions from different industries are divided into key carbon emissions and basic carbon emissions. Key carbon emissions include carbon emissions from industry, construction, processing and conversion, and industrial production processes. Basic carbon emissions include carbon emissions from agriculture, forestry, animal husbandry and fishery, transportation, services, and residential life.
[0015] Furthermore, the prior value of the historical average annual growth rate of carbon emissions in S3 The given data should at least consider the average annual carbon emission growth rate of regions with the same climate type, development stage, similar area and city size as the study region, both domestically and internationally.
[0016] Furthermore, the Bayesian hierarchical model includes, in sequence, an adaptive observation layer, a prior and parameter generation layer, a prediction and posterior evolution layer, a macro-equilibrium and redline verification layer, and an endogenous game and intensity inference layer.
[0017] Furthermore, step S4 specifically includes the following steps:
[0018] S41. Input the historical data of each industry's sub-indicators into the adaptive observation layer of the Bayesian hierarchical model, and fit the observed carbon emission value of industry i in year t. and the observed activity level k in year t. ;
[0019] S42. Observation of carbon emissions of industry i in year t. and the observed activity level k in year t. The data is input into the prior and parameter generation layer to obtain the historical data distribution of carbon emissions and activity levels in various industries. Then, based on the historical data distribution of carbon emissions and activity levels in various industries, the historical average annual growth rate of each sub-indicator and carbon emissions is obtained, which serves as the basic supporting data for the prior values of the future average annual growth rate of each industry sub-indicator and the prior values of the future average annual growth rate of carbon emissions.
[0020] S43. Based on historical data of various industry sub-indicators and prior values of historical average annual growth rates of carbon emissions. The distribution of the posterior growth rate of carbon emissions was calculated. Then, the distribution of the posterior growth rate of carbon emissions random variable... The data is input into the prediction and posterior evolution layer, and after MCMC sampling, the posterior carbon emission prediction distribution is obtained, including the pre-adjustment original emission predictions for multiple different industries i in the next T years. Then, the posterior prediction distribution of carbon emissions is arranged to obtain the 95% confidence interval of the posterior prediction distribution of carbon emissions.
[0021] S44. Based on historical data of industry sub-indicators and priori growth rates of activity levels The distribution of the posterior growth rate of the activity level random variable was calculated. Then, the distribution of the posterior growth rate random variable of the activity level. The input is fed into the prediction and posterior evolution layer, and the posterior prediction distribution of the activity level obtained through MCMC sampling includes autonomous extrapolation values of multiple different activity levels k for the next T years. Then, the posterior prediction distribution of the activity level is sorted to construct the 95% confidence interval of the posterior prediction distribution of the activity level.
[0022] S45. Introduce the maximum permissible carbon emission amount for the study area into the macro-equilibrium and redline check layer of the Bayesian hierarchical model. As a macroeconomic constraint, and based on meeting the macroeconomic constraint, the original emission forecasts before adjustment are used. Calculate the system balance factor Then, based on the system balance factor and the original emission forecasts before adjustment Calculate the final balanced forecast value of industry i in the next T years. and final predicted value The 95% confidence interval;
[0023] S46. Final balanced forecasts for all industries in year T. Summing is performed to obtain the sum of the key carbon emissions and basic carbon emissions, adjusted for balance, for the predicted carbon emissions in the next T years, H; then, based on the maximum permissible total carbon emissions for the study area... The sum of the projected carbon emissions for the next T years, H, is used to calculate the new carbon emissions. Introducing a dual balance constraint, when new carbon emissions... When the dual balance constraint is not satisfied, adjust the parameters in the prior and parameter generation layers, the prediction and posterior evolution layers, and the system balance factor in the macro-balance and redline verification layers. To satisfy the dual balance constraint;
[0024] S47. Based on satisfying the dual balance constraints, the final predicted value will be... and autonomously extrapolated values The input is fed into the endogenous game and intensity extrapolation layers of the Bayesian hierarchical model to calculate the endogenous carbon emission intensity of industry i in the next T years. and endogenous derivation of carbon emission intensity The 95% confidence interval is used to quantify the emission reduction pressure that various industries must bear to meet their targets. Then, based on expert experience, it is judged whether the obtained endogenous derivation of carbon emission intensity results meet the development requirements. If they are unreasonable, the bottom parameters of the first 4 layers of the Bayesian hierarchical model are adjusted.
[0025] Furthermore, the observed carbon emissions of industry i in year t. It follows a normal distribution, as detailed below:
[0026] Normal( );
[0027] ;
[0028] Where Normal represents the normal distribution function; This represents the mathematical expectation of carbon emissions for industry i in year t. This represents the observation variance of carbon emissions for industry i. Indicates industry i in the starting year Raw carbon emission data; t represents the year index;
[0029] The observed activity level k in year t It follows a normal distribution, as detailed below:
[0030] Normal( );
[0031] ;
[0032] in, This represents the mathematical expectation of activity level k in year t. The observed variance represents the activity level k; Indicates the starting year Raw data for activity level k; This represents the prior of the rate of increase in activity level k;
[0033] Historical average annual growth rate of carbon emissions It follows the normal distribution:
[0034] Normal( , );
[0035] in, Let j be the prior mean of the industry category j; The prior variance representing the rate of increase in carbon emissions;
[0036] Activity level growth prior It follows the normal distribution:
[0037] Normal( , );
[0038] in, This represents the historical average rate of increase in inertia at activity level k; This represents the prior variance of the rate of increase in activity level.
[0039] Furthermore, in S43, the pre-adjustment raw emission forecasts for industry i in the next year T. The expression is as follows:
[0040] ;
[0041] The autonomous projection value of activity level k in S44 for the next T years The expression is as follows:
[0042] ;
[0043] The maximum permissible carbon emissions in the study area mentioned in S45 The expression is as follows:
[0044] ;
[0045] in, This represents the total carbon emissions of the study region in year t. This represents the total GDP of the study region in year t. This represents the target value for reducing carbon emission intensity in the study area within year Tt; This indicates the expected average annual growth rate of the studied region within year Tt;
[0046] System balance factor in S45 The expression is as follows:
[0047] ;
[0048] in, This indicates the maximum permissible total carbon emissions for the studied region; This represents the additional carbon emissions, which are reserved for carbon emission allowances for future major new production capacity. It is the sum of the mathematical expectations of the original forecasts for all industries, that is, the sum of the predicted carbon emissions for the next T years before balance adjustment of key carbon emissions and base carbon emissions;
[0049] The final predicted value of industry i in S45 after balance adjustment in the next T years. The expression is as follows:
[0050] .
[0051] Furthermore, the specific components of the dual balance constraint include:
[0052] a. Bottom-up carbon emissions totals are the binding carbon emissions totals calculated from top to bottom, expressed as follows:
[0053] ;
[0054] ;
[0055] in, This represents the sum of bottom-up carbon emission forecasts for each industry. , These represent the minimum and maximum new carbon emissions, respectively.
[0056] b. New carbon emissions As a balancing factor, the following condition must be met: Total Emissions - Basic Carbon Emissions - Key Carbon Emissions = New Carbon Emissions The expression is as follows:
[0057] ;
[0058] in, This indicates the projected carbon emissions for year T after balance adjustment for key carbon emissions; This represents the base carbon emissions adjusted for balance, and the projected carbon emissions for year T. and The sum of these equals the sum of the predicted carbon emissions for the next T years, H.
[0059] Furthermore, the endogenously derived carbon emission intensity in S47 The expression is as follows:
[0060] .
[0061] In a second aspect, the present invention also provides a comprehensive carbon emission budget allocation system, which is configured to or executes the aforementioned comprehensive carbon emission budget allocation method.
[0062] The beneficial effects of this invention are:
[0063] 1. This invention discloses and uses a Bayesian hierarchical model, which has the following advantages:
[0064] a. The Bayesian hierarchical model can capture the macro trends of major industries such as transportation and industry, and accurately identify the operational characteristics of various industrial products and different types of transportation.
[0065] b. In the modeling process, the emission reduction target is set as the "model prior", so that the prediction results are no longer a simple statistical extrapolation, but a deep integration of historical patterns and policies;
[0066] c. Simultaneous modeling of carbon emissions and activity levels is adopted to obtain the results of the data game between total emissions and activity levels, and to deeply analyze the impact of the two-factor perspective on carbon emissions.
[0067] d. It can quantify the risks of the entire process. Unlike traditional single numerical prediction, the Bayesian hierarchical model gives a 95% confidence interval, which can quantify the uncertainty of the prediction results and intuitively show the potential volatility risks in the future carbon budget execution. Attached Figure Description
[0068] Figure 1 This is a flowchart of the carbon emission budget allocation method for the entire field in this invention. Detailed Implementation
[0069] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many other different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.
[0070] Reference Figure 1 This application provides a method for allocating carbon emission budgets across the entire domain based on dual-stream Bayesian hierarchical structure, comprising the following steps:
[0071] S1. Extract sub-indicators for each industry from different industries;
[0072] S2. Based on the economic development forecast and cumulative carbon emission intensity reduction target for the study area in the next T years, calculate the maximum permissible total carbon emissions for the study area in the next T years. ;
[0073] S3. Collect historical data for various industry sub-indicators, and combine this data with the planning and actual conditions of the study area to provide a priori values for the historical average annual growth rate of carbon emissions for each industry. ;
[0074] S4. Combine the historical data of each industry's sub-indicators with the prior values of each industry's historical average annual carbon emission growth rate. Input into the Bayesian hierarchical model (i.e. Figure 1 In the dual-stream Bayesian inference model, the predicted values and 95% confidence intervals of each industry sub-indicator are calculated. The predicted values of each industry sub-indicator include the final predicted value of each industry after balance adjustment in the next T years, the autonomously extrapolated values of all activity levels in the next T years (e.g., 2030), the endogenously derived carbon emission intensity of each industry in the next T years, and the newly added carbon emissions. .
[0075] In some embodiments, the following steps are further included after step S4:
[0076] S5. Visualize and output the predicted values of various industry sub-indicators to achieve a visual display of the prediction results.
[0077] In some embodiments, S1 specifically includes the following steps:
[0078] S11. Extract sub-indicators for each industry from industry, construction, agriculture, forestry, animal husbandry and fishery, transportation, service industry, residents' life, processing and transformation, industrial production process and other industries, so as to make comparison and improvement with other regions in multiple dimensions;
[0079] The indicators for each industry cover multiple sectors, including industry, construction, agriculture, forestry, animal husbandry and fishery, transportation, services, residential life, and other industries. Specifically, in agriculture, forestry, animal husbandry and fishery, indicators include total power of agricultural machinery and carbon emissions per unit of total power of agricultural machinery; in transportation, indicators include turnover carbon emission intensity, turnover, number of vehicles, and carbon emission intensity per vehicle; in residential life, indicators include the number of permanent residents and per capita residential carbon emissions; in services and other industries, indicators include carbon emissions per unit of commercial and public building area and commercial and public building area; in industry, indicators include industrial energy consumption, carbon emission factors of energy types, output of industrial production processes, carbon emission factors of product types, carbon emission base of processing and conversion processes, and processing and conversion intensity; and in construction, indicators include construction added value and carbon emissions per unit of added value in construction.
[0080] S12. Carbon emissions from different industries are divided into key carbon emissions and basic carbon emissions. Key carbon emissions include carbon emissions from industry, construction, processing and conversion, and industrial production processes. Basic carbon emissions include carbon emissions from agriculture, forestry, animal husbandry and fishery, transportation, services, and residential life.
[0081] In some embodiments, the prior value of the historical annual average growth rate of carbon emissions in S3 The given data should at least consider the average annual carbon emission growth rate of regions with the same climate type, development stage, similar area and city size as the study region, both domestically and internationally.
[0082] In some embodiments, the Bayesian hierarchical model includes, in sequence, an adaptive observation layer, a prior and parameter generation layer, a prediction and posterior evolution layer, a macro-equilibrium and redline verification layer, and an endogenous game and intensity inference layer. To satisfy the final balance and verification relationship of the model, there are iterative and game-like processes between each layer.
[0083] In some embodiments, S4 specifically includes the following steps:
[0084] S41. Input the historical data of each industry's sub-indicators into the adaptive observation layer of the Bayesian hierarchical model, and fit the observed carbon emission value of industry i in year t. and the observed activity level k in year t. ;
[0085] S42. Observation of carbon emissions of industry i in year t. and the observed activity level k in year t. The data is input into the prior and parameter generation layer to obtain the historical data distribution of carbon emissions and activity levels in various industries. Then, based on the historical data distribution of carbon emissions and activity levels in various industries, the historical average annual growth rate of each sub-indicator and carbon emissions is obtained, which serves as the basic supporting data for the prior values of the future average annual growth rate of each industry sub-indicator and the prior values of the future average annual growth rate of carbon emissions.
[0086] S43. Based on historical data of various industry sub-indicators and prior values of historical average annual growth rates of carbon emissions. The distribution of the posterior growth rate of carbon emissions was calculated. Then, the distribution of the posterior growth rate of carbon emissions random variable... The data is input into the prediction and posterior evolution layer, and after MCMC sampling, the posterior carbon emission prediction distribution is obtained, including the pre-adjustment original emission predictions for multiple different industries i in the next T years. Then, the posterior prediction distribution of carbon emissions is arranged to obtain the 95% confidence interval of the posterior prediction distribution of carbon emissions.
[0087] Specifically, the distribution of the posterior growth rate of carbon emissions is first calculated. Then, the distribution of the posterior growth rate of carbon emissions was obtained through MCMC sampling. Q samples are extracted from the model. Each sample is substituted into the prediction and posterior evolution layer of the Bayesian hierarchical model to calculate the corresponding carbon emission prediction value, resulting in Q carbon emission prediction values. These Q carbon emission prediction values constitute the posterior prediction distribution of carbon emissions. The Q carbon emission prediction values are sorted, and the 2.5th quantile is taken as the lower limit, and the 97.5th quantile is taken as the upper limit. The 95% confidence interval is represented as [E_P2.5, E_P97.5], where E_P2.5 represents the 2.5th quantile of carbon emissions, and E_P97.5 represents the 97.5th quantile of carbon emissions.
[0088] S44. Based on historical data of industry sub-indicators and priori growth rates of activity levels The distribution of the posterior growth rate of the activity level random variable was calculated. Then, the distribution of the posterior growth rate random variable of the activity level. The input is fed into the prediction and posterior evolution layer, and the posterior prediction distribution of the activity level obtained through MCMC sampling includes autonomous extrapolation values of multiple different activity levels k for the next T years. Then, the posterior prediction distribution of the activity level is sorted to construct the 95% confidence interval of the posterior prediction distribution of the activity level.
[0089] Among them, MCMC (Markov Chain Monte Carlo) sampling is a powerful statistical simulation technique, mainly used to extract samples from complex, high-dimensional probability distributions.
[0090] S45. Introduce the maximum permissible carbon emission amount for the study area into the macro-equilibrium and redline check layer of the Bayesian hierarchical model. As a macroeconomic constraint, and based on meeting the macroeconomic constraint, the original emission forecasts before adjustment are used. Calculate the system balance factor Then, based on the system balance factor and the original emission forecasts before adjustment Calculate the final balanced forecast value of industry i in the next T years. and final predicted value 95% confidence interval; final predicted value The calculation method for the 95% confidence interval is similar to the construction method of the 95% confidence interval in S43 and S44, and will not be elaborated in detail below;
[0091] S46. Final balanced forecasts for all industries in year T. Summing is performed to obtain the sum of the key carbon emissions and basic carbon emissions, adjusted for balance, for the predicted carbon emissions in the next T years, H; then, based on the maximum permissible total carbon emissions for the study area... The sum of the projected carbon emissions for the next T years, H, is used to calculate the new carbon emissions. Introducing a dual balance constraint, when new carbon emissions... When the dual balance constraint is not satisfied, adjust the parameters in the prior and parameter generation layers, the prediction and posterior evolution layers, and the system balance factor in the macro-balance and redline verification layers. To satisfy the dual balance constraint;
[0092] S47. Based on satisfying the dual balance constraints, the final predicted value will be... and autonomously extrapolated values The input is fed into the endogenous game and intensity extrapolation layers of the Bayesian hierarchical model to calculate the endogenous carbon emission intensity of industry i in the next T years. and endogenous derivation of carbon emission intensity The 95% confidence interval is used to quantify the emission reduction pressure that various industries must bear to meet their targets. Then, based on expert experience, it is judged whether the obtained endogenous derivation of carbon emission intensity results meet the development requirements. If they are unreasonable, the bottom parameters of the first 4 layers of the Bayesian hierarchical model are adjusted.
[0093] In some embodiments, the carbon emission observation of industry i in year t It follows a normal distribution, as detailed below:
[0094] Normal( );
[0095] ;
[0096] Where Normal represents the normal distribution function; This represents the mathematical expectation of carbon emissions for industry i in year t. This represents the observation variance of carbon emissions for industry i. Indicates industry i in the starting year Raw carbon emission data; t represents the year index;
[0097] The observed activity level k in year t It follows a normal distribution, as detailed below:
[0098] Normal( );
[0099] ;
[0100] in, This represents the mathematical expectation of activity level k in year t. The observed variance represents the activity level k; Indicates the starting year Raw data for activity level k; This represents the prior of the rate of increase in activity level k;
[0101] Historical average annual growth rate of carbon emissions It follows the normal distribution:
[0102] Normal( , );
[0103] in, Let j be the prior mean of the industry category j; The prior variance representing the rate of increase in carbon emissions;
[0104] Activity level growth prior It follows the normal distribution:
[0105] Normal( , );
[0106] in, This represents the historical average rate of increase in inertia at activity level k; This represents the prior variance of the rate of increase in activity level.
[0107] In some embodiments, the pre-adjustment raw emission forecasts for industry i in S43 for the next T years The expression is as follows:
[0108] ;
[0109] The autonomous projection value of activity level k in S44 for the next T years The expression is as follows:
[0110] ;
[0111] The maximum permissible carbon emissions in the study area mentioned in S45 The expression is as follows:
[0112] ;
[0113] in, This represents the total carbon emissions of the study region in year t. This represents the total GDP of the study region in year t. This represents the target value for reducing carbon emission intensity in the study area within year Tt; This indicates the expected average annual growth rate of the studied region within year Tt;
[0114] System balance factor in S45 The expression is as follows:
[0115] ;
[0116] in, This indicates the maximum permissible total carbon emissions for the studied region; This represents the additional carbon emissions, which are reserved for carbon emission allowances for future major new production capacity. It is the sum of the mathematical expectations of the original forecasts for all industries, that is, the sum of the predicted carbon emissions for the next T years before balance adjustment of key carbon emissions and base carbon emissions;
[0117] The final predicted value of industry i in S45 after balance adjustment in the next T years. The expression is as follows:
[0118] .
[0119] In some embodiments, the dual balance constraint specifically includes:
[0120] a. Bottom-up carbon emissions totals are the binding carbon emissions totals calculated from top to bottom, expressed as follows:
[0121] ;
[0122] ;
[0123] in, This represents the sum of bottom-up carbon emission forecasts for each industry. , These represent the minimum and maximum new carbon emissions, respectively.
[0124] b. New carbon emissions As a balancing factor, the following condition must be met: Total Emissions - Basic Carbon Emissions - Key Carbon Emissions = New Carbon Emissions The expression is as follows:
[0125] ;
[0126] in, This represents the projected carbon emissions for year T after balance adjustment for key carbon emissions, corresponding to... Figure 1 In ; This represents the predicted carbon emissions for year T after the base carbon emissions have been balanced, corresponding to... Figure 1 In , and The sum of these equals the sum of the predicted carbon emissions for the next T years, H.
[0127] In some embodiments, the endogenously derived carbon emission intensity in S47 The expression is as follows:
[0128] .
[0129] This invention discloses the use of a Bayesian hierarchical model, the innovation of which is reflected in the following three aspects:
[0130] a. Mechanism integration: The Bayesian hierarchical model achieves organic coupling of "top-down" macro constraints and "bottom-up" industry-specific dynamic forecasting within a unified framework.
[0131] The "top-down" approach primarily uses macroeconomic indicators to predict energy systems and carbon emissions. Its advantages include high data availability and a simple, intuitive reflection of the impact of various driving factors on energy consumption and carbon emissions. However, its disadvantages include a macro-level focus on forecasting and analysis, which cannot be broken down to specific industries or energy types, and cannot reflect the specific optimization process of energy consumption and carbon emissions.
[0132] "Bottom-up" refers to using data from each stage to conduct detailed analysis and simulation of the impact of technological progress. Data from each stage includes energy production, conversion, and end-use consumption in various industries. The disadvantage is that data availability is poor, and the conclusions are easily distorted after using empirical methods to decompose the data. It is also easy to ignore the impact of changes in other fields on the research field.
[0133] b. Dynamic intelligent adjustment: The Bayesian hierarchical model has the ability to dynamically and flexibly adjust the total amount of new carbon emissions and carbon quotas between industries to respond to uncertainties and real-time needs.
[0134] c. Precise implementation orientation: Specifically designed for the data characteristics, industrial structure and planning process at the provincial level, it is committed to improving the scientific nature, operability and fit of the decomposition results with local planning.
[0135] A second aspect of the present invention also provides a comprehensive carbon emission budget allocation system, which is configured to or executes the aforementioned comprehensive carbon emission budget allocation method.
[0136] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A comprehensive carbon emission budget allocation method based on two-stream Bayesian hierarchical structure, characterized in that, Includes the following steps: S1. Extract sub-indicators for each industry from different industries; S2. Based on the economic development forecast and cumulative carbon emission intensity reduction target for the study area in the next T years, calculate the maximum permissible total carbon emissions for the study area in the next T years. ; S3. Collect historical data for various industry sub-indicators, and combine this data with the planning and actual conditions of the study area to provide a priori values for the historical average annual growth rate of carbon emissions for each industry. ; S4. Combine the historical data of each industry's sub-indicators with the prior values of each industry's historical average annual carbon emission growth rate. The data is input into a Bayesian hierarchical model to calculate the predicted values and 95% confidence intervals for each industry's sub-indicators. The predicted values for each industry's sub-indicators include the final, balanced predicted value for the next T years, the autonomously extrapolated values of all activity levels for the next T years, the endogenously derived carbon emission intensity for each industry in the next T years, and the newly added carbon emissions. ; The Bayesian hierarchical model includes, in sequence, an adaptive observation layer, a prior and parameter generation layer, a prediction and posterior evolution layer, a macro-equilibrium and redline verification layer, and an endogenous game and intensity inference layer. S4 specifically includes the following steps: S41. Input the historical data of each industry's sub-indicators into the adaptive observation layer of the Bayesian hierarchical model, and fit the observed carbon emission value of industry i in year t. and the observed activity level k in year t. ; S42. Observation of carbon emissions of industry i in year t. and the observed activity level k in year t. The data is input into the prior and parameter generation layer to obtain the historical data distribution of carbon emissions and activity levels in various industries. Then, based on the historical data distribution of carbon emissions and activity levels in various industries, the historical average annual growth rate of each sub-indicator and carbon emissions is obtained, which serves as the basic supporting data for the prior values of the future average annual growth rate of each industry sub-indicator and the prior values of the future average annual growth rate of carbon emissions. S43. Based on historical data of various industry sub-indicators and prior values of historical average annual growth rates of carbon emissions. The distribution of the posterior growth rate of carbon emissions was calculated. Then, the distribution of the posterior growth rate of carbon emissions random variable... The data is input into the prediction and posterior evolution layer, and after MCMC sampling, the posterior carbon emission prediction distribution is obtained, including the pre-adjustment original emission predictions for multiple different industries i in the next T years. Then, the posterior prediction distribution of carbon emissions is arranged to obtain the 95% confidence interval of the posterior prediction distribution of carbon emissions. S44. Based on historical data of industry sub-indicators and priori growth rates of activity levels The distribution of the posterior growth rate of the activity level random variable was calculated. Then, the distribution of the posterior growth rate random variable of the activity level. The input is fed into the prediction and posterior evolution layer, and the posterior prediction distribution of the activity level obtained through MCMC sampling includes autonomous extrapolation values of multiple different activity levels k for the next T years. Then, the posterior prediction distribution of the activity level is sorted to construct the 95% confidence interval of the posterior prediction distribution of the activity level. S45. Introduce the maximum permissible carbon emission amount for the study area into the macro-equilibrium and redline check layer of the Bayesian hierarchical model. As a macroeconomic constraint, and based on meeting the macroeconomic constraint, the original emission forecasts before adjustment are used. Calculate the system balance factor Then, based on the system balance factor and the original emission forecasts before adjustment Calculate the final balanced forecast value of industry i in the next T years. and final predicted value The 95% confidence interval; S46. Final balanced forecasts for all industries in year T. Summing is performed to obtain the sum of the key carbon emissions and basic carbon emissions, adjusted for balance, for the predicted carbon emissions in the next T years, H; then, based on the maximum permissible total carbon emissions for the study area... The sum of the projected carbon emissions for the next T years, H, is used to calculate the new carbon emissions. Introducing a dual balance constraint, when new carbon emissions... When the dual balance constraint is not satisfied, adjust the parameters in the prior and parameter generation layers, the prediction and posterior evolution layers, and the system balance factor in the macro-balance and redline verification layers. To satisfy the dual balance constraint; S47. Based on satisfying the dual balance constraints, the final predicted value will be... and autonomously extrapolated values The input is fed into the endogenous game and intensity extrapolation layers of the Bayesian hierarchical model to calculate the endogenous carbon emission intensity of industry i in the next T years. and endogenous derivation of carbon emission intensity The 95% confidence interval is used to quantify the emission reduction pressure that various industries must bear to meet the target. Then, based on expert experience, it is judged whether the obtained endogenous derivation of carbon emission intensity results meet the development requirements. If it is unreasonable, the bottom parameters of the first 4 layers of the Bayesian hierarchical model are adjusted. The pre-adjustment raw emission forecasts for industry i in S43 for the next T years. The expression is as follows: ; The autonomous projection value of activity level k in S44 for the next T years The expression is as follows: ; The maximum permissible carbon emissions in the study area mentioned in S45 The expression is as follows: ; in, This represents the total carbon emissions of the study region in year t. This represents the total GDP of the study region in year t. This represents the target value for reducing carbon emission intensity in the study area within year Tt; This indicates the expected average annual growth rate of the studied region within year Tt; System balance factor in S45 The expression is as follows: ; in, This indicates the maximum permissible total carbon emissions for the studied region; This represents the additional carbon emissions, which are reserved for carbon emission allowances for future major new production capacity. It is the sum of the mathematical expectations of the original forecasts for all industries, that is, the sum of the predicted carbon emissions for the next T years before balance adjustment of key carbon emissions and base carbon emissions; The final predicted value of industry i in S45 after balance adjustment in the next T years. The expression is as follows: 。 2. The method for allocating a comprehensive carbon emission budget based on dual-stream Bayesian hierarchical structure according to claim 1, characterized in that, S1 specifically includes the following steps: S11. Extract sub-indicators for each industry from the processes of industry, construction, agriculture, forestry, animal husbandry and fishery, transportation, service industry, residents' life, processing and conversion and industrial production. S12. Carbon emissions from different industries are divided into key carbon emissions and basic carbon emissions. Key carbon emissions include carbon emissions from industry, construction, processing and conversion, and industrial production processes. Basic carbon emissions include carbon emissions from agriculture, forestry, animal husbandry and fishery, transportation, services, and residential life.
3. The method for allocating a comprehensive carbon emission budget based on dual-stream Bayesian hierarchical structure according to claim 1, characterized in that, The prior value of the historical average annual growth rate of carbon emissions in S3 Given the data, we also need to consider the average annual growth rate of carbon emissions in regions with the same climate type, development stage, similar area and city size as the study region, both domestically and internationally.
4. The method for allocating a comprehensive carbon emission budget based on dual-stream Bayesian hierarchical structure according to claim 1, characterized in that, The carbon emission observation value of industry i in year t It follows a normal distribution, as detailed below: Normal( ); ; Where Normal represents the normal distribution function; This represents the mathematical expectation of carbon emissions for industry i in year t. This represents the observation variance of carbon emissions for industry i. Indicates industry i in the starting year Raw carbon emission data; t represents the year index; The observed activity level k in year t It follows a normal distribution, as detailed below: Normal( ); ; in, This represents the mathematical expectation of activity level k in year t. The observed variance represents the activity level k; Indicates the starting year Raw data for activity level k; This represents the prior of the rate of increase in activity level k; Historical average annual growth rate of carbon emissions It follows the normal distribution: Normal( , ); in, Let j be the prior mean of the industry category j; The prior variance representing the rate of increase in carbon emissions; Activity level growth prior It follows the normal distribution: Normal( , ); in, This represents the historical average rate of increase in inertia at activity level k; This represents the prior variance of the rate of increase in activity level.
5. The method for allocating a comprehensive carbon emission budget based on dual-stream Bayesian hierarchical structure according to claim 4, characterized in that, The specific components of the dual balance constraints include: a. Bottom-up carbon emissions totals are the binding carbon emissions totals calculated from top to bottom, expressed as follows: ; ; in, This represents the sum of bottom-up carbon emission forecasts for each industry. , These represent the minimum and maximum new carbon emissions, respectively. b. New carbon emissions As a balancing factor, the following condition must be met: Total Emissions - Basic Carbon Emissions - Key Carbon Emissions = New Carbon Emissions The expression is as follows: ; in, This indicates the projected carbon emissions for year T after balance adjustment for key carbon emissions; This represents the base carbon emissions adjusted for balance, and the projected carbon emissions for year T. and The sum of these equals the sum of the predicted carbon emissions for the next T years, H.
6. The method for allocating a comprehensive carbon emission budget based on dual-stream Bayesian hierarchical structure according to claim 5, characterized in that, The endogenously derived carbon emission intensity in S47 The expression is as follows: 。 7. A comprehensive carbon emission budget allocation system, characterized in that, The method for allocating carbon emission budgets across the entire domain as described in any one of claims 1 to 6 is configured or implemented.