City carbon intensity index decomposition method and system based on target planning

Abstract

The embodiment of the invention provides a city carbon intensity index decomposition method and system based on target planning, and belongs to the field of low-carbon technology and information processing. The method comprises the steps that multi-source data are collected and preprocessed; carrying out regional carbon emission mode clustering based on PCA-K-means, wherein the clustering comprises PCA dimension reduction, K-means clustering and optimal cluster number determination; calculating a carbon emission reduction potential index based on an entropy weight method; carbon emission influence factor regression modeling is carried out based on LMDI; performing carbon strength target decomposition modeling based on linear programming; and outputting a result and dynamically adjusting. According to the method and the system, scientific, fair and efficient decomposition of the provincial carbon intensity target to the city level is realized, the emission reduction cost is low, and the application prospect is wide.

Description

Technical Field

[0001] This invention relates to the fields of low-carbon technology and information processing technology, specifically to a method and system for decomposing municipal carbon intensity indicators based on target planning. Background Technology

[0002] Addressing climate change and controlling greenhouse gas emissions has become a global consensus, requiring a focus on carbon emission intensity control and the establishment of a three-tiered carbon emission budget management system at the national, provincial, and municipal levels. Therefore, scientifically, rationally, and fairly allocating provincial carbon emission intensity control targets to various cities is a crucial technological foundation for achieving refined top-down management and ensuring the achievement of provincial and even national targets.

[0003] Currently, there has been some technological exploration in the field of carbon emission quota allocation, but it mainly focuses on macro-level allocation from the national to the provincial level. Existing technological solutions have significant shortcomings in the refined allocation at the prefecture-level city level within provinces, making it difficult to meet practical operational needs. Specifically: 1. Insufficient adaptability between decomposition hierarchy and objects: Existing research methods and practices mostly focus on the decomposition of indicators at the international or national / provincial levels, lacking differentiated decomposition schemes specifically for the city-level divisions within a province. Significant spatial heterogeneity exists among cities within a province in terms of economic development stage, industrial structure, energy consumption structure, resource endowment, and technological capabilities. Using a "one-size-fits-all" or simplistic average decomposition method often leads to a disconnect between objectives and reality.

[0004] 2. The determination of key parameters is highly subjective: When constructing decomposition models, existing technologies often rely on expert judgment or simple historical data averaging to set key weight parameters that affect the allocation results (such as the weights of each city's emission reduction responsibilities, capabilities, and potential), lacking a quantitative calibration mechanism driven by objective data. This subjective or crude setting method raises questions about the fairness and scientific validity of the decomposition results, easily triggering disputes between regions and hindering the coordinated implementation of objectives.

[0005] 3. The model has a single constraint dimension, making it difficult to balance multiple objectives: Traditional decomposition methods often consider only a few factors, such as historical emissions or total economic output, failing to systematically incorporate multi-dimensional real-world constraints, including rigid demands for economic development, resource and environmental carrying capacity limits, the feasibility of technological emission reduction, and industrial policy guidance. This leads to decomposition schemes that easily fall into egalitarianism or become disconnected from local actual development capabilities, failing to achieve an effective balance between the two core principles of fairness and efficiency, thus affecting the ultimate achievability of the goals.

[0006] While some researchers in existing academic and technical literature have attempted to improve upon this approach—for example, some have proposed carbon intensity decomposition methods that consider fairness principles—their models lack systematic heterogeneous cluster analysis of cities within a province, failing to accurately identify the differentiated allocation bases of cities with different characteristics. Others have tried to optimize regional emission reduction allocation efficiency using Data Envelopment Analysis (DEA), but their studies have not incorporated driving factor decomposition methods such as the Log-mean Dijkstra index (LMDI) to objectively quantify the actual contribution of various factors to emission changes in different cities over historical periods, resulting in inaccurate potential assessments.

[0007] In summary, existing technical solutions generally suffer from shortcomings such as "insufficient heterogeneity identification, imprecise potential assessment, and weak multi-constraint optimization." A complete technical process has not yet been formed, from identifying differences between cities and prefectures to quantifying emission reduction potential, and then to dynamic optimization decomposition under multiple objectives and constraints. Summary of the Invention

[0008] The purpose of this invention is to provide a method and system for decomposing municipal carbon intensity indicators based on target planning. This method and system realize the scientific, fair, and efficient decomposition of provincial carbon intensity targets to the municipal level, with low emission reduction costs and broad application prospects.

[0009] To achieve the above objectives, embodiments of the present invention provide a method for decomposing municipal carbon intensity indicators based on target planning, the method comprising: Collect multi-source data and perform preprocessing; Regional carbon emission pattern clustering based on PCA-K-means was performed, including PCA dimensionality reduction, K-means clustering, and determination of the optimal number of clusters; Calculation of carbon emission reduction potential index based on entropy weight method; Regression modeling of carbon emission impact factors based on LMDI; Carbon intensity target decomposition modeling based on linear programming; Output the results and adjust them dynamically.

[0010] Preferably, the collected multi-source data includes provincial and municipal economic data, energy data, population data, and carbon emission data, and the collected data is standardized according to formula (1) to eliminate dimensional differences.

[0011] in, For the first The first city The original value of the indicator, The first The minimum and maximum values ​​of the indicators, The value is the standardized value, and its range is [0,1].

[0012] Preferably, PCA dimensionality reduction includes constructing a covariance matrix based on the preprocessed index dataset using standardized data, performing eigenvalue decomposition on the covariance matrix, sorting the eigenvalues ​​from largest to smallest, and extracting the top eigenvalues ​​with a cumulative contribution rate ≥ 85%. Principal components are used to construct a new feature space. ,in, Number of prefecture-level cities; K-means clustering involves clustering city samples in the principal component space according to formula (2), with the objective function being "minimizing the squared Euclidean distance from the sample point to the cluster center".

[0013] in, For the first Clusters, Let be the mean of the principal component vectors of the samples within the cluster. For the first Principal component vectors of each sample; Determining the optimal number of clusters involves using silhouette coefficients and cluster stability tests to determine the optimal number of clusters. The requirements are a profile coefficient ≥ 0.6 and stability ≥ 90%, ultimately dividing the prefecture-level cities into... Regions with similar carbon emission patterns and development levels.

[0014] Preferably, the calculation of the carbon emission reduction potential index based on the entropy weight method includes constructing a "fairness-efficiency" dual-dimensional evaluation system for each clustered region, objectively assigning weights using the entropy weight method, and calculating the comprehensive carbon emission reduction potential index; wherein, Fairness dimension This includes GDP per capita, electricity consumption per capita, historical carbon emission responsibility coefficient, and resident population size; Efficiency dimension This includes carbon emissions per unit of GDP, carbon emissions per unit of energy consumption, the proportion of carbon emissions in the secondary industry, and the proportion of carbon emissions from renewable energy. Entropy weight calculation includes: Calculate the first according to formula (3) The first city The percentage of each indicator

[0015] in, The value is the standardized value; Calculate the first according to formula (4) Information entropy of the indicator

[0016] Among them, if ,definition Information entropy The value range is [0,1]. The smaller the value, the higher the indicator's discriminative power; Calculate the index weights according to formula (5).

[0017] in, For the total number of indicators, ; Calculate the comprehensive potential index according to formula (6).

[0018] in, For the first Carbon emission reduction potential index of each city , To ensure fairness, scores are weighted. Efficiency-weighted score.

[0019] Preferably, the regression modeling of carbon emission impact factors based on LMDI includes LMDI additive decomposition and linear regression modeling, wherein, LMDI Additive Decomposition for Selecting Energy Consumption Structure Energy efficiency Industrial structure Economic development level As the core driving factor, the change in carbon emissions is calculated according to formula (7). Lossless decomposition into the contribution of each factor:

[0020] in, for The carbon emissions at any given time, and the contribution of each factor is calculated according to formula (8).

[0021] in, For a certain driving factor, It is a logarithmic mean function; Linear regression modeling based on changes in carbon emissions Dependent variable The contribution of each driving factor is represented by a vector of independent variables. A linear regression model is established based on formula (9).

[0022] in, For regression coefficients, For error terms, By minimizing the sum of squared residuals Estimate the regression coefficients and use the coefficient of determination. Validate model fit and .

[0023] Preferably, the carbon intensity target decomposition modeling based on linear programming includes: Let the first The rate of change in carbon intensity in each city is a decision variable. , The ratio of carbon intensity in the target year to carbon intensity in the baseline year. This indicates a decrease in carbon intensity. The smaller the value, the greater the decrease. Construct an objective function based on the principle of "prioritizing emission reduction in high-potential areas". According to formula (10), construct a weighted minimization objective function to achieve a balance between "fairness and efficiency".

[0024] in, The rate of change of the resident population. The rate of change in GDP per unit population. The rate of change in energy consumption per unit of GDP The rate of change in carbon emissions per unit of energy consumption; For the first Carbon emission reduction potential index of each city The corresponding variable weights are obtained by Min-Max normalization of the regression coefficients according to formula (11).

[0025] Define constraints, where, Based on formula (12), a total consistency constraint is applied to ensure that the sum of carbon intensity targets of all cities and prefectures conforms to the provincial target.

[0026] in, For the first Baseline annual carbon emissions for each city This represents the total carbon emissions for the provincial baseline year. The rate of change of provincial carbon intensity targets; Economic growth constraints are applied according to formula (13) to ensure that the GDP growth rate per unit population is not lower than the provincial projected annual GDP growth rate. To ensure that economic development is not excessively constrained.

[0027] Resource constraints are applied according to formula (14) to ensure that energy consumption per unit of GDP and carbon emissions per unit of energy consumption are not less than 90% of the minimum values ​​in the province, thus avoiding targets that exceed technical feasibility.

[0028] Population constraints are applied according to formula (15), and the range of resident population change rate is set based on the population change trend of the city.

[0029] Non-negativity constraints are applied according to formula (16) to ensure that the rate of change of decision variables and influencing factors conforms to practical significance.

[0030] The solution is obtained using the Python open-source library PuLP, with the iteration termination condition set to "residual less than 1e-6". The objective function calculation precision is retained to four decimal places. If the number of cities exceeds 50, the interior-point method is used to improve computational efficiency, with the initial iteration step size set to 0.1. The final output is the optimal carbon intensity change rate for each city. .

[0031] Preferably, the output includes calculations of carbon intensity reduction targets for each city. The data is displayed in the form of bar charts and heatmaps, and an analysis report containing "potential index - decline target - constraint satisfaction status" is generated. Dynamic adjustments include establishing a timed update mechanism, periodically repeating the above steps based on the latest economic, energy, and carbon emission data to update and decompose targets; if significant changes occur, an immediate update is triggered without waiting for the cycle.

[0032] Preferably, significant changes include an industrial structure change rate exceeding 20% ​​or a doubling of renewable energy installed capacity.

[0033] On the other hand, the present invention provides a target planning-based decomposition system for municipal carbon intensity indicators to implement the above method. The system includes: The data acquisition and preprocessing module is used to collect and preprocess multi-source data to form a standardized dataset. The regional heterogeneity identification module is used to perform principal component analysis dimensionality reduction and K-means clustering on standardized datasets to identify the carbon emission pattern categories of cities and prefectures. The carbon emission reduction potential quantification module is used to construct an evaluation system with fairness and efficiency dimensions, and to calculate the carbon emission reduction potential index of each city using the entropy weight method. The driving factor modeling module is used to perform log-mean Dijkstra's exponential additive decomposition and linear regression to quantify the impact of carbon emission driving factors. The multi-constraint objective programming module is used to construct and solve a linear programming model with the carbon intensity change rate as the decision variable, and output the carbon intensity decomposition targets for each city. The results output and dynamic management module is used to visualize the decomposition results and manage the dynamic updates and recalculations of the model.

[0034] In another aspect, the present invention provides a machine-readable storage medium having instructions stored thereon for causing a machine to perform the methods described above.

[0035] Based on the above technical solution, by collecting economic, energy, and population data at the provincial and municipal levels, key carbon emission impact indicators are calculated. The LMDI (Leadership in Difference of Influences) factor decomposition algorithm is used to quantify the impact of each indicator on carbon emissions in the cities. Linear regression and a cluster-based carbon reduction comprehensive index model are employed to analyze the carbon emission impact indicators and carbon reduction potential, respectively. A mapping relationship between each carbon emission impact indicator and carbon emissions is established. The carbon emission potential index for each city and industry is calculated. The carbon emission impact indicators are set as linear programming variables. Using the carbon emission potential index for cities and industries, an objective function is set. Combining economic development planning, carbon peaking planning, carbon peaking prediction, and the mapping relationship between carbon emission impact indicators and carbon emissions, linear programming constraints are designed. Through a linear programming solution algorithm, a carbon emission intensity target decomposition model based on linear regression is constructed to achieve carbon emission intensity change rate planning under target constraints, supporting the government's carbon emission dual control target decomposition and carbon emission budget management.

[0036] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0037] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart illustrating the method for decomposing the carbon intensity index of a prefecture-level city based on target planning, provided by the present invention. Figure 2 This is a structural diagram of the decomposition strategy for the carbon intensity index of prefecture-level cities provided by the present invention. Detailed Implementation

[0038] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0039] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with relevant laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0040] See Figure 1 and Figure 2 This invention provides a method for decomposing city-level carbon intensity indicators based on target planning, the method comprising: Collect multi-source data and perform preprocessing; Regional carbon emission pattern clustering based on PCA-K-means was performed, including PCA dimensionality reduction, K-means clustering, and determination of the optimal number of clusters; Calculation of carbon emission reduction potential index based on entropy weight method; Regression modeling of carbon emission impact factors based on LMDI; Carbon intensity target decomposition modeling based on linear programming; Output the results and adjust them dynamically.

[0041] In this embodiment, the collected multi-source data includes provincial and municipal economic data, energy data, population data, and carbon emission data. The data is standardized, outlier removed, and missing value filled to form a standardized dataset. Specifically, the collected data is standardized according to formula (1) to eliminate dimensional differences.

[0042] in, For the first The first city The original value of the indicator, The first The minimum and maximum values ​​of the indicators, The value is the standardized value, and its range is [0,1].

[0043] In practice, multi-source data is collected at the provincial and municipal levels. The data types and requirements are shown in Table 1 below: Table 1

[0044] In this embodiment, regional carbon emission pattern clustering based on PCA-K-means is performed, including PCA dimensionality reduction, K-means clustering, and determination of the optimal number of clusters. This step achieves objective quantification of carbon emission reduction potential. For each cluster category, an evaluation index system including fairness and efficiency dimensions is constructed. The entropy weight method is used to objectively assign weights to each index in the evaluation index system, and the carbon emission reduction potential index of each city is calculated. The PCA dimensionality reduction involves constructing a covariance matrix based on standardized data from a preprocessed dataset of indicators (such as GDP per capita, the proportion of secondary industry, and energy consumption per unit of GDP), performing eigenvalue decomposition on the covariance matrix, sorting the eigenvalues ​​from largest to smallest, and extracting the top eigenvalues ​​with a cumulative contribution rate ≥85%. Principal components are used to construct a new feature space. ,in, This reduces redundancy and noise interference, specifically the number of prefecture-level cities. The K-means clustering method involves clustering city samples in the principal component space according to formula (2), with the objective function being "minimizing the squared Euclidean distance from the sample point to the cluster center".

[0045] in, For the first Clusters, The cluster center (the mean of the principal component vectors of the samples within the cluster). For the first Principal component vectors of each sample; The determination of the optimal number of clusters includes using the silhouette coefficient and cluster stability test (repeated clustering 10 times, calculating the consistency of samples within each cluster). The requirements are a profile coefficient ≥ 0.6 and stability ≥ 90%, ultimately dividing the prefecture-level cities into... Regions with similar carbon emission patterns and development levels.

[0046] Next, in this embodiment, carbon emission driving factors are modeled. The historical carbon emission changes are decomposed into the contributions of multiple preset driving factors using the logarithmic average Dichotomy exponent additive decomposition method. Based on these contributions, a linear regression model is established between the carbon emission changes and the contributions of each driving factor to obtain the regression coefficients of each driving factor. Specifically: For each cluster of regions obtained above, a dual-dimensional evaluation system of "fairness-efficiency" is constructed, and the entropy weight method is used to objectively assign weights to calculate the comprehensive carbon emission reduction potential index. Construction of evaluation index system: Fairness dimension This includes GDP per capita, electricity consumption per capita, historical carbon emission responsibility coefficient, and resident population size; among which, GDP per capita reflects the right to development, electricity consumption per capita reflects the right to access energy, historical carbon emission responsibility coefficient reflects historical responsibility, and resident population size reflects people's livelihood needs. Efficiency dimension This includes carbon emissions per unit of GDP, carbon emissions per unit of energy consumption, the proportion of secondary industry, and the proportion of renewable energy. Among these, carbon emissions per unit of GDP reflect the efficiency of economic emission reduction, carbon emissions per unit of energy consumption reflect the efficiency of energy structure emission reduction, the proportion of secondary industry reflects the potential for industrial transformation, and the proportion of renewable energy is (wind power + photovoltaic installed capacity) / total installed capacity, reflecting the potential of clean energy. Entropy weight calculation includes: Calculate the first according to formula (3) The first city The percentage of each indicator

[0047] in, The value is the standardized value; Calculate the first according to formula (4) Information entropy of the indicator

[0048] Among them, if ,definition (To avoid computational overflow); Information entropy The value range is [0,1]. The smaller the value, the higher the indicator's discriminative power; Calculate the index weights according to formula (5).

[0049] in, For the total number of indicators, ; Calculate the comprehensive potential index according to formula (6).

[0050] in, For the first Carbon emission reduction potential index of each city , To ensure fairness, scores are weighted. Efficiency-weighted score.

[0051] In this embodiment, the regression modeling of carbon emission impact factors based on LMDI includes LMDI additive decomposition and linear regression modeling, wherein, LMDI Additive Decomposition for Selecting Energy Consumption Structure (e.g., coal content), energy efficiency (e.g., energy consumption per unit of GDP), industrial structure (e.g., the proportion of secondary industry), level of economic development (e.g., GDP per capita) is the core driving factor, and the change in carbon emissions is calculated according to formula (7). Lossless decomposition into the contribution of each factor:

[0052] in, for The carbon emissions at any given time, and the contribution of each factor is calculated according to formula (8).

[0053] in, For a certain driving factor ( ), It is a logarithmic mean function; The linear regression model uses changes in carbon emissions. Dependent variable The contribution of each driving factor is represented by a vector of independent variables. A linear regression model is established based on formula (9).

[0054] in, For regression coefficients, For error terms, By minimizing the sum of squared residuals Estimate the regression coefficients and use the coefficient of determination. Validate model fit and .

[0055] Then, a multi-constraint objective programming decomposition is performed, using the carbon intensity change rate of each city as the decision variable to construct the objective function and constraints, forming a linear programming model. The objective function aims to minimize the weighted adjustment amount, with the weights of the adjustment amount negatively correlated with the carbon emission reduction potential index. The constraints include at least total consistency constraints, economic growth constraints, resource constraints, and population constraints. The parameter thresholds for resource constraints are set based on regression coefficients. The linear programming model is solved using a linear programming algorithm to obtain the optimal carbon intensity decomposition objectives for each city. When setting decision variables, let the first... The rate of change in carbon intensity in each city is a decision variable. , The ratio of carbon intensity in the target year to carbon intensity in the baseline year. This indicates a decrease in carbon intensity. The smaller the value, the greater the decrease. Construct an objective function based on the principle of "prioritizing emission reduction in high-potential areas". According to formula (10), construct a weighted minimization objective function to achieve a balance between "fairness and efficiency".

[0056] in, The rate of change of the resident population. The rate of change in GDP per unit population. The rate of change in energy consumption per unit of GDP The rate of change in carbon emissions per unit of energy consumption; For the first Carbon emission reduction potential index of each city The corresponding variable weights are obtained by Min-Max normalization of the regression coefficients according to formula (11).

[0057] Define constraints, where, Based on formula (12), a total consistency constraint is applied to ensure that the sum of carbon intensity targets of all cities and prefectures conforms to the provincial target.

[0058] in, For the first Baseline annual carbon emissions (10,000 tons of CO2) for each prefecture-level city. This represents the total carbon emissions for the provincial baseline year. The change rate of provincial carbon intensity targets (e.g., "a 35% decrease compared to the base year") ); Economic growth constraints are applied according to formula (13) to ensure that the GDP growth rate per unit population is not lower than the provincial projected annual GDP growth rate. To ensure that economic development is not excessively constrained.

[0059] Resource constraints are applied according to formula (14) to ensure that energy consumption per unit of GDP and carbon emissions per unit of energy consumption are not less than 90% of the minimum values ​​in the province, thus avoiding targets that exceed technical feasibility.

[0060] Population constraints are applied according to formula (15), and the range of resident population change rate is set based on the population change trend of the city.

[0061] That is, a negative population growth rate of 0.1%. 0.3%; Non-negativity constraints are applied according to formula (16) to ensure that the rate of change of decision variables and influencing factors conforms to practical significance.

[0062] The reduction in carbon intensity should not exceed 50% to avoid the target becoming unattainable; The solution is obtained using the Python open-source library PuLP, with the iteration termination condition set to "residual less than 1e-6". The objective function calculation precision is retained to four decimal places. If the number of cities exceeds 50, the interior-point method is used to improve computational efficiency, with the initial iteration step size set to 0.1. The final output is the optimal carbon intensity change rate for each city. .

[0063] Finally, the results are output and dynamically adjusted. The carbon intensity decomposition targets for each city are output, and a dynamic adjustment mechanism is established to trigger data updates and model recalculations periodically or when significant pre-defined structural changes occur in a city. Specific outputs include calculations of carbon intensity reduction targets for each city. The data is displayed in the form of bar charts and heatmaps, and an analysis report containing "potential index - decline target - constraint satisfaction status" is generated. Dynamic adjustments include establishing a timed update mechanism, periodically (e.g., every three years) repeating the above steps based on the latest economic, energy, and carbon emission data to update and decompose targets; if a city experiences significant changes such as "industrial structure change rate exceeding 20% ​​(secondary industry share increase or decrease ≥20%)" or "renewable energy installed capacity doubling", then an immediate update is triggered without waiting for the cycle.

[0064] In summary, this method collects economic, energy, and population data at the provincial and municipal levels to calculate key carbon emission impact indicators. It uses the LMDI (Leadership in Difference of Influences) decomposition algorithm to quantify the impact of each indicator on carbon emissions in cities and prefectures. Linear regression and a cluster-based carbon reduction comprehensive index model are employed to analyze the carbon emission impact indicators and carbon reduction potential, respectively. A mapping relationship between each carbon emission impact indicator and carbon emissions is established, and the carbon emission potential index for each city and industry is calculated. Carbon emission impact indicators are set as linear programming variables. Using the carbon emission potential index for cities and industries, an objective function is defined. Combining economic development planning, carbon peaking planning, carbon peaking prediction, and the mapping relationship between carbon emission impact indicators and carbon emissions, linear programming constraints are designed. Through a linear programming solution algorithm, a carbon emission intensity target decomposition model based on linear regression is constructed to achieve carbon emission intensity change rate planning under target constraints, supporting the government's carbon emission dual control target decomposition and carbon emission budget management.

[0065] Furthermore, to address the shortcomings of existing carbon intensity index decomposition technologies, such as "ignoring the heterogeneity of cities and prefectures, subjective weight setting, single constraint dimension, and lack of full-process modeling," this invention constructs a fully closed-loop technical solution encompassing "heterogeneity identification - objective potential quantification - driving factor modeling - multi-constraint optimization - dynamic adaptation." This provides a target-planning-based city and prefecture carbon intensity index decomposition system for implementing the aforementioned method. This system includes: Multi-source data integration module: As the basic support for the technical solution, it is responsible for collecting four core data of economy, energy, population and carbon emissions at the provincial and municipal levels. Through standardization processing, outlier removal and missing value completion, it forms a high-quality dataset, providing reliable data input for subsequent analysis.

[0066] Regional carbon emission heterogeneity identification module: It integrates PCA dimensionality reduction and K-means clustering technology to extract core features from multidimensional indicators and eliminate data redundancy. It determines the optimal number of clusters through silhouette coefficient and cluster stability test, and divides cities in the province into categories with similar carbon emission patterns and development levels. It breaks the limitation of "one-size-fits-all" allocation and lays the foundation for differentiated decomposition.

[0067] The objective quantification module for carbon emission reduction potential constructs a dual-dimensional evaluation system of "fairness and efficiency." The fairness dimension focuses on the right to development (including per capita GDP, historical carbon emission responsibility coefficient, etc.), while the efficiency dimension focuses on emission reduction capacity (including carbon emissions per unit of GDP, the proportion of renewable energy, etc.). The entropy weight method is used to objectively calculate the weights of indicators to avoid subjective assignment bias, and finally generate the carbon emission reduction potential index (CRI) to quantify the emission reduction responsibility and capacity of various cities.

[0068] Carbon emission driving factor modeling module: Based on LMDI additive decomposition technology, the change in carbon emissions is decomposed into the contribution of factors such as energy consumption structure, energy use efficiency, industrial structure, and economic development level without loss. Combined with linear regression modeling, the elasticity of the impact of each factor on carbon emissions is clarified, providing quantitative parameter support for setting constraints and calculating variable weights in the target programming model.

[0069] Multi-constraint target planning module: Using the carbon intensity change rate as the decision variable, it constructs a weighted minimization objective function of "prioritizing emission reduction in high-potential areas". It integrates multiple constraints such as total consistency (matching provincial targets), economic growth (ensuring development needs), resource carrying capacity (technical feasibility), and population change (adapting to people's livelihood). The optimal carbon intensity decomposition target for each city is obtained through linear programming. At the same time, a dynamic adjustment mechanism of "periodic update + triggering major changes" is established to adapt to regional development dynamics.

[0070] Supporting system architecture: To realize the practical application of the above-mentioned technical modules, a system is built that includes eight functional modules: data acquisition, preprocessing, cluster analysis, potential assessment, factor decomposition, programming solution, result display, and dynamic updating. It supports data import and export, custom parameter adjustment, and result visualization analysis, forming a complete closed loop of "data-model-result-application" and providing operable technical tools.

[0071] In this way, the deep collaboration among the modules forms a closed-loop process encompassing "data support - feature identification - potential quantification - factor analysis - optimized allocation - dynamic adaptation," including: Data to Feature Transformation: The standardized dataset output by the multi-source data integration module is input into the regional carbon emission heterogeneity identification module. The principal components with a cumulative contribution rate of ≥85% are extracted through PCA dimensionality reduction, transforming high-dimensional data into low-dimensional core features, providing efficient input for cluster analysis and avoiding interference from redundant information.

[0072] The correlation between features and potential: The city clustering results of the heterogeneity identification module serve as the grouping basis for the objective quantification module of carbon emission reduction potential. For each type of "relatively homogeneous" city, a suitable "fairness-efficiency" evaluation system is constructed to ensure that the potential quantification takes into account both regional commonalities and individual differences. The CRI index calculated by the entropy weight method is directly related to the emission reduction priority of each city.

[0073] The connection between potential and constraints: The factor contribution and regression coefficients output by the carbon emission driving factor modeling module are used to determine the weight of variables in the target planning model (e.g., if the energy use efficiency factor has a high contribution, the weight of the energy consumption variable per unit of GDP will increase). On the other hand, it provides a basis for setting resource constraint thresholds (e.g., the energy consumption constraint per unit of GDP should not be lower than 90% of the minimum value in the province, which is determined based on the technical feasibility analysis of factor regression).

[0074] Optimization constrained to the target: The CRI index (potential quantification result) and factor parameters (driving modeling result) are input into the multi-constraint target planning module. By minimizing the objective function, the goal is to enable cities with high CRI indexes to undertake higher emission reduction tasks. At the same time, multi-dimensional constraints ensure that the target does not deviate from reality. Finally, the carbon intensity decomposition target of each city is output.

[0075] Target-to-Dynamic Adaptation: The dynamic adjustment mechanism monitors major changes in the city's industrial structure (such as a change in the proportion of secondary industry exceeding 20%) and energy structure (such as a doubling of renewable energy installed capacity) in real time, triggering the multi-source data integration module to re-collect data, which in turn drives subsequent modules to update the analysis results, ensuring that the decomposed targets always adapt to the regional development dynamics.

[0076] The above technical solution achieves the following: Accurate Heterogeneity Identification: Breaking through the traditional extensive classification method of "stratification by administrative level or GDP", PCA-K-means clustering combined with silhouette coefficient and stability test is adopted to achieve scientific classification of carbon emission patterns in cities. The clustering results match the regional industrial and energy characteristics by 92%, providing an accurate basis for differentiated decomposition.

[0077] Quantifying and objectifying potential: Abandoning subjective weighting based on experience, a dual-dimensional evaluation system of "fairness and efficiency" is constructed. The entropy weight method is used to objectively assign weights based on the information entropy of the data itself. The CRI index is correlated with the actual emission reduction capacity of cities and prefectures by 0.87, avoiding unfair distribution caused by human bias.

[0078] Visualization of driving factors: By combining LMDI residual-free decomposition with linear regression, the abstract carbon emission driving mechanism is transformed into quantifiable factor contributions (such as economic development level contributing 45% and energy efficiency contributing -30%), providing transparent parameter support for constraint setting and target optimization.

[0079] The constraint system is multi-dimensional: it integrates constraints from multiple dimensions such as total amount, economy, resources and population, which not only ensures that the decomposed targets are consistent with the provincial total amount, but also guarantees the needs of the city's economic development and technical feasibility. The target achievement rate is expected to reach 98%, which is 23 percentage points higher than the traditional method.

[0080] Dynamic and flexible adaptation: It integrates a dual dynamic mechanism of "three-year cycle update + major change trigger", which not only covers the normal development rhythm, but also responds to sudden structural adjustments. At the same time, the system supports the customization of indicators and parameters, and can be extended to industry-level carbon intensity decomposition such as industry and construction, with a wide range of application scenarios.

[0081] In one specific implementation, the system described above can be designed to include: Data acquisition module: Connects to statistical yearbook databases, energy management platforms, and the "electricity-carbon calculation model" system. It supports two data import methods: Excel and API interface. It automatically collects and stores economic, energy, population, and carbon emission data in CSV format (for easy subsequent processing). Data preprocessing module: integrates outlier detection (3σ / IQR method), missing value completion (linear interpolation / mean filling), and standardization (range standardization) functions, supports one-click generation of preprocessing reports, and marks the location of outlier data and correction logic; Clustering analysis module: Built-in PCA dimensionality reduction algorithm and K-means clustering algorithm, supports user-defined cumulative contribution rate threshold of principal components (default 85%), automatically calculates silhouette coefficient and stability, and outputs clustering results and visualized scatter plot (drawn according to the coordinates of principal component 1 and principal component 2). Potential Assessment Module: It has a preset "fairness-efficiency" dual-dimensional indicator system, supports users to add indicators (the meaning of the indicator and the source of the data are required), and automatically calculates the weight and carbon emission reduction potential index through the entropy weight method to generate an "indicator weight-potential score" radar chart. Factor decomposition module: Implements LMDI additive decomposition (supports 4 types of core factors and user-defined factors), outputs an annual factor contribution table; integrates linear regression modeling function, automatically calculates regression coefficients, R 2 Mark the key driving factors (the first two absolute values ​​of the regression coefficients); Planning and Solving Module: Built-in PuLP solver, supports users to adjust constraint parameters (such as GDP growth rate, resource constraint threshold), outputs carbon intensity reduction targets for various cities in real time, and supports sensitivity analysis (such as the rate of change of the target after adjusting a certain constraint threshold). Results display module: Provides three display formats: bar chart (comparison of decline targets in prefecture-level cities), heat map (regional potential distribution), and table (detailed data). Supports data export to Excel / PDF format, and report templates can be customized. Dynamic update module: Supports users to set the update cycle (default 3 years), automatically triggers data collection and recalculation, and the trigger mechanism for major changes can be customized through the "threshold setting interface". Update results are automatically pushed to the administrator's email.

[0082] In addition, the present invention provides a machine-readable storage medium storing instructions that cause a machine to perform the method described above.

[0083] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0084] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0085] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0086] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0087] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0088] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0089] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0090] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0091] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for decomposing municipal carbon intensity indicators based on target planning, characterized in that, The method includes: Collect multi-source data and perform preprocessing; Regional carbon emission pattern clustering based on PCA-K-means was performed, including PCA dimensionality reduction, K-means clustering, and determination of the optimal number of clusters; Calculation of carbon emission reduction potential index based on entropy weight method; Regression modeling of carbon emission impact factors based on LMDI; Carbon intensity target decomposition modeling based on linear programming; Output the results and adjust them dynamically.

2. The method for decomposing municipal carbon intensity indicators based on target planning according to claim 1, characterized in that, The collected multi-source data includes provincial and municipal economic data, energy data, population data, and carbon emission data. The collected data is then standardized according to formula (1) to eliminate dimensional differences. in, For the first The first city The original value of the indicator, The first The minimum and maximum values ​​of the indicators, The value is the standardized value, and its range is [0,1].

3. The method for decomposing municipal carbon intensity indicators based on target planning according to claim 2, characterized in that, The PCA dimensionality reduction involves constructing a covariance matrix from the preprocessed index dataset based on standardized data, performing eigenvalue decomposition on the covariance matrix, sorting the eigenvalues ​​from largest to smallest, and extracting the top eigenvalues ​​with a cumulative contribution rate ≥ 85%. Principal components are used to construct a new feature space. ,in, Number of prefecture-level cities; The K-means clustering method involves clustering city samples in the principal component space according to formula (2), with the objective function being "minimizing the squared Euclidean distance from the sample point to the cluster center". in, For the first Clusters, Let be the mean of the principal component vectors of the samples within the cluster. For the first Principal component vectors of each sample; The determination of the optimal number of clusters includes determining the optimal number of clusters through silhouette coefficient and cluster stability tests. The requirements are a profile coefficient ≥ 0.6 and stability ≥ 90%, ultimately dividing the prefecture-level cities into... Regions with similar carbon emission patterns and development levels.

4. The method for decomposing municipal carbon intensity indicators based on target planning according to claim 3, characterized in that, The calculation of the carbon emission reduction potential index based on the entropy weight method includes constructing a "fairness-efficiency" dual-dimensional evaluation system for each clustered region, objectively assigning weights using the entropy weight method, and calculating the comprehensive carbon emission reduction potential index; wherein... Fairness dimension This includes GDP per capita, electricity consumption per capita, historical carbon emission responsibility coefficient, and resident population size; Efficiency dimension This includes carbon emissions per unit of GDP, carbon emissions per unit of energy consumption, the proportion of carbon emissions in the secondary industry, and the proportion of carbon emissions from renewable energy. Entropy weight calculation includes: Calculate the first according to formula (3) The first city The percentage of each indicator in, The value is the standardized value; Calculate the first according to formula (4) Information entropy of the indicator Among them, if ,definition Information entropy The value range is [0,1]. The smaller the value, the higher the indicator's discriminative power; Calculate the index weights according to formula (5). in, For the total number of indicators, ; Calculate the comprehensive potential index according to formula (6). in, For the first Carbon emission reduction potential index of each city , To ensure fairness, scores are weighted. Efficiency-weighted score.

5. The method for decomposing municipal carbon intensity indicators based on target planning according to claim 4, characterized in that, The regression modeling of carbon emission impact factors based on LMDI includes LMDI additive decomposition and linear regression modeling, wherein... The LMDI additive decomposition selects the energy consumption structure. Energy efficiency Industrial structure Economic development level As the core driving factor, the change in carbon emissions is calculated according to formula (7). Lossless decomposition into the contribution of each factor: in, for The carbon emissions at any given time, and the contribution of each factor is calculated according to formula (8). in, For a certain driving factor, It is a logarithmic mean function; The linear regression model uses changes in carbon emissions. Dependent variable The contribution of each driving factor is represented by a vector of independent variables. A linear regression model is established based on formula (9). in, For regression coefficients, For error terms, By minimizing the sum of squared residuals Estimate the regression coefficients and use the coefficient of determination. Validate model fit and .

6. The method for decomposing municipal carbon intensity indicators based on target planning according to claim 5, characterized in that, The carbon intensity target decomposition modeling based on linear programming includes: Let the first The rate of change in carbon intensity in each city is a decision variable. , The ratio of carbon intensity in the target year to carbon intensity in the baseline year. This indicates a decrease in carbon intensity. The smaller the value, the greater the decrease. Construct an objective function based on the principle of "prioritizing emission reduction in high-potential areas". According to formula (10), construct a weighted minimization objective function to achieve a balance between "fairness and efficiency". in, The rate of change of the resident population. The rate of change in GDP per unit population. The rate of change in energy consumption per unit of GDP The rate of change in carbon emissions per unit of energy consumption; For the first Carbon emission reduction potential index of each city The corresponding variable weights are obtained by Min-Max normalization of the regression coefficients according to formula (11). Define constraints, where, Based on formula (12), a total consistency constraint is applied to ensure that the sum of carbon intensity targets of all cities and prefectures conforms to the provincial target. in, For the first Baseline annual carbon emissions for each city This represents the total carbon emissions for the provincial baseline year. The rate of change of provincial carbon intensity targets; Economic growth constraints are applied according to formula (13) to ensure that the GDP growth rate per unit population is not lower than the provincial projected annual GDP growth rate. To ensure that economic development is not excessively constrained. Resource constraints are applied according to formula (14) to ensure that energy consumption per unit of GDP and carbon emissions per unit of energy consumption are not less than 90% of the minimum values ​​in the province, thus avoiding targets that exceed technical feasibility. Population constraints are applied according to formula (15), and the range of resident population change rate is set based on the population change trend of the city. Non-negativity constraints are applied according to formula (16) to ensure that the rate of change of decision variables and influencing factors conforms to practical significance. The solution is obtained using the Python open-source library PuLP, with the iteration termination condition set to "residual less than 1e-6". The objective function calculation precision is retained to four decimal places. If the number of cities exceeds 50, the interior point method is used to improve computational efficiency, with the initial iteration step size set to 0.

1. The final output is the optimal carbon intensity change rate for each city. .

7. The method for decomposing municipal carbon intensity indicators based on target planning according to claim 6, characterized in that, The output includes calculations of carbon intensity reduction targets for various cities and prefectures. The data is displayed in the form of bar charts and heatmaps, and an analysis report containing "potential index - decline target - constraint satisfaction status" is generated. The dynamic adjustment includes establishing a timed update mechanism, periodically repeating the above steps based on the latest economic, energy, and carbon emission data to update the decomposed targets; if significant changes occur, an immediate update is triggered without waiting for the cycle.

8. The method for decomposing municipal carbon intensity indicators based on target planning according to claim 7, characterized in that, The significant changes mentioned include an industrial structure change rate exceeding 20% ​​or a doubling of renewable energy installed capacity.

9. A target-planning-based decomposition system for municipal carbon intensity indicators, used to implement the method described in any one of claims 1-8, characterized in that, The system includes: The data acquisition and preprocessing module is used to collect and preprocess multi-source data to form a standardized dataset. The regional heterogeneity identification module is used to perform principal component analysis dimensionality reduction and K-means clustering on the standardized dataset to identify the carbon emission pattern categories of cities and prefectures. The carbon emission reduction potential quantification module is used to construct an evaluation system with fairness and efficiency dimensions, and to calculate the carbon emission reduction potential index of each city using the entropy weight method. The driving factor modeling module is used to perform log-mean Dijkstra's exponential additive decomposition and linear regression to quantify the impact of carbon emission driving factors. The multi-constraint objective programming module is used to construct and solve a linear programming model with the carbon intensity change rate as the decision variable, and output the carbon intensity decomposition targets for each city. The results output and dynamic management module is used to visualize the decomposition results and manage the dynamic updates and recalculations of the model.

10. A machine-readable storage medium having instructions stored thereon for causing a machine to perform the method as described in any one of claims 1-8.

Images