Low-carbon evaluation method and system for cyclic utilization of renewable resources of power grid based on deep learning

Abstract

The invention discloses a power grid renewable resource cyclic utilization low-carbon evaluation method and system based on deep learning, and relates to the technical field of low-carbon circulation of deep learning, and the method comprises the steps: firstly obtaining recycled resource record data, and building a first mapping relation and a second mapping relation according to the recycled resource record data; based on the first mapping relation and the second mapping relation, calculating a first benefit result and a second benefit result through a multi-task system optimization path method; obtaining a carbon emission reduction predicted value and an economic benefit predicted value for the first benefit result and the second benefit result by using a variational mode decomposition and multi-source feature fusion mechanism; a game equilibrium weight vector is solved through an iterative optimization algorithm of a negotiation game model, and a continuous and stable low-carbon evaluation index is calculated according to a carbon emission reduction predicted value and an economic benefit predicted value, so that a fine management strategy and green low-carbon transformation of a power grid enterprise under a low-carbon energy-saving target are effectively solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of low-carbon recycling of deep learning, in particular to a low-carbon evaluation method and system for recycling of renewable resources in power grids based on deep learning. BACKGROUND

[0002] With the continuous expansion of the power grid system, the amount of renewable resources generated in the process of power grid equipment updating and decommissioning increases sharply. How to scientifically and effectively evaluate the low-carbon benefits in the recycling process of these resources has become a key technical problem for promoting the green and low-carbon transformation of power grids.

[0003] The prior art usually processes recyclable resource data, waste resource data, target production data and energy production data as independent information islands, lacking effective correlation mechanisms. Due to the diverse types and characteristics of renewable resources in power grids, there are significant differences in their recycling paths, processing methods and energy conversion approaches. Traditional methods can only accurately quantify key parameters such as conversion rate, processing efficiency and quality loss rate through simple data matching, resulting in a lack of reliable data basis for subsequent benefit evaluation. The low-carbon benefits and economic benefits of power grid resource recycling are dynamically affected by multiple factors such as market fluctuations, policy adjustments and technological progress, showing complex non-stationary time series characteristics. Traditional statistical analysis methods are difficult to effectively decompose and capture the trend, periodicity and randomness components, and conventional machine learning models are not sufficient in the fusion of static and dynamic characteristics, resulting in large deviations in prediction results and failing to provide accurate forward-looking guidance for decision-making. In addition, existing evaluation methods mostly use fixed weight allocation schemes, which fail to effectively balance the internal relationship and contradiction between carbon emission reduction and economic benefits. In particular, in the process of combining subjective and objective weights, traditional combination weighting methods often ignore the game relationship between the two, either over-relying on expert experience and lacking data support, or completely relying on objective data and ignoring actual business needs, resulting in evaluation results that cannot truly reflect the low-carbon benefits of resource recycling and are difficult to meet the needs of actual management and decision-making. SUMMARY

[0004] The technical problem solved by the present application is that it is difficult to establish a mapping correlation between renewable data and energy production paths, current technologies lack sufficient analysis of the time series regularity of carbon emission reduction and economic benefits, and there is no accurate quantification of low-carbon evaluation results for recycling of renewable resources in power grids.

[0005] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, a low-carbon evaluation method for recycling of renewable resources in power grids based on deep learning includes the following steps: Step S1: Obtain recycled resource record data, which includes recyclable resource data, waste resource data, target production data, and energy output data. Establish a first mapping relationship and a second mapping relationship based on the recycled resource record data. Step S2: Based on the first mapping relationship and the second mapping relationship, calculate the first benefit result and the second benefit result using the multi-task system optimization path method; Step S3: Using variational mode decomposition and multi-source feature fusion mechanism, carbon emission reduction prediction value and economic benefit prediction value are obtained from the first benefit result and the second benefit result. Step S4: Solve the game equilibrium weight vector through the iterative optimization algorithm of the negotiation game model, and calculate the low-carbon evaluation index based on the carbon emission reduction prediction value and the economic benefit prediction value.

[0006] Preferably, step S1 specifically includes: The system collects and recycles resources from various locations, and obtains recycling record data based on statistics from the unit management systems of each location. The recycling record data includes data on recyclable resources, waste resources, target production data, and energy output data. The recyclable resource data includes material ID, material type, recycling quantity, source, processing date, unit raw material cost, unit recycled material cost, and current recycling status, which is used to record information on recyclable materials with reuse value; The waste resource data includes waste ID, waste type, waste quantity, processing date, processing method, processing cost, and potential carbon emissions, and is used to record waste materials that cannot be directly utilized or require special treatment. The target production data includes target ID, product type, required materials, planned quantity, potential carbon emissions, and production date, and is used to record target information and production plans for the reuse of recyclable resource data. The energy output data includes energy ID, energy type, energy output, carbon emission factor, energy label, and alternative energy type, and is used to record the power generation or energy utilization pathways for the reuse of waste resources.

[0007] Preferably, a first mapping relationship and a second mapping relationship are established based on the recycled resource record data: Establish the first mapping relationship: The data on recyclable resources is associated with the target production data. Based on the matching of material type and product type, the conversion rate, processing efficiency and recyclable quality loss rate are calculated using historical data. The first mapping relationship includes material ID, target ID, conversion rate, processing efficiency and recyclable quality loss rate. The conversion rate is the proportion of recyclable materials converted into the target product. The processing efficiency refers to the efficiency of the processing procedure; The quality loss rate is the proportion of losses due to quality problems. Establish a second mapping relationship: By associating waste resource data with energy production data, and based on the matching of waste type and energy type, the corresponding energy production efficiency, energy conversion rate, treatment efficiency and quality loss rate are calculated through a multiple regression model. The second mapping relationship includes: waste ID, energy ID, energy production efficiency, energy conversion rate, treatment efficiency and waste resource quality loss rate. The energy production efficiency is the energy generated per unit of waste material; The energy conversion rate is the proportion of waste materials converted into energy. The processing efficiency refers to the efficiency of the processing procedure.

[0008] Preferably, step S2 specifically includes: A multi-task optimization path method is constructed to simultaneously optimize carbon emission reduction and economic benefits. The processing logic of the multi-task optimization path method is as follows: The first benefit is obtained based on the carbon emission reduction benefit of resource conservation, and the second benefit is obtained based on the economic benefits of resource conservation. The processing logic for obtaining the primary benefit result based on the carbon emission reduction benefits of resource conservation is as follows: Based on the first mapping relationship, the material ID of the recyclable resource data is mapped to the target ID of the target production data to obtain the first mapping pair I( ), where i is the i-th first mapping pair. For the first mapping pair and the first mapping relationship, the effective recovery quantity is calculated. The expression for calculating the effective recovery quantity is: ; Using the carbon emission factor method, the reduction in carbon emissions from recycling is calculated based on the carbon emission factors corresponding to the effective recycling quantity and material type. The expression for calculating the reduction in carbon emissions from recycling is as follows: ; Based on the second mapping relationship, the waste ID of the waste resource data and the energy ID of the energy production data are mapped to obtain the second mapping pair J( ), where j is the j-th second mapping pair. For the second mapping pair and the second mapping relationship, the amount of waste utilization is calculated. The expression for calculating the amount of waste utilization is:

[0009] The reduction in carbon emissions from waste is calculated based on the reduction in carbon emissions from replacing primary energy sources with waste materials. The formula for calculating the reduction in carbon emissions from waste is as follows: ; The carbon emission factor is obtained from energy output data and is used to represent the reduction in carbon emissions per unit of energy output; The first benefit result is obtained by adding the reduction in carbon emissions from recycling and the reduction in carbon emissions from waste.

[0010] The preferred approach to obtaining the second benefit result based on the economic benefits of resource conservation is as follows: The unit raw material cost and recycling material cost of recyclable resources are obtained. Based on the first mapping relationship, the recycling economic benefit is calculated for each first mapping pair. The calculation expression for the recycling economic benefit is as follows: ; The unit cost of raw materials mentioned is the market cost of purchasing new materials; The unit cost of recycled materials is the total cost of the material's unit recycling cost, dismantling labor cost, and remanufacturing cost. For all first mapping pairs, the total regression economic benefit is calculated, and the expression for the total regression economic benefit is as follows: ; The waste disposal cost and unit energy sales revenue of the acquired energy waste resources are used to calculate the energy economic benefits based on the second mapping relationship and the amount of waste utilization for each second mapping pair. The calculation expression for the energy economic benefits is as follows:

[0011] The unit energy sales revenue mentioned is the market sales price of energy output; The waste disposal cost is the total cost of waste collection, transportation, and manual processing. For all second mapping pairs, calculate the total energy economic benefit, expressed as follows: ; The second benefit result is obtained by adding the total economic benefits of recycling and the total economic benefits of energy.

[0012] Preferably, step S3 specifically includes: Step S31: Perform variational mode decomposition on the first and second benefit results, setting the number of modes K and the penalty coefficient. K intrinsic mode components are extracted to represent the periodic, trend, and random fluctuation characteristics of carbon emission reduction and economic benefits; Step S32: Construct a dual-channel feature vector system. Extract material type, unit raw material cost, unit recycled material cost, and energy label for feature encoding to generate a first feature vector. Extract K intrinsic mode components, recycling quantity, waste utilization quantity, and energy output efficiency according to the processing date. Then, through normalization, concatenate to generate a second feature vector. Use a gated attention fusion mechanism to integrate the dual-channel features. Pass the first feature vector through a linear transformation layer to generate a gated weight vector r. Pass the second feature vector through a linear transformation layer to generate a candidate feature vector c. Finally, obtain the fused feature vector through gated weighted calculation. The gated weighted calculation expression is: ; in, To fuse feature vectors, and For learnable projection matrices, The first eigenvector, Let g be the second feature vector, and g be the gating weight coefficient. Step S33: Construct time series samples by sliding the fused feature vectors according to the time window to obtain the time series sample matrix. Traverse each time window and take the last time point of the current time window as the corresponding labeled time of the time series sample matrix of this stage. Step S34: Use the variational mode decomposition multi-objective prediction network to predict the time series sample matrix and generate prediction result values ​​and auxiliary information values.

[0013] Preferably, the prediction results and auxiliary information include: The forecast results include predicted carbon emission reductions and predicted economic benefits; Supporting information includes prediction confidence, uncertainty indicators, and consistency metrics.

[0014] Preferably, step S4 specifically includes: The fusion feature vector, prediction results, first benefit results, second benefit results and auxiliary information are standardized and then concatenated to generate a standardized decision matrix. A dual-criteria judgment matrix of carbon emission reduction and economic benefits is constructed using the analytic hierarchy process (AHP). A threshold for consistency indicators is set, and when the consistency indicator falls below the threshold, a subjective weight vector is derived. ; Using the improved entropy weighting method, information entropy is calculated for four indicators in the standardized decision matrix: the fusion feature vector, the prediction result, the first benefit result, and the second benefit result. An objective weight vector is obtained by applying weight reduction penalties using uncertainty indicators and consistency indicators. ; A negotiation game model is constructed, with subjective and objective weight vectors representing the two negotiating parties. The final weight vector W is then solved. The processing logic for constructing the negotiation game model is as follows: Set a negotiation breakdown point. When neither party can obtain their ideal weights, reset the breakdown point payoff of both the subjective and objective weight vectors to zero, and set the utility function accordingly. ; ; in, For subjective utility, For objective utility, Let the squared Euclidean distance between the weight vector and the subjective weight vector be the standard distance. Based on the subjective and objective utility, the negotiation game constraint is defined as maximizing the Nash product. Construct the objective function to maximize the Nash product: ; Take the arithmetic mean of the subjective weight vector and the objective weight vector as the initial weights. The objective function of maximizing the Nash product is solved iteratively, corresponding to each iteration. Perform the following sub-steps: Calculate the current weight vector Nash product objective function value at point And calculate the gradient vector. The expression for calculating the objective function value of the Nash product is: ; Construct a quadratic programming subproblem, wherein the quadratic programming subproblem is in The original objective function is approximated using a quadratic function, and the search direction is determined by the gradient vector. Along the search direction Perform a one-dimensional search and set the step size. Make the objective function value of the Nash product Satisfy the sufficient increase condition, while ensuring the updated weight vector Given that the values ​​are non-negative and conform to normalization constraints, calculate the change norm of the weight vector. If the current change norm is less than the preset threshold, the iteration is terminated and the final weight vector W is output as the game equilibrium weight vector. The compromise solution for each resource recycling batch *s* is calculated based on the game equilibrium weight vector and the standardized decision matrix. The compromise solution is then used to generate a low-carbon evaluation index through reverse mapping. The confidence level of the compromise solution is corrected using a consistency index to obtain the final low-carbon evaluation index. The calculation expression for the confidence correction is as follows: ; in, As a low-carbon evaluation indicator. , as a consistency indicator This is a sign of uncertainty; Low-carbon evaluation indicators are classified into levels according to set thresholds to obtain low-carbon evaluation levels.

[0015] Preferably, the processing logic for classifying low-carbon evaluation indicators into levels according to evaluation rules to obtain low-carbon evaluation levels is as follows: The threshold is set as a first threshold and a second threshold. When the low-carbon evaluation index is less than the first threshold, the current low-carbon evaluation index is determined to be at the red level. When the first threshold is less than the low-carbon evaluation index and the second threshold is less than the second threshold, the current low-carbon evaluation index is determined to be at the yellow level. When the second threshold is less than the low-carbon evaluation index, the current low-carbon evaluation index is determined to be at the green level. The red level indicates a high-carbon, low-efficiency pathway. The yellow level indicates paths where resource allocation needs to be optimized. The yellow level indicates a dual-optimal path for a low-carbon economy.

[0016] The second aspect is a low-carbon evaluation system for the recycling of renewable resources in the power grid based on deep learning, which includes a relationship mapping module, a benefit calculation module, a target prediction module, and a low-carbon evaluation module. The relationship mapping module acquires recycled resource record data, which includes recyclable resource data, waste resource data, target production data, and energy output data. Based on the recycled resource record data, a first mapping relationship and a second mapping relationship are established. The benefit calculation module is used to calculate the first benefit result and the second benefit result by using the multi-task system optimization path method to calculate the first and second benefit results of the first and second mapping relationships of the relationship mapping module. The target prediction module is used to obtain carbon emission reduction prediction values ​​and economic benefit prediction values ​​from the first benefit result and the second benefit result by using variational mode decomposition and multi-source feature fusion mechanism. The low-carbon evaluation module is used to solve the game equilibrium weight vector through the iterative optimization algorithm of the negotiation game model, and to calculate the low-carbon evaluation index based on the carbon emission reduction prediction value and the economic benefit prediction value.

[0017] The beneficial effects of this invention are as follows: By establishing a mapping relationship between recycled resource record data, which includes recyclable resource data, waste resource data, target production data, and energy output data, the comprehensiveness of the data is ensured, covering the source and destination of resource cycles, obtaining complete path tracking of resource cycles, achieving full life cycle data coverage, and solving the problem of information silos. Multi-task optimization is employed to calculate carbon emission benefit values ​​and economic benefit values, providing a data foundation for subsequent synergistic consideration of environmental and economic benefits. Variational modal hierarchical and multi-source feature fusion technologies are used to effectively capture the complex temporal characteristics of carbon emission reduction and economic benefits. A gating attention mechanism is used to divide the acquired data into static and dynamic features. Intelligent weighting is applied based on the mathematical characteristics of both factors, ultimately outputting more accurate carbon emission reduction and economic benefit predictions. Furthermore, this invention employs a negotiation game model and a weight optimization method based on this model, overcoming the arbitrariness of subjective weighting and the mechanical nature of objective weighting. By constructing an objective function and iteratively optimizing the game equilibrium weight vector through multiple iterations, the equilibrium weight vector is rationalized without the need for manual weight setting. Simultaneously, considering the numerical characteristics of both subjective and objective features, iterative constraint calculations make the weight allocation values ​​more accurate. Subsequently, a dual-path benefit model is established, comprehensively covering both recycling and energy optimization resource cycles, objectively yielding low-carbon evaluation results and providing practical decision-making for the recycling of renewable resources in the power grid. Attached Figure Description

[0018] Figure 1 A flowchart illustrating the steps of a deep learning-based low-carbon evaluation method for the recycling of renewable resources in power grids, provided in one embodiment of the present invention. Figure 2 This is a basic flowchart of a deep learning-based low-carbon evaluation system for the recycling of renewable resources in power grids, provided as an embodiment of the present invention. Figure 3 This is a schematic diagram of the processing flow of a negotiation game model. Detailed Implementation

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0020] Example 1, referring to Figure 1 This paper provides a low-carbon evaluation method for the recycling of renewable resources in power grids based on deep learning, which includes the following steps: Step S1: Obtain the recycling resource record data, which includes recyclable resource data, waste resource data, target production data, and energy output data. Establish a first mapping relationship and a second mapping relationship based on the recycling resource record data. Step S2: Based on the first mapping relationship and the second mapping relationship, calculate the first benefit result and the second benefit result using the multi-task system optimization path method; Step S3: Using variational mode decomposition and multi-source feature fusion mechanism, carbon emission reduction prediction value and economic benefit prediction value are obtained from the first benefit result and the second benefit result. Step S4: Solve the game equilibrium weight vector through the iterative optimization algorithm of the negotiation game model, and calculate the low-carbon evaluation index based on the carbon emission reduction prediction value and the economic benefit prediction value.

[0021] By establishing a mapping relationship between recycled resource record data, which includes recyclable resource data, waste resource data, target production data, and energy output data, the comprehensiveness of the data is ensured. This covers the source and destination of the resource cycle, enabling complete path tracking of the resource cycle and achieving full life-cycle data coverage, thus solving the problem of information silos. Multi-task optimization is employed to calculate carbon emission benefit values ​​and economic benefit values, providing a data foundation for subsequent synergistic consideration of environmental and economic benefits. Variational modal classification and multi-source feature fusion techniques are used to effectively capture the complex temporal characteristics of carbon emission reduction and economic benefits. A gating attention mechanism is used to divide the acquired data into static and dynamic features, and based on these features… The invention employs intelligent weighting based on mathematical features to output more accurate carbon emission reduction and economic benefit predictions. Furthermore, it utilizes a negotiation game model and a weight optimization method based on this model. This overcomes the arbitrariness of subjective weighting and the mechanical nature of objective weighting. By constructing an objective function and iteratively optimizing the game equilibrium weight vector, the invention rationalizes the game equilibrium weight vector without requiring manual weight setting. It also considers the numerical characteristics of both subjective and objective features, and iterative constraint calculations make the weight allocation more accurate. Finally, a dual-path benefit model is established to comprehensively cover both recycling and energy optimization resource cycles, objectively yielding low-carbon evaluation results and providing practical decision-making support for the recycling of renewable resources in the power grid.

[0022] Step S1 specifically includes: We collect recycled materials from various locations and obtain recycled resource record data based on the unit management systems of each location. The recycled resource record data includes data on recyclable resources, waste resources, target production data, and energy output data. Recyclable resource data includes material ID, material type, recycling quantity, origin, processing date, unit raw material cost, unit recycled material cost, and current recycling status, used to record information on recycled materials with reuse value; Waste resource data includes waste ID, waste type, waste quantity, treatment date, treatment method, treatment cost, and potential carbon emissions, used to record waste materials that cannot be directly utilized or require special treatment; Target production data includes target ID, product type, required materials, planned quantity, potential carbon emissions, and production date, and is used to record target information and production plans for the reuse of recyclable resource data; Energy output data includes energy ID, energy type, energy output, carbon emission factor, energy label, and alternative energy type, used to record the power generation or energy utilization pathways of waste resource reuse.

[0023] The system compiles and statistically analyzes four main categories of data from the management systems of various units located in different regions: recyclable resource data, waste resource data, target production data, and energy output data. The compilation of these four categories of data ensures the comprehensiveness of the data and covers the source and destination of resource recycling. In this embodiment, a power grid company recycled a batch of scrapped cables. The recyclable resource data at this time is as follows: Material ID: CU-001, Material Type: Copper, Recycling Quantity: 10 tons, Unit Raw Material Cost: 60,000 yuan / ton, Unit Recycling Material Cost: 20,000 yuan / ton... The target production data is as follows: Target ID: P-101, Product Type: New Type Distribution Cabinet, Required Material: Copper, Planned Quantity: 8 tons... The first mapping relationship can be established by automatically associating CU-001 with P-101 through the material type copper. Based on historical data, the conversion rate is calculated to be 85%, the processing efficiency is 95%, and the recyclable quality loss rate is 5%, indicating that 7.67 tons of copper material that can be used in the new distribution cabinet can be produced from 10 tons of recycled copper. Waste resource data: Waste ID: W-202, Waste type: Composite insulator, Waste quantity: 5 tons, Processing cost: 1000 yuan / ton... Energy output data: Energy ID: E-301, Energy type: Cement kiln co-processing, Energy output: Coal replacement, Carbon emission factor: 0.8 tCO2 / ton of waste... The second mapping relationship is established by matching waste type with energy conversion type, associating W-202 with E-301, and calculating an energy conversion rate of 98% and a processing efficiency of 90%, indicating that 5 tons of waste insulators can achieve a carbon emission reduction of 3.53tCO2 through this approach. In this embodiment, a set of first and second mapping pairs is defined, and the calculated conversion rate and efficiency parameters are the core coefficients for subsequent calculations of the effective recycling quantity and the waste utilization quantity.

[0024] Establish a first mapping relationship and a second mapping relationship based on the recycled resource record data: Establish the first mapping relationship: The data on recyclable resources is linked with the target production data. Based on the matching of material type and product type, the conversion rate, processing efficiency and recyclable quality loss rate are calculated using historical data. The first mapping relationship includes material ID, target ID, conversion rate, processing efficiency and recyclable quality loss rate. The conversion rate is the proportion of recyclable materials that are converted into the target product. Processing efficiency refers to the efficiency of the processing procedure. The quality loss rate is the proportion of losses due to quality problems. Establish a second mapping relationship: By linking waste resource data with energy output data, and based on the matching of waste type and energy type, the corresponding energy output efficiency, energy conversion rate, treatment efficiency and quality loss rate are calculated through a multiple regression model. The second mapping relationship includes: waste ID, energy ID, energy output efficiency, energy conversion rate, treatment efficiency and waste resource quality loss rate. Energy output efficiency is the energy generated per unit of waste material; Energy conversion rate is the proportion of waste materials converted into energy; Processing efficiency refers to the efficiency of the processing procedure.

[0025] In this embodiment, by establishing two mapping paths, the originally static data points are connected into a dynamic and traceable resource flow map, transforming the resource allocation decision-making based on experience into a quantitative path planning based on data. The calculated internal values ​​of the first and second mapping relationships can accurately reflect the actual parameters of process efficiency and truly reflect the actual production energy consumption, making the subsequent benefit calculation results more in line with reality and significantly improving the credibility of low-carbon evaluation indicators.

[0026] Step S2 specifically includes: A multi-task optimization path method is constructed to simultaneously optimize carbon emission reduction and economic benefits. The processing logic of the multi-task optimization path method is as follows: The first benefit is obtained based on the carbon emission reduction benefit of resource conservation, and the second benefit is obtained based on the economic benefits of resource conservation. The processing logic for obtaining the primary benefit result based on the carbon emission reduction benefits of resource conservation is as follows: Based on the first mapping relationship, the material ID of the recyclable resource data is mapped to the target ID of the target production data to obtain the first mapping pair I( ), where i is the i-th first mapping pair. For the first mapping pair and the first mapping relationship, the effective recovery quantity is calculated. The expression for calculating the effective recovery quantity is: ; Using the carbon emission factor method, the reduction in carbon emissions from recycling is calculated based on the carbon emission factor corresponding to the effective recycling quantity and material type. The expression for calculating the reduction in carbon emissions from recycling is as follows: ; Based on the second mapping relationship, the waste ID of the waste resource data and the energy ID of the energy production data are mapped to obtain the second mapping pair J( ), where j is the j-th second mapping pair. For the second mapping pair and the second mapping relationship, the amount of waste utilization is calculated. The expression for calculating the amount of waste utilization is:

[0027] The reduction in carbon emissions from waste is calculated based on the carbon emission reduction from replacing primary energy sources with waste. The formula for calculating the reduction in carbon emissions from waste is as follows: ; The carbon emission factor is obtained from energy output data and is used to represent the reduction in carbon emissions per unit of energy output; The first benefit result is obtained by adding the reduction in carbon emissions from recycling and the reduction in carbon emissions from waste.

[0028] By introducing multi-stage efficiency coefficients, calculating effective quantities, and using the most suitable carbon emission factors for both recycling and waste paths, the final carbon emission reduction is no longer a rough theoretical value, but a precise estimate close to engineering reality. The method of calculating by mapping the bipartite paths allows the system to understand the source and composition of benefits. Managers can not only obtain the total carbon emission reduction, but also which recycling path or energy conversion method contributes the most, providing clear direction for subsequent optimization. For example, the path with the highest carbon emission reduction per unit cost can be prioritized.

[0029] The processing logic for obtaining the second benefit result based on the economic benefits of resource conservation is as follows: Obtain the unit raw material cost and recycling material cost of recyclable resources. Based on the first mapping relationship, calculate the recycling economic benefit for each first mapping pair. The expression for calculating the recycling economic benefit is as follows: ; The unit cost of raw materials is the market cost of purchasing new materials; The unit cost of recycled materials is the total cost of the material's unit recycling cost, dismantling labor cost, and remanufacturing cost. For all first mapping pairs, calculate the total regression economic benefit. The expression for calculating the total regression economic benefit is as follows: ; The waste disposal cost and unit energy sales revenue of the acquired energy waste resources are obtained. Based on the second mapping relationship, the energy economic benefits are calculated for each second mapping pair according to the amount of waste utilized. The calculation expression for energy economic benefits is as follows:

[0030] Unit energy sales revenue is the market sales price of energy products; Waste disposal cost is the total cost of waste collection, transportation, and labor processing. For all second mapping pairs, calculate the total energy economic benefit. The expression for calculating the total energy economic benefit is as follows: ; The second benefit result is obtained by adding the total economic benefits of recycling and the total economic benefits of energy.

[0031] By employing full-cost accounting and incremental analysis, this method avoids the potential overestimation of benefits that can occur with traditional methods. For example, it may only focus on the savings in raw material costs while ignoring the high labor costs associated with dismantling. This method is a prerequisite for scientific evaluation. By decomposing and calculating economic benefits along different paths, it allows managers to clearly identify which recycling or energy conversion paths are economically and environmentally beneficial, and which are economically detrimental. This provides direct data support for optimal resource allocation and a solid economic dimension input for the game theory model in step S4.

[0032] Step S3 specifically includes: Step S31: Perform variational mode decomposition on the first and second benefit results, setting the number of modes K and the penalty coefficient. K intrinsic mode components are extracted to represent the periodic, trend, and random fluctuation characteristics of carbon emission reduction and economic benefits; Step S32: Construct a dual-channel feature vector system. Extract material type, unit raw material cost, unit recycled material cost, and energy label for feature encoding to generate the first feature vector. Extract K intrinsic mode components, recycling quantity, waste utilization quantity, and energy output efficiency according to the processing date. Then, through normalization, concatenate to generate the second feature vector. Use a gated attention fusion mechanism to integrate the dual-channel features. Pass the first feature vector through a linear transformation layer to generate a gated weight vector r. Pass the second feature vector through a linear transformation layer to generate a candidate feature vector c. Finally, obtain the fused feature vector through gated weighted calculation. The gated weighted calculation expression is: ; in, To fuse feature vectors, and For learnable projection matrices, The first eigenvector, Let g be the second feature vector, and g be the gating weight coefficient. Step S33: Construct time series samples by sliding the fused feature vectors according to the time window to obtain the time series sample matrix. Traverse each time window and take the last time point of the current time window as the corresponding labeled time of the time series sample matrix of this stage. Step S34: Use the variational mode decomposition multi-objective prediction network to predict the time series sample matrix and generate prediction result values ​​and auxiliary information values.

[0033] set up Using a 30-day time window, variational mode decomposition is first employed to decompose the non-stationary and noisy benefit time series data into smooth subsequences with different center frequencies, achieving deep decoupling and feature enhancement of the time series signal. This is equivalent to equipping the prediction model with a microscope, enabling it to see both short-term fluctuations and grasp long-term trends, fundamentally improving prediction accuracy. Secondly, dual-channel features are used in gating fusion to distinguish and integrate the inherent attributes of resources and the dynamic changes in the market. The gating mechanism g determines under what circumstances which feature should be believed, and the value of g ranges from 0 to 1. When g approaches 1, for example, when g is between 0.85 and 0.95, static characteristics dominate, and invariant attributes such as material cost and type play a decisive role in decision-making. In the case of recycling rare metals, cost difference is the primary factor determining the path. When g approaches 0, for example, when g is between 0.05 and 0.15, dynamic characteristics dominate, and the temporal fluctuations of VMD decomposition play a decisive role. On the day of sudden changes in carbon trading policies, short-term carbon price fluctuations have a greater impact on LCS scores than the materials themselves. When g approaches 0.5, the dual-channel equilibrium is reached, and the weights of static and dynamic information are roughly equivalent to those of a regular batch, with both cost and carbon price near their historical averages. The gated fusion architecture in this embodiment, combined with material embedding similarity learning and VMD component normalization strategies, enables static and dynamic features to achieve information decoupling and on-demand fusion at the channel level, effectively distinguishing it from the creative contribution of simple splicing.

[0034] The prediction results and auxiliary information include: The forecast results include predicted carbon emission reductions and predicted economic benefits; Supporting information includes prediction confidence, uncertainty indicators, and consistency metrics.

[0035] The range of prediction confidence values ​​in this embodiment is: ; Uncertainty Standard The range of values ​​for the consistency index is: .

[0036] Step S4 specifically includes: The fusion feature vector, prediction results, first benefit results, second benefit results and auxiliary information are standardized and then concatenated to generate a standardized decision matrix. A dual-criteria judgment matrix of carbon emission reduction and economic benefits is constructed using the analytic hierarchy process (AHP). A threshold for consistency indicators is set, and when the consistency indicator falls below the threshold, a subjective weight vector is derived. ; Using the improved entropy weighting method, information entropy is calculated for four indicators in the standardized decision matrix: the fusion feature vector, the prediction result, the first benefit result, and the second benefit result. An objective weight vector is obtained by applying weight reduction penalties using uncertainty indicators and consistency indicators. ; A negotiation game model is constructed, with subjective and objective weight vectors representing the two negotiating parties. The final weight vector W is solved, and the processing logic of the negotiation game model is as follows: Set a negotiation breakdown point. When neither party can obtain their ideal weights, reset the breakdown point payoff of both the subjective and objective weight vectors to zero, and set the utility function accordingly. ; ; in, For subjective utility, For objective utility, Let the squared Euclidean distance between the weight vector and the subjective weight vector be the standard distance. Based on the subjective and objective utility, the negotiation game constraint is defined as maximizing the Nash product. Construct the objective function to maximize the Nash product: ; Take the arithmetic mean of the subjective weight vector and the objective weight vector as the initial weights. The objective function of maximizing the Nash product is solved iteratively, corresponding to each iteration. Perform the following sub-steps: Calculate the current weight vector Nash product objective function value at point And calculate the gradient vector. The expression for calculating the objective function value of the Nash product is: ; Construct a quadratic programming subproblem, and the quadratic programming subproblem is in The original objective function is approximated using a quadratic function, and the search direction is determined by the gradient vector. Along the search direction Perform a one-dimensional search and set the step size. Make the objective function value of the Nash product Satisfy the sufficient increase condition, while ensuring the updated weight vector Given that the values ​​are non-negative and conform to normalization constraints, calculate the change norm of the weight vector. If the current change norm is less than the preset threshold, the iteration is terminated and the final weight vector W is output as the game equilibrium weight vector. The compromise solution for each resource recycling batch *s* is calculated based on the game equilibrium weight vector and the standardized decision matrix. This compromise solution is then used to generate a low-carbon evaluation index through reverse mapping. Finally, the confidence level of the compromise solution is corrected using a consistency index to obtain the low-carbon evaluation index. The expression for the confidence correction is as follows: ; in, As a low-carbon evaluation indicator. As a consistency indicator, This is a sign of uncertainty; Low-carbon evaluation indicators are classified into levels according to set thresholds to obtain low-carbon evaluation levels.

[0037] The principle of negotiation game theory posits that the solution is the point that maximizes the product of the utilities of both parties. This point simultaneously satisfies axioms such as Pareto optimality, symmetry, and linear transformation invariance. It transforms the weight allocation problem into an optimizable mathematical problem. The obtained consistency index (CI) measures the logical consistency of expert judgments. A lower CI value indicates less reliable subjective judgments. For example, the equilibrium weight W might be [0.62, 0.38], which is closer to the subjective weight. This is because the negotiation model finds that completely favoring objective weights leads to a sharp decline in the subjective party's utility. However, it doesn't completely adopt subjective weights; it takes into account the demands of objective data. This effectively prevents misjudgments caused by accidental errors in a single prediction model or data quality issues, greatly improving the system's robustness and the reliability of its output in complex real-world environments.

[0038] The low-carbon evaluation indicators are classified into levels according to the evaluation rules, and the processing logic for obtaining the low-carbon evaluation level is as follows: The thresholds include a first threshold and a second threshold. When the low-carbon evaluation index is less than the first threshold, the current low-carbon evaluation index is determined to be at the red level. When the first threshold is less than the low-carbon evaluation index and the second threshold is less than the second threshold, the current low-carbon evaluation index is determined to be at the yellow level. When the second threshold is less than the low-carbon evaluation index, the current low-carbon evaluation index is determined to be at the green level. The red level indicates a high-carbon, low-efficiency pathway; Yellow levels indicate resource allocation paths that need optimization. The yellow level indicates a dual-optimal path for a low-carbon economy.

[0039] In this embodiment, the low-carbon evaluation index of the quantitative results is quantified into three levels: red, yellow, and green. This provides managers at different levels with extremely intuitive decision-making signals. Red paths should be reviewed or eliminated immediately, yellow paths need to be optimized, and green paths should be promoted. This enables the theoretical model to be transformed into a simple and clear management tool, achieving an effective transformation from technical solutions to management practices.

[0040] Example 2, refer to Figure 2 It provides a low-carbon evaluation system for the recycling of renewable resources in the power grid based on deep learning, including a relationship mapping module, a benefit calculation module, a target prediction module, and a low-carbon evaluation module; The relationship mapping module acquires recycled resource record data, which includes recyclable resource data, waste resource data, target production data, and energy output data. Based on the recycled resource record data, it establishes a first mapping relationship and a second mapping relationship. The benefit calculation module is used to calculate the first and second benefit results by using the multi-task system optimization path method, based on the first and second mapping relationships of the relationship mapping module. The target prediction module is used to obtain carbon emission reduction predictions and economic benefit predictions from the first benefit result and the second benefit result by using variational mode decomposition and multi-source feature fusion mechanism. The low-carbon evaluation module is used to solve the equilibrium weight vector of the game through the iterative optimization algorithm of the negotiation game model, and to calculate the low-carbon evaluation index based on the carbon emission reduction forecast and the economic benefit forecast.

[0041] In this embodiment, the process that originally relied on manual data collection, decentralized calculation, and experience-based judgment is integrated into a highly efficient automated decision support system. The relationship mapping module automatically constructs a resource path network, the benefit calculation module quantifies the dual benefits in real time, the target prediction module dynamically updates trends, and the low-carbon evaluation module outputs evaluation levels instantly. This enables enterprises to conduct rapid and batch evaluations of resource recycling projects, shortening the decision-making cycle from weeks to hours or even minutes, greatly improving management efficiency and providing continuous impetus for precise management and strategic planning of enterprises.

[0042] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. 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.

[0043] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A low-carbon evaluation method for the recycling of renewable resources in power grids based on deep learning, characterized in that, Includes the following steps: Step S1: Obtain recycled resource record data, which includes recyclable resource data, waste resource data, target production data, and energy output data. Establish a first mapping relationship and a second mapping relationship based on the recycled resource record data. Step S2: Based on the first mapping relationship and the second mapping relationship, calculate the first benefit result and the second benefit result using the multi-task system optimization path method; Step S3: Using variational mode decomposition and multi-source feature fusion mechanism, carbon emission reduction prediction value and economic benefit prediction value are obtained from the first benefit result and the second benefit result. Step S4: Solve the game equilibrium weight vector through the iterative optimization algorithm of the negotiation game model, and calculate the low-carbon evaluation index based on the carbon emission reduction prediction value and the economic benefit prediction value.

2. The low-carbon evaluation method for the recycling of renewable resources in power grids based on deep learning as described in claim 1, characterized in that, Step S1 specifically includes: The system collects and recycles resources from various locations, and obtains recycling record data based on statistics from the unit management systems of each location. The recycling record data includes data on recyclable resources, waste resources, target production data, and energy output data. The recyclable resource data includes material ID, material type, recycling quantity, source, processing date, unit raw material cost, unit recycled material cost, and current recycling status, which is used to record information on recyclable materials with reuse value; The waste resource data includes waste ID, waste type, waste quantity, processing date, processing method, processing cost, and potential carbon emissions, and is used to record waste materials that cannot be directly utilized or require special treatment. The target production data includes target ID, product type, required materials, planned quantity, potential carbon emissions, and production date, and is used to record target information and production plans for the reuse of recyclable resource data. The energy output data includes energy ID, energy type, energy output, carbon emission factor, energy label, and alternative energy type, and is used to record the power generation or energy utilization pathways for the reuse of waste resources.

3. The low-carbon evaluation method for the recycling of renewable resources in power grids based on deep learning as described in claim 2, characterized in that, Establish a first mapping relationship and a second mapping relationship based on the recycled resource record data: Establish the first mapping relationship: The data on recyclable resources is associated with the target production data. Based on the matching of material type and product type, the conversion rate, processing efficiency and recyclable quality loss rate are calculated using historical data. The first mapping relationship includes material ID, target ID, conversion rate, processing efficiency and recyclable quality loss rate. The conversion rate is the proportion of recyclable materials converted into the target product. The processing efficiency refers to the efficiency of the processing procedure; The quality loss rate is the proportion of losses due to quality problems. Establish a second mapping relationship: By associating waste resource data with energy production data, and based on the matching of waste type and energy type, the corresponding energy production efficiency, energy conversion rate, treatment efficiency and quality loss rate are calculated through a multiple regression model. The second mapping relationship includes: waste ID, energy ID, energy production efficiency, energy conversion rate, treatment efficiency and waste resource quality loss rate. The energy production efficiency is the energy generated per unit of waste material; The energy conversion rate is the proportion of waste materials converted into energy. The processing efficiency refers to the efficiency of the processing procedure.

4. The low-carbon evaluation method for the recycling of renewable resources in power grids based on deep learning as described in claim 3, characterized in that, Step S2 specifically includes: A multi-task optimization path method is constructed to simultaneously optimize carbon emission reduction and economic benefits. The processing logic of the multi-task optimization path method is as follows: The first benefit is obtained based on the carbon emission reduction benefit of resource conservation, and the second benefit is obtained based on the economic benefits of resource conservation. The processing logic for obtaining the primary benefit result based on the carbon emission reduction benefits of resource conservation is as follows: Based on the first mapping relationship, the material ID of the recyclable resource data is mapped to the target ID of the target production data to obtain the first mapping pair I( ), where i is the i-th first mapping pair. For the first mapping pair and the first mapping relationship, the effective recovery quantity is calculated. The expression for calculating the effective recovery quantity is: ； Using the carbon emission factor method, the reduction in carbon emissions from recycling is calculated based on the carbon emission factors corresponding to the effective recycling quantity and material type. The expression for calculating the reduction in carbon emissions from recycling is as follows: ； Based on the second mapping relationship, the waste ID of the waste resource data and the energy ID of the energy production data are mapped to obtain the second mapping pair J( ), where j is the j-th second mapping pair. For the second mapping pair and the second mapping relationship, the amount of waste utilization is calculated. The expression for calculating the amount of waste utilization is: ; The reduction in carbon emissions from waste is calculated based on the reduction in carbon emissions from replacing primary energy sources with waste materials. The formula for calculating the reduction in carbon emissions from waste is as follows: ； The carbon emission factor is obtained from energy output data and is used to represent the reduction in carbon emissions per unit of energy output; The first benefit result is obtained by adding the reduction in carbon emissions from recycling and the reduction in carbon emissions from waste.

5. The low-carbon evaluation method for the recycling of renewable resources in power grids based on deep learning as described in claim 4, characterized in that, The processing logic for obtaining the second benefit result based on the economic benefits of resource conservation is as follows: The unit raw material cost and recycling material cost of recyclable resources are obtained. Based on the first mapping relationship, the recycling economic benefit is calculated for each first mapping pair. The calculation expression for the recycling economic benefit is as follows: ； The unit cost of raw materials mentioned is the market cost of purchasing new materials; The unit cost of recycled materials is the total cost of the material's unit recycling cost, dismantling labor cost, and remanufacturing cost. For all first mapping pairs, the total regression economic benefit is calculated, and the expression for the total regression economic benefit is as follows: ； The waste disposal cost and unit energy sales revenue of the acquired energy waste resources are obtained. Based on the second mapping relationship, for each second mapping pair, the energy economic benefits are calculated according to the amount of waste utilization. The calculation expression for the energy economic benefits is as follows: ; The unit energy sales revenue mentioned is the market sales price of energy output; The waste disposal cost is the total cost of waste collection, transportation, and manual processing. For all second mapping pairs, calculate the total energy economic benefit, expressed as follows: ； The second benefit result is obtained by adding the total economic benefits of recycling and the total economic benefits of energy.

6. The low-carbon evaluation method for the recycling of renewable resources in power grids based on deep learning as described in claim 5, characterized in that, Step S3 specifically includes: Step S31: Perform variational mode decomposition on the first and second benefit results, setting the number of modes K and the penalty coefficient. K intrinsic mode components are extracted to represent the periodic, trend, and random fluctuation characteristics of carbon emission reduction and economic benefits; Step S32: Construct a dual-channel feature vector system. Extract material type, unit raw material cost, unit recycled material cost, and energy label for feature encoding to generate a first feature vector. Extract K intrinsic mode components, recycling quantity, waste utilization quantity, and energy output efficiency according to the processing date. Then, through normalization, concatenate to generate a second feature vector. Use a gated attention fusion mechanism to integrate the dual-channel features. Pass the first feature vector through a linear transformation layer to generate a gated weight vector r. Pass the second feature vector through a linear transformation layer to generate a candidate feature vector c. Finally, obtain the fused feature vector through gated weighted calculation. The gated weighted calculation expression is: ； in, To fuse feature vectors, and For learnable projection matrices, The first eigenvector, Let g be the second feature vector, and g be the gating weight coefficient. Step S33: Construct time series samples by sliding the fused feature vectors according to the time window to obtain the time series sample matrix. Traverse each time window and take the last time point of the current time window as the corresponding labeled time of the time series sample matrix of this stage. Step S34: Use the variational mode decomposition multi-objective prediction network to predict the time series sample matrix and generate prediction result values ​​and auxiliary information values.

7. The low-carbon evaluation method for the recycling of renewable resources in power grids based on deep learning as described in claim 6, characterized in that, The prediction results and auxiliary information include: The forecast results include predicted carbon emission reductions and predicted economic benefits; Supporting information includes prediction confidence, uncertainty indicators, and consistency metrics.

8. The low-carbon evaluation method for the recycling of renewable resources in power grids based on deep learning as described in claim 7, characterized in that, Step S4 specifically includes: The fusion feature vector, prediction results, first benefit results, second benefit results and auxiliary information are standardized and then concatenated to generate a standardized decision matrix. A dual-criteria judgment matrix of carbon emission reduction and economic benefits is constructed using the analytic hierarchy process (AHP). A threshold for consistency indicators is set, and when the consistency indicator falls below the threshold, a subjective weight vector is derived. ; Using the improved entropy weighting method, information entropy is calculated for four indicators in the standardized decision matrix: the fusion feature vector, the prediction result, the first benefit result, and the second benefit result. An objective weight vector is obtained by applying weight reduction penalties using uncertainty indicators and consistency indicators. ; A negotiation game model is constructed, with subjective and objective weight vectors representing the two negotiating parties. The final weight vector W is then solved. The processing logic for constructing the negotiation game model is as follows: Set a negotiation breakdown point. When neither party can obtain their ideal weights, reset the breakdown point payoff of both the subjective and objective weight vectors to zero, and set the utility function accordingly. ； ； in, For subjective utility, For objective utility, Let the squared Euclidean distance between the weight vector and the subjective weight vector be the standard distance. Based on the subjective and objective utility, the negotiation game constraint is defined as maximizing the Nash product. Construct the objective function to maximize the Nash product: ； The initial weights are the arithmetic mean of the subjective and objective weight vectors. The objective function of maximizing the Nash product is solved iteratively, corresponding to each iteration. Perform the following sub-steps: Calculate the current weight vector Nash product objective function value at point And calculate the gradient vector. The expression for calculating the objective function value of the Nash product is: ； Construct a quadratic programming subproblem, wherein the quadratic programming subproblem is in The original objective function is approximated using a quadratic function, and the search direction is determined by the gradient vector. Along the search direction Perform a one-dimensional search and set the step size. Make the objective function value of the Nash product Satisfy the sufficient increase condition, while ensuring the updated weight vector Given that the values ​​are non-negative and conform to normalization constraints, calculate the change norm of the weight vector. If the current change norm is less than the preset threshold, the iteration is terminated and the final weight vector W is output as the game equilibrium weight vector. The compromise solution for each resource recycling batch *s* is calculated based on the game equilibrium weight vector and the standardized decision matrix. The compromise solution is then used to generate a low-carbon evaluation index through reverse mapping. The confidence level of the compromise solution is corrected using a consistency index to obtain the final low-carbon evaluation index. The calculation expression for the confidence correction is as follows: ； in, As a low-carbon evaluation indicator. As a consistency indicator, This is a sign of uncertainty; Low-carbon evaluation indicators are classified into levels according to set thresholds to obtain low-carbon evaluation levels.

9. The low-carbon evaluation method for the recycling of renewable resources in power grids based on deep learning as described in claim 8, characterized in that, The low-carbon evaluation indicators are classified into levels according to the evaluation rules, and the processing logic for obtaining the low-carbon evaluation level is as follows: The threshold is set as a first threshold and a second threshold. When the low-carbon evaluation index is less than the first threshold, the current low-carbon evaluation index is determined to be at the red level. When the first threshold is less than the low-carbon evaluation index and the second threshold is less than the second threshold, the current low-carbon evaluation index is determined to be at the yellow level. When the second threshold is less than the low-carbon evaluation index, the current low-carbon evaluation index is determined to be at the green level. The red level indicates a high-carbon, low-efficiency pathway. The yellow level indicates paths where resource allocation needs to be optimized. The yellow level indicates a dual-optimal path for a low-carbon economy.

10. A deep learning-based low-carbon evaluation system for the recycling of renewable resources in power grids, applied in any one of the deep learning-based low-carbon evaluation methods for the recycling of renewable resources in power grids as described in claims 1-9, characterized in that, It includes a relationship mapping module, a benefit calculation module, a target prediction module, and a low-carbon evaluation module; The relationship mapping module acquires recycled resource record data, which includes recyclable resource data, waste resource data, target production data, and energy output data. Based on the recycled resource record data, a first mapping relationship and a second mapping relationship are established. The benefit calculation module is used to calculate the first benefit result and the second benefit result by using the multi-task system optimization path method to calculate the first and second benefit results of the first and second mapping relationships of the relationship mapping module. The target prediction module is used to obtain carbon emission reduction prediction values ​​and economic benefit prediction values ​​from the first benefit result and the second benefit result by using variational mode decomposition and multi-source feature fusion mechanism. The low-carbon evaluation module is used to solve the game equilibrium weight vector through the iterative optimization algorithm of the negotiation game model, and to calculate the low-carbon evaluation index based on the carbon emission reduction prediction value and the economic benefit prediction value.

Images