Carbon emission calculation method and device supporting urban pollution reduction and carbon reduction collaborative development

Abstract

The invention discloses a carbon emission calculation method and device supporting urban pollution reduction and carbon reduction collaborative development, and the method comprises the steps: firstly, carrying out the quantitative observation and analysis of the correlation between each impact factor of carbon emission and the carbon emission through a gray correlation analysis method improved based on a distance analysis method, and screening main impact factors; secondly, constructing an LSTM carbon emission prediction model based on the analyzed and determined main factors, optimizing hyper-parameters of the LSTM carbon emission prediction model through a GCS algorithm, forming a GCS-LSTM carbon emission prediction model, and realizing annual prediction of urban carbon emission based on the GCS-LSTM carbon emission prediction model; and finally, on the basis of the obtained carbon emission predicted value, constructing a composite system collaboration degree model to measure and calculate the future pollution reduction and carbon reduction collaboration degree. The device comprises a processor and a memory.

Description

Technical Field

[0001] The present invention relates to the field of carbon emission calculation, and in particular to a carbon emission calculation method and device that supports the coordinated development of urban pollution reduction and carbon reduction. Background Art

[0003] Climate change is a global concern that could lead to rising sea levels, more frequent and more intense weather events, and significant impacts on ecosystems and human well-being. Carbon dioxide emissions are the primary cause of global warming, so predicting future carbon emissions is a pressing issue.

[0004] The current methods for calculating carbon emissions have the following shortcomings:

[0005] (1) The research on influencing factors has not fully considered the interaction of multiple dimensions and multiple factors, resulting in the lack of comprehensiveness of the carbon emission model or the introduction of redundancy due to too many dependent variables; (2) It is difficult for the prediction model of carbon emissions to capture nonlinear complex relationships, and hyperparameter optimization mostly uses grid search or random search methods, which have low search efficiency and are prone to falling into local optimality. Summary of the invention

[0006] The present invention provides a carbon emission calculation method and device that supports the coordinated development of urban pollution reduction and carbon reduction. The present invention fully explores and analyzes the multi-dimensional carbon emission influencing factors involving environmental protection, clarifies the correlation between carbon emissions and pollutant emissions, and assists cities in achieving sustainable ecological environment improvement. The GCS-LSTM model is used to train a large amount of historical data such as environmental pollution to realize the calculation of carbon emissions. See the following description for details:

[0007] In a first aspect, a carbon emission calculation method supporting the coordinated development of urban pollution reduction and carbon reduction is provided, the method comprising:

[0008] The influencing factors of urban carbon emissions are selected from four perspectives: economy, macro level, high-energy-consuming industries and transportation. The grey correlation analysis method based on the distance analysis method is used to quantitatively observe and analyze the correlation between the various influencing factors of carbon emissions and carbon emissions, and the main factors are screened out.

[0009] Based on the main factors analyzed and determined, an LSTM carbon emission prediction model is constructed, and the hyperparameters of the LSTM carbon emission prediction model are optimized through the GCS algorithm to form a GCS-LSTM carbon emission prediction model. Based on the GCS-LSTM carbon emission prediction model, the carbon emissions of the city are predicted;

[0010] Based on the obtained carbon emission prediction values, a composite system synergy model is constructed to calculate the future synergy of pollution reduction and carbon reduction.

[0011] The grey relational analysis method improved based on the distance analysis method is specifically as follows:

[0012] Collect relevant data, determine carbon emissions as the reference series, each influencing factor as the comparison series, and calculate the correlation coefficient ξ between the reference series and the comparison series i (k), the formula is as follows:

[0013]

[0014] Where k is the kth data in the sequence; △i(k) represents the absolute value of the difference between a certain element of the carbon emission sequence and the corresponding element of a comparison sequence; △min and △max represent the minimum and maximum values ​​of these absolute values, respectively. ρ is the resolution coefficient; calculate the reference sequence x0 and the comparison sequence x i When calculating the correlation between carbon emissions, each influencing factor is weighted to measure the correlation between each influencing factor and carbon emissions. The formula is as follows:

[0015]

[0016] Among them, ξ i is the correlation degree, α is the weight, and N is the number of influencing factors; the weight is calculated as follows: determine the best factor and the worst factor, and use the best factor and the worst factor as reference factors, use Euclidean distance to calculate the distance from each influencing factor to the reference factor, comprehensively evaluate the distance of each factor by combining the positive and negative distances, and normalize the data to obtain all weight vectors α(k);

[0017] Finally, according to the grey correlation result ξ i The influencing factors with significant correlation with carbon emissions were screened out as input variables of the carbon emission prediction model.

[0018] The steps for constructing the GCS-LSTM carbon emission prediction model are as follows:

[0019] Initialize the search range of the three hyperparameters h, learning rate lr and training times n of the LSTM carbon emission prediction model, the number of bird nests and the maximum number of iterations of the GCS algorithm; then initialize the population, assuming that the probability p of cuckoo eggs being found in the nest by the host bird a = 0.25, initialize the location of the bird's nest, and randomly generate n bird's nest locations Right now Each bird's nest location corresponds to a three-dimensional vector (h, lr, n). The location of each bird's nest is determined by the number of hidden layer neurons, learning rate and training times. The root mean square error formula is used to calculate the fitness of each bird's nest location. The LSTM carbon emission prediction model is trained using the parameter combination and a predicted value is generated. The predicted value of the model is compared with the actual value, and the prediction error is calculated to obtain the optimal nest location in the contemporary era. And the optimal fitness F min ;

[0020] Keep the best nest position and update other nest positions through Levy flight; compare the new nest position with the previous generation position p according to the fitness i-1 For comparison, use the previous generation position p i-1 Update the new bird's nest position with a better position in the , and get a new bird's nest position sequence:

[0021] The random number r and p t The probability of each nest location being found is p a Compare and simulate the probability of random events, retaining p t The nest positions with a low probability of being found are randomly updated, and the fitness of the new nests is calculated and compared with the fitness of the previous generation to obtain a set of nest positions with better fitness.

[0022] Add Gaussian perturbation to the above changes in the optimal nest position to determine a set of new positions, namely p′ t Each nest position in p t The better ones are retained to obtain a set of better nest locations p″ t , determine the optimal nest location based on the results obtained at this time, and calculate its fitness to determine whether it meets the requirements; if it does, output the global optimal nest location to obtain the optimal hyperparameters in the LSTM carbon emission prediction model; if it does not meet the requirements, return to continue to update other nest locations through Levy flight; according to the optimal nest location The corresponding hyperparameter values ​​are used as the optimal parameters of the LSTM neural network to establish the GCS-LSTM carbon emission prediction model.

[0023] Furthermore, the calculation method of the composite system synergy model is as follows:

[0024] The two subsystems of the composite system synergy model are pollution reduction subsystem S1 and carbon reduction subsystem S2. Each subsystem consists of several order parameters. wk is the kth order parameter of the wth subsystem. Order parameters can be divided into positive and negative indicators;

[0025] For the pollution reduction subsystem, the decline rates of PM2.5 and SO2 concentrations are selected as order parameters; for the carbon reduction subsystem, the decline rate of total carbon emissions, carbon emission intensity and per capita carbon emissions are selected as order parameters; the order parameters used in this method are all positive indicators; the order parameter data of the carbon reduction subsystem are calculated from the carbon emission prediction value solved by the GCS-LSTM carbon emission prediction model, and the order parameter data of the pollution reduction subsystem are all based on the reasonable data of the existing prediction algorithm;

[0026] According to the above order parameters, calculate the order of each order parameter:

[0027]

[0028] Where: u w (P wk ) is the order parameter P wk The larger the order of the wk The greater the contribution to the order of the subsystem; w1 , P w2 ,……,P wm is a positive indicator, P wm+1 , P wm+2 ,……,P wn is a negative indicator; α wk and β wk They represent the minimum and maximum values ​​of the order parameter during the study period respectively; u is expressed as follows w (P wk ) weighted sum, we can get the subsystem S w The order of:

[0029]

[0030] Where: u w (S w ) is the subsystem S w The degree of order; θ k is the order parameter P wk The weight of θ is calculated using the correlation coefficient matrix method. k , assuming that subsystem S w It consists of n order parameters, and its correlation coefficient matrix A is as follows:

[0031]

[0032] Where: A k represents the kth order parameter P wk The total effect on the other n-1 order parameters is A k Normalization can get the order parameter P wk The weight θ k :

[0033]

[0034] At the initial time t, the synergy between the pollution reduction and carbon reduction subsystems is As the system evolves to time t+1, the degree of coordination between the two becomes The carbon synergy of pollution reduction from time t to time t+1 is:

[0035]

[0036] Where: Synergy is the synergy degree of pollution reduction and carbon reduction; the value of parameter λ determines whether pollution reduction and carbon reduction can achieve synergy; if and only if When λ = 1, that is, when both the pollution reduction subsystem S1 and the carbon reduction subsystem S2 evolve in a more orderly direction, the value of Synergy is positive, and pollution reduction and carbon reduction achieve synergy; otherwise, it is negative, indicating that pollution reduction and carbon reduction do not achieve synergy; the parameter η w For subsystem S w Weight, Synergy∈[-1,1], the larger its value, the higher the degree of synergy in pollution reduction and carbon reduction.

[0037] In a second aspect, a carbon emission calculation device supporting the coordinated development of urban pollution reduction and carbon reduction is provided, the device comprising:

[0038] The screening module is used to select the influencing factors of urban carbon emissions from four perspectives: economy, macro level, high-energy-consuming industries and transportation. The grey correlation analysis method based on the distance analysis method is used to quantitatively observe and analyze the correlation between various influencing factors of carbon emissions and carbon emissions, and screen out the main factors.

[0039] The carbon emission module is used to build an LSTM carbon emission prediction model based on the main factors analyzed and determined, and optimize the hyperparameters of the LSTM carbon emission prediction model through the GCS algorithm to form a GCS-LSTM carbon emission prediction model. Based on the GCS-LSTM carbon emission prediction model, the carbon emissions of the city are predicted;

[0040] The calculation module is used to construct a composite system synergy model based on the obtained carbon emission prediction value to calculate the future pollution reduction and carbon reduction synergy.

[0041] The screening module includes: a screening submodule, and the screening submodule includes: a grey relational analysis improved based on a distance analysis method, specifically:

[0042] Collect relevant data, determine carbon emissions as the reference series, each influencing factor as the comparison series, and calculate the correlation coefficient ξ between the reference series and the comparison series i (k), the formula is as follows:

[0043]

[0044] Where k is the kth data in the sequence; △i(k) represents the absolute value of the difference between a certain element of the carbon emission sequence and the corresponding element of a comparison sequence; △min and △max represent the minimum and maximum values ​​of these absolute values, respectively. ρ is the resolution coefficient; calculate the reference sequence x0 and the comparison sequence x i When calculating the correlation between carbon emissions, each influencing factor is weighted to measure the correlation between each influencing factor and carbon emissions. The formula is as follows:

[0045]

[0046] Among them, ξ i is the correlation degree, α is the weight, and N is the number of influencing factors; the weight is calculated as follows: determine the best factor and the worst factor, and use the best factor and the worst factor as reference factors, use Euclidean distance to calculate the distance from each influencing factor to the reference factor, comprehensively evaluate the distance of each factor by combining the positive and negative distances, and normalize the data to obtain all weight vectors α(k);

[0047] Finally, according to the grey relational result ξ i The influencing factors with significant correlation with carbon emissions were screened out as input variables of the carbon emission prediction model.

[0048] The steps for constructing the GCS-LSTM carbon emission prediction model are as follows:

[0049] Initialize the search range of the three hyperparameters h, learning rate lr and training times n of the LSTM carbon emission prediction model, the number of bird nests and the maximum number of iterations of the GCS algorithm; then initialize the population, assuming that the probability p of cuckoo eggs being found in the nest by the host bird a = 0.25, initialize the location of the bird's nest, and randomly generate n bird's nest locations Right now Each bird's nest location corresponds to a three-dimensional vector (h, lr, n). The location of each bird's nest is determined by the number of hidden layer neurons, learning rate and training times. The root mean square error formula is used to calculate the fitness of each bird's nest location. The LSTM carbon emission prediction model is trained using the parameter combination and a predicted value is generated. The predicted value of the model is compared with the actual value, and the prediction error is calculated to obtain the optimal nest location in the contemporary era. And the optimal fitness F min ;

[0050] Keep the best nest position and update other nest positions through Levy flight; compare the new nest position with the previous generation position p according to the fitnessi-1 For comparison, use the previous generation position p i-1 Update the new bird's nest position with a better position in the , and get a new bird's nest position sequence:

[0051] The random number r and p t The probability of each nest location being found is p a Compare and simulate the probability of random events, retaining p t The nest positions with a low probability of being found are randomly updated, and the fitness of the new nests is calculated and compared with the fitness of the previous generation to obtain a set of nest positions with better fitness.

[0052] Add Gaussian perturbation to the above changes in the optimal nest position to determine a set of new positions, namely p′ t Each nest position in p t The better ones are retained to obtain a set of better nest locations p″ t , determine the optimal nest location based on the results obtained at this time, and calculate its fitness to determine whether it meets the requirements; if it does, output the global optimal nest location to obtain the optimal hyperparameters in the LSTM carbon emission prediction model; if it does not meet the requirements, return to continue to update other nest locations through Levy flight; according to the optimal nest location The corresponding hyperparameter values ​​are used as the optimal parameters of the LSTM neural network to establish the GCS-LSTM carbon emission prediction model.

[0053] In the third aspect, a carbon emission calculation device that supports the coordinated development of urban pollution reduction and carbon reduction, the device comprises: a processor and a memory, the memory stores program instructions, and the processor calls the program instructions stored in the memory to enable the device to execute any one of the methods described in the first aspect.

[0054] In a fourth aspect, a computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor executes any one of the methods described in the first aspect.

[0055] The beneficial effects of the technical solution provided by the present invention are:

[0056] (1) The present invention improves the grey correlation analysis algorithm based on the distance analysis method and performs weighted processing on each feature. It is applicable to systems with a small number of samples and systems with or without obvious regularities in the samples. It will not cause the situation where the quantitative results are inconsistent with the qualitative analysis, and effectively improves the accuracy of selecting carbon emission factors and reduces redundancy.

[0057] (2) The present invention proposes a cuckoo search algorithm based on Gaussian perturbation to determine the appropriate parameter values ​​in the LSTM model; the Gaussian variation helps to improve the search accuracy, jump out of the local extreme point to perform a global search, and greatly optimize the algorithm's search speed and optimization accuracy;

[0058] (3) The more refined, faster, and more accurate prediction model constructed by this method can provide strong data support for cities to formulate more effective environmental management policies. This accurate prediction can provide governments and enterprises with more scientific decision-making basis, optimize emission reduction strategies, and improve policy implementation effects; help identify and prioritize high pollution sources, thereby achieving more targeted pollution control and significantly reducing the concentration of pollutants in the air;

[0059] (4) Through the improved grey correlation analysis method and GCS-LSTM model, the present invention effectively screens the key influencing factors of carbon emissions and overcomes the deficiencies of traditional prediction methods in data redundancy and precision, thereby improving the accuracy of prediction; accurate carbon emission prediction can provide quantitative support for cities in formulating emission reduction strategies, making pollution reduction and carbon reduction work more scientifically based and targeted;

[0060] (5) Based on the accuracy of carbon emission prediction, cities can more effectively formulate and optimize air pollution prevention and control measures. By calculating the future coordination degree of pollution reduction and carbon reduction in cities, development strategies can be adjusted in advance. The present invention can provide support for the precise control of major pollution sources, reduce the total amount of pollutant emissions, and help reduce the concentration level of pollutants such as PM2.5 and SO2 in the air, thereby achieving a significant improvement in air quality.

[0061] (6) Predicted carbon emission data can not only assist cities in achieving carbon reduction targets in the short term, but also provide long-term support for urban ecological environment governance in the medium and long term; accurate carbon emission forecast data can help optimize systematic management measures such as industrial structure adjustment, energy structure optimization and land use planning, and gradually promote the development of green and low-carbon industries and the increase of urban green space, thereby achieving comprehensive improvement of the urban ecological environment and improving the quality of life of residents. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 A flowchart of a carbon emission calculation method that supports the coordinated development of urban pollution reduction and carbon reduction;

[0063] Figure 2 Schematic diagram of the training and prediction process of the GCS-LSTM network;

[0064] Figure 3 This is the structure diagram of the LSTM memory unit;

[0065] Figure 4This is an analysis chart of the predicted values ​​and actual values ​​of carbon emissions by the GCS-LSTM model and other models;

[0066] Figure 5 To predict the carbon emissions of Tianjin from 2023 to 2025 based on the GCS-LSTM model. DETAILED DESCRIPTION

[0067] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention are described in further detail below.

[0068] The embodiment of the present invention adopts a combination of two methods to predict carbon emissions from a new perspective. First, a variety of influencing factors are selected based on previous studies and screened using an improved grey correlation analysis method, which expands the scope of research on carbon emission influencing factors. Secondly, the LSTM neural network optimized by the cuckoo algorithm based on Gaussian perturbation is used to predict carbon emissions, and the future coordination degree of pollution reduction and carbon reduction in the city is calculated based on the predicted carbon emissions.

[0069] The embodiment of the present invention provides a carbon emission prediction method that supports the coordinated development of urban pollution reduction and carbon reduction. The embodiment of the present invention fully explores and analyzes the influencing factors of urban carbon emissions, and uses an improved GCS-LSTM model to predict carbon emissions, providing support for "carbon peak" and "carbon neutrality". See the following description for details:

[0070] Example 1

[0071] A carbon emission calculation method to support the coordinated development of urban pollution reduction and carbon reduction, see Figure 1 , the method comprises the following steps:

[0072] Step S101: Select influencing factors of urban carbon emissions from four perspectives: economy, macro level, high energy consumption industry and transportation, use the improved grey correlation analysis method based on distance analysis to quantitatively observe and analyze the correlation between various influencing factors of carbon emissions and carbon emissions, and screen out the main factors;

[0073] Through the above processing, the main factors with strong correlation with carbon emissions are screened out from many influencing factors for prediction, avoiding redundancy caused by too much data.

[0074] Among them, the main factors are: population, energy consumption per unit of GDP, cement production and urbanization rate;

[0075] Step S102: Based on the main factors analyzed and determined, an LSTM carbon emission prediction model is constructed, and the hyperparameters of the LSTM carbon emission prediction model are optimized through the GCS algorithm to form a GCS-LSTM carbon emission prediction model, and the carbon emissions of the city are predicted based on the GCS-LSTM carbon emission prediction model;

[0076] Step S103: Based on the obtained carbon emission prediction value, a composite system synergy model is constructed to calculate the future pollution reduction and carbon reduction synergy.

[0077] For step S101, the grey correlation analysis method improved based on the distance analysis method is used to quantitatively observe and analyze the correlation between various special influencing factors of carbon emissions and carbon emissions, and then the principal component analysis method is used to screen out four main influencing factors from the influencing factors determined by the grey correlation analysis, and the four main influencing factors are used as input variables of the GCS-LSTM prediction model.

[0078] For step S102, based on the influencing factors determined in step S102, a GCS-LSTM prediction model is constructed to predict carbon emissions, including the following steps:

[0079] Normalize the data set, take the influencing factors of carbon emissions as input variables, carbon emissions as output variables, and divide them into training sets and test sets;

[0080] The LSTM neural network is trained and the Gaussian perturbation-based cuckoo algorithm (GCS algorithm) is used to optimize the hyperparameters of the LSTM neural network, including the number of neurons in the hidden layer, the learning rate, and the number of training times, so as to improve the network weights.

[0081] According to the optimal nest location The corresponding hyperparameter values ​​are used as the optimal parameters of the LSTM neural network. The model is used to train the training set data to obtain the predicted values ​​of the training set.

[0082] For step S103, two subsystems are defined as the pollution reduction subsystem S1 and the carbon reduction subsystem S2, which together constitute a composite system for pollution reduction and carbon reduction.

[0083] For the pollution reduction subsystem, in order to accurately reflect the effectiveness of pollution reduction work and ensure the comparability of indicators, PM 2.5 The decrease rate of SO2 concentration is the order parameter. For the carbon reduction subsystem, considering the huge differences in economic development and population size across China, the decrease rate of total carbon emissions, carbon emission intensity and per capita carbon emissions are selected as order parameters.

[0084] First, the order of the order parameter is quantified, and the weight of the order parameter is further calculated using the correlation coefficient matrix method. The order of the subsystem's order parameter is weighted and summed to obtain the order of the subsystem. Finally, the order of the pollution reduction and carbon reduction subsystems is combined to calculate the synergy of the overall composite system. The synergy is evaluated by the change in order at two moments. When both subsystems evolve in a more ordered direction and the synergy value is positive, it indicates that pollution reduction and carbon reduction have achieved synergy. The higher the synergy value, the more significant the synergy effect.

[0085] Example 2

[0086] The scheme in Example 1 is further introduced below in conjunction with a specific calculation formula, and Tianjin data is used for analysis, as described below for details:

[0087] Step S201: Select factors that have a greater impact on carbon emissions in the city dimension from the four perspectives of economy, macro level, high-energy-consuming industries and transportation, use the improved grey correlation analysis method based on the distance analysis method to quantitatively observe and analyze the correlation between various influencing factors of carbon emissions and carbon emissions, and select the most important main factors as input variables of the GCS-LSTM prediction model;

[0088] Step S202: Based on the analyzed and determined influencing factors, an LSTM carbon emission prediction model is constructed, and the model hyperparameters are optimized through the GCS algorithm, such as Figure 2 As shown, the carbon emissions of the city can be predicted.

[0089] Step S203: Based on the calculated carbon emission prediction value, a composite system synergy model is constructed to predict the future coordinated development of pollution reduction and carbon reduction in the city.

[0090] For step S201, the correlation between various influencing factors of carbon emissions and carbon emissions is quantitatively observed and analyzed using the improved grey correlation analysis method based on the distance analysis method, and then the principal component analysis method is used to screen out the main influencing factors from the influencing factors determined by the grey correlation analysis, and use them as input variables of the GCS-LSTM prediction model, including the following steps:

[0091] Among them, the influencing factors of carbon emissions include sixteen influencing factors at four levels: economic factors, macro factors, high-energy-consuming industries, and transportation factors. They are population, total power generation, thermal power generation, urbanization rate, primary industry GDP, secondary industry GDP, tertiary industry GDP, unit GDP energy consumption, total fixed asset investment, integrated circuit output, cement output, flat glass output, automobile output, highway mileage, green area and freight volume.

[0092] Among them, the operation steps of the grey relational analysis method improved based on the distance analysis method are as follows:

[0093] Collect relevant data based on the research content, determine carbon emissions as a reference series, and sixteen carbon emission influencing factors as comparison series:

[0094] x′0(k)=(x′0(1),x′0(2),…,x′0(k))

[0095] x′ i (k) = (x′ i (1),x′i (2),…,x′ i (k))

[0096] Among them, x′0(k) represents the sequence of carbon emissions, k is the total number of samples, and i represents the number of comparison sequences. x′1(k), x′2(k), x′3(k)…x′ 15 (k), x′ 16 (k) respectively represent population, total power generation, thermal power generation, urbanization rate, primary industry GDP, secondary industry GDP, tertiary industry GDP, total exports, total fixed asset investment, integrated circuit output, cement output, flat glass output, automobile output, highway mileage, railway operating mileage and freight volume. Due to the different physical meanings of each variable, the initialization method is used to perform dimensionless processing on the reference sequence and comparison series:

[0097] Compute the correlation coefficient between a reference series and a comparison series:

[0098]

[0099] Where △i(k) represents the absolute value of the difference between an element of the carbon emission sequence and the corresponding element of a comparison sequence; △min and △max represent the minimum and maximum values ​​of these absolute values, respectively. ρ is the resolution coefficient, which ranges from (0,1) and is usually 0.5.

[0100] When calculating the correlation between the reference series and the comparison series, considering that the influence of each feature on the result is different, in order to avoid losing the potential features hidden in the data, the distance analysis method is used to improve the grey correlation analysis algorithm, and each feature is weighted to measure the correlation between each influencing factor and the reference data. The formula is as follows:

[0101]

[0102] Among them, ξ i is the correlation degree, and α is the weight.

[0103] The weight calculation method is as follows: take the best factor and the worst factor as reference factors, calculate the distance between each factor and the reference factor, and consider the samples close to the best factor point and far from the worst sample as the overall better factor. Suppose the data is an m*n matrix, m represents the number of samples. n represents the number of features or influencing factors of each sample.

[0104] First, determine the best and worst factors as follows in, Then the Euclidean distance is used to calculate the distance from each feature point to the reference feature point. Based on this, the distance of each element is comprehensively evaluated by combining the positive and negative distances. k The numerator and denominator are zero, and S k Add a bias β (where β is generally 1) to both the numerator and denominator of Finally, the data is normalized to obtain all weight vectors: α=(α1,α2,…,α m ) T .

[0105] Finally, the grey correlation results are tested and analyzed. According to the grey correlation analysis process, when ρ = 0.5, the correlation coefficient between each influencing factor and carbon emissions is obtained, and the P value test analysis is performed on each influencing factor. The P value test mainly tests the similarity between the influencing factors and carbon emissions, that is, the correlation between each factor and carbon emissions. The larger the P value, the more likely the difference between the two sets of data is to be an essential difference, that is, the smaller the correlation. According to the above process, the grey correlation between China's carbon emissions and each influencing factor is shown in the following table. According to experience, the grey correlation level is divided into: 0.8-1 is a strong correlation, 0.6-0.8 is a general correlation, and 0-0.6 is a weak correlation. According to the results obtained, the influencing factors with significant correlation to carbon emissions are judged. The correlation of the 15 influencing factors except the integrated circuit output is greater than 0.6, and there is a significant correlation with carbon emissions. Therefore, the influencing factors with a correlation greater than 0.99 are selected as the input of the carbon emission prediction model.

[0106]

[0107] The main influencing factors of carbon emissions determined by the grey correlation analysis method improved by the distance analysis method are population, energy consumption per unit GDP, cement production and urbanization rate, which are used as input variables of the GCS-LSTM prediction model.

[0108] For step S202, based on the influencing factors determined in step S201, a GCS-LSTM prediction model is constructed to predict carbon emissions, including the following steps:

[0109] First, the data set is normalized, the influencing factors of carbon emissions are taken as input variables, carbon emissions are taken as output variables, and the training set and test set are divided.

[0110] See also Figure 3, train the LSTM neural network, and use the CS algorithm (GCS algorithm) based on Gaussian perturbation to optimize the hyperparameters of the LSTM neural network. The hyperparameters include: the number of hidden layer neurons, learning rate, and number of training times, so that the network weights are improved. The specific steps for constructing the GCS-LSTM model are as follows:

[0111] Design of the original LSTM model structure. Design the number of LSTM network layers, the number of neurons in the input layer and hidden layer, the optimizer, the loss function, etc., and construct a multivariate multidimensional single-step LSTM neural network model with an initial input of 18 dimensions, a time step of 1, an output of 1 dimension, and 50 neurons in the hidden layer.

[0112] Initialize the search range of the hyperparameters of the LSTM neural network and the parameters of the GCS algorithm respectively, and then initialize the population. Assume that the probability of a cuckoo egg being found in the nest by the host bird is p a = 0.25, initialize the positions of n bird nests, that is, The position of each bird's nest is determined by the number of hidden layer neurons, learning rate and number of training times. The fit of each bird's nest position is calculated to obtain the current optimal bird's nest position. And the optimal fitness F min The root mean square error is selected as the fitness function, and its formula is as follows:

[0113]

[0114] Among them, y t is the true value of carbon emissions, is the model predicted value for the current parameters.

[0115] Update the position to optimize the parameters. Keep the best nest position and update other nest positions through Levy flight. Then generate a new set of nest positions and calculate the corresponding fitness F. According to the fitness, compare the new nest position with the previous generation position p i-1 Compare and update the bird's nest position with a better position, so that a new bird's nest position sequence can be obtained:

[0116] The random number r and p a For comparison, keep p t The nests with a low probability of being found are randomly updated. The fitness of the new nests is calculated and compared with that of the previous generation. The nests with better fitness are replaced with nests with better fitness. t The worse one among them, thus obtaining a set of better bird nest locations, which is still recorded as p t .

[0117] Add the Gaussian perturbation to the change of the nest position and determine a set of new positions, namely p′ t Each nest position in p t The better ones are retained to obtain a set of better nest locations p″ t , and start the next iteration. Determine the optimal nest location based on the results obtained at this time The fitness function is calculated to determine whether it meets the requirements. If it does, the search process is stopped and the global optimal nest position is output, thereby obtaining the optimal hyperparameters in the LSTM model; if it does not meet the requirements, the update position is returned to continue parameter optimization.

[0118] According to the optimal nest location The corresponding hyperparameter values ​​are used as the optimal parameters of the LSTM neural network to establish a carbon emission prediction model, train the training set data, and use the mean absolute percentage error (MAPE) calculation to test the model. The test results are compared with the traditional LSTM, GPR and BP models. The test results are shown in the following table.

[0119] Unit: 10,000 tons

[0120]

[0121]

[0122] As can be seen from the table above, the MAPE value of the improved GCS-LSTM model is lower than that of the traditional LSTM, GPR and BPNN models. Figure 4 As shown in the figure, it shows that the improved GCS-LSTM model more accurately captures the intrinsic connection between carbon emissions and influencing factors such as pollutants, and the carbon emission prediction results are more accurate. The model is used to predict the carbon emissions of Tianjin from 2023 to 2025. The prediction results are as follows: Figure 5 As shown. It provides quantitative support for Tianjin in formulating emission reduction strategies, making pollution reduction and carbon reduction work more scientifically based and targeted. Based on the accuracy of carbon emission prediction, cities can more effectively formulate and optimize air pollution prevention and control measures.

[0123] For step S203: based on the carbon emission data predicted in step S202, combined with the order parameters of the pollution reduction subsystem and the carbon reduction subsystem, the city's future pollution reduction and carbon reduction coordination degree is calculated. The specific process includes:

[0124] Calculation of order parameter order degree: For the pollution reduction subsystem, the PM2.5 and SO2 concentration decline rates are selected as order parameters; for the carbon reduction subsystem, the total carbon emission decline rate, carbon emission intensity decline rate and per capita carbon emission decline rate are selected as order parameters. Calculate the order degree of each order parameter. The formula is as follows:

[0125]

[0126] The correlation coefficient matrix method is used to determine the weight of the order parameter and calculate the order degree of the subsystem. The formula is as follows:

[0127]

[0128] Synergy calculation: According to the order of pollution reduction and carbon reduction subsystems, the synergy of the composite system is calculated. The higher the synergy value, the stronger the synergy effect of pollution reduction and carbon reduction. The formula is as follows:

[0129]

[0130] In summary, the embodiments of the present invention fully explore and analyze the influencing factors of carbon emissions in multiple dimensions such as environmental protection, clarify the correlation between carbon emissions and pollutant emissions, assist cities in achieving sustainable ecological environment improvement, and use the GCS-LSTM model to train a large amount of historical data such as environmental pollution to realize the calculation of carbon emissions.

[0131] Example 3

[0132] A carbon emission calculation device supporting the coordinated development of urban pollution reduction and carbon reduction, the device comprising:

[0133] The screening module is used to select the influencing factors of urban carbon emissions from four perspectives: economy, macro level, high-energy-consuming industries and transportation. The grey correlation analysis method based on the distance analysis method is used to quantitatively observe and analyze the correlation between various influencing factors of carbon emissions and carbon emissions, and screen out the main factors.

[0134] The carbon emission module is used to build an LSTM carbon emission prediction model based on the main factors analyzed and determined, and optimize the hyperparameters of the LSTM carbon emission prediction model through the GCS algorithm to form a GCS-LSTM carbon emission prediction model. Based on the GCS-LSTM carbon emission prediction model, the carbon emissions of the city are predicted;

[0135] The calculation module is used to construct a composite system synergy model based on the obtained carbon emission prediction value to calculate the future pollution reduction and carbon reduction synergy.

[0136] Among them, the screening module includes: a screening submodule, and the screening submodule includes: a grey correlation analysis improved based on the distance analysis method, specifically:

[0137] Collect relevant data, determine carbon emissions as the reference series, and each influencing factor as the comparison series, and calculate the correlation coefficient ξ(k) between the reference series and the comparison series. The formula is as follows:

[0138]

[0139] Where k is the kth data in the sequence; △i(k) represents the absolute value of the difference between a certain element of the carbon emission sequence and the corresponding element of a comparison sequence; △min and △max represent the minimum and maximum values ​​of these absolute values, respectively. ρ is the resolution coefficient; calculate the reference sequence x0 and the comparison sequence x i When calculating the correlation between carbon emissions, each influencing factor is weighted to measure the correlation between each influencing factor and carbon emissions. The formula is as follows:

[0140]

[0141] Among them, ξ i is the correlation degree, α is the weight, and N is the number of influencing factors; the weight is calculated as follows: determine the best factor and the worst factor, and use the best factor and the worst factor as reference factors, use Euclidean distance to calculate the distance from each influencing factor to the reference factor, comprehensively evaluate the distance of each factor by combining the positive and negative distances, and normalize the data to obtain all weight vectors α(k);

[0142] Finally, according to the grey correlation result ξ i The influencing factors with significant correlation with carbon emissions were screened out as input variables of the carbon emission prediction model.

[0143] Among them, the construction steps of the GCS-LSTM carbon emission prediction model are:

[0144] Initialize the search range of the three hyperparameters h, learning rate lr and training times n of the LSTM carbon emission prediction model, the number of bird nests and the maximum number of iterations of the GCS algorithm; then initialize the population, assuming that the probability p of cuckoo eggs being found in the nest by the host bird a = 0.25, initialize the location of the bird's nest, and randomly generate n bird's nest locations Right now Each bird's nest location corresponds to a three-dimensional vector (h, lr, n). The location of each bird's nest is determined by the number of hidden layer neurons, learning rate and training times. The root mean square error formula is used to calculate the fitness of each bird's nest location. The LSTM carbon emission prediction model is trained using the parameter combination and a predicted value is generated. The predicted value of the model is compared with the actual value, and the prediction error is calculated to obtain the optimal nest location in the contemporary era. And the optimal fitness F min ;

[0145] Keep the best nest position and update other nest positions through Levy flight; compare the new nest position with the previous generation position p according to the fitness i-1For comparison, use the previous generation position p i-1 Update the new bird's nest position with a better position in the , and get a new bird's nest position sequence:

[0146] The random number r and p t The probability of each nest location being found is p a Compare and simulate the probability of random events, retaining p t The nest positions with a low probability of being found are randomly updated, and the fitness of the new nests is calculated and compared with the fitness of the previous generation to obtain a set of nest positions with better fitness.

[0147] Add Gaussian perturbation to the above changes in the optimal nest position to determine a set of new positions, namely p′ t Each nest position in p t The better ones are retained to obtain a set of better nest locations p″ t , determine the optimal nest location based on the results obtained at this time, and calculate its fitness to determine whether it meets the requirements; if it does, output the global optimal nest location to obtain the optimal hyperparameters in the LSTM carbon emission prediction model; if it does not meet the requirements, return to continue to update other nest locations through Levy flight; according to the optimal nest location The corresponding hyperparameter values ​​are used as the optimal parameters of the LSTM neural network to establish the GCS-LSTM carbon emission prediction model.

[0148] In summary, the embodiments of the present invention fully explore and analyze the influencing factors of carbon emissions in multiple dimensions such as environmental protection, clarify the correlation between carbon emissions and pollutant emissions, assist cities in achieving sustainable ecological environment improvement, and use the GCS-LSTM model to train a large amount of historical data such as environmental pollution to realize the calculation of carbon emissions.

[0149] Example 4

[0150] A carbon emission calculation device supporting the coordinated development of urban pollution reduction and carbon reduction, the device comprising: a processor and a memory, the memory storing program instructions, the processor calling the program instructions stored in the memory to enable the device to execute the following method steps in Example 1:

[0151] The influencing factors of urban carbon emissions are selected from four perspectives: economy, macro level, high-energy-consuming industries and transportation. The grey correlation analysis method based on the distance analysis method is used to quantitatively observe and analyze the correlation between the various influencing factors of carbon emissions and carbon emissions, and the main factors are screened out.

[0152] Based on the main factors analyzed and determined, an LSTM carbon emission prediction model is constructed, and the hyperparameters of the LSTM carbon emission prediction model are optimized through the GCS algorithm to form a GCS-LSTM carbon emission prediction model. Based on the GCS-LSTM carbon emission prediction model, the carbon emissions of the city are predicted;

[0153] Based on the obtained carbon emission prediction values, a composite system synergy model is constructed to calculate the future synergy of pollution reduction and carbon reduction.

[0154] Among them, the grey relational analysis method improved based on the distance analysis method is as follows:

[0155] Collect relevant data, determine carbon emissions as the reference series, each influencing factor as the comparison series, and calculate the correlation coefficient ξ between the reference series and the comparison series i (k), the formula is as follows:

[0156]

[0157] Where k is the kth data in the sequence; △i(k) represents the absolute value of the difference between a certain element of the carbon emission sequence and the corresponding element of a comparison sequence; △min and △max represent the minimum and maximum values ​​of these absolute values, respectively. ρ is the resolution coefficient; calculate the reference sequence x0 and the comparison sequence x i When calculating the correlation between carbon emissions, each influencing factor is weighted to measure the correlation between each influencing factor and carbon emissions. The formula is as follows:

[0158]

[0159] Among them, ξ i is the correlation degree, α is the weight, and N is the number of influencing factors; the weight is calculated as follows: determine the best factor and the worst factor, and use the best factor and the worst factor as reference factors, use Euclidean distance to calculate the distance from each influencing factor to the reference factor, comprehensively evaluate the distance of each factor by combining the positive and negative distances, and normalize the data to obtain all weight vectors α(k);

[0160] Finally, according to the grey relational result ξ i The influencing factors with significant correlation with carbon emissions were screened out as input variables of the carbon emission prediction model.

[0161] Among them, the construction steps of the GCS-LSTM carbon emission prediction model are:

[0162] Initialize the search range of the three hyperparameters h, learning rate lr and training times n of the LSTM carbon emission prediction model, the number of bird nests and the maximum number of iterations of the GCS algorithm; then initialize the population, assuming that the probability p of cuckoo eggs being found in the nest by the host bird a = 0.25, initialize the location of the bird's nest, and randomly generate n bird's nest locations Right now Each bird's nest location corresponds to a three-dimensional vector (h, lr, n). The location of each bird's nest is determined by the number of hidden layer neurons, learning rate and training times. The root mean square error formula is used to calculate the fitness of each bird's nest location. The LSTM carbon emission prediction model is trained using the parameter combination and a predicted value is generated. The predicted value of the model is compared with the actual value, and the prediction error is calculated to obtain the optimal nest location in the contemporary era. And the optimal fitness F min ;

[0163] Keep the best nest position and update other nest positions through Levy flight; compare the new nest position with the previous generation position p according to the fitness i-1 For comparison, use the previous generation position p i-1 Update the new bird's nest position with a better position in the , and get a new bird's nest position sequence:

[0164] The random number r and p t The probability of each nest location being found is p a Compare and simulate the probability of random events, retaining p t The nest positions with a low probability of being found are randomly updated, and the fitness of the new nests is calculated and compared with the fitness of the previous generation to obtain a set of nest positions with better fitness.

[0165] Add Gaussian perturbation to the above changes in the optimal nest position to determine a set of new positions, namely p′ t Each nest position in p t The better ones are retained to obtain a set of better nest locations p″ t , determine the optimal nest location based on the results obtained at this time, and calculate its fitness to determine whether it meets the requirements; if it does, output the global optimal nest location to obtain the optimal hyperparameters in the LSTM carbon emission prediction model; if it does not meet the requirements, return to continue to update other nest locations through Levy flight; according to the optimal nest location The corresponding hyperparameter values ​​are used as the optimal parameters of the LSTM neural network to establish the GCS-LSTM carbon emission prediction model.

[0166] Furthermore, the calculation method of the composite system synergy model is as follows:

[0167] The two subsystems of the composite system synergy model are pollution reduction subsystem S1 and carbon reduction subsystem S2. Each subsystem consists of several order parameters. wk is the kth order parameter of the wth subsystem. Order parameters can be divided into positive and negative indicators;

[0168] For the pollution reduction subsystem, the decline rates of PM2.5 and SO2 concentrations are selected as order parameters; for the carbon reduction subsystem, the decline rate of total carbon emissions, carbon emission intensity and per capita carbon emissions are selected as order parameters; the order parameters used in this method are all positive indicators; the order parameter data of the carbon reduction subsystem are calculated from the carbon emission prediction value solved by the GCS-LSTM carbon emission prediction model, and the order parameter data of the pollution reduction subsystem are all based on the reasonable data of the existing prediction algorithm;

[0169] According to the above order parameters, calculate the order of each order parameter:

[0170]

[0171] Where: u w (P wk ) is the order parameter P wk The larger the order of the wk The greater the contribution to the order of the subsystem; w1 , P w2 ,……,P wm is a positive indicator, P wm+1 , P wm+2 ,……,P wn is a negative indicator; α wk and β wk They represent the minimum and maximum values ​​of the order parameter during the study period respectively; u is expressed as follows w (P wk ) weighted sum, we can get the subsystem S w The order of:

[0172]

[0173] Where: u w (S w ) is the subsystem S w The degree of order; θ k is the order parameter P wk The weight of θ is calculated using the correlation coefficient matrix method. k , assuming that subsystem S w It consists of n order parameters, and its correlation coefficient matrix A is as follows:

[0174]

[0175] Where: A k represents the kth order parameter P wk The total effect on the other n-1 order parameters is A k Normalization can get the order parameter P wk The weight θ k :

[0176]

[0177] At the initial time t, the synergy between the pollution reduction and carbon reduction subsystems is As the system evolves to time t+1, the degree of coordination between the two becomes The carbon synergy of pollution reduction from time t to time t+1 is:

[0178]

[0179] Where: Synergy is the synergy degree of pollution reduction and carbon reduction; the value of parameter λ determines whether pollution reduction and carbon reduction can achieve synergy; if and only if When λ = 1, that is, when both the pollution reduction subsystem S1 and the carbon reduction subsystem S2 evolve in a more orderly direction, the value of Synergy is positive, and pollution reduction and carbon reduction achieve synergy; otherwise, it is negative, indicating that pollution reduction and carbon reduction do not achieve synergy; the parameter η w For subsystem S w Weight, Synergy∈[-1,1], the larger its value, the higher the degree of synergy in pollution reduction and carbon reduction.

[0180] It should be pointed out here that the device description in the above embodiment corresponds to the method description in the embodiment, and the embodiment of the present invention will not be described in detail here.

[0181] The execution subjects of the above-mentioned processor and memory can be devices with computing functions such as computers, single-chip microcomputers, and microcontrollers. In specific implementation, the embodiments of the present invention do not limit the execution subjects and are selected according to the needs of actual applications.

[0182] The data signal is transmitted between the memory and the processor via a bus, which is not described in detail in the embodiment of the present invention.

[0183] Based on the same inventive concept, an embodiment of the present invention further provides a computer-readable storage medium, the storage medium includes a stored program, and when the program is running, the device where the storage medium is located is controlled to execute the method steps in the above embodiment.

[0184] The computer-readable storage medium includes but is not limited to a flash memory, a hard disk, a solid-state drive, and the like.

[0185] It should be pointed out here that the description of the readable storage medium in the above embodiment corresponds to the description of the method in the embodiment, and the embodiment of the present invention will not be described in detail here.

[0186] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions according to the embodiments of the present invention are generated.

[0187] The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions may be stored in a computer-readable storage medium or transmitted via a computer-readable storage medium. The computer-readable storage medium may be any available medium that can be accessed by the computer or a data storage device such as a server or a data center that includes one or more available media. The available medium may be a magnetic medium or a semiconductor medium, etc.

[0188] Unless otherwise specified, the models of the components in the embodiments of the present invention are not limited, and any device that can perform the above functions may be used.

[0189] Those skilled in the art will appreciate that the accompanying drawing is only a schematic diagram of a preferred embodiment, and the serial numbers of the embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0190] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A carbon emission calculation method that supports the coordinated development of urban pollution reduction and carbon reduction, characterized in that: The method comprises: The influencing factors of urban carbon emissions are selected from four perspectives: economy, macro level, high-energy-consuming industries and transportation. The grey correlation analysis method based on the distance analysis method is used to quantitatively observe and analyze the correlation between the various influencing factors of carbon emissions and carbon emissions, and the main factors are screened out. Based on the main factors analyzed and determined, an LSTM carbon emission prediction model is constructed, and the hyperparameters of the LSTM carbon emission prediction model are optimized through the GCS algorithm to form a GCS-LSTM carbon emission prediction model. Based on the GCS-LSTM carbon emission prediction model, the carbon emissions of the city are predicted; Based on the obtained carbon emission prediction values, a composite system synergy model is constructed to calculate the future synergy of pollution reduction and carbon reduction.

2. A carbon emission calculation method supporting the coordinated development of urban pollution reduction and carbon reduction according to claim 1, characterized in that: The improved grey relational analysis method based on distance analysis is specifically as follows: Collect relevant data, determine carbon emissions as the reference series, and each influencing factor as the comparison series, and calculate the correlation coefficient ξ(k) between the reference series and the comparison series. The formula is as follows: Where k is the kth data in the sequence; △i(k) represents the absolute value of the difference between a certain element of the carbon emission sequence and the corresponding element of a comparison sequence; △min and △max represent the minimum and maximum values ​​of these absolute values, respectively. ρ is the resolution coefficient; calculate the reference sequence x0 and the comparison sequence x i When calculating the correlation between carbon emissions, each influencing factor is weighted to measure the correlation between each influencing factor and carbon emissions. The formula is as follows: Among them, ξ i is the correlation, α is the weight, and N is the number of influencing factors; the weight is calculated as follows: Determine the best and worst factors, and use the best and worst factors as reference factors. Use Euclidean distance to calculate the distance from each influencing factor to the reference factor. Comprehensively evaluate the distance of each factor by combining positive and negative distances. Normalize the data to obtain all weight vectors α(k). Finally, according to the grey relational result ξ i The influencing factors with significant correlation with carbon emissions were screened out as input variables of the carbon emission prediction model.

3. A carbon emission calculation method supporting the coordinated development of urban pollution reduction and carbon reduction according to claim 1, characterized in that: The steps for constructing the GCS-LSTM carbon emission prediction model are as follows: Initialize the search range of the three hyperparameters h, learning rate lr and training times n of the LSTM carbon emission prediction model, the number of bird nests and the maximum number of iterations of the GCS algorithm; then initialize the population, assuming that the probability p of cuckoo eggs being found in the nest by the host bird a = 0.25, initialize the location of the bird's nest, and randomly generate n bird's nest locations Right now Each bird's nest location corresponds to a three-dimensional vector (h, lr, n). The location of each bird's nest is determined by the number of hidden layer neurons, learning rate and training times. The root mean square error formula is used to calculate the fitness of each bird's nest location. The LSTM carbon emission prediction model is trained using the parameter combination and a predicted value is generated. The predicted value of the model is compared with the actual value, and the prediction error is calculated to obtain the optimal nest location in the contemporary era. And the optimal fitness F min ; Keep the best nest position and update other nest positions through Levy flight; compare the new nest position with the previous generation position p according to the fitness i-1 For comparison, use the previous generation position p i-1 Update the new bird's nest position with a better position in the , and get a new bird's nest position sequence: The random number r and p t The probability of each nest location being found is p a Compare and simulate the probability of random events, retaining p t The nest positions with a low probability of being found are randomly updated, and the fitness of the new nests is calculated and compared with the fitness of the previous generation to obtain a set of nest positions with better fitness. Add Gaussian perturbation to the above changes in the optimal nest position to determine a set of new positions, namely p′ t Each nest position in p t The better ones are retained to obtain a set of better nest locations p″ t , determine the optimal nest location based on the results obtained at this time, and calculate its fitness to determine whether it meets the requirements; if it does, output the global optimal nest location to obtain the optimal hyperparameters in the LSTM carbon emission prediction model; if it does not meet the requirements, return to continue to update other nest locations through Levy flight; according to the optimal nest location The corresponding hyperparameter values ​​are used as the optimal parameters of the LSTM neural network to establish the GCS-LSTM carbon emission prediction model.

4. A carbon emission calculation method supporting the coordinated development of urban pollution reduction and carbon reduction according to claim 1, characterized in that: The calculation method of the composite system synergy model is as follows: The two subsystems of the composite system synergy model are pollution reduction subsystem S1 and carbon reduction subsystem S2. Each subsystem consists of several order parameters. wk is the kth order parameter of the wth subsystem. Order parameters can be divided into positive and negative indicators; For the pollution reduction subsystem, the decline rates of PM2.5 and SO2 concentrations are selected as order parameters; for the carbon reduction subsystem, the decline rate of total carbon emissions, carbon emission intensity and per capita carbon emissions are selected as order parameters; the order parameters used in this method are all positive indicators; the order parameter data of the carbon reduction subsystem are calculated from the carbon emission prediction value solved by the GCS-LSTM carbon emission prediction model, and the order parameter data of the pollution reduction subsystem are all based on the reasonable data of the existing prediction algorithm; According to the above order parameters, calculate the order of each order parameter: Where: u w (P wk ) is the order parameter P wk The larger the order of the wk The greater the contribution to the order of the subsystem; w1 , P w2 ,……,P wm is a positive indicator, P wm+1 , P wm+2 ,……,P wn is a negative indicator; α wk and β wk They represent the minimum and maximum values ​​of the order parameter during the study period respectively; u is expressed as follows w (P wk ) weighted sum, we can get the subsystem S w The order of: Where: u w (S w ) is the subsystem S w The degree of order; θ k is the order parameter P wk The weight of θ is calculated using the correlation coefficient matrix method. k , assuming that subsystem S w It consists of n order parameters, and its correlation coefficient matrix A is as follows: Where: A k represents the kth order parameter P wk The total effect on the other n-1 order parameters is A k Normalization can get the order parameter P wk The weight θ k : At the initial time t, the synergy between the pollution reduction and carbon reduction subsystems is As the system evolves to time t+1, the degree of coordination between the two becomes The carbon synergy degree of pollution reduction from time t to time t+1 is: Where: Synergy is the synergy degree of pollution reduction and carbon reduction; the value of parameter λ determines whether pollution reduction and carbon reduction can achieve synergy; if and only if When λ = 1, that is, when both the pollution reduction subsystem S1 and the carbon reduction subsystem S2 evolve in a more orderly direction, the value of Synergy is positive, and pollution reduction and carbon reduction achieve synergy; otherwise, it is negative, indicating that pollution reduction and carbon reduction do not achieve synergy; the parameter η w For subsystem S w Weight, Synergy∈[-1,1], the larger its value, the higher the degree of synergy in pollution reduction and carbon reduction.

5. A carbon emission calculation device that supports the coordinated development of urban pollution reduction and carbon reduction, characterized in that: The device comprises: The screening module is used to select the influencing factors of urban carbon emissions from four perspectives: economy, macro level, high-energy-consuming industries and transportation. The grey correlation analysis method based on the distance analysis method is used to quantitatively observe and analyze the correlation between various influencing factors of carbon emissions and carbon emissions, and screen out the main factors. The carbon emission module is used to build an LSTM carbon emission prediction model based on the main factors analyzed and determined, and optimize the hyperparameters of the LSTM carbon emission prediction model through the GCS algorithm to form a GCS-LSTM carbon emission prediction model. Based on the GCS-LSTM carbon emission prediction model, the carbon emissions of the city are predicted; The calculation module is used to construct a composite system synergy model based on the obtained carbon emission prediction value to calculate the future pollution reduction and carbon reduction synergy.

6. A carbon emission calculation device supporting the coordinated development of urban pollution reduction and carbon reduction according to claim 5, characterized in that: The screening module includes: a screening submodule, and the screening submodule includes: a grey relational analysis improved based on a distance analysis method, specifically: Collect relevant data, determine carbon emissions as the reference series, each influencing factor as the comparison series, and calculate the correlation coefficient ξ between the reference series and the comparison series i (k), the formula is as follows: Where k is the kth data in the sequence; Δi(k) represents the absolute value of the difference between a certain element of the carbon emission sequence and the corresponding element of a comparison sequence; Δmin and Δmax represent the minimum and maximum values ​​of these absolute values, respectively. ρ is the resolution coefficient; calculate the reference sequence x0 and the comparison sequence x i When calculating the correlation between carbon emissions, each influencing factor is weighted to measure the correlation between each influencing factor and carbon emissions. The formula is as follows: Among them, ξ i is the correlation, α is the weight, and N is the number of influencing factors; the weight is calculated as follows: Determine the best and worst factors, and use the best and worst factors as reference factors. Use Euclidean distance to calculate the distance from each influencing factor to the reference factor. Comprehensively evaluate the distance of each factor by combining positive and negative distances. Normalize the data to obtain all weight vectors α(k). Finally, according to the grey correlation result ξ i The influencing factors with significant correlation with carbon emissions were screened out as input variables of the carbon emission prediction model.

7. A carbon emission calculation device supporting the coordinated development of urban pollution reduction and carbon reduction according to claim 5, characterized in that: The steps for constructing the GCS-LSTM carbon emission prediction model are as follows: Initialize the search range of the three hyperparameters h, learning rate lr and training times n of the LSTM carbon emission prediction model, the number of bird nests and the maximum number of iterations of the GCS algorithm; then initialize the population, assuming that the probability p of cuckoo eggs being found in the nest by the host bird a = 0.25, initialize the location of the bird's nest, and randomly generate n bird's nest locations Right now Each bird's nest location corresponds to a three-dimensional vector (h, lr, n). The location of each bird's nest is determined by the number of hidden layer neurons, learning rate and training times. The root mean square error formula is used to calculate the fitness of each bird's nest location. The LSTM carbon emission prediction model is trained using the parameter combination and a predicted value is generated. The predicted value of the model is compared with the actual value, and the prediction error is calculated to obtain the optimal nest location in the contemporary era. And the optimal fitness F min ; Keep the best nest position and update other nest positions through Levy flight; compare the new nest position with the previous generation position p according to the fitness i-1 For comparison, use the previous generation position p i-1 Update the new bird's nest position with a better position in the , and get a new bird's nest position sequence: The random number r and p t The probability of each nest location being found is p a Compare and simulate the probability of random events, retaining p t The nest positions with a low probability of being found are randomly updated, and the fitness of the new nests is calculated and compared with the fitness of the previous generation to obtain a set of nest positions with better fitness. Add Gaussian perturbation to the above changes in the optimal nest position to determine a set of new positions, namely p′ t Each nest position in p t The better ones are retained to obtain a set of better nest locations p″ t , determine the optimal nest location based on the results obtained at this time, and calculate its fitness to determine whether it meets the requirements; if it does, output the global optimal nest location to obtain the optimal hyperparameters in the LSTM carbon emission prediction model; if it does not meet the requirements, return to continue to update other nest locations through Levy flight; according to the optimal nest location The corresponding hyperparameter values ​​are used as the optimal parameters of the LSTM neural network to establish the GCS-LSTM carbon emission prediction model.

8. A carbon emission calculation device that supports the coordinated development of urban pollution reduction and carbon reduction, characterized in that: The device comprises: a processor and a memory, wherein program instructions are stored in the memory, and the processor calls the program instructions stored in the memory to enable the device to execute the method according to any one of claims 1 to 4.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor is enabled to perform the method according to any one of claims 1 to 4.

Images