A method for predicting carbon emissions from urban systems based on input-output analysis and Bayesian neural networks

Abstract

This invention discloses a method for predicting carbon emissions in urban systems based on input-output analysis and Bayesian neural networks. The method includes: determining the carbon emissions of each sector in the city; calculating a carbon emission input-output table; identifying key carbon-emitting sectors and major energy consumption factors; calculating the economic share and energy consumption factors of these sectors; establishing an urban carbon emission prediction model considering economic development and energy consumption; designing high, medium, and low future scenarios based on historical data of key carbon-emitting sectors and energy consumption factors, as well as the city's future development plan; solving the coupled model; quantifying the impact of different scenarios on urban carbon emissions; and determining under what development model the city's carbon emissions can meet predetermined requirements. This invention introduces a reasonable and scientific analytical prediction model into the urban energy and economic system, providing more effective technical support for the generation of decision-making solutions.

Description

Technical Field

[0001] This invention belongs to the field of energy economic systems, specifically relating to a method for predicting carbon emissions from urban systems based on input-output analysis and Bayesian neural networks. Background Technology

[0002] With the rapid growth of fossil fuel consumption, global warming caused by greenhouse gas emissions has attracted widespread attention. Carbon dioxide (CO2) is the main contributor to the increasingly severe greenhouse effect, accounting for approximately 70% of greenhouse gases. As the world's largest energy consumer and largest carbon emitter, China faces energy shortages, excessive CO2 emissions, and related resource and ecological problems. In response to the international community's call for carbon emission reduction, China pledged at the 21st United Nations Climate Change Conference that "carbon emissions will peak by 2030, and efforts will be made to reach the peak as soon as possible." Low-carbon development has become a crucial strategic priority. Therefore, under the constraint of peak carbon emissions, the rational planning of energy consumption in urban economic development is of great significance to ensuring the sustainable development of urban energy and economy.

[0003] Currently, while existing technologies have conducted extensive analysis and experimentation in urban economy, energy, and carbon emissions, they still have certain limitations. For example, they can only analyze historical conditions and cannot predict future situations, nor can they effectively predict under conditions of multi-scenario uncertainty. Furthermore, the selection of factors is highly subjective and lacks scientific basis, resulting in calculations that are not universally applicable.

[0004] Purpose of the invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by providing a method for predicting carbon emissions from urban systems based on input-output analysis and Bayesian neural networks. This method involves the rational analysis of uncertainties across multiple scenarios and the scientific selection of influencing factors during the prediction process. As a result, a reasonable and scientific analytical and predictive model is introduced into the urban energy and economic system, which can more effectively provide technical support for the generation of decision-making solutions. Summary of the Invention

[0006] This invention provides a method for predicting carbon emissions from urban systems based on input-output analysis and Bayesian neural networks, comprising the following steps:

[0007] Step A: Determine the carbon emissions of each sector in the city, calculate the carbon emission input-output table, identify the key carbon emission sectors and major energy consumption factors, and calculate the economic share and energy consumption factors of these sectors.

[0008] Step B: Establish a city carbon emission prediction model that considers economic development and energy consumption. The specific process is as follows:

[0009] The transformation relationship between the value stream matrix and the actual logistics matrix is ​​expressed as shown in equation (1):

[0010] T+eV=eD (1);

[0011] Carbon emission intensity is expressed as shown in equation (2):

[0012] e = T(DV) -1 (2);

[0013] Carbon inflows and outflows between different sectors of the city can be represented as shown in equation (3):

[0014]

[0015] The direct emission factor is expressed as shown in equation (4):

[0016]

[0017] The network utility analysis is expressed as shown in equations (5) and (6):

[0018]

[0019] U = U 0 +U 1 +U 2 +…U l +…=(ID) -1 (6),

[0020] The network control analysis is represented as shown in equations (7), (8), and (9):

[0021] cn ji =n ji / n′ ij (7),

[0022]

[0023]

[0024] In equations (1)-(9), V is the value flow between sectors; D is the diagonal matrix of total output of each sector; T is the initial resources or waste flowing into or out of the sector; ε represents the carbon emission intensity f of each sector. ij G is the carbon dioxide flow rate from department i to department j; i in G is the total amount of carbon dioxide flowing into sector i. i out This refers to the total amount of carbon dioxide flowing out of department i; z i and y iLet represent the boundary input and output of sector i inflow and outflow from the network, respectively; γ is the direct emission coefficient; Tj is the input of sector j when the system is in a steady state; and dij is the net flow from sector i to sector j.

[0025] Step C: Based on historical data of key carbon emission sectors and energy consumption factors, as well as the city's future development plan, design three future scenarios: high, medium, and low.

[0026] Step D: Solve the coupled model, quantify the impact of different scenarios on urban carbon emissions, and determine under what development model urban carbon emissions can meet the predetermined requirements.

[0027] Preferably, the method for determining key carbon-emitting sectors and major energy-consuming factors in step A includes:

[0028] The Pearson correlation coefficient is expressed as shown in equation (10):

[0029]

[0030] The Kendall correlation coefficient is expressed as shown in equation (11):

[0031]

[0032] The Spearman correlation coefficient is expressed as shown in equation (12):

[0033]

[0034] In equations (10)-(12), x i It is the value of the independent variable, y i P is the value of the dependent variable; P is the number of logarithms of objects with two identical attribute values; n is the number of objects. It is the average value of the independent variable. It is the average value of the dependent variable.

[0035] Preferably, the process of establishing an urban carbon emission prediction model for economic development and energy consumption in step B further includes:

[0036] The objective function is set as shown in equations (13)-(15):

[0037] F(w)=βF D (w)+αF w (w) (13),

[0038]

[0039]

[0040] Among them, FD (w) is the error function, F w (w) is the weight decay term; D is the training set with N samples, D = {x} k ,t k ;k=1…N};w is the weight, including the bias value;x k The k-th input vector; y k It is the actual output of the network; t k This represents the target output; M is the weight number; t, α, and β are the coefficients of the error function, respectively.

[0041] The optimizations of α and β are shown in equations (16), (17), and (18):

[0042]

[0043]

[0044] γ=M-αTraceA -1 (18),

[0045] Where γ is the effective parameter required by the network; A is the Hessian matrix of the objective function F(W);

[0046] The posterior distribution of the weight w is expressed as shown in equation (19):

[0047]

[0048] Among them, Z M (α,β) is a normalization factor independent of w;

[0049] The output distribution of the network is represented as shown in equations (20) and (21):

[0050]

[0051]

[0052] In equations (20)-(21), w mp It is the most likely value of the weight vector in the posterior probability of S(w); y(x,w) mp ) is the average of the output values; σ t It is the standard deviation of the output value; g is the standard deviation of y(x,w) in w mp The gradient below;

[0053] The 95% confidence interval of the output equation is expressed as shown in equations (22)-(23):

[0054] y(x) low =y(x,w mp -1.96σt (twenty two),

[0055] y(x) up =y(x,w mp )+1.96σ t (twenty three), Where, y(x) low It is the lower boundary of the 95% confidence interval; y(x) up It is the upper boundary of the 95% confidence interval.

[0056] Preferably, the method for quantifying the impact of different scenarios on urban carbon emissions in step D, and determining under which development model urban carbon emissions can meet the predetermined requirements, includes:

[0057] The mean squared error (MSE) is expressed as shown in equation (24):

[0058]

[0059] The coefficient of determination R 2 It is represented as shown in equation (25):

[0060]

[0061] Where N is the number of samples; NDVI obs,i These are the observed values ​​of NDVI; NDVI sim,i It is an analog value of NDVI. Attached Figure Description

[0062] Figure 1 This is a schematic diagram of the urban system carbon emission prediction method based on input-output and Bayesian neural networks described in this invention.

[0063] Figure 2 This is a schematic diagram illustrating the carbon emission scenarios under different development scenarios in the embodiments of the present invention.

[0064] Figure 3 This is a schematic diagram of the model effect under different numbers of hidden layers in the embodiments of the present invention. Detailed Implementation

[0065] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0066] Those skilled in the art will understand that the step numbers used herein are for convenience of description only and are not intended to limit the order in which the steps are performed. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise. The terms “comprising” and “including” indicate the presence of the described features, integrals, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof. The term “and / or” refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations.

[0067] Figure 1 This is a schematic diagram of the urban system carbon emission prediction method based on input-output and Bayesian neural networks described in this invention. As shown in the figure, an urban carbon emission prediction model considering economic development and energy consumption is established, including determining the carbon emissions of each sector of the city, calculating the carbon emission input-output table, identifying key carbon emission sectors and major energy consumption factors, and calculating the economic share and energy consumption factors of these sectors.

[0068] The prediction model sequentially includes input-output description, factor screening, carbon dioxide emissions, and Bayesian neural network, while also considering sectoral development scenario input and energy consumption factor scenario input. The sectoral development scenario input includes different economic growth rates and different industrial transformation efforts; the energy consumption factor scenario input includes different energy transformation efficiencies and the development potential of clean energy.

[0069] Example

[0070] The per capita energy consumption and primary energy consumption of the major carbon-emitting sectors of the secondary industry (manufacturing and construction) and the major carbon-emitting sectors of the tertiary industry (transportation and services) in the urban system were set as high, medium, and low scenarios, with a total of 81 different scenarios. The specific changes are shown in Table 1.

[0071] Table 1. Data for the secondary and tertiary industries under three scenarios: high, medium, and low.

[0072]

[0073] Figure 2This diagram illustrates the carbon emission scenarios under different development scenarios in the embodiments. As shown, based on the method described in this invention, predictions are made to obtain the results under different development scenarios, which are categorized into lenient scenarios (peaking in 2035), normal scenarios (peaking in 2030), and severe scenarios (peaking in 2025) according to the timing of the carbon emission peak. Specifically, under the lenient scenario, Guangdong Province's carbon emission peak could reach 620 million tons, and the trough could reach 500 million tons; under the normal scenario, Guangdong Province's carbon emission peak could reach 600 million tons, and the trough could reach 350 million tons; under the severe scenario, Guangdong Province's carbon emission peak could reach 570 million tons, and the trough could reach 250 million tons.

[0074] Figure 3 This figure illustrates the performance of the prediction model under different numbers of hidden layers in this embodiment. As shown, it displays the carbon emission prediction results for 1988-2018 under four different numbers of hidden layers. (a) to (b) represent the results with 1-4 hidden layers, respectively. The results show that the model performs best when there is 1 hidden layer, and the training period R... 2 =0.99716, verification period R 2 =0.87019.

[0075] This embodiment is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting carbon emissions from urban systems based on input-output analysis and Bayesian neural networks, characterized in that, Includes the following steps: Step A: Determine the carbon emissions of each sector in the city, calculate the carbon emission input-output table, identify the key carbon emission sectors and major energy consumption factors, and calculate the economic share and energy consumption factors of these sectors. Step B: Establish a city carbon emission prediction model that considers economic development and energy consumption. The specific process is as follows: The transformation relationship between the value stream matrix and the actual logistics matrix is ​​expressed as shown in equation (1): （1）； Carbon emission intensity is expressed as shown in equation (2): （2）； Carbon inflows and outflows between different sectors of the city can be represented as shown in equation (3): （3）； The direct emission factor is expressed as shown in equation (4): （4）； The network utility analysis is expressed as shown in equations (5) and (6): （5）， （6）， The network control analysis is represented as shown in equations (7), (8), and (9): （7）， （8）， （9）， In equations (1)-(9), V is the value flow between sectors; D is the diagonal matrix of total output of each sector; T is the initial resources or waste flowing into or out of the sector; ε represents the carbon emission intensity of each sector; f ij G is the carbon dioxide flow rate from department i to department j; i in G is the total amount of carbon dioxide flowing into sector i. i out This refers to the total amount of carbon dioxide flowing out of department i; z i y represents the boundary input of sector i flowing into the network. i Represents the boundary output of sector i emanating from the network; r is the direct emission coefficient; T j It is the input quantity of system j when the system is in a steady state; d ji It is the net flow from department j to department i; The process of establishing an urban carbon emission prediction model for economic development and energy consumption, as described in step B, further includes: The objective function is set as shown in equations (13)-(15): （13）， （14）， （15）， Among them, F D (w) is the error function, F w (w) is the weight decay term; w is the weight, including the bias value; x k The k-th input vector; y k It is the actual output of the network; t k This represents the target output; M is the weight number; α and β are the coefficients of the error function, respectively. The optimizations of α and β are shown in equations (16), (17), and (18): （16）， （17）， （18）， Where γ is the network parameter; A is the Hessian matrix of the objective function F(W); The posterior distribution of the weight w is expressed as shown in equation (19): （19）， Among them, Z M (α, β) is a normalization factor independent of w; The output distribution of the network is represented as shown in equations (20) and (21): （20）， （21）， In equations (20)-(21), w mp It is the most likely value of the weight vector in the posterior probability; y(x,w) mp ) is the average of the output values; σ t It is the standard deviation of the output value; g is the standard deviation of y(x,w) in w mp The gradient below; The 95% confidence interval of the output equation is expressed as shown in equations (22)-(23): （22）， （23）， Where, y(x) low It is the lower boundary of the 95% confidence interval; y(x) up It is the upper boundary of the 95% confidence interval; Step C: Based on historical data of key carbon emission sectors and energy consumption factors, as well as the city's future development plan, design three future scenarios: high, medium, and low. Step D: Solve the coupled model, quantify the impact of different scenarios on urban carbon emissions, and determine under what development model urban carbon emissions can meet the predetermined requirements.

2. The method for predicting carbon emissions from urban systems based on input-output analysis and Bayesian neural networks according to claim 1, characterized in that, The method for identifying key carbon-emitting sectors and major energy-consuming factors described in step A includes: The Pearson correlation coefficient is expressed as shown in equation (10): （10）， The Kendall correlation coefficient is expressed as shown in equation (11): （11）， The Spearman correlation coefficient is expressed as shown in equation (12): （12）， In equations (10)-(12), x i It is the value of the independent variable, y i P is the value of the dependent variable; P is the number of logarithms of objects with two identical attribute values; n is the number of objects. It is the average value of the independent variable. It is the average value of the dependent variable.

3. The method for predicting carbon emissions from urban systems based on input-output analysis and Bayesian neural networks according to claim 1, characterized in that, The method described in step D for quantifying the impact of different scenarios on urban carbon emissions and determining under which development model urban carbon emissions can meet predetermined requirements includes: The mean squared error (MSE) is expressed as shown in equation (24): （24）， The coefficient of determination R 2 It is represented as shown in equation (25): （25）， Where N is the number of samples; NDVI obs,i These are the observed values ​​of NDVI; NDVI sim,i It is an analog value of NDVI.

Images