Method for measuring and calculating enterprise carbon emission by using electrical carbon model
By constructing and applying the electric carbon model, the accuracy of enterprise carbon emission calculations has been solved, and the carbon emission forecasting and monitoring of high-energy-consuming industries has been improved.
Patent Information
- Application Number
- CN202411851788.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-16
- Publication Date
- 2025-03-28
AI Technical Summary
The existing technology lacks effective methods to construct and apply electric carbon models for accurate calculation of enterprise carbon emissions, especially in high-energy-consuming industries.
By collecting the annual electricity consumption, energy consumption and product output data of the enterprise, the frequency conversion and sequence decomposition of the data is carried out, and an electric carbon model of carbon emissions brought by energy consumption and carbon emissions generated by the production process is selected. A model with high goodness of fit, small mean square error and small rolling prediction error is selected as the optimal model.
It realizes accurate calculation of corporate carbon emissions, provides a reference for carbon emission forecasts in high-energy consumption industries, and improves corporate carbon monitoring and verification capabilities.
Smart Images
Figure SMS_2 
Figure SMS_3 
Figure SMS_5
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of carbon emission calculation, and in particular to a method for calculating carbon emission of an enterprise by using an electric carbon model. Background Art
[0002] In order to achieve sustainable economic development, carbon emission modeling and prediction technology will become a hot topic in research. The carbon emission management of enterprises is becoming more and more important. In order to standardize the implementation of enterprise electricity carbon model monitoring and guide the reasonable implementation of enterprise electricity carbon model data collection, model construction and detection, there is no existing experience to learn from the construction and application of electricity carbon model.
[0003] Electricity is an important component of energy consumption. The production process of enterprises is closely related to electricity consumption, and electricity consumption data is highly accurate, easy to obtain, and real-time. Especially for high-energy-consuming enterprises, studying the electricity-carbon relationship of production enterprises and building an electricity-carbon model are of positive significance for carbon emission prediction of high-energy-consuming enterprises. Summary of the invention
[0004] In order to overcome the shortcomings of the prior art, the present invention proposes a method for calculating corporate carbon emissions using an electric carbon model. By constructing an electric carbon model, accurate calculation of corporate carbon emissions is achieved.
[0005] The present invention provides a method for calculating carbon emissions of an enterprise using an electric carbon model, comprising the following steps:
[0006] Step 1: Data collection: collect the company's annual electricity consumption data, energy consumption data, and product output data;
[0007] The electricity consumption data is monthly data; the energy consumption data is annual data, and at least 5 years of data should be collected; the product output data is annual data, and at least 5 years of data should be collected;
[0008] Step 2: Perform frequency transformation of data: Use the Denton model to convert all annual indicators into monthly indicators.
[0009] Step 3: Data sequence decomposition: Using the X-12-ARIMA model, the original sequence is decomposed into S sequence, TC sequence and TCI sequence.
[0010] Step 4: Conduct unit root test and cointegration test on the enterprise's monthly total energy consumption data, monthly output data and monthly electricity consumption data.
[0011] Step five, model construction, guided by the long-term equilibrium model theory, construct an electric-carbon model of carbon emissions caused by energy consumption and an electric-carbon model of carbon emissions generated in the production process.
[0012] Step six, selection of electric carbon model: using goodness of fit (R2), mean square error (MSE), and model rolling prediction error as judgment criteria, compare and screen the electric carbon models of carbon emissions caused by the above energy consumption and the electric carbon models of carbon emissions generated by the production process, and finally select the model with higher goodness of fit, smaller MSE, and smaller rolling prediction error as the optimal model.
[0013] Step 7: Utilize the selected optimal model and combine the monthly electricity consumption data of the enterprise to calculate the monthly total energy consumption data and the monthly output data of the enterprise.
[0014] Step eight, multiply the calculated monthly total energy consumption data of the enterprise by the carbon emission coefficient of standard coal to obtain the monthly carbon emission data of the enterprise; multiply the calculated monthly output data of the enterprise by the carbon emission parameters of the enterprise's production process to obtain the monthly carbon emission data of the enterprise's production process.
[0015] The step 2 is specifically as follows: taking the monthly electricity consumption of the enterprise as a reference, the annual energy consumption of the enterprise is converted into the monthly energy consumption of the enterprise by using the Denton model; similarly, taking the monthly electricity consumption of the enterprise as a reference, the annual output data of the enterprise is converted into the monthly output data of the enterprise by using the Denton model;
[0016] The step three is to use the X-12-ARIMA model to decompose the monthly electricity consumption data of the enterprise into the enterprise's monthly electricity consumption TC series, the enterprise's monthly electricity consumption TCI series, and the enterprise's monthly electricity consumption S series; similarly, the X-12-ARIMA model is used to decompose the monthly energy consumption data of the enterprise into the enterprise's monthly energy consumption TC series, the enterprise's monthly energy consumption TCI series, and the enterprise's monthly energy consumption S series; the X-12-ARIMA model is used to decompose the monthly output data of the enterprise into the enterprise's monthly output TC series, the enterprise's monthly output TCI series, and the enterprise's monthly output S series.
[0017] In the above step 4, the specific construction of the electric carbon model of carbon emissions caused by energy consumption is as follows: guided by the theory of long-term equilibrium model, considering the large fluctuation of enterprise data, the TC series (Y1_TC) of the enterprise's monthly energy consumption is selected to model the TC series (X_TC) of the enterprise's monthly electricity consumption. The specific model is expressed as follows:
[0018] B(L)(Y1_TC t )=β0+A(L)(X_TC t )+ε t ,B(L)=1-γ1L-...-γ4L 4 ,A(L)=α0+α1L+...+α4L 4 .
[0019] Here Y1_TC t 、X_TC t are the TC series of the enterprise energy consumption series and the TC series of the enterprise monthly electricity consumption series, L is the lag operator, B(L) and A(L) are the polynomials of the lag operator L, β0, γ i (i=1,...,4), α j (j=0,1,...,4) is the parameter to be estimated, ε t is white noise.
[0020] In the above step 4, the electric carbon model of carbon emissions generated in the production process is constructed as follows: guided by the theory of long-term equilibrium model, considering that the data of enterprises fluctuate greatly, the TC series (Y2_TC) of the monthly output of enterprises is used to model the TC series (X_TC) of the monthly electricity consumption of enterprises. The specific model is expressed as follows:
[0021] B(L)(Y2_TC t )=β0+A(L)(X_TC t )+ε t ,B(L)=1-γ1L-...-γ4L 4 ,A(L)=α0+α1L+...+α4L 4 .
[0022] Here Y2_TC t 、X_TC t are the TC sequence of the enterprise product output sequence and the TC sequence of the enterprise monthly electricity consumption sequence, L is the lag operator, B(L) and A(L) are the polynomials of L of the lag operator, β0, γ i (i=1,...,4), α j (j=0,1,...,4) is the parameter to be estimated, ε t is white noise.
[0023] The enterprise electricity-carbon model is a more general model form, which is used as an alternative model in the future. The specific model is expressed as follows:
[0024] B(L)(Y_TC t )=β0+A(L)(X_TC t )+ε t ,B(L)=1-γ1L-...-γ p L p ,A(L)=α0+α1L+...+α q L q .
[0025] where Y_TC t is the dependent variable, X_TC tis the independent variable, L is the lag operator, B(L) and A(L) are the polynomials of the lag operator L, β0, γ i (i=1,...,p), α j (j=0,1,...,q) is the parameter to be estimated, ε t is white noise.
[0026] The carbon emission parameters mentioned above are derived from the IPCC (2006, 2019) standards and the latest national or provincial standards.
[0027] The beneficial effects of the present invention are: from a technical perspective, it provides reference and inspiration for other high-energy-consuming industries to calculate the carbon emissions of the industry based on power big data; from an application perspective, it provides application guidance for the carbon emission calculation technology and implementation of high-energy-consuming industries; and it provides reference and inspiration for carbon monitoring and carbon verification of enterprises. DETAILED DESCRIPTION
[0028] Example 1, the present invention provides a method for calculating carbon emissions generated during the energy consumption process of an enterprise using an electric carbon model,
[0029] Step 1: Data collection: collect the annual electricity consumption data and energy consumption data of the enterprise;
[0030] Step 2: Perform frequency transformation of data: Use the Denton model to convert all annual indicators into monthly indicators.
[0031] Step 3: Data sequence decomposition: Using the X-12-ARIMA model, the original sequence is decomposed into S sequence, TC sequence and TCI sequence.
[0032] Step 4: Conduct unit root test and cointegration test on the monthly total energy consumption data and monthly electricity consumption data of enterprises.
[0033] Step five: Model construction: guided by the long-term equilibrium model theory, construct an electric-carbon model of carbon emissions caused by energy consumption.
[0034] Step six, selection of electricity-carbon model: using goodness of fit (R2), mean square error (MSE), and model rolling prediction error as judgment criteria, compare and screen the electricity-carbon models of carbon emissions caused by the above energy consumption, and finally select the model with higher goodness of fit, smaller MSE, and smaller rolling prediction error as the optimal model.
[0035] Step 7: Calculate the monthly total energy consumption data by using the selected best model and combining it with the company's monthly electricity consumption data.
[0036] Step 8: Multiply the calculated monthly total energy consumption data of the enterprise by the carbon emission coefficient of standard coal to obtain the monthly carbon emission data of the enterprise.
[0037] The second step is specifically as follows: taking the enterprise's monthly electricity consumption as a reference, converting the enterprise's annual energy consumption into the enterprise's monthly energy consumption using the Denton model.
[0038] The step three is to use the X-12-ARIMA model to decompose the monthly electricity consumption data of the enterprise into the enterprise monthly electricity consumption TC series, the enterprise monthly electricity consumption TCI series, and the enterprise monthly electricity consumption S series; similarly, the monthly energy consumption data of the enterprise is decomposed into the enterprise monthly energy consumption TC series, the enterprise monthly energy consumption TCI series, and the enterprise monthly energy consumption S series by using the X-12-ARIMA model;
[0039] In the above step 4, the specific construction of the electric carbon model of carbon emissions caused by energy consumption is as follows: guided by the theory of long-term equilibrium model, considering the large fluctuation of enterprise data, the TC series (Y1_TC) of the enterprise's monthly energy consumption is selected to model the TC series (X_TC) of the enterprise's monthly electricity consumption. The specific model is expressed as follows:
[0040] B(L)(Y1_TC t )=β0+A(L)(X_TC t )+ε t ,B(L)=1-γ1L-...-γ4L 4 ,A(L)=α0+α1L+...+α4L 4 .
[0041] Here Y1_TC t 、X_TC t are the TC series of the enterprise energy consumption series and the TC series of the enterprise monthly electricity consumption series, L is the lag operator, B(L) and A(L) are the polynomials of the lag operator L, β0, γ i (i=1,...,4), α j (j=0,1,...,4) is the parameter to be estimated, ε t is white noise.
[0042] Example 2, the present invention provides a method for calculating carbon emissions generated in the production process of an enterprise using an electric carbon model, comprising the following steps:
[0043] Step 1: Data collection: collect the annual electricity consumption data and product output data of the enterprise;
[0044] Step 2: Perform frequency transformation of data: Use the Denton model to convert all annual indicators into monthly indicators.
[0045] Step 3: Data sequence decomposition: Using the X-12-ARIMA model, the original sequence is decomposed into S sequence, TC sequence and TCI sequence.
[0046] Step 4: Conduct unit root test and cointegration test on the company's monthly output data and monthly electricity consumption data.
[0047] Step five, model construction, guided by the long-term equilibrium model theory, constructs an electric carbon model of carbon emissions generated in the production process.
[0048] Step six, selection of electric-carbon model: using goodness of fit (R2), mean square error (MSE) and model rolling prediction error as judgment criteria, compare and screen the electric-carbon models of carbon emissions generated by the above production process, and finally select the model with higher goodness of fit, smaller MSE and smaller rolling prediction error as the optimal model.
[0049] Step seven, using the selected best model and combining the monthly electricity consumption data of the enterprise, calculate the monthly output data of the enterprise.
[0050] Step eight, multiplying the calculated monthly output data of the enterprise by the carbon emission parameters of the enterprise's production process to obtain the carbon emission data of the enterprise's monthly production process.
[0051] The step 2 is specifically as follows: taking the monthly electricity consumption of the enterprise as a reference, converting the annual production data of the enterprise into the monthly production data of the enterprise using the Denton model;
[0052] The step three is to use the X-12-ARIMA model to decompose the monthly electricity consumption data of the enterprise into the enterprise's monthly electricity consumption TC series, the enterprise's monthly electricity consumption TCI series, and the enterprise's monthly electricity consumption S series; similarly, the X-12-ARIMA model is used to decompose the enterprise's monthly output data into the enterprise's monthly output TC series, the enterprise's monthly output TCI series, and the enterprise's monthly output S series.
[0053] In the above step 4, the electric carbon model of carbon emissions generated in the production process is constructed as follows: guided by the theory of long-term equilibrium model, considering that the data of enterprises fluctuate greatly, the TC series (Y2_TC) of the monthly output of enterprises is used to model the TC series (X_TC) of the monthly electricity consumption of enterprises. The specific model is expressed as follows:
[0054] B(L)(Y2_TC t )=β0+A(L)(X_TC t )+ε t ,B(L)=1-γ1L-...-γ4L 4 ,A(L)=α0+α1L+...+α4L 4 .
[0055] Here Y2_TC t 、X_TC t are the TC series of the enterprise product output series and the TC series of the enterprise monthly electricity consumption series, L is the lag operator, B(L) and A(L) are the polynomials of L of the lag operator, β0, γ i (i=1,...,4), α j (j=0,1,...,4) is the parameter to be estimated, ε t is white noise.
[0056] Adding the company's monthly energy consumption carbon emission data to the monthly production process carbon emission data will yield the company's high-frequency carbon emission calculation data, which is the company's total carbon emissions.
[0057] Logical verification: Annual total verification. The calculated monthly total energy consumption data of enterprises are summed up to obtain the annual data, which is verified with the annual total energy consumption data of enterprises from official statistics or third-party verification. The error rate is controlled within 5%, and the calculation result verification effect is good.
[0058] Intensity verification: The carbon emission intensity of companies of similar size in the industry is compared, and the error rate is controlled within 5%, and the calculation results are well verified.
[0059] Denton model: A frequency conversion model that decomposes a low-frequency data sequence into a high-frequency data sequence to further establish a related time series model.
[0060] Denton model data splitting algorithm:
[0061] Let y=(y1,y2,…,y N )' is a known low-frequency sequence, the number of which is N, that is, y k ,(k=1,2,…,N). Let w=(w1,w2,…,w T )' is the high-frequency sequence to be estimated, and the number of data is T, that is, w t , (t=1,2,…,T). Let z=(z1,z2,…,z T )' is a known high-frequency sequence related to sequence w, and the number of its data is T, that is, z t ,(t=1,2,…,T).
[0062] Assume that each data in the low-frequency sequence can be divided into j segments when converted to a high-frequency sequence, that is, T = jN. Among them, when the sequence y is converted from annual to quarterly frequency, j = 4; when the sequence y is converted from annual to monthly frequency, j = 12; when the sequence y is converted from quarterly to monthly frequency, j = 3;
[0063] The matrix form of the constraint condition can be expressed as: y = Bw, where B is an N × T-order weight matrix, which is the expression The expression form of the weight n in , its specific form is shown in formula (1). In particular, when the conversion sequence is a stock sequence, the form of the weight matrix is formula (2).
[0064]
[0065]
[0066] By solving the minimization of P(w,z), we can get the high-frequency data sequence This method is also called the Denton ratio method.
[0067]
[0068] in is a T×T matrix based on the correlation sequence z; and in general, the matrix The specific form of is as shown in formula (4). In particular, when the high-frequency data sequence z related to w is not used, It is a T-order diagonal matrix with elements equal to 1. In addition, D is a (T-1)×T-order matrix used to calculate the first-order difference of a vector, and its specific form is shown in formula (6).
[0069]
[0070] Using Denton's method to estimate high frequency sequences When , it can be transformed into the problem of minimizing equation (3) under the constraint y = Bw, by changing the equation relative to w t The partial derivatives of and the Lagrange multipliers of the system are equal to zero to solve. That is:
[0071]
[0072] Finally, through simple inverse operations and multiplication operations of the involved matrices, a vector containing the target estimation sequence can be obtained, as shown in formula (7), thereby estimating the target high-frequency data sequence
[0073]
[0074] Where 0 N is an N-order zero matrix, Bz is a T×1-dimensional vector, and λ is an N-dimensional vector.
[0075] Seasonal adjustment: A method of estimating and decomposing seasonal factors from time series data. The purpose is to adjust the seasonal impact so that the established statistical model can better reveal the characteristics and trends of the series.
[0076] X-12-ARIMA model: an autoregressive combined moving average model for seasonal adjustment. It introduces the idea of stochastic modeling, estimates seasonal factors in data series, and has a certain forecasting ability. Compared with X-11, it adds several model forms and seasonal adjustment test methods.
[0077] X-12-ARIMA model seasonal adjustment algorithm:
[0078] X-12-ARIMA uses the signal-to-noise ratio method to select between fixed sets of moving average filters (commonly referred to as X-11 type filters). The theoretical principle of X-12-ARIMA is still the sliding average method, but the combination of its model and use is more complete, and it also includes a variety of new diagnostic methods, which can help users discover and correct irregular factors under the influence of seasonal and calendar factors, and adjust different time series by selecting different programs, so as to better improve the fit of the sequence end values and breakpoints (generally caused by special time points such as holidays and stock trading days) and improve the prediction effect. The main steps are as follows:
[0079] The first step is to make monthly adjustments. Considering that the actual number of working days in each month is different, it is necessary to construct a monthly adjustment factor, and use this adjustment factor to adjust the original sequence to eliminate the difference in the number of working days in a given month between different years caused by holidays or other reasons. This adjustment is only for monthly data. There are two points to note here: First, the monthly adjustment actually distinguishes between the working days of the week (Monday to Friday) and the rest days (Saturday and Sunday), and at the same time deducts the impact of the number of days of other holidays in each month, and then summarizes the number of working days in a month. Therefore, the monthly adjustment can be regarded as a preliminary adjustment of both the working day effect and the holiday effect; second, the monthly adjustment is an adjustment made before the formal repeated moving average adjustment of X-12-ARIMA, so it can be regarded as a pre-adjustment. If this adjustment is made, the user needs to provide a monthly adjustment factor sequence P (PriorMonthly Adjustment Factors), where P = {p1, p2, …, p n}, where p j is the adjustment factor for month j, and n is the total number of months. Under the multiplicative model, for the time series Y = {y1, y2, ..., y n} to make monthly adjustments, and the monthly adjustment factor is P = {p1, p2, ..., p n}, is the adjusted sequence, then:
[0080]
[0081] The P sequence can be the actual working days sequence of each month after deducting holidays, rest days or shutdown days, so that the adjusted monthly The sequence is the monthly average daily value sequence of working days. This adjustment method is called the monthly average daily value method. The monthly average daily value method is only suitable for the correction of the growth rate sequence.
[0082] The second step is to adjust trading days. Trading day adjustment is related to the different number of days from Monday to Sunday in each month of the year. In addition to the four weeks in each month, there are usually 1 to 3 more days, which means that there are always at least 4 Mondays, 4 Tuesdays, ..., 4 Sundays in each month, but some days will appear 5 times. For example, since there are 5 weekends in January instead of the usual 4 weekends, for the retail industry, there may be more commodity turnover and consumer expenditure in January, which will affect the changes in the commodity retail price index and the consumer price index. Since the weekly working day adjustment can be regarded as a special case of "trading day adjustment", the two are usually considered together when making seasonal adjustments. Trading day adjustment takes into account the different impacts of weekend leisure days and Monday to Friday working days on economic activities, and uses qualitative subjective judgment or quantitative regression analysis to obtain the impact weights of weekend leisure days and ordinary working days, so as to correct irregular factors. In general, there are several methods for trading day adjustment, namely, a priori trading day adjustment, trading day adjustment based on regression analysis, Young (1965) trading day adjustment model, Findley et al. (1998) trading day adjustment model, and TRAMO trading day adjustment model. Due to space limitations, these sub-type models will not be introduced in detail here.
[0083] The third step is to calculate the final modified weights of the weekly working day variation factors and idiosyncratic items. This part uses the previously calculated weekly working day adjusted sequence (if no weekly working day adjustment is made) to calculate the final modified weights of the weekly working day variation factors and idiosyncratic items.
[0084] The fourth step is to calculate the final seasonal elements, seasonally adjusted series, trend and cycle elements, and irregular elements. This part calculates the final seasonal elements, seasonally adjusted series, trend and cycle elements, and irregular elements. In the seasonal adjustment method of the X series, every time new data is added, all past data must be seasonally adjusted, and the seasonally adjusted series must be revised by looking back at several years of historical data. However, since it is not only troublesome to perform such seasonal adjustment calculations and revise past seasonally adjusted data every month, and it also causes changes in past monthly data, seasonal adjustment is generally performed once a year. Since the seasonal change elements of the next year are not yet known, the X-11 method simply calculates the predicted values of the seasonal elements of the next year, and uses this temporary predicted value to perform seasonal adjustment for the next year. When a full year of data is collected, when using X-11 for seasonal adjustment at the beginning of the next year, the collected data is used to revise the past seasonally adjusted series, and the revised series is used to calculate the predicted values of the seasonal change elements of the next year.
[0085] The fifth step is a concise statistical analysis of the results. In X-11, a set of concise statistical analysis results are given for various decomposition factors, so that users can use them to analyze the characteristics of economic indicators and the effects of seasonal adjustments. Generally speaking, the simplest method is for users to calculate the mean and standard deviation of the rate of change (quantity) of the interval j months (quarters) in X-11 to compare the effects before and after adjustment. The following is a concise statistical analysis method for reference in technical implementation. Here, the rate of change sequence compared to (difference) of the previous month (quarter) is calculated. First, calculate the rate of change sequence of the original sequence Y compared to (difference) of the previous month (quarter):
[0086]
[0087] (R Y ) i =Y i -Y i-1 (i=1,2,…,n) (addition model)
[0088] Next, calculate the rate of change of the seasonally adjusted TCI series to the previous month (season) (difference) series R TCI :
[0089]
[0090] (R TCI ) i =TCI i -TCI i-1 (i=1,2,…,n) (addition model)
[0091] Finally, compare R Yand R TCI Generally speaking, the latter of these two change rate series should have a smaller change range than the former.
[0092] Unit root test: A test method in time series analysis, used to test whether a time series has a unit root, that is, to test the stationarity of the time series.
[0093] Perform unit root test on the sequence to determine the stationarity of the sequence. Unit root test is a problem of random process. Define a random sequence, t=1,2,… is a unit root process, if
[0094] x t =ρx t-1 +εt=1,2...
[0095] where |ρ|<1, {ε t} is a stationary series (white noise), and
[0096] E[ε t ]=0,V(ε t )=σ<∞,Cov(ε t ,ε τ )=μ<∞Here t, τ=1,2…. In particular, if ρ=1, the above formula becomes a random walk sequence, so the random walk sequence is the simplest unit root process. Rewrite the definition into the following form: (1-ρL)x t =ε t , t=1,2,… where L is the lag operator, 1-ρL is the lag operator polynomial, and its characteristic equation is 1-ρz=0, with root z=1 / ρ. When ρ=1, the time series has a unit root, and at this time {x t} is a unit root process. When ρ<1, {x t} is a stationary sequence. When ρ>1, {x t} is a type of non-stationary process with the so-called exploding root. It is still a non-stationary process after differencing, so it is not a single integral process.
[0097] Data cointegration test: A test method in time series analysis used to test whether there is a long-term stable relationship between two or more time series.
[0098] EG two-step data cointegration test: Generally speaking, for two non-stationary sequences, the cointegration test needs to be passed before long-term equilibrium relationship modeling can be performed. The EG two-step method is a relatively simple cointegration test method, which tests whether the two variables Yt and Xt are cointegrated, referred to as the EG test. The main steps are as follows:
[0099] In the first step, the OLS method is used to estimate the equation Yt = α0 + α1Xt + μt and calculate the unbalanced error, obtaining:
[0100]
[0101] The second step is to test the integrity of et. If et is a stable sequence, then the variables Yt and Xt are considered to be (1,1) order cointegrated; if et is 1st order cointegrated, then the variables Yt and Xt are considered to be (2,1) order cointegrated. The integrity test of et can be done by ADF or DF test.
[0102] Granger Causality Test:
[0103] The Granger causality test assumes that all the information about the prediction of each variable, y and x, is contained in the time series of these variables. The test requires estimating the following regression:
[0104]
[0105] The test is discussed in four cases: First, x is the cause of the change in y, that is, there is a one-way causal relationship from x to y. If the coefficient estimate of the lagged x in the first equation is statistically significantly non-zero, and the coefficient estimate of the lagged y in the second equation is statistically significantly zero, then x is said to be the cause of the change in y. Second, y is the cause of the change in x, that is, there is a one-way causal relationship from y to x. If the coefficient estimate of the lagged y in the second equation is statistically significantly non-zero, and the coefficient estimate of the lagged x in the first equation is statistically significantly zero, then y is said to be the cause of the change in x. Third, x and y are mutually causal, that is, there is a one-way causal relationship from x to y, and there is also a one-way causal relationship from y to x. If the coefficient estimate of the lagged x in the first equation is statistically significantly non-zero, and the coefficient estimate of the lagged y in the second equation is statistically significantly non-zero, then there is a feedback relationship or a two-way causal relationship between x and y. Fourth, x and y are independent, or there is no causal relationship between x and y. If the coefficient estimate of the lagged x in the first equation is statistically significantly zero overall, and the coefficient estimate of the lagged y in the second equation is statistically significantly zero overall, then there is no causal relationship between x and y.
Claims
1. A method for calculating corporate carbon emissions using an electric carbon model, characterized in that: The following steps are involved: Step 1: Data collection: collect the company's annual electricity consumption data, energy consumption data, and product output data; Step 2: Perform frequency transformation of data: Use the Denton model to transform all annual indicators into monthly indicators; Step 3: Data sequence decomposition: Using the X-12-ARIMA model, the original sequence is decomposed into S sequence, TC sequence and TCI sequence; Step 4: Conduct unit root test and cointegration test on the monthly total energy consumption data, monthly output data and monthly electricity consumption data of enterprises; Step 5: Model construction: guided by the long-term equilibrium model theory, construct an electric carbon model for carbon emissions caused by energy consumption and an electric carbon model for carbon emissions generated by the production process; Step 6: Selection of electric carbon model: Using goodness of fit, mean square error, and model rolling prediction error as the judgment criteria, compare and screen the electric carbon model of carbon emissions caused by the above energy consumption and the electric carbon model of carbon emissions generated by the production process, and finally select the model with higher goodness of fit, smaller MSE, and smaller rolling prediction error as the optimal model; Step 7: Calculate the monthly total energy consumption data and the monthly output data of the enterprise by using the selected best model and combining it with the monthly electricity consumption data of the enterprise; Step eight, multiply the calculated monthly total energy consumption data of the enterprise by the carbon emission coefficient of standard coal to obtain the monthly carbon emission data of the enterprise; multiply the calculated monthly output data of the enterprise by the carbon emission parameters of the enterprise's production process to obtain the monthly carbon emission data of the enterprise's production process.
2. The method for calculating corporate carbon emissions using an electric carbon model according to claim 1, characterized in that: The step 2 is specifically as follows: taking the enterprise's monthly electricity consumption as a reference, the enterprise's annual energy consumption is converted into the enterprise's monthly energy consumption using the Denton model; similarly, taking the enterprise's monthly electricity consumption as a reference, the enterprise's annual output data is converted into the enterprise's monthly output data using the Denton model.
3. The method for calculating corporate carbon emissions using an electric carbon model according to claim 2, characterized in that: The step three is to use the X-12-ARIMA model to decompose the monthly electricity consumption data of the enterprise into the enterprise's monthly electricity consumption TC series, the enterprise's monthly electricity consumption TCI series, and the enterprise's monthly electricity consumption S series; similarly, the X-12-ARIMA model is used to decompose the monthly energy consumption data of the enterprise into the enterprise's monthly energy consumption TC series, the enterprise's monthly energy consumption TCI series, and the enterprise's monthly energy consumption S series; the X-12-ARIMA model is used to decompose the monthly output data of the enterprise into the enterprise's monthly output TC series, the enterprise's monthly output TCI series, and the enterprise's monthly output S series.
4. The method for calculating corporate carbon emissions using an electric carbon model according to claim 3, characterized in that: In the above step 4, the specific construction of the electric carbon model of carbon emissions caused by energy consumption is as follows: guided by the theory of long-term equilibrium model, considering the large fluctuation of enterprise data, the TC series (Y1_TC) of the enterprise's monthly energy consumption is selected to model the TC series (X_TC) of the enterprise's monthly electricity consumption. The specific model is expressed as follows: B(L)(Y1_TC t )=β0+A(L)(X_TC t )+e t ,B(L)=1-γ1L-...-γ4L 4 ,A(L)=α0+α1L+...+α4L 4 . Here Y1_TC t 、X_TC t are the TC series of the enterprise energy consumption series and the TC series of the enterprise monthly electricity consumption series, L is the lag operator, B(L) and A(L) are the polynomials of the lag operator L, β0, γ i (i=1,...,4), α j (j=0,1,...,4) is the parameter to be estimated, ε t is white noise.
5. The method for calculating corporate carbon emissions using an electric carbon model according to claim 3, characterized in that: In the above step 4, the electric carbon model of carbon emissions generated in the production process is constructed as follows: guided by the theory of long-term equilibrium model, considering that the data of enterprises fluctuate greatly, the TC series (Y2_TC) of the monthly output of enterprises is used to model the TC series (X_TC) of the monthly electricity consumption of enterprises. The specific model is expressed as follows: B(L)(Y2_TC t )=β0+A(L)(X_TC t )+e t ,B(L)=1-γ1L-...-γ4L 4 ,A(L)=α0+α1L+...+α4L 4 . Here Y2_TC t 、X_TC t are the TC sequence of the enterprise product output sequence and the TC sequence of the enterprise monthly electricity consumption sequence, L is the lag operator, B(L) and A(L) are the polynomials of L of the lag operator, β0, γ i (i=1,...,4), α j (j=0,1,...,4) is the parameter to be estimated, ε t is white noise.