Carbon emission data analysis method

Abstract

The invention provides a carbon emission data quality analysis method which comprises the following steps: acquiring various activity level data influencing product / enterprise carbon emission, calculating carbon emission intensity, longitudinally comparing the carbon emission intensity with historical data and transversely comparing the carbon emission intensity with an industry reference value, and acquiring first quality evaluation data of the carbon emission activity level data; according to the production / business logic of the enterprise, analyzing the correlation of the activity level data in the production process, and obtaining second quality evaluation data of the carbon emission activity level data; constructing a multiple regression model, performing multiple regression analysis on correlation between molecules and denominators of the carbon emission intensity indexes and noise influence on time, places, processes / technologies and quality of the carbon emission intensity indexes, and obtaining third quality evaluation data of the carbon emission activity level data; and the authenticity of the carbon emission data is judged according to the first, second and third quality evaluation data, so that the carbon emission data quality can be improved, and disordered filling, wrong filling and counterfeiting of the data are avoided.

Description

Technical Field

[0001] The present invention relates to the field of carbon emission data analysis, and in particular to a carbon emission data analysis method. Background Art

[0002] In today's era of energy conservation and emission reduction, carbon emissions are a hotly debated issue, including the authenticity of carbon emissions data.

[0003] Accurate carbon emission data is an important basis for the country to formulate and evaluate emission reduction policies. It can also encourage enterprises to take more effective emission reduction measures and promote technological innovation and industrial upgrading. The authenticity of carbon emission data is closely related to market credit and the quality of corporate carbon assets. False data will damage the reputation and market trust of enterprises and make it difficult to achieve emission reduction targets.

[0004] Therefore, it is necessary to strengthen the supervision and verification of carbon emission data to ensure its authenticity and accuracy. Summary of the invention

[0005] In order to overcome the defects existing in the above-mentioned prior art, an object of the present invention is to provide a carbon emission data quality analysis method.

[0006] In order to achieve the above-mentioned object of the present invention, the present invention provides a carbon emission data quality analysis method, comprising the following steps:

[0007] Obtain data on the activity levels that affect product / enterprise carbon emissions, calculate carbon emission intensity, compare the carbon emission intensity with historical data vertically, and compare it with industry benchmarks horizontally to obtain first quality evaluation data on carbon emission activity levels;

[0008] According to the production / business logic of the enterprise, analyze the correlation of various activity level data in the production process and obtain the second quality evaluation data of carbon emission activity level data;

[0009] Construct a multivariate regression model and conduct a multivariate regression analysis on the influence of time, location, process / technology, and quality noise on carbon emission intensity to obtain the third quality evaluation data of carbon emission activity level data;

[0010] When the first quality evaluation data, the second quality evaluation data, and the third quality evaluation data all exceed their corresponding threshold ranges, it is determined that the carbon emission data is incorrect.

[0011] In an optional scheme of the carbon emission data analysis method, the carbon emission intensity is the ratio of carbon emissions to production, or the carbon emissions per kilowatt-hour of the battery, or the carbon emissions per unit mass of output.

[0012] In an optional scheme of the carbon emission data analysis method, the carbon emission intensity of the past n years is calculated by the moving average method, n ≥ 3, and the annual and monthly data are compared with it longitudinally;

[0013] The carbon emission intensity of the past m months is calculated using the moving average method, where m ≥ 12, and the monthly data is compared with it longitudinally;

[0014] For products or production links with the same process / technology, calculate the carbon emission intensity of other enterprises / products with the same process / technology in the past n years as the carbon emission intensity reference value, or import the industry benchmark value and compare the carbon emission intensity of the past n years with the carbon emission intensity reference value or industry benchmark value;

[0015] If both the longitudinal comparison and the lateral comparison results exceed the corresponding setting ranges, the first quality evaluation data exceeds its setting range.

[0016] In an optional solution of the carbon emission data analysis method, the volatility tolerance value compared with the moving average is set between 15% and 35%;

[0017] Or, the volatility tolerance value can be calculated based on the standard deviation of the variable data after removing the upper and lower 1% extreme values;

[0018] Or, the modification can be carried out based on test verification on a large number of enterprises, that is, the overall test pass rate of enterprises under this fault tolerance rate can be seen.

[0019] In an optional scheme of the carbon emission data analysis method, the linear correlation between carbon emissions and production is analyzed. When carbon emissions and production have a linear relationship, linear regression is used to fit the linear relationship between production and carbon emissions, and the linear trend of carbon emission intensity is tested. If the carbon emission intensity of a certain period deviates from the linear trend, the second quality evaluation data exceeds the set range;

[0020] If there is no linear correlation, then analyze whether there is a monotonic relationship between carbon emissions and production. If there is, fit a nonlinear curve and use the curve to evaluate the correlation between carbon emissions and production. If the carbon emission intensity in a certain period deviates from the nonlinear curve, the second quality evaluation data exceeds the set range;

[0021] Analyze the correlation between carbon emissions and production in the time dimension to obtain the trend of the correlation between carbon emissions and production over time; if the carbon emission intensity in a certain period deviates from the trend of the correlation between carbon emissions and production over time, the second quality evaluation data exceeds the set range.

[0022] In an optional scheme of the carbon emission data analysis method, the linear correlation between carbon emission and production is determined according to the following formula:

[0023]

[0024] Among them, x i represents the production volume or unit production volume at the i-th time point, y i Represents x i The corresponding carbon emissions, and are the means of production and carbon emissions, respectively;

[0025] If r is close to 1, carbon emissions are positively correlated with production; if r is close to -1, carbon emissions are negatively correlated with production; if r is close to 0, there is no obvious linear relationship between carbon emissions and production.

[0026] In an optional scheme of the carbon emission data analysis method, it is verified whether there is a monotonic relationship between carbon emission and production according to the following formula:

[0027]

[0028] Among them, d i is the rank difference of the ith pair of carbon emission and production data, and n is the number of carbon emission or production data.

[0029] In an optional scheme of the carbon emission data analysis method, the steps of analyzing the correlation between carbon emission and production in the time dimension are:

[0030] Arrange the data of carbon emissions and production in time series;

[0031] The sliding window technique was used to calculate the correlation coefficients for different periods;

[0032] Draw a sliding correlation diagram to analyze the trend of the correlation between carbon emissions and production over time.

[0033] In an optional scheme of the carbon emission data analysis method, a multiple regression model is constructed and multiple regression analysis is used to explain the changes in carbon emission intensity by introducing multiple independent variables;

[0034] The independent variables include: production time, production location, process / technology, and product quality;

[0035] The regression model is as follows:

[0036] I carbon =β 0 +β 1 P time +β 2 P location +β 3 P tech +β 4 Pquality +∈

[0037] Among them, I carbon represents carbon emission intensity; P time is the production time factor; P location is the production location factor; P tech It is a process / technical factor; P quality is the product quality factor; ε is the error term, which is used to explain the unobservable noise and is a constant; β 0 , β 1 , β 2 , β 3 , β 4 All are weights.

[0038] In an optional solution of the carbon emission data analysis method, the multivariate regression model is optimized by using one or any combination of the following:

[0039] Check for multicollinearity: In a multiple regression model, test and analyze the collinearity or correlation between independent variables; if there is a collinearity problem, use the variance inflation factor to evaluate it and remove highly collinear variables; and / or use regularized regression methods to reduce the impact of collinearity by adding penalty terms and stabilize the regression coefficients of the model;

[0040] Perform residual analysis: Analyze the residuals of the multivariate regression model, evaluate the model's fitting effect, and identify possible outliers or noise effects; if abnormal residuals that do not conform to the normal distribution or have significant serial correlation are found, further optimize the model or use robust regression methods;

[0041] Interaction analysis: Add interaction terms to the regression model.

[0042] The beneficial effects of the present invention are:

[0043] The present invention uses a variety of comprehensive methods to monitor product / enterprise data, comprehensively and multi-dimensionally grasp and test data logic to improve data quality control and avoid random filling, wrong filling, and falsification of data. These analysis tools and visualization methods help enterprises and competent departments to quickly identify potential problems in data and optimize related business and management. Specifically, it has the following effects:

[0044] 1. Improve data quality control: Through the automated ratio calculation and monitoring system, ensure that the data reported by the enterprise during the production process is accurate. The present invention effectively improves the accuracy and traceability of data through comparative analysis of historical data and industry standards.

[0045] 2. Reduce carbon emission risks: Through real-time monitoring of carbon emission intensity, enterprises can promptly detect abnormal emission behaviors and take measures to reduce carbon emissions, comply with environmental protection standards and reduce environmental protection costs.

[0046] 3. Improve production efficiency: The system automatically generates analysis reports and provides trend analysis, which can help enterprises optimize production processes, reduce resource waste, and improve overall production efficiency.

[0047] 4. Support decision-making optimization: The present invention provides horizontal and vertical data comparison to help enterprises conduct self-assessment and optimization adjustments in the same industry, while enhancing the transparency of the production process.

[0048] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. DETAILED DESCRIPTION

[0049] The embodiments of the present invention are described in detail below.

[0050] The present invention provides a carbon emission data quality analysis method, which can be used to supervise and verify the authenticity of carbon emission data of products and enterprises, and specifically includes the following steps:

[0051] 1. Obtain data on the levels of various activities that affect the carbon emissions of products / enterprises and calculate the carbon emission intensity. Generally speaking, carbon emission intensity is the ratio of carbon emissions to production, that is,

[0052]

[0053] Among them, L carbon Represents carbon emission intensity, C cmission is carbon emissions, P production For production volume.

[0054] However, there are differences in different products and different situations. For example, for batteries, carbon emissions per kilowatt-hour are often used. In addition, carbon emissions per unit of output mass are also used as carbon emission intensity indicators.

[0055] The calculated carbon emission intensity is compared vertically with historical data and horizontally with industry benchmark values ​​to obtain first-quality evaluation data on carbon emission activity levels.

[0056] Specifically, this embodiment adopts the moving average method to calculate the carbon emission intensity of the past n years, n≥3, and compares the annual and monthly data with it vertically; the moving average method is used to calculate the carbon emission intensity of the past m months, m≥12, and compares the monthly data with it vertically.

[0057] For products or production links with the same process / technology, the carbon emission intensity of other companies / products using this process / technology in the past n years is used as a reference value for carbon emission intensity, or the industry benchmark value is imported, and the carbon emission intensity in the past n years is compared horizontally with the carbon emission intensity reference value or industry benchmark value.

[0058] You can also monitor the ranking trends and volatility in the same industry and conduct horizontal comparisons on another level.

[0059] In this embodiment, the volatility tolerance value compared with the moving average is set to between 15% and 35%; or, the volatility tolerance value is calculated based on the standard deviation of the variable data (such as activity level data) after removing the extreme values ​​above and below 1%; or, it is modified based on testing and verification of a large number of enterprises, that is, looking at the overall test pass rate of enterprises under this tolerance rate.

[0060] If both the longitudinal comparison and the lateral comparison results exceed the corresponding range, the first quality evaluation data exceeds its setting range. If neither the longitudinal comparison nor the lateral comparison results exceed its corresponding range, the first quality evaluation data is within the setting range.

[0061] 2. According to the production / business logic of the enterprise, analyze the correlation of various activity level data in the production process (evaluate the quality of production and carbon emission data in the enterprise's production process) and obtain the second quality evaluation data of carbon emission activity level data.

[0062] Specifically, the linear correlation between carbon emissions and production is analyzed. The linear correlation between carbon emissions and production is determined according to the following formula:

[0063]

[0064] Among them, x i represents the production volume or unit production volume at the i-th time point, y i Represents x i The corresponding carbon emissions, and are the means of production and carbon emissions, respectively;

[0065] If r is close to 1, carbon emissions are positively correlated with production; if r is close to -1, carbon emissions are negatively correlated with production; if r is close to 0, there is no obvious linear relationship between carbon emissions and production.

[0066] When carbon emissions have a linear relationship with production, linear regression is used to fit the linear relationship between production and carbon emissions, to test the linear trend of carbon emission intensity, and to visualize it. The specific operations are as follows:

[0067] A scatter plot is used to show the distribution of production volume and carbon emissions, and a linear fitting trend line is superimposed to show the linear relationship between the two.

[0068] Scatter plot: With production as the horizontal axis and carbon emissions as the vertical axis, draw / generate a scatter plot to show the distribution of the two. If the scatter distribution is relatively linear, it means that the relationship between carbon emissions and production is strong.

[0069] Linear fitting: Linear regression is performed through the least squares method to fit an optimal linear trend line and display its fitting equation (in the form of): y = a·x+b, where a is the slope, representing the change in carbon emissions caused by each unit increase in production, and b is the intercept.

[0070] If the carbon emission intensity in a certain period deviates from the linear trend, the second quality evaluation data exceeds the set range.

[0071] If there is no linear correlation, analyze whether there is a monotonic relationship between carbon emissions and production. If so, fit a nonlinear curve and use the curve to evaluate the correlation between carbon emissions and production. If the carbon emission intensity in a certain period deviates from the nonlinear curve, the second quality evaluation data exceeds the set range.

[0072] Verify whether there is a monotonic relationship between carbon emissions and production according to the following formula:

[0073]

[0074] Among them, d i is the rank difference of the ith pair of carbon emission and production data, and n is the number of carbon emission or production data.

[0075] If there is a monotonic relationship between carbon emissions and production, the distribution of production and carbon emissions can also be displayed through a scatter plot, but the nonlinear regression curve is superimposed to show a more complex trend relationship. Specifically, a scatter plot is drawn and a variety of nonlinear fitting curves, such as polynomial, logarithmic or exponential, are tried to find the best matching data relationship; an appropriate nonlinear model (such as logarithmic or exponential regression) is selected for fitting. Based on the results of the fitting model, the nonlinear relationship between carbon emissions and production is determined.

[0076] The relationship between carbon emissions and production may change in different periods. When carbon emissions and production data are provided in time series, time series analysis can effectively capture the changes in the relationship between the two over time. Therefore, time series analysis can be used to analyze the correlation between carbon emissions and production in the time dimension, and obtain the trend of the correlation between carbon emissions and production over time, especially to identify seasonal fluctuations or long-term trends in the data. Commonly used tools include moving correlation and cross-correlation.

[0077] A way to visualize this is:

[0078] Sliding window chart: displays the correlation coefficient over time, helping to analyze the fluctuation of carbon emission intensity in different time periods.

[0079] Cross-correlation plot: used to assess whether there is a lag effect between carbon emissions and production, that is, whether changes in carbon emissions lag behind changes in production.

[0080] The specific steps are:

[0081] Arrange the data of carbon emissions and production in time series;

[0082] The sliding window technique was used to calculate the correlation coefficients for different periods;

[0083] Draw a sliding correlation diagram to analyze the trend of the correlation between carbon emissions and production over time;

[0084] If there is a lag effect, use cross-correlation analysis to find the optimal lag period for carbon emissions relative to production.

[0085] If the carbon emission intensity in a certain period deviates from the trend of the correlation between the carbon emission and the production volume over time, the second quality evaluation data exceeds the set range.

[0086] 3. Construct a multivariate regression model and conduct a multivariate regression analysis on the impact of noise, paying special attention to the noise impact of time, location, process / technology, and quality on carbon emission intensity, and obtain third-party quality evaluation data on carbon emission activity levels.

[0087] When the first quality evaluation data, the second quality evaluation data, and the third quality evaluation data all exceed their corresponding threshold ranges, the carbon emission data is judged to be incorrect and untrue. When any of the first quality evaluation data, the second quality evaluation data, and the third quality evaluation data is within its corresponding range, the carbon emission data is judged to be true and highly credible.

[0088] Specifically, a multiple regression model was constructed and multiple regression analysis was used to explain the changes in carbon emission intensity by introducing multiple independent variables;

[0089] The independent variables include: production time, production location, process / technology and product quality.

[0090] The regression model is as follows:

[0091] I carbon =β 0 +β 1 P time +β 2 P location +β 3 P tech +β 4 P qualty ∈

[0092] Among them, I carbon represents carbon emission intensity; P time is the production time factor; P location is the production location factor; P tech It is a process / technical factor; P quality is the product quality factor; ε is the error term, which is used to explain the unobservable noise and is a constant; β 0 , β 1 , β 2 , β 3 , β 4 All are weights.

[0093] The multivariate regression model is optimized by using one or any combination of the following:

[0094] Check for multicollinearity issues: In a multiple regression model, test and analyze the collinearity or correlation between independent variables; if there is a significant collinearity problem, use the variance inflation factor (VIF) to evaluate and remove highly collinear variables; and / or use regularized regression methods to reduce the impact of collinearity by adding penalty terms and stabilize the regression coefficients of the model.

[0095] Perform residual analysis: Analyze the residuals of the multivariate regression model, evaluate the model's fitting effect, and identify possible outliers or noise effects; if abnormal residuals that do not conform to the normal distribution or have significant serial correlation are found, further optimize the model or use robust regression methods.

[0096] Interaction analysis: Add interaction terms to the regression model to enhance the interaction between two independent variables.

[0097] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0098] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A carbon emission data quality analysis method, characterized in that: The following steps are involved: S1, obtaining data on the activity levels that affect the carbon emissions of products / enterprises, calculating the carbon emission intensity, comparing the carbon emission intensity with historical data vertically, and comparing it with the industry benchmark value horizontally, obtaining first quality evaluation data of the carbon emission activity level data, if the first quality evaluation data does not exceed the first threshold, the carbon emission data quality is qualified, otherwise, executing step S2; S2, according to the production / business logic of the enterprise, analyze the correlation of various activity level data in the production process, and obtain the second quality evaluation data of the carbon emission activity level data. If the second quality evaluation data does not exceed the second threshold, the carbon emission data quality is qualified, otherwise, execute step S3; S3, construct a multivariate regression model, conduct a multivariate regression analysis on the influence of time, location, process / technology, and quality noise on carbon emission intensity, and obtain the third quality evaluation data of carbon emission activity level data. If the third quality evaluation data does not exceed the third threshold, the carbon emission data quality is qualified; When the first quality evaluation data, the second quality evaluation data, and the third quality evaluation data all exceed their corresponding threshold ranges, it is determined that the carbon emission data is incorrect.

2. The carbon emission data analysis method according to claim 1, characterized in that: The carbon emission intensity is the ratio of carbon emissions to production, or the carbon emissions per kilowatt-hour of the battery, or the carbon emissions per unit mass of output.

3. The carbon emission data analysis method according to claim 1, characterized in that: The carbon emission intensity of the past n years is calculated using the moving average method, n ≥ 3, and the annual and monthly data are compared with it longitudinally; The carbon emission intensity of the past m months is calculated using the moving average method, where m ≥ 12, and the monthly data is compared with it longitudinally; For products or production links with the same process / technology, calculate the carbon emission intensity of other enterprises / products with the same process / technology in the past n years as the carbon emission intensity reference value, or import the industry benchmark value and compare the carbon emission intensity of the past n years with the carbon emission intensity reference value or industry benchmark value; If both the longitudinal comparison and the lateral comparison results exceed the corresponding setting ranges, the first quality evaluation data exceeds its setting range.

4. The carbon emission data analysis method according to claim 3, characterized in that: The volatility tolerance compared to the moving average is set between 15% and 35%; Or, the volatility tolerance value can be calculated based on the standard deviation of the variable data after removing the upper and lower 1% extreme values; Or, the modification can be carried out based on test verification on a large number of enterprises, that is, the overall test pass rate of enterprises under this fault tolerance rate can be seen.

5. The carbon emission data analysis method according to claim 1, characterized in that: Analyze the linear correlation between carbon emissions and production volume. When carbon emissions and production volume have a linear relationship, use linear regression to fit the linear relationship between production volume and carbon emissions, and check the linear trend of carbon emission intensity. If the carbon emission intensity in a certain period deviates from the linear trend, the second quality evaluation data exceeds the set range. If there is no linear correlation, then analyze whether there is a monotonic relationship between carbon emissions and production. If there is, fit a nonlinear curve and use the curve to evaluate the correlation between carbon emissions and production. If the carbon emission intensity in a certain period deviates from the nonlinear curve, the second quality evaluation data exceeds the set range; Analyze the correlation between carbon emissions and production in the time dimension, and obtain the trend of the correlation between carbon emissions and production over time; If the carbon emission intensity in a certain period deviates from the trend of the correlation between the carbon emission and the production volume over time, the second quality evaluation data exceeds the set range.

6. The carbon emission data analysis method according to claim 5, characterized in that: The linear correlation between carbon emissions and production volume is determined by the following formula: Among them, x i represents the production volume or unit production volume at the i-th time point, y i Represents x i The corresponding carbon emissions, and are the means of production and carbon emissions, respectively; If r is close to 1, carbon emissions are positively correlated with production; if r is close to -1, carbon emissions are negatively correlated with production; if r is close to 0, there is no obvious linear relationship between carbon emissions and production.

7. The carbon emission data analysis method according to claim 5, characterized in that: Verify whether there is a monotonic relationship between carbon emissions and production according to the following formula: Among them, d i is the rank difference of the ith pair of carbon emission and production data, and n is the number of carbon emission or production data.

8. The carbon emission data analysis method according to claim 5, characterized in that: The steps to analyze the correlation between carbon emissions and production in the time dimension are: Arrange the data of carbon emissions and production in time series; The sliding window technique was used to calculate the correlation coefficients for different periods; Draw a sliding correlation diagram to analyze the trend of the correlation between carbon emissions and production over time.

9. The carbon emission data analysis method according to claim 1, characterized in that: A multivariate regression model was constructed and multivariate regression analysis was used to explain the changes in carbon emission intensity by introducing multiple independent variables; The independent variables include: production time, production location, process / technology, and product quality; The regression model is as follows: I carbon =β0+β1P time +β2P location +β3P tech +β4β quality +∈ Among them, I carbon represents carbon emission intensity; P time is the production time factor; P location is the production location factor; P tech It is a process / technical factor; P quality is the product quality factor; ε is the error term, which is used to explain the unobservable noise and is a constant; β0, β1, β2, β3, and β4 are all weights.

10. The carbon emission data analysis method according to claim 1, characterized in that: The multivariate regression model is optimized by using one or any combination of the following: Check for multicollinearity: In a multiple regression model, test and analyze the collinearity or correlation between independent variables; if there is a collinearity problem, use the variance inflation factor to evaluate it and remove highly collinear variables; and / or use regularized regression methods to reduce the impact of collinearity by adding penalty terms and stabilize the regression coefficients of the model; Perform residual analysis: Analyze the residuals of the multivariate regression model, evaluate the model's fitting effect, and identify possible outliers or noise effects; if abnormal residuals that do not conform to the normal distribution or have significant serial correlation are found, further optimize the model or use robust regression methods; Interaction analysis: Add interaction terms to the regression model.

Images