Carbon market data quality evaluation method based on Bayesian updating mechanism

Abstract

The invention discloses a carbon market data quality evaluation method based on a Bayesian updating mechanism, and the method comprises the following steps: 1, recognizing input parameters in a carbon emission accounting model, employing a sensitivity analysis method to evaluate the influence degree of each parameter on the total carbon emission, and determining main parameters; step 2, performing grade evaluation on the data quality of the main parameters from four dimensions of numerical value accuracy, source reliability, evidence storage normalization and evidence storage digitization degree to form an MRV data quality index, and converting a grading result into a parameter uncertainty U value according to a preset uncertainty mapping table; according to the carbon market data quality evaluation method, a data quality evaluation system combining multi-dimensional subjective scoring and objective numerical modeling is constructed, empirical judgment and statistical expression are considered, and conversion of carbon emission data quality from'non-quantifiable 'to'measurable and traceable' is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of carbon market MRV system data quality assessment, and in particular to a carbon market data quality assessment method based on a Bayesian update mechanism. Background Art

[0002] At present, the MRV (Monitoring, Reporting and Verification) system of mainstream carbon markets such as the EU ETS, as well as the uncertainty modeling methods recommended by the IPCC's "Guidelines for National Greenhouse Gas Inventories", are mainly used in the calculation stage of carbon emission data, focusing on error control of key parameters such as emission factors and activity level data. Such methods often use default coefficients, uncertainty intervals or expert estimates to convert data errors into standardized values ​​to improve the accuracy and consistency of the overall accounting. However, the modeling focus of existing methods is on quantitative control of numerical errors in the calculation model, which essentially serves the accurate calculation of total emissions. A technical framework for identifying data authenticity issues, preventing systematic subjective falsification, and quantitatively evaluating data quality has not yet been established.

[0003] Prior art CN202411712639.2, a carbon emissions data quality analysis method, determines the authenticity of carbon emissions data based on first, second, and third quality evaluation data, improving the quality of carbon emissions data and preventing data mis-entry, errors, and falsification. Prior art CN202210166468.2, a solution that collects data from three different sources: power generators, electricity users, and electricity traders, and uploads it to the electricity trading blockchain, preventing data falsification or tampering. Prior art CN202311545823.8, a carbon emissions verification method, device, electronic device, and storage medium, obtains activity level data of carbon source flows based on a verification model; based on the verification model and activity level data, determines the carbon emissions of the verified object to obtain verification results. This approach enables automated verification and reduces manual workload, thereby improving the efficiency and quality of carbon verification and addressing issues such as low verification quality and data falsification caused by the varying competence of verifiers during manual verification. Existing technology, CN202310172198.0, a carbon market carbon emission data quality evaluation method, the method includes: receiving evaluation requests from multiple evaluation objects, selecting multiple evaluation indicators for the carbon emission data quality of the evaluation objects; using Bendeford's law to measure the index value of each evaluation indicator of each evaluation object; according to the index value of each evaluation indicator of each evaluation object, using the linear weighted sum method, TOPSIS approximation ideal solution method and VIKOR compromise method to calculate the integrated value of the evaluation index of the carbon emission data quality of each evaluation object; performing gray statistical evaluation and grading on each carbon emission location to obtain the quality grade of the carbon emission data quality of each carbon emission location. This implementation method can quickly, efficiently and accurately evaluate the quality grade of carbon emission data, reduce the subjectivity of carbon emission data supervision, provide a reference for the refined management of carbon emission data, and ensure the standardized operation of the carbon market.

[0004] The above existing technologies can identify obvious fraud or errors, but cannot identify minor or systematic data adjustments. Based on this, we further developed a model to address these issues.

[0005] As the practice of carbon markets increases, the use of electronic ledgers, instrumentation, cross-entity data transfer, and manual reporting can lead to authenticity risks during data recording, storage, transmission, and review due to irregular processes or inadequate management. Traditional uncertainty models are unable to effectively characterize and address these risks. Existing uncertainty systems also struggle to capture structural characteristics of data quality, such as the reliability of data sources, temporal consistency, completeness of archiving, and the degree of digitization of stored evidence. These aspects pose significant risks to the compliance and transparency of carbon market accounting, yet are difficult to characterize and assess using fixed coefficients or confidence intervals. Summary of the Invention

[0006] The purpose of the present invention is to provide a carbon market data quality assessment method based on the Bayesian updating mechanism, which belongs to the field of data quality management technology of the carbon market MRV (monitoring, reporting and verification) system. It can integrate a combination method of multi-dimensional subjective scoring mechanism, mapping conversion mechanism, uncertainty Monte Carlo propagation mechanism and Bayesian statistical inference mechanism to quantitatively and systematically evaluate data uncertainty, provide technical support for data supervision, compliance verification and policy formulation in the carbon market, and improve the scientificity, transparency and operability of carbon emission data quality management.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for carbon market data quality assessment based on a Bayesian update mechanism, comprising the following steps: Step 1: Identify the input parameters in the carbon emission accounting model, use sensitivity analysis to assess the impact of each parameter on the total carbon emissions, and determine the main parameters; Step 2: The data quality of the main parameters is graded from four dimensions: numerical accuracy, source reliability, evidence standardization, and degree of evidence digitization. This forms the MRV data quality index, and the scoring results are converted into parameter uncertainty U values ​​based on the preset uncertainty mapping table. Step 3: Perform Monte Carlo simulation based on the parameter uncertainty U value to obtain the mean and standard deviation of the main parameters. Perform Monte Carlo simulation again to obtain the mean and standard deviation of carbon emissions and construct a likelihood function. Step 4: Based on multiple periods of historical carbon emission data and corresponding quality assessments, calculate the historical uncertainty U value and obtain the prior probability distribution of uncertainty through fitting; Step five: Apply the Bayesian update formula to fuse the likelihood function corresponding to the current observation with the prior probability distribution, derive the posterior distribution of the uncertainty of the current carbon emission data, and determine whether the current carbon emission data meets the quality requirements.

[0008] Furthermore, in step 1, the sensitivity analysis method is a single-factor sensitivity analysis method, which determines the main parameters by calculating the sensitivity of each parameter to carbon emissions. The sensitivity calculation formula is: ,in For sensitivity, is the relative change in total carbon emissions, is the relative change of the parameter.

[0009] Furthermore, in step 2, the rating is from 1 to 5 points, and the four dimensions of the data quality index are: Numerical accuracy: assessing whether there are errors in the raw data or problems with the algorithm logic; Source reliability: assess whether the data comes from an authoritative system or traceable institution; Standardization of evidence storage: assess whether there are complete original records, ledgers or written supporting materials; The degree of digitization of evidence: assess whether it has features such as electronic archiving and system accessibility.

[0010] Furthermore, in step 2, the scoring criteria from 1 to 5 are: Numerical accuracy: 1 point indicates serious calculation errors or data problems; 2 points indicate minor calculation errors with limited impact; 3 points indicate controversial algorithms or data; 4 points indicate minor data inaccuracies but accurate algorithms; 5 points indicate reliable algorithms, complete data, and accurate results; Source reliability: 1 indicates an unclear and unsupported source; 2 indicates a questionable and low-reliability source; 3 indicates a generally reliable but unsupported source; 4 indicates a generally reliable and supportable source; and 5 indicates an authoritative, traceable, and well-supported source. Standardization of evidence storage: 1 point indicates that the evidence storage process is seriously flawed; 2 points indicate that the evidence storage is not standardized and lacks important links; 3 points indicate that the evidence storage is basically complete but still controversial; 4 points indicate that the evidence storage is relatively standardized and has written support; 5 points indicate that the evidence storage process is standardized and the information is complete; The degree of digitization of evidence: 1 point means there are no digital records and the paper files are scattered or missing; 2 points means that they are basically paper files with a low degree of digitization; 3 points means that they are partially digitized and the archiving standards are average; 4 points means that they have complete electronic data support; 5 points means that they are fully digitally archived and support system calls.

[0011] Furthermore, in the uncertainty mapping table, the mapping relationship between the score and the uncertainty U value satisfies the following formula: ,in, It is Dimension The mapping function of the main parameters can be any one of linear function, logarithmic function, exponential function or custom function.

[0012] Furthermore, the mapping relationship of the mapping function is: a score of 5 points corresponds to an uncertainty of 2.5%, 4 points corresponds to 5%, 3 points corresponds to 10%, 2 points corresponds to 15%, and 1 point corresponds to 20%.

[0013] Furthermore, in step two, the weights of the four dimensions are: numerical accuracy 20%, source reliability 30%, evidence standardization 30%, and degree of evidence digitization 20%.

[0014] Furthermore, in step three, the number of iterations of the Monte Carlo simulation is no less than 1000 times. The first Monte Carlo simulation iterates the four-dimensional U values ​​and corresponding weights to obtain the mean and standard deviation of the main parameters; the second Monte Carlo simulation obtains the mean and standard deviation of carbon emissions based on the uncertainty of the main parameters.

[0015] Furthermore, in step 3, the likelihood function is constructed based on the relative standard deviation of carbon emissions, and assuming that the observation error follows a normal distribution, the calculation formula for the relative standard deviation RSD is: ,in, is the standard deviation, is the mean; standard error The calculation formula is ,in, is the number of observation samples.

[0016] Furthermore, in step 4, the prior probability distribution is a Gamma distribution, a lognormal distribution, a triangular distribution, or a normal distribution, and the shape parameter α and the scale parameter β of the Gamma distribution are obtained by fitting based on the historical uncertainty U value.

[0017] Furthermore, the selection criteria for the Gamma distribution include: the domain is (0, +∞) to constrain the uncertainty to be positive, the right-skewed distribution can be fitted by adjusting the shape parameter α and the scale parameter β, it has clear expression for expectation and variance, and it is a conjugate prior distribution to facilitate Bayesian updating.

[0018] Furthermore, in step 5, the Bayesian update formula is in, is the posterior distribution, is the likelihood function constructed based on the current observation data, is a prior distribution constructed based on historical data.

[0019] Compared with the prior art, the present invention provides a carbon market data quality assessment method based on the Bayesian updating mechanism, which has the following beneficial effects: (1) the carbon market data quality assessment method adopts sensitivity analysis to screen the main parameters as the quantitative scoring mapping objects, thereby improving the assessment efficiency and being more in line with practical applications; (2) This carbon market data quality assessment method constructs a data quality assessment system that combines multi-dimensional subjective scoring with objective numerical modeling, taking into account both empirical judgment and statistical expression, and realizing the transformation of data quality from "unquantifiable" to "measurable and traceable"; (3) This carbon market data quality assessment method innovatively introduces the Bayesian mechanism to achieve dynamic correction of current uncertainty based on historical credibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is an overall flow chart of the carbon market data quality assessment method based on the Bayesian update mechanism of the present invention; Figure 2 This is the result of the single-factor sensitivity analysis of the input parameters of Company A from January 2024 to January 2025; Figure 3 This is a flowchart of data quality assessment based on the Bayesian updating mechanism; Figure 4 This is the parameter uncertainty transmission result diagram after the first Monte Carlo simulation; Figure 5 This is the uncertainty distribution diagram of carbon emissions obtained from the second Monte Carlo simulation; Figure 6 This is the Gamma distribution curve obtained by fitting based on historical RSD data. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0022] Example 1: Please refer to the attached figure. The present invention provides a carbon market data quality assessment method based on the Bayesian update mechanism. The method first performs expert scoring on the input parameters from multiple data quality dimensions (such as numerical accuracy, source reliability, evidence standardization, evidence digitization, etc.) to construct an initial scoring system for subjective credibility; then, the scoring results are converted into uncertainty values ​​through a preset mapping table and incorporated into the parameter uncertainty modeling system. On this basis, combined with the Bayesian statistical inference method, historical data is introduced to probabilistically update the scoring results, and the posterior distribution of credibility is constructed, so as to systematically reflect the risk level of the data in terms of authenticity and compliance. At the same time, the uncertainty formed by the scoring is calculated by root mean square superposition to synthesize the total uncertainty, and finally participates in the Monte Carlo uncertainty propagation of the carbon emission model through the normal distribution expression. In this way, the originally subjective and decentralized data quality assessment system is incorporated into a set of mathematically operable, repeatable and transferable unified modeling framework, which significantly improves the coverage and transparency of carbon emission data quality control. A carbon market data quality assessment method based on a Bayesian update mechanism, comprising the following steps: Step 1: Identify the input parameters in the carbon emission accounting model, use sensitivity analysis to assess the impact of each parameter on the total carbon emissions, and determine the main parameters; Optionally, a single-factor sensitivity analysis method is used to evaluate the sensitivity of each parameter to carbon emissions and determine the main influencing parameters.

[0023] Step 2: The data quality of the main parameters is graded from 1 to 5 based on the four dimensions of numerical accuracy, source reliability, evidence standardization, and degree of digitization to form the MRV data quality index (MRV_DQI). The scoring results are converted into parameter uncertainty U values ​​based on the preset uncertainty mapping table, and different grade scores are mapped to different uncertainties. The data quality index MRV_DQI level is a 1-5 grade assigned by the system rules. The DQI is mapped to the uncertainty U value, and the mapping relationship satisfies the following formula: .in, It is the mapping function of the jth main parameter in the i-th dimension.

[0024] Furthermore, the MRV_DQI scoring dimensions are: (1) Numerical accuracy: whether there are errors in the original data or algorithm logic problems; (2) Source reliability: whether the data comes from an authoritative system or a traceable institution; (3) Standardization of evidence storage: whether there are complete original records, ledgers or written supporting materials; (4) The degree of digitalization of evidence: whether it has the characteristics of electronic archiving and system callability.

[0025] Furthermore, the MRV_DQI data quality index scoring criteria are shown in Table 1 below: Table 1

[0026] Optionally, the mapping function can be a linear function, a logarithmic function, an exponential function, or a user-defined function set according to actual needs, including but not limited to an empirical rule function, a piecewise nonlinear function, or a machine learning generated function. .

[0027] Step 3: Mapping uncertainty values ​​based on the aforementioned main parameters , perform the first Monte Carlo simulation to estimate the observation uncertainty of each main parameter under the current environment, and use the mean μ and relative standard deviation , the calculation formula of RSD is ,in, is the standard deviation, is the mean; On this basis, based on the observation uncertainty results of each main parameter, a second round of Monte Carlo simulation was performed to quantify the overall uncertainty of the target carbon emissions, expressed in RSD, and a credibility assessment model with RSD as input was constructed.

[0028] Preferably, the number of Monte Carlo simulation iterations is set to no less than 1000 times to improve the confidence level of the results and reduce the impact of sample variance on the evaluation results.

[0029] Furthermore, the likelihood function of the current data is constructed using the mean RSD value of the carbon emissions observations obtained in the second round of simulation, and it is assumed that the observation error obeys a normal distribution, thereby forming a credibility assessment model with the current observation data as input, which is used to characterize the rationality and credibility of the data under the constructed statistical model.

[0030] Preferably, in the credibility assessment model, the observation error is assumed to follow a normal distribution, and the standard error , can be calculated as follows: Where RSD is the relative standard deviation and n is the number of observed samples.

[0031] Step 4: Based on the RSD corresponding to multiple periods of historical carbon emission data, construct the prior uncertainty distribution of the main parameters. Preferably, the uncertainty distribution is modeled using the Gamma distribution. The reason is: (1) Non-negativity: The domain of the Gamma distribution is (0, +∞), which can naturally constrain the uncertainty to be positive; (2) Right deviation and adjustability , the scale parameter β can flexibly fit different degrees of skewness: the uncertainty usually presents a right-skewed distribution with "most concentrated in low values ​​and a few high"; (3) Mathematically friendly: The Gamma distribution has clear expression for expectation and variance, which makes it easy to estimate distribution parameters through historical samples; (4) Bayesian adaptability: Gamma distribution is one of the many conjugate prior distributions, which is convenient for subsequent Bayesian update analysis. In particular, it has good prior-posterior convergence characteristics when dealing with uncertainty or variance-type random variables.

[0032] Optionally, in addition to the Gamma distribution, other continuous probability distributions may be selected for modeling based on the actual distribution form of the uncertainty index, including but not limited to: lognormal distribution, triangular distribution, normal distribution, etc.

[0033] Step five: Apply the Bayesian update formula to fuse the likelihood function corresponding to the current observation data with the historically constructed prior uncertainty distribution to derive the posterior distribution of the uncertainty of the current carbon emission data, which is then used to determine whether the current carbon emission data meets the data quality requirements.

[0034] Furthermore, the Bayesian update is performed based on the following formula:

[0035] in, is the updated posterior distribution, which represents the reasonable range of uncertainty of the current data under the historical statistical law. P(obs|RSD) is the likelihood function constructed based on the current observation data, and P(RSD) is the uncertainty prior distribution constructed based on the historical data.

[0036] Preferably, the upper limit of the 95% confidence interval of uncertainty does not exceed 5%.

[0037] The carbon market data quality assessment method of the present invention is specifically described through the following examples in conjunction with the accompanying drawings.

[0038] This embodiment takes the carbon emission data submitted by Company A to the national carbon market in January 2025 as the object, and illustrates the specific application process of the carbon market data quality assessment method based on the Bayesian update mechanism of the present invention.

[0039] The formula for calculating the carbon emissions from coal combustion of Company A in January 2025 is:

[0040] The meaning and units of the parameters are shown in Table 2 below: Table 2

[0041] First, the single-factor sensitivity analysis method was used to increase each input parameter by 1% and calculate its relative change in the total carbon emissions E. The sensitivity calculation formula is: , the calculation results are shown in Table 3: Table 3

[0042] It can be seen that FC coal 、C d 、M ar It is the main parameter affecting carbon emissions.

[0043] The data quality index MRV_DQI level is a 1-5 grade assigned by the system rules. The DQI is mapped to the uncertainty U value with the following conversion relationship: 5 corresponds to 2.5%, 4 corresponds to 5%, 3 corresponds to 10%, 2 corresponds to 15%, and 1 corresponds to 20%; as shown in Table 4 below: Table 4

[0044] A company's main parameters FC coal 、C d 、M ar The scores are shown in Table 5 below: Table 5

[0045] Therefore, the parameter FCcoal 、C d 、M ar The U value in four dimensions (U Q1 ,U Q2 ,U Q3 ,U Q4 ) is as follows:

[0046] Combined with the above U value, Monte Carlo simulation (1000 iterations) was performed on the three main parameters to conduct the uncertainty of carbon emissions. The simulation results are shown in Table 6 below: Table 6

[0047] Taking the above parameter uncertainties as input, a second round of Monte Carlo simulation (1000 iterations) was performed to quantify the uncertainty of carbon emissions. The results are as follows: the mean of carbon emissions is μ = 465837, the standard deviation is 18726, and the relative standard deviation RSD is 4.02%.

[0048] Calculate the standard error (SE) from this result:

[0049] Construct the likelihood coefficient of the observed RSD:

[0050] Prior distribution: Based on the historical data submitted from January to December 2024, the above evaluation process is repeated to obtain the historical RSD data as shown in Table 7 below: Table 7

[0051] After fitting the Gamma distribution, we get the prior distribution: ,in, is the shape parameter and β is the scale parameter.

[0052] Taking the current observation value as the likelihood and combining it with the historical Gamma prior distribution, the posterior distribution characteristics are calculated as shown in Table 8: Table 8

[0053] The results show that after combining historical data, the Bayesian assessment believes that the most likely value of the current data uncertainty is 4.10%, with a fluctuation range of 3.86% to 4.34%, which is generally in line with the expected credibility range.

[0054] In summary, this example demonstrates the application process and effect of the method of the present invention in actual carbon market data evaluation, and verifies its feasibility and accuracy in the fusion of different data quality dimensions, parameter sensitivity identification, uncertainty quantification and Bayesian posterior inference.

[0055] In the description of the present invention, unless otherwise specified, "plurality" means two or more; terms such as "upper," "lower," "left," "right," "inner," "outer," "front end," "rear end," "head," and "tail" indicate positions or relationships based on those shown in the accompanying drawings. These terms are intended solely to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or components referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, terms such as "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0056] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "connected" and "connection" should be understood in a broad sense. For example, they can refer to fixed connection, detachable connection, or integral connection; mechanical connection, electrical connection; direct connection, or indirect connection through an intermediary. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

Claims

1. A carbon market data quality assessment method based on a Bayesian updating mechanism, characterized by: The following steps are involved: Step 1: Identify the input parameters in the carbon emission accounting model, use sensitivity analysis to assess the impact of each parameter on the total carbon emissions, and determine the main parameters; Step 2: The data quality of the main parameters is graded from four dimensions: numerical accuracy, source reliability, evidence standardization, and degree of evidence digitization. This forms the MRV data quality index, and the scoring results are converted into parameter uncertainty U values ​​based on the preset uncertainty mapping table. Step 3: Perform Monte Carlo simulation based on the parameter uncertainty U value to obtain the mean and standard deviation of the main parameters. Perform Monte Carlo simulation again to obtain the mean and standard deviation of carbon emissions and construct a likelihood function. Step 4: Based on multiple periods of historical carbon emission data and corresponding quality assessments, calculate the historical uncertainty U value and obtain the prior probability distribution of uncertainty through fitting; Step five: Apply the Bayesian update formula to fuse the likelihood function corresponding to the current observation with the prior probability distribution, derive the posterior distribution of the uncertainty of the current carbon emission data, and determine whether the current carbon emission data meets the quality requirements.

2. The carbon market data quality assessment method based on the Bayesian updating mechanism according to claim 1 is characterized in that: In step 1, the sensitivity analysis method is a single-factor sensitivity analysis method, which determines the main parameters by calculating the sensitivity of each parameter to carbon emissions. The sensitivity calculation formula is: ,in, For sensitivity, is the relative change in total carbon emissions, is the relative change of the parameter.

3. The carbon market data quality assessment method based on the Bayesian updating mechanism according to claim 2 is characterized in that: In step 2, the rating scale is 1 to 5, and the four dimensions of the data quality index are: Numerical accuracy: assessing whether there are errors in the raw data or problems with the algorithm logic; Source reliability: assess whether the data comes from an authoritative system or traceable institution; Standardization of evidence storage: assess whether there are complete original records, ledgers or written supporting materials; The degree of digitization of evidence: assess whether it has the features of electronic archiving and can be called by the system.

4. The carbon market data quality assessment method based on the Bayesian updating mechanism according to claim 3 is characterized in that: In step 2, the scoring scale from 1 to 5 is: Numerical accuracy: 1 point indicates serious calculation errors or data problems; 2 points indicate minor calculation errors with limited impact; 3 points indicate controversial algorithms or data; 4 points indicate minor data inaccuracies but accurate algorithms; 5 points indicate reliable algorithms, complete data, and accurate results; Source reliability: 1 indicates an unclear and unsupported source; 2 indicates a questionable and low-reliability source; 3 indicates a generally reliable but unsupported source; 4 indicates a generally reliable and supportable source; and 5 indicates an authoritative, traceable, and well-supported source. Standardization of evidence storage: 1 point indicates that the evidence storage process is seriously flawed; 2 points indicate that the evidence storage is not standardized and lacks important links; 3 points indicate that the evidence storage is basically complete but still controversial; 4 points indicate that the evidence storage is relatively standardized and has written support; 5 points indicate that the evidence storage process is standardized and the information is complete; The degree of digitization of evidence: 1 point means there are no digital records and the paper files are scattered or missing; 2 points means that they are basically paper files with a low degree of digitization; 3 points means that they are partially digitized and the archiving standards are average; 4 points means that they have complete electronic data support; 5 points means that they are fully digitally archived and support system calls.

5. The carbon market data quality assessment method based on the Bayesian updating mechanism according to claim 4 is characterized in that: In step 2, in the uncertainty mapping table, the mapping relationship between the score and the uncertainty U value satisfies the following formula: ,in, It is the mapping function of the jth main parameter in the i-th dimension. The mapping function can be any one of a linear function, a logarithmic function, an exponential function, or a custom function.

6. The carbon market data quality assessment method based on the Bayesian updating mechanism according to claim 5 is characterized in that: The mapping relationship of the mapping function is: a score of 5 points corresponds to an uncertainty of 2.5%, 4 points corresponds to 5%, 3 points corresponds to 10%, 2 points corresponds to 15%, and 1 point corresponds to 20%.

7. The carbon market data quality assessment method based on the Bayesian updating mechanism according to claim 6 is characterized in that: In step two, the weights of the four dimensions are: numerical accuracy 20%, source reliability 30%, evidence standardization 30%, and degree of evidence digitization 20%.

8. The carbon market data quality assessment method based on the Bayesian updating mechanism according to claim 7 is characterized in that: In step three, the number of iterations of the Monte Carlo simulation is no less than 1,000. The first Monte Carlo simulation iterates the four-dimensional U values ​​and corresponding weights to obtain the mean and standard deviation of the main parameters; the second Monte Carlo simulation obtains the mean and standard deviation of carbon emissions based on the uncertainty of the main parameters.

9. The carbon market data quality assessment method based on the Bayesian updating mechanism according to claim 8 is characterized in that: In step 3, the likelihood function is constructed based on the relative standard deviation of carbon emissions, and it is assumed that the observation error follows a normal distribution. The calculation formula is ,in, is the standard deviation, is the mean; standard error The calculation formula is ,in, is the number of observation samples.

10. The carbon market data quality assessment method based on the Bayesian updating mechanism according to claim 9 is characterized in that: In step 4, the prior probability distribution is any one of the gamma distribution, lognormal distribution, triangular distribution or normal distribution, and the shape parameter α and scale parameter β of the gamma distribution are fitted based on the historical uncertainty U value.

11. The carbon market data quality assessment method based on the Bayesian updating mechanism according to claim 10 is characterized in that: The selection criteria for the Gamma distribution include: the domain is (0, +∞) to constrain the uncertainty to be positive, the shape parameter α and scale parameter β can be adjusted to fit the right-skewed distribution, it has clear expression for expectation and variance, and it is a conjugate prior distribution to facilitate Bayesian updating.

12. The carbon market data quality assessment method based on the Bayesian updating mechanism according to claim 11 is characterized in that: In step 5, the Bayesian update formula is ,in, is the posterior distribution, is the likelihood function constructed based on the current observation data, is a prior distribution constructed based on historical data.

Images