A method and system for calculating carbon emissions during the life cycle of electricity

Abstract

The present invention provides a method and system for calculating carbon emissions during the life cycle of electric power. The method includes the following steps: obtaining carbon footprint assessment parameters and basic data related to carbon emissions in an electric power project; calculating a data deviation value between the basic data and actual carbon emission data using a Bayesian network method; constructing a program review and evaluation technique distribution based on the basic data and upper and lower limits of the data deviation value to fit the data distribution of the basic data; uniformly simplifying the continuously changing carbon footprint assessment parameters into a triangular distribution to simulate the parameter distribution of the carbon footprint assessment parameters; combining the data distribution and the parameter distribution and calculating using a Monte Carlo simulation method to obtain the carbon footprint distribution in the electric power project; and calculating the carbon emissions of the electric power project based on the carbon footprint distribution. The present invention has the effect of making the calculated carbon emission results repeatable and more accurate.

Description

Technical Field

[0001] The present invention belongs to the technical field of power generation carbon emission analysis, and specifically relates to a method and system for calculating carbon emissions during the life cycle of electricity. Background Art

[0002] The carbon emissions from power generation in large-scale power projects are often calculated using a life cycle approach. Due to the uncertainty surrounding the numerous parameters and input data involved, the calculated results often have significant uncertainty. Uncertainty, defined in life cycle assessment as "the impact on the output caused by inaccurate measurements, missing data, model assumptions, and other factors during the process," is widely present in the life cycle assessment process. This is particularly problematic for research projects with numerous sub-processes and long durations, as uncertainty accumulates during the analysis and accumulates in the final assessment results. This uncertainty can even lead to inaccurate or even completely erroneous conclusions.

[0003] Life cycle assessments of the carbon footprint of large-scale power generation projects are generally calculated based on design data and parameters. Errors between design data and actual data are inevitable, and the accuracy of the design document is affected by multiple factors, resulting in significant uncertainty. The parameter, the unit activity intensity carbon footprint, is the fundamental parameter in the carbon footprint assessment process. Parameter uncertainty is reflected in multiple factors, including technology, time, region, and data acquisition methods, directly impacting the assessment results. Uncertainty is particularly prominent in the carbon footprint assessment process for large-scale power generation projects.

[0004] Currently, methods such as data quality scoring and Monte Carlo simulation have been developed to address data uncertainty. Data quality indicators can be used to evaluate data quality, and life cycle assessment data can be qualitatively or quantitatively evaluated using data quality scores. Data quality requirements include the following aspects: temporal representativeness, geographical representativeness, technical representativeness, data precision, completeness, and representativeness. Monte Carlo is a stochastic simulation method based on probability and statistical theory. It uses random numbers (or more commonly, pseudo-random numbers) to perform sampling and then calculate settlement results. This allows data uncertainty to propagate and be superimposed on the output during calculations, ultimately quantifying the uncertainty of the calculation results.

[0005] However, data quality scoring only provides a qualitative evaluation of data quality, and the scoring process is affected by the subjectivity of the operator. Different subjects will have different views on different indicators, which in turn affects the final evaluation results. This limitation is mainly reflected in the subjectivity of the operator, and also leads to weak reproducibility of the evaluation results. Monte Carlo analysis can only obtain trend distribution, which only takes into account the uncertainty of the data. However, in the entire calculation process, in addition to data uncertainty, there are other sources of uncertainty such as parameter uncertainty and model uncertainty. Therefore, the carbon emission calculation results calculated by the above-mentioned existing calculation methods are relatively inaccurate and not repeatable. Summary of the Invention

[0006] The present invention provides a method and system for calculating carbon emissions during the life cycle of electricity, so as to solve the problem that the carbon emissions calculation results are relatively inaccurate and non-repeatable.

[0007] In a first aspect, the present invention provides a method for calculating carbon emissions over the life cycle of electricity, the method comprising the following steps:

[0008] Obtain carbon footprint assessment parameters and basic data related to carbon emissions in target projects;

[0009] Calculating the data deviation between the basic data and the actual carbon emission data using a Bayesian network method;

[0010] Constructing a program review and evaluation technical distribution by combining the basic data and the upper and lower limits of the data deviation value to fit the data distribution of the basic data;

[0011] Simplifying the continuously changing carbon footprint assessment parameters into a uniform triangular distribution to simulate the parameter distribution of the carbon footprint assessment parameters;

[0012] Combining the data distribution and the parameter distribution and calculating by Monte Carlo simulation method, the carbon footprint distribution of the target project is obtained;

[0013] The carbon emissions of the target project are calculated based on the carbon footprint distribution.

[0014] Optionally, the calculating of the data deviation value between the basic data and the actual carbon emission data by using the Bayesian network method comprises the following steps:

[0015] Constructing Bayesian network nodes based on the objective influencing factors of the basic data;

[0016] Constructing a topological structure model between all the Bayesian network nodes by using an interpretive structural model method;

[0017] Obtaining historical basic data related to the carbon emissions in the target project;

[0018] Calculating the conditional probability of each of the Bayesian network nodes in the topological structure model based on the historical basic data analysis;

[0019] Constructing a Bayesian network by combining the topological structure model and the conditional probability;

[0020] The data deviation value between the basic data and the real carbon emission data is calculated based on the transmission effect of the conditional probability in the Bayesian network.

[0021] Optionally, constructing a topological structure model among all the objective influencing factors by using an interpretive structure model method comprises the following steps:

[0022] Constructing an adjacency matrix between all the objective influencing factors based on the correlations between all the objective influencing factors;

[0023] Combining the adjacency matrix and a preset identity matrix to calculate the reachability matrix between all the objective influencing factors;

[0024] Solving the reachability matrix and calculating the set classification results of the different types of objective influencing factors in the reachability matrix;

[0025] Performing inter-level decomposition on all the objective influencing factors in the reachable matrix based on the set classification result to obtain an inter-level decomposition result;

[0026] A topological structure model among all the objective influencing factors is constructed by combining the set classification result and the inter-level decomposition result.

[0027] Optionally, the calculation formula of the data deviation value is as follows:

[0028]

[0029] Where: A i represents the Bayesian network node, ±P represents the Bayesian network node A i The data deviation values ​​corresponding to the basic data, X1, X2, ..., X n Indicates the various deviation levels of the basic data, Represents the Bayesian network node A i The conditional probability in the nth level of the deviation

[0030] The conditional probability The calculation formula is as follows:

[0031]

[0032] Where:

[0033]

[0034] Where: k represents A i There are k influencing factors A j exist, Influencing factor A j Under the influence of A i The conditional probability of y1, y2, ..., ym represents the m value states of the Bayesian network nodes, f ym，j Indicates A j is the probability of the mth value state, f ym|ym，j Indicates that in A j When A is the mth value state i The probability of the mth value state occurring.

[0035] Optionally, the calculating the conditional probability of each Bayesian network node in the topological structure model based on the historical basic data analysis includes the following steps:

[0036] Performing data preprocessing on the historical basic data to obtain preprocessed historical data;

[0037] Calculating the frequency of each Bayesian network node taking values ​​of the pre-processed historical data respectively;

[0038] Calculate the marginal probability distribution and conditional probability distribution of the Bayesian network nodes based on the value frequency;

[0039] The conditional probability of each of the Bayesian network nodes is calculated based on the topological structure model and in combination with the edge probability distribution and the conditional probability distribution.

[0040] In a second aspect, the present invention further provides a system for calculating carbon emissions during the life cycle of electricity, the system comprising:

[0041] The data acquisition subsystem is used to obtain carbon footprint assessment parameters and basic data related to carbon emissions in the target project;

[0042] a deviation value calculation subsystem, configured to calculate the data deviation value between the basic data and the actual carbon emission data using a Bayesian network method;

[0043] A data distribution simulation subsystem is used to construct a program review and evaluation technology distribution by combining the basic data and the upper and lower limits of the data deviation value to fit the data distribution of the basic data;

[0044] a parameter distribution simulation subsystem, configured to uniformly simplify the continuously changing carbon footprint assessment parameters into a triangular distribution, so as to simulate the parameter distribution of the carbon footprint assessment parameters;

[0045] A carbon footprint distribution simulation subsystem, configured to combine the data distribution and the parameter distribution and calculate the carbon footprint distribution of the target project using a Monte Carlo simulation method;

[0046] The carbon emission calculation subsystem is used to calculate the carbon emission of the target project according to the carbon footprint distribution.

[0047] Optionally, the deviation value calculation subsystem includes:

[0048] A node construction module, used to construct Bayesian network nodes according to the objective influencing factors of the basic data;

[0049] A model building module, used for constructing a topological structure model between all the Bayesian network nodes by using an interpretive structural model method;

[0050] A historical data acquisition module, used to acquire historical basic data related to the carbon emissions in the target project;

[0051] A probability calculation module, configured to calculate the conditional probability of each of the Bayesian network nodes in the topological structure model based on the historical basic data analysis;

[0052] A network construction module, configured to construct a Bayesian network by combining the topological structure model and the conditional probability;

[0053] The deviation value calculation module is used to calculate the data deviation value between the basic data and the real carbon emission data based on the transmission effect of the conditional probability in the Bayesian network.

[0054] Optionally, the model building module includes:

[0055] A first matrix construction unit is configured to construct an adjacency matrix between all the objective influencing factors according to the correlations between all the objective influencing factors;

[0056] A second matrix construction unit is configured to calculate a reachable matrix between all the objective influencing factors by combining the adjacency matrix and a preset identity matrix;

[0057] A matrix solving unit, configured to solve the reachable matrix and calculate a set classification result of different types of objective influencing factors in the reachable matrix;

[0058] A factor decomposition unit, configured to perform inter-level decomposition on all the objective influencing factors in the reachable matrix according to the set classification result to obtain an inter-level decomposition result;

[0059] A model building unit is used to construct a topological structure model between all the objective influencing factors by combining the set classification result and the inter-level decomposition result.

[0060] Optionally, the calculation formula of the data deviation value is as follows:

[0061]

[0062] Where: A i represents the Bayesian network node, ±P represents the Bayesian network node A i The data deviation values ​​corresponding to the basic data, X1, X2, ..., X n Indicates the various deviation levels of the basic data, Represents the Bayesian network node A i The conditional probability in the nth level of the deviation

[0063] The conditional probability The calculation formula is as follows:

[0064]

[0065] Where:

[0066]

[0067] Where: k represents A i There are k influencing factors A j exist, Influencing factor A j Under the influence of A i The conditional probability of y1, y2, ..., ym represents the m value states of the Bayesian network nodes, f ym，j Indicates A j is the probability of the mth value state, f ym|ym，j Indicates that in A j When A is the mth value state i The probability of the mth value state occurring.

[0068] Optionally, the probability calculation module includes:

[0069] A preprocessing unit, configured to perform data preprocessing on the historical basic data to obtain preprocessed historical data;

[0070] A frequency calculation unit, configured to respectively calculate the frequency of values ​​taken by each of the Bayesian network nodes for the pre-processed historical data;

[0071] A probability distribution calculation unit, configured to calculate the marginal probability distribution and conditional probability distribution of the Bayesian network nodes according to the value frequencies;

[0072] A conditional probability calculation unit is used to calculate the conditional probability of each of the Bayesian network nodes based on the topological structure model in combination with the edge probability distribution and the conditional probability distribution.

[0073] The beneficial effects of the present invention are:

[0074] This method divides the carbon emission calculation process into two parts: parameters and data, and evaluates the distribution of the two separately. It then uses a random simulation method to fit and calculate the uncertainty of the global data. A Bayesian network model is constructed for the input basic data, incorporating various factors that affect the accuracy of the input data, thereby calculating the data deviation value between the basic data and the actual carbon emission data. On the other hand, the carbon footprint assessment parameters are processed to obtain the values ​​of the parameters under various circumstances, replacing the single parameter value with an interval parameter value; then the interval distribution value of the parameter is fitted with a triangular distribution, and finally a Monte Carlo simulation is performed with multiple sampling fits to obtain the distribution of the output data. Compared with existing data quality scoring methods, this method reduces the subjectivity of subjective evaluation of data quality to a certain extent, making the calculation results repeatable. Moreover, compared with the calculation and analysis using only the Monte Carlo method, the uncertainty of the global data is taken into account, making the calculation results more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 Schematic diagram of the flow chart of the method for calculating carbon emissions during the life cycle of electricity in the present invention.

[0076] Figure 2 Schematic diagram for explaining the principle of the structural model method in the present invention. DETAILED DESCRIPTION

[0077] The invention discloses a method for calculating carbon emissions during the life cycle of electricity.

[0078] Reference Figure 1 The calculation method of carbon emissions during the electricity life cycle includes the following steps:

[0079] S101. Obtain carbon footprint assessment parameters and basic data related to carbon emissions in the target project.

[0080] Target projects primarily refer to a series of engineering projects, such as large-scale power projects, that require carbon emissions accounting. In the carbon footprint assessment process, unit activity intensity refers to the carbon footprint assessment parameter generated by unit output or unit consumption in a specific activity or process. This parameter can be used to measure the contribution of a specific activity to carbon emissions. The carbon footprint assessment parameter for unit activity intensity can take various forms, the most common of which include the following:

[0081] Carbon footprint per unit of energy consumption: This measures the amount of carbon emissions generated per unit of energy consumed in a specific activity. For example, the carbon footprint per unit of electricity consumption represents the amount of carbon emissions generated per kilowatt-hour of electricity consumed.

[0082] Carbon footprint per unit of output value: This measures the carbon emissions generated for each product or service of a certain value produced in a specific activity. For example, the carbon footprint per unit of GDP represents the carbon emissions generated for every 10,000 yuan of GDP produced.

[0083] Carbon footprint per unit of transport distance: This measures the carbon emissions generated per distance transported in a specific transport activity. For example, the carbon footprint per kilometer represents the carbon emissions generated per kilometer of goods or passengers transported.

[0084] Carbon footprint per unit area: This measures the amount of carbon emissions generated per unit area by a specific land use activity. For example, the carbon footprint per hectare represents the amount of carbon emissions generated per hectare of land use.

[0085] The basic data related to carbon emissions mainly include electricity consumption data, fuel consumption data, industrial activity data, transportation data, waste disposal data, energy consumption structure data, greenhouse gas emission factor data, and design data at the beginning of project design.

[0086] S102. Use the Bayesian network method to calculate the data deviation value between the basic data and the actual carbon emission data.

[0087] By constructing a Bayesian network model, the parent nodes and topological structure of each node can be determined. Based on the baseline data and actual carbon emission data, a conditional probability table is estimated for each node in the Bayesian network model. This conditional probability table reflects the relationship between the baseline data and the actual carbon emission data. Based on the Bayesian network model and the conditional probability table, the data deviation between the baseline data and the actual carbon emission data is calculated. By calculating the transmission effect of the baseline data in the Bayesian network model, the probability difference between the baseline data and the actual carbon emission data can be obtained. The calculated data deviation values ​​can be analyzed to understand the extent of the discrepancy between the baseline data and the actual carbon emission data and the influencing factors. Statistical analysis and visualization of the data deviation values ​​can be used for further exploration and interpretation.

[0088] S103. Construct a plan review and approval technical distribution by combining the basic data and the upper and lower limits of the data deviation value to fit the data distribution of the basic data.

[0089] The deviation value of each basic data is calculated using the basic data and the data deviation value of the Program Evaluation and Review Technique (PERT). The deviation value can be the difference between the actual execution data and the planned data, or other measurement indicators. The distribution type of the data is determined based on the distribution of the data deviation value. Common distribution types include normal distribution, exponential distribution, uniform distribution, etc. The distribution parameters are estimated based on the upper and lower limits of the data deviation value and the determined distribution type. For normal distribution, the mean and standard deviation can be estimated; for exponential distribution, the parameter λ can be estimated; for uniform distribution, the upper and lower limits can be estimated. Based on the estimated distribution parameters, the PERT distribution is constructed. The distribution can be represented by a probability density function or a cumulative distribution function. The basic data is compared and fitted with the PERT distribution. The distribution of the basic data can be evaluated by calculating the probability density or cumulative probability of the basic data in the PERT distribution.

[0090] S104. Simplify the continuously changing carbon footprint assessment parameters into a uniform triangular distribution to simulate the parameter distribution of the carbon footprint assessment parameters.

[0091] Among them, the carbon footprint assessment parameters are the direct calculation data of the carbon footprint assessment process, and their accuracy directly determines the accuracy of the calculation results. Due to differences in time, geography, technology, etc., the carbon footprint assessment parameters cannot accurately represent the carbon footprint of specific activity units. Especially for large-scale power generation project stations, which have long construction periods and complex processes, a single parameter can hardly simulate the carbon footprint of an activity. Since the carbon footprint assessment parameters are not static and are in constant change with different actual conditions, a single parameter is not applicable to the calculation process. The carbon footprint assessment parameters are uniformly simplified into a triangular distribution. Through the triangular distribution, the maximum, minimum and most likely values ​​can be predicted. The values ​​close to the maximum and minimum values ​​are less likely to appear than the values ​​close to the most likely value. This can simulate the distribution of the parameters and reduce the uncertainty of the parameters to a certain extent.

[0092] S105. Combining the data distribution and parameter distribution and calculating through Monte Carlo simulation method, obtain the carbon footprint distribution of the target project.

[0093] Among them, Monte Carlo simulation (MC) links the problem to be solved with a certain probability model, and uses an electronic computer to implement statistical simulation or sampling to obtain an approximate solution to the problem. Its principle is to construct an approximate distribution of data, perform calculations by sampling a large number of samples in the distribution interval, and finally obtain a calculated value close to the true value. The Monte Carlo simulation method is often used in uncertainty analysis. Because it can obtain the distribution of output data, it has strong visibility. In this embodiment, after fitting the basic distribution of basic data and parameters with the PERT distribution, the calculation process is calculated using the Monte Carlo simulation method. After multiple samplings, a result close to the actual situation is finally obtained, which effectively avoids the error caused by a single data, and finally obtains the output data distribution and clarifies the distribution of carbon footprint.

[0094] S106. Calculate the carbon emissions of the target project based on the carbon footprint distribution.

[0095] In this embodiment, the calculation process of carbon emissions is divided into two parts: parameters and data, and the distribution of the two is evaluated separately. After fitting using a random simulation method, the uncertainty of the global data is calculated. A Bayesian network model is constructed for the input basic data, and various factors that affect the accuracy of the input data are included, so as to calculate the data deviation value between the basic data and the actual carbon emission data. On the other hand, the carbon footprint assessment parameters are processed to obtain the values ​​of the parameters under various circumstances, and the single parameter value is replaced by the interval parameter value; then the interval distribution value of the parameter is fitted with a triangular distribution, and finally the Monte Carlo simulation is performed multiple sampling fittings to obtain the distribution of the output data. Compared with the existing data quality scoring method, the subjectivity of the subjective evaluation of data quality is reduced to a certain extent, so that the calculation results have a certain degree of repeatability. Moreover, compared with the calculation and analysis using only the Monte Carlo method, the uncertainty of the global data is taken into account, making the calculation results more accurate.

[0096] In one embodiment, step S102, i.e., calculating the data deviation value between the basic data and the actual carbon emission data using the Bayesian network method, specifically includes the following steps:

[0097] Construct Bayesian network nodes based on the objective influencing factors of basic data;

[0098] Construct a topological structure model between all Bayesian network nodes through the interpretation structure model method;

[0099] Obtain historical basic data related to carbon emissions in target projects;

[0100] Calculate the conditional probability of each Bayesian network node in the topological structure model based on historical basic data analysis;

[0101] Combining topological structure models and conditional probabilities to construct Bayesian networks;

[0102] The data deviation value between the basic data and the actual carbon emission data is calculated based on the transmission effect of conditional probability in the Bayesian network.

[0103] In this embodiment, the following three steps are mainly included: node determination, Bayesian network structure learning, and Bayesian network parameter learning. In the node determination stage, factors affecting the accuracy of the input data are selected as basic nodes and defined according to their characteristics; in the network structure learning stage, the Interpretative Structural Modeling Method (ISM method) is used to analyze each node, clarify the cause-effect relationship, and construct a directed acyclic graph; in the parameter learning stage, the probability of the node under various states is investigated based on the directed acyclic graph to obtain a set of conditional probabilities.

[0104] Each Bayesian network node represents a random variable. In this implementation, Bayesian network nodes represent the various objective factors influencing the input data, including project complexity, project scale, capacity utilization, energy structure, and power consumption. Bayesian network structure learning can clarify the relationships between these objective factors, explore their structural relationships, construct a multi-level hierarchical structure model, and determine the topological structure model.

[0105] The calculation formula for the data deviation value is as follows:

[0106]

[0107] Where: A i represents the Bayesian network node, ±P represents the Bayesian network node A i The data deviation value corresponding to the basic data, X1, X2, ..., X n Indicates the various deviation levels of the basic data, Represents Bayesian network node A i Conditional probability at the nth level of bias

[0108] Conditional probability The calculation formula is as follows:

[0109]

[0110] Where:

[0111]

[0112] Where: k represents A i There are k influencing factors Aj exist, Influencing factor A j Under the influence of A i The conditional probability of y1, y2, ..., ym represents the m value states of the Bayesian network nodes, f ym，j Indicates A j is the probability of the mth value state, f ym|ym，j Indicates that in A j When A is the mth value state i The probability of the mth value state occurring.

[0113] In one embodiment, the steps of constructing a topological structure model between all Bayesian network nodes by interpreting the structural model method specifically include the following steps:

[0114] Construct an adjacency matrix between all objective influencing factors based on the correlation between them;

[0115] Combine the adjacency matrix and the preset identity matrix to calculate the reachability matrix between all objective influencing factors;

[0116] Solve the reachability matrix and calculate the set classification results of different types of objective influencing factors in the reachability matrix;

[0117] Based on the set classification results, all objective influencing factors in the reachable matrix are decomposed inter-level to obtain the inter-level decomposition results;

[0118] The topological structure model among all objective influencing factors is constructed by combining the set classification results and the inter-level decomposition results.

[0119] In this embodiment, referring to Figure 2 , Figure 2 Schematic diagram to explain the principle of Interpretative Structural Modeling Method (ISM).

[0120] For the matrix element a in the i-th row and j-th column of the adjacency matrix ij , when a ij When it is 0, it means the objective influencing factor A i and objective influencing factors A j There is no correlation between ij When it is 1, it means A i to A j There is a channel of length 1 between them, which can be reached directly, indicating the objective influencing factor A i and objective influencing factors A jThere is a correlation between them. The reachability matrix indicates whether there are channels that can be connected between elements, and can effectively reflect the correlation between influencing factors.

[0121] The reachable matrix has the property of transfer, if A i to A j is 1, A j to A n is 1, then A i to A n There must be a channel. Therefore, the reachability matrix can be obtained by analyzing the push property, or it can be obtained by adding the adjacency matrix to the preset identity matrix and performing certain operations. The identity matrix refers to a matrix in which only the diagonal elements are 1 and the rest of the elements are 0. Assuming that the identity matrix is ​​L and the adjacency matrix is ​​A, according to the Boolean operation rules, it can be proved that:

[0122] (A+L) 2 =L+A+A 2

[0123] ↓

[0124] (A+L) k =L+A+A 2 +...+A k

[0125] If the adjacency matrix A satisfies the condition: (A+L) k-1 ≠(A+L) k =(A+L) k+1 =M, then M is the reachability matrix of the adjacency matrix A.

[0126] The set classification results include the reachable set R(A i ), lookahead set A(A i ), common set T and underlying element N. Among them, the reachable set R(A i )={A i =1}, indicating that the element A in the reachable matrix i The column element displayed as 1 in the corresponding row indicates that the element A i Reachable elements. Antecedent set A(A i )={A j =1}, indicating that the element A in the reachable matrix i The row element with 1 in the corresponding column indicates that element A can be reached i Elements of. Common set T = {A i ∈R(A i )∩A(A i )}. The bottom element N = {A i |R(A i )∩A(A i )=A(A i)}, indicating that the element is at the bottom level and no element can reach it.

[0127] Regional decomposition is required before inter-level decomposition. When element A i With A j Belong to the same area, otherwise they do not belong to the same area. All elements in the matrix belong to the same area, and there are no subsystems in different areas. Inter-level decomposition is mainly to clearly understand the hierarchical relationship between the elements in the system. The top level represents the ultimate goal of the system, and the next level is the influencing factor of the previous level. The steps of inter-level decomposition are as follows:

[0128] Let L0 = φ, L j ={A i ∈P-L0-L1-...-L j-1 |R i-1 (A i )∩A j-1 (A i )=R i-1 (A i )}, where j = 1, 2, ..., n, L j represents the objective influencing factors of the jth layer, P represents the total number of objective influencing factors, and φ represents the empty set. j =φ, the interstage decomposition is completed. During the interstage decomposition period, if R(A i )∩A(A i )=R(A i ), then the element A i Extract and continue until the last element. Through inter-level decomposition, multiple objective influencing factors are divided into a multi-layer structure model, which is the topological structure model between all objective influencing factors.

[0129] In one embodiment, the step of calculating the conditional probability of each Bayesian network node in the topological structure model based on historical basic data analysis specifically includes the following steps:

[0130] Perform data preprocessing on historical basic data to obtain preprocessed historical data;

[0131] Calculate the frequency of each Bayesian network node's value for the pre-processed historical data;

[0132] Calculate the marginal probability distribution and conditional probability distribution of Bayesian network nodes based on the value frequency;

[0133] The conditional probability of each Bayesian network node is calculated based on the topological structure model and combined with the marginal probability distribution and conditional probability distribution.

[0134] In this embodiment, data preprocessing mainly includes a data cleansing step and a missing value processing step. Data cleaning can check for errors or outliers in the data and correct or delete them. For missing values, you can choose to fill missing values, delete missing values, or use interpolation methods to fill them. Preprocessed historical data is obtained through data preprocessing, and then for each Bayesian network node, the frequency of occurrence of each value is calculated. At the same time, the value frequency of each node is counted, and the probability of each value is calculated. For each node, the marginal probability distribution is calculated based on the value frequency. The marginal probability distribution represents the probability distribution of the node without considering other nodes. Based on the topological structure model and the marginal probability distribution, the conditional probability distribution is calculated. That is, for each node, the probability distribution of the node value is calculated under the condition of the given parent node value. Finally, the conditional probability of each Bayesian network node is calculated by combining the topological structure model, the marginal probability distribution, and the conditional probability distribution. That is, based on the node's parent node and the conditional probability distribution, the probability of each node under the condition of the given parent node value is calculated.

[0135] In one embodiment, the step of calculating the conditional probability of each Bayesian network node in the topological structure model based on historical basic data analysis may further include the following steps:

[0136] Obtain historical data quality scores for historical basic data;

[0137] Based on the historical data quality score and by constructing the Beta distribution, the random distribution of the historical basic data is fitted;

[0138] The conditional probability of each Bayesian network node in the topological structure model is calculated based on the random distribution of data.

[0139] In this embodiment, no matter how accurate the source of basic data is, there is always a certain deviation between the basic data and the actual situation. Therefore, the data has a certain distribution range, and the probability distribution can be used to fit the data.

[0140] The beta distribution is a density function that is a conjugate prior distribution of the Bernoulli distribution and the binomial distribution. The probability density function of the beta distribution is determined by the endpoints a, b and the morphological parameters α, β. Under different morphological parameters, the beta distribution presents different distribution forms. Therefore, different endpoints and morphological parameters can almost approximate various forms of distribution such as normal, lognormal, uniform and exponential distribution. Therefore, after obtaining the historical data quality score of the historical basic data, the historical basic data can be converted into various forms of probability distribution based on the historical data quality score, so as to approximate the actual distribution of the data and reduce the uncertainty of the data. The specific formula is as follows:

[0141]

[0142] Where: x is the historical basic data, f represents the probability distribution of the historical basic data, and Γ represents the Γ function.

[0143] The present invention also discloses a system for calculating carbon emissions during the life cycle of electricity, the system comprising:

[0144] The data acquisition subsystem is used to obtain carbon footprint assessment parameters and basic data related to carbon emissions in the target project;

[0145] Deviation value calculation subsystem, used to calculate the data deviation value between the basic data and the actual carbon emission data using the Bayesian network method;

[0146] The data distribution simulation subsystem is used to construct the plan review and evaluation technical distribution by combining the basic data and the upper and lower limits of the data deviation value to fit the data distribution of the basic data;

[0147] The parameter distribution simulation subsystem is used to uniformly simplify the continuously changing carbon footprint assessment parameters into a triangular distribution to simulate the parameter distribution of the carbon footprint assessment parameters;

[0148] The carbon footprint distribution simulation subsystem is used to combine data distribution and parameter distribution and calculate the carbon footprint distribution of the target project through Monte Carlo simulation method;

[0149] The carbon emission calculation subsystem is used to calculate the carbon emissions of the target project based on the carbon footprint distribution.

[0150] In this embodiment, the calculation process of carbon emissions is divided into two parts: parameters and data, and the distribution of the two is evaluated separately. After fitting using a random simulation method, the uncertainty of the global data is calculated. A Bayesian network model is constructed for the input basic data, and various factors that affect the accuracy of the input data are included, so as to calculate the data deviation value between the basic data and the actual carbon emission data. On the other hand, the carbon footprint assessment parameters are processed to obtain the values ​​of the parameters under various circumstances, and the single parameter value is replaced by the interval parameter value; then the interval distribution value of the parameter is fitted with a triangular distribution, and finally the Monte Carlo simulation is performed multiple sampling fittings to obtain the distribution of the output data. Compared with the existing data quality scoring method, the subjectivity of the subjective evaluation of data quality is reduced to a certain extent, so that the calculation results have a certain degree of repeatability. Moreover, compared with the calculation and analysis using only the Monte Carlo method, the uncertainty of the global data is taken into account, making the calculation results more accurate.

[0151] In one embodiment, the deviation value calculation subsystem includes:

[0152] Node construction module, used to construct Bayesian network nodes based on the objective influencing factors of basic data;

[0153] A model building module is used to construct a topological structure model between all Bayesian network nodes through the interpretation structure model method;

[0154] Historical data acquisition module, used to obtain historical basic data related to carbon emissions in target projects;

[0155] The probability calculation module is used to calculate the conditional probability of each Bayesian network node in the topological structure model based on historical basic data analysis;

[0156] Network construction module, used to combine topological structure model and conditional probability to construct Bayesian network;

[0157] The deviation value calculation module is used to calculate the data deviation value between the basic data and the actual carbon emission data based on the transmission effect of conditional probability in the Bayesian network.

[0158] In this implementation, each Bayesian network node represents a random variable. In this implementation, Bayesian network nodes represent various objective factors influencing the input data, including project complexity, project scale, capacity utilization, energy structure, and power consumption. Bayesian network structure learning can clarify the relationships between these objective factors, explore their structural relationships, construct a multi-level hierarchical model, and determine the topological structure model.

[0159] The calculation formula for the data deviation value is as follows:

[0160]

[0161] Where: A i represents the Bayesian network node, ±P represents the Bayesian network node A i The data deviation value corresponding to the basic data, X1, X2, ..., X n Indicates the various deviation levels of the basic data, Represents Bayesian network node A i Conditional probability at the nth level of bias

[0162] Conditional probability The calculation formula is as follows:

[0163]

[0164] Where:

[0165]

[0166] Where: k represents Ai There are k influencing factors A j Existence, FA i |A j Influencing factor A j Under the influence of A i The conditional probability of y1, y2, ..., ym represents the m value states of the Bayesian network nodes, f ym，j Indicates A j is the probability of the mth value state, fym| ym，j Indicates that in A j When A is the mth value state i The probability of the mth value state occurring.

[0167] In one embodiment, the model building module includes:

[0168] A first matrix construction unit is used to construct an adjacency matrix between all objective influencing factors according to the correlation between all objective influencing factors;

[0169] The second matrix construction unit is used to combine the adjacency matrix and the preset identity matrix to calculate the reachability matrix between all objective influencing factors;

[0170] A matrix solving unit is used to solve the reachability matrix and calculate the set classification results of different types of objective influencing factors in the reachability matrix;

[0171] A factor decomposition unit is used to perform inter-level decomposition on all objective influencing factors in the reachable matrix according to the set classification result to obtain an inter-level decomposition result;

[0172] The model building unit is used to combine the set classification results and the inter-level decomposition results to construct a topological structure model among all objective influencing factors.

[0173] In this embodiment, for the matrix element a in the i-th row and j-th column of the adjacency matrix ij , when a ij When it is 0, it means the objective influencing factor A i and objective influencing factors A j There is no correlation between ij When it is 1, it means A i to A j There is a channel of length 1 between them, which can be reached directly, indicating the objective influencing factor A i and objective influencing factors A j There is a correlation between them. The reachability matrix indicates whether there are channels that can be connected between elements, and can effectively reflect the correlation between influencing factors.

[0174] The reachable matrix has the property of transfer, if A ito A j is 1, A j to A n is 1, then A i to A n There must be a channel. Therefore, the reachability matrix can be obtained by analyzing the push property, or it can be obtained by adding the adjacency matrix to the preset identity matrix and performing certain operations. The identity matrix refers to a matrix in which only the diagonal elements are 1 and the rest of the elements are 0. Assuming that the identity matrix is ​​L and the adjacency matrix is ​​A, according to the Boolean operation rules, it can be proved that:

[0175] (A+L) 2 =L+A+A 2

[0176] ↓

[0177] (A+L) k =L+A+A 2 +…+A k

[0178] If the adjacency matrix A satisfies the condition: (A+L) k-1 ≠(A+L) k =(A+L) k+1 =M, then M is the reachability matrix of the adjacency matrix A.

[0179] The set classification results include the reachable set R(A i ), lookahead set A(A i ), common set T and underlying element N. Among them, the reachable set R(A i )={A i =1}, indicating that the element A in the reachable matrix i The column element displayed as 1 in the corresponding row indicates that the element A i Reachable elements. Antecedent set A(A i )={A j =1}, indicating that the element A in the reachable matrix i The row element with 1 in the corresponding column indicates that element A can be reached i Elements of. Common set T = {A i ∈R(A i )∩A(A i )}. The bottom element N = {A i |R(A i )∩A(A i )=A(A i )}, indicating that the element is at the bottom level and no element can reach it.

[0180] Before inter-level decomposition, regional decomposition must be performed first. i )∩R(A j)≠φ, element A i With A j Belong to the same area, otherwise they do not belong to the same area. All elements in the matrix belong to the same area, and there are no subsystems in different areas. Inter-level decomposition is mainly to clearly understand the hierarchical relationship between the elements in the system. The top level represents the ultimate goal of the system, and the next level is the influencing factor of the previous level. The steps of inter-level decomposition are as follows:

[0181] Let L0 = φ, L j ={A i ∈P-L0-L1-...-L j-1 |R i-1 (A i )∩A j-1 (A i )=R i-1 (A i )}, where j = 1, 2, ..., n, L j represents the objective influencing factors of the jth layer, P represents the total number of objective influencing factors, and φ represents the empty set. j =φ, the interstage decomposition is completed. During the interstage decomposition period, if R(A i )∩A(A i )=R(A i ), then the element A i Extract and continue until the last element. Through inter-level decomposition, multiple objective influencing factors are divided into a multi-layer structure model, which is the topological structure model between all objective influencing factors.

[0182] In one embodiment, the probability calculation module includes:

[0183] A preprocessing unit, used for performing data preprocessing on historical basic data to obtain preprocessed historical data;

[0184] A frequency calculation unit, used to calculate the frequency of each Bayesian network node taking values ​​for pre-processed historical data;

[0185] A probability distribution calculation unit, used to calculate the marginal probability distribution and conditional probability distribution of Bayesian network nodes according to the value frequency;

[0186] The conditional probability calculation unit is used to calculate the conditional probability of each Bayesian network node based on the topological structure model and in combination with the edge probability distribution and the conditional probability distribution.

[0187] In this embodiment, data preprocessing mainly includes a data cleansing step and a missing value processing step. Data cleaning can check for errors or outliers in the data and correct or delete them. For missing values, you can choose to fill missing values, delete missing values, or use interpolation methods to fill them. Preprocessed historical data is obtained through data preprocessing, and then for each Bayesian network node, the frequency of occurrence of each value is calculated. At the same time, the value frequency of each node is counted, and the probability of each value is calculated. For each node, the marginal probability distribution is calculated based on the value frequency. The marginal probability distribution represents the probability distribution of the node without considering other nodes. Based on the topological structure model and the marginal probability distribution, the conditional probability distribution is calculated. That is, for each node, the probability distribution of the node value is calculated under the condition of the given parent node value. Finally, the conditional probability of each Bayesian network node is calculated by combining the topological structure model, the marginal probability distribution, and the conditional probability distribution. That is, based on the node's parent node and the conditional probability distribution, the probability of each node under the condition of the given parent node value is calculated.

[0188] In one embodiment, the probability calculation module may further include:

[0189] The historical score acquisition unit is used to obtain the historical data quality score of the historical basic data.

[0190] The random distribution fitting unit is used to fit the random distribution of historical basic data based on the quality score of historical data and by constructing a beta distribution.

[0191] The probability calculation unit is used to calculate the conditional probability of each Bayesian network node in the topological structure model according to the random distribution of data.

[0192] In this embodiment, the beta distribution is a density function that is a conjugate prior distribution of the Bernoulli distribution and the binomial distribution. The probability density function of the beta distribution is determined by the endpoints a, b and the morphological parameters α, β. Under different morphological parameters, the beta distribution presents different distribution forms. Therefore, different endpoints and morphological parameters can almost approximate the distribution of various forms such as normal, lognormal, uniform and exponential distributions. Therefore, after obtaining the historical data quality score of the historical basic data, the historical basic data can be converted into various forms of probability distribution based on the historical data quality score, so as to approximately simulate the actual distribution of the data, thereby reducing the uncertainty of the data. The specific formula is as follows:

[0193]

[0194] Where: x is the historical basic data, f represents the probability distribution of the historical basic data, and Γ represents the Γ function.

[0195] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of protection of the present application is limited to these examples. In line with the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of different aspects of one or more embodiments of the present application as above, which are not provided in detail for the sake of simplicity.

[0196] The one or more embodiments of this application are intended to encompass all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of this application should be included in the scope of protection of this application.

Claims

1. A method for calculating carbon emissions during the life cycle of electricity, characterized in that: The steps include: Obtain carbon footprint assessment parameters and basic data related to carbon emissions in target projects; The Bayesian network method is used to calculate the data deviation value between the basic data and the actual carbon emission data, including the following steps: Constructing Bayesian network nodes based on the objective influencing factors of the basic data; constructing a topological structure model between all the Bayesian network nodes using the interpretive structural model method; obtaining historical basic data related to the carbon emissions in the target project; and analyzing and calculating the conditional probability of each Bayesian network node in the topological structure model based on the historical basic data; Constructing a Bayesian network by combining the topological structure model and the conditional probability; Calculating a data deviation value between the basic data and the actual carbon emission data based on the transmission effect of the conditional probability in the Bayesian network; Among them, constructing a topological structure model between all the objective influencing factors by the explanatory structural model method includes the following steps: constructing an adjacency matrix between all the objective influencing factors based on the correlation between all the objective influencing factors; combining the adjacency matrix and a preset unit matrix to calculate the reachable matrix between all the objective influencing factors; solving the reachable matrix and calculating the set classification results of the objective influencing factors of different types in the reachable matrix; performing inter-level decomposition on all the objective influencing factors in the reachable matrix based on the set classification result to obtain an inter-level decomposition result; combining the set classification result and the inter-level decomposition result to construct a topological structure model between all the objective influencing factors; The calculation formula of the data deviation value is as follows: Where: represents the Bayesian network node, Represents the Bayesian network node The data deviation value corresponding to the basic data, Indicates the various deviation levels of the basic data, Represents the Bayesian network node The conditional probability in the nth level of the deviation ; The conditional probability The calculation formula is as follows: Where: Where: k express have k Influencing factors exist, Influencing factors Under the influence The conditional probability of Represents the Bayesian network node m Value state, express For the m The probability of a value state, Indicates For the m When the value is in the state Appear m The probability of a value state; Constructing a program review and evaluation technical distribution by combining the basic data and the upper and lower limits of the data deviation value to fit the data distribution of the basic data; Simplifying the continuously changing carbon footprint assessment parameters into a uniform triangular distribution to simulate the parameter distribution of the carbon footprint assessment parameters; Combining the data distribution and the parameter distribution and calculating by Monte Carlo simulation method, the carbon footprint distribution of the target project is obtained; The carbon emissions of the target project are calculated based on the carbon footprint distribution.

2. The method for calculating carbon emissions during the life cycle of electricity according to claim 1, characterized in that: The step of analyzing and calculating the conditional probability of each Bayesian network node in the topological structure model based on the historical basic data comprises the following steps: Performing data preprocessing on the historical basic data to obtain preprocessed historical data; Calculating the frequency of each Bayesian network node taking values ​​of the pre-processed historical data respectively; Calculate the marginal probability distribution and conditional probability distribution of the Bayesian network nodes based on the value frequency; The conditional probability of each of the Bayesian network nodes is calculated based on the topological structure model and in combination with the edge probability distribution and the conditional probability distribution.

3. A system for calculating carbon emissions during the life cycle of electricity, characterized in that: The system comprises: The data acquisition subsystem is used to obtain carbon footprint assessment parameters and basic data related to carbon emissions in the target project; a deviation value calculation subsystem, configured to calculate the data deviation value between the basic data and the actual carbon emission data using a Bayesian network method; The deviation value calculation subsystem includes: A node construction module, used to construct Bayesian network nodes according to the objective influencing factors of the basic data; A model building module, used for constructing a topological structure model between all the Bayesian network nodes by using an interpretive structural model method; A historical data acquisition module, used to acquire historical basic data related to the carbon emissions in the target project; A probability calculation module, configured to calculate the conditional probability of each of the Bayesian network nodes in the topological structure model based on the historical basic data analysis; A network construction module, configured to construct a Bayesian network by combining the topological structure model and the conditional probability; a deviation value calculation module, configured to calculate a data deviation value between the basic data and the actual carbon emission data based on the transfer effect of the conditional probability in the Bayesian network; Wherein, the model building module includes: A first matrix construction unit is configured to construct an adjacency matrix between all the objective influencing factors according to the correlations between all the objective influencing factors; A second matrix construction unit is configured to calculate a reachable matrix between all the objective influencing factors by combining the adjacency matrix and a preset identity matrix; A matrix solving unit, configured to solve the reachable matrix and calculate a set classification result of different types of objective influencing factors in the reachable matrix; A factor decomposition unit, configured to perform inter-level decomposition on all the objective influencing factors in the reachable matrix according to the set classification result to obtain an inter-level decomposition result; A model building unit, configured to construct a topological structure model among all the objective influencing factors by combining the set classification result and the inter-level decomposition result; The calculation formula of the data deviation value is as follows: Where: represents the Bayesian network node, Represents the Bayesian network node The data deviation value corresponding to the basic data, Indicates the various deviation levels of the basic data, Represents the Bayesian network node The conditional probability in the nth level of the deviation ; The conditional probability The calculation formula is as follows: Where: Where: k express have k Influencing factors exist, Influencing factors Under the influence The conditional probability of Represents the Bayesian network node m Value state, express For the m The probability of a value state, Indicates For the m When the value is in the state Appear m The probability of a value state; A data distribution simulation subsystem is used to construct a program review and evaluation technology distribution by combining the basic data and the upper and lower limits of the data deviation value to fit the data distribution of the basic data; a parameter distribution simulation subsystem, configured to uniformly simplify the continuously changing carbon footprint assessment parameters into a triangular distribution, so as to simulate the parameter distribution of the carbon footprint assessment parameters; A carbon footprint distribution simulation subsystem, configured to combine the data distribution and the parameter distribution and calculate the carbon footprint distribution of the target project using a Monte Carlo simulation method; The carbon emission calculation subsystem is used to calculate the carbon emission of the target project according to the carbon footprint distribution.

4. The system for calculating carbon emissions during the life cycle of electricity according to claim 3, characterized in that: The probability calculation module includes: A preprocessing unit, configured to perform data preprocessing on the historical basic data to obtain preprocessed historical data; A frequency calculation unit, configured to respectively calculate the frequency of values ​​taken by each of the Bayesian network nodes for the pre-processed historical data; A probability distribution calculation unit, configured to calculate the marginal probability distribution and conditional probability distribution of the Bayesian network nodes according to the value frequencies; A conditional probability calculation unit is used to calculate the conditional probability of each of the Bayesian network nodes based on the topological structure model in combination with the edge probability distribution and the conditional probability distribution.

Images