Search

CN-122023088-A - Full-period carbon emission statistical method based on project financial expenditure

CN122023088ACN 122023088 ACN122023088 ACN 122023088ACN-122023088-A

Abstract

The invention discloses a full-period carbon emission statistical method based on project financial expenditure, and belongs to the technical field of carbon emission accounting and environmental information. Aiming at the technical problem that the method is difficult to convert and count the full-period carbon emission amount efficiently and accurately directly from project financial expenditure data in the prior art, the method is provided that the expenditure detail data in the full life cycle of the project is collected and collected to the carbon emission activity category according to the application, the text analysis module is further called to extract the material attribute and map the material attribute to the standard mark, the expenditure value is converted into the physical consumption amount by dynamically matching the physical parameter according to the expenditure application classification through the expenditure-physical quantity conversion model, the physical consumption amount and the carbon emission activity category are input into the multi-level emission factor matching module, the carbon emission factor is obtained through the three-level sequential matching process, and the total full-period carbon emission amount is finally accumulated. The method is mainly used for realizing automation and standardized full-cycle carbon emission accounting based on project financial data.

Inventors

  • LAI JINSA

Assignees

  • 河北建工集团有限责任公司

Dates

Publication Date
20260512
Application Date
20251226

Claims (10)

  1. 1. The full-period carbon emission statistical method based on project financial expenditure is characterized by comprising the following steps of: Collecting financial expenditure detail data of a target item in a full life cycle through a financial data interface, wherein the financial expenditure detail data comprises expenditure data, expenditure object names, expenditure occurrence dates and expenditure purpose classification codes; Invoking a text analysis module to process the expenditure object name, extracting the material key attribute and mapping the material key attribute into a standardized material identification name; aiming at the collected financial expenditure data, a financial-physical quantity mapping model is called, expenditure values are converted into physical consumption data, and the conversion process determines applicable physical parameter types according to the expenditure use classification codes; matching and obtaining corresponding unit physical parameter values from a corresponding physical parameter database according to the standardized material identification name and the expenditure occurrence date; Inputting the physical consumption data, the standardized material identification names and the carbon emission activity categories into a multi-level emission factor matching module, executing a three-level sequence matching process, performing name complete matching from a general factor library at the first level, performing category and specification matching from a subdivision factor library at the second level, performing similar material replacement matching from a correlation factor library at the third level, generating a confidence score by each level matching, stopping matching when the confidence score of a certain level matching reaches a threshold of 75% -95%, and outputting carbon emission factor data corresponding to the level; The method comprises the steps of obtaining physical consumption data, obtaining carbon emission factor data, calculating carbon emission data corresponding to financial expenditure detail data according to the physical consumption data and the carbon emission factor data obtained by matching, accumulating the carbon emission data in the same stage according to the stage division of the project life cycle to generate carbon emission sub-item summarized data of each life cycle stage, and summing the carbon emission sub-item summarized data of each life cycle stage to generate full-cycle carbon emission total data of a target project.
  2. 2. The project financial expenditure based full cycle carbon emission statistical method of claim 1, wherein the multi-level emission factor matching module performs a three-level sequential matching process specifically comprising: Establishing a local cache database comprising a general factor library, a subdivision factor library and an association factor library, and setting a main matching channel and a standby matching channel; Starting a main matching channel, sequentially executing the complete matching of the first-stage names and the matching of the second-stage category specifications, and outputting a matching result by the channel when the confidence score of any one-stage matching result reaches a threshold value of 75% -95%; When the two-stage matching of the main matching channel does not reach the threshold value, starting a standby matching channel, and executing third-stage similar material substitution matching, wherein the stage matching selects an emission factor corresponding to a material with similarity within the range of 0.7-0.9 from a correlation factor library as a candidate output by calculating cosine similarity of a material key attribute vector; And the module is internally provided with a decision unit for receiving the matching result of the main channel and the standby channel, the result is directly adopted when only one channel outputs the result, and when both channels output the result, the decision unit calls a preset rule base, and one rule is automatically selected from the rule base to execute according to the classification code of the current matching material, so that finally output carbon emission factor data is determined.
  3. 3. The full-cycle carbon emission statistical method based on project financial expenditure according to claim 2, wherein a dynamic threshold setting module is further provided in the multi-level emission factor matching module; The dynamic threshold setting module is called before each starting of a matching process, and performs the following operations of receiving the input standardized material identification name and the carbon emission activity category, inquiring a preset threshold reference mapping table according to the carbon emission activity category to obtain an initial threshold reference value, wherein the threshold reference value covers 75% -95% of the range of values among different activity categories, and calling a level threshold adjustment coefficient to correct the initial threshold reference value according to the current matching level to be executed, wherein the adjustment coefficient of the first level matching is 1.0, the adjustment coefficient of the second level matching is 0.95-0.98, the adjustment coefficient of the third level matching is 0.90-0.94, and outputting a corrected level exclusive confidence threshold value for controlling stopping judgment of the level matching.
  4. 4. The project financial expenditure based full cycle carbon emission statistical method of claim 2, further comprising an intelligent routing pre-decision step before the multi-level emission factor matching module initiates the matching process: the intelligent routing prejudging step inquires a material characteristic prejudging knowledge base according to the standardized material identification name to obtain a prejudging matchable level corresponding to the current material identification name; According to the pre-judging matchable level, executing channel selection weight calculation, namely distributing 100% of execution weight for the main matching channel if the pre-judging is a first level or a second level, distributing 0% -30% of execution weight for the main matching channel if the pre-judging is a third level, and distributing 70% -100% of execution weight for the standby matching channel; In the subsequent matching process, whether each channel is started and the execution sequence thereof are dynamically determined from high to low according to the execution weight.
  5. 5. The full-cycle carbon emission statistical method based on project financial expenditure according to claim 2, wherein an arbitration decision rule is stored in the preset rule base; When the arbitration decision rule is selected and executed by the decision unit, triggering the following operations of calculating a confidence score difference value of output results of the main matching channel and the standby matching channel, adopting a channel result with high confidence score if the difference value is more than or equal to 10%, further querying a history matching record library if the difference value is less than 10%, acquiring verification accuracy of history matching results of each channel under the current material classification coding, and adopting a result output by a channel with high history verification accuracy.
  6. 6. The project financial expenditure based full cycle carbon emission statistical method of claim 5 wherein when said arbitration decision rule is selected for execution, a data sufficiency check step is first performed prior to said step of calculating a confidence score difference; Inquiring the history matching record library to obtain all the number of history matching records which are the same as the current material classification codes; If the number of the history matching records is greater than or equal to a preset statistical sample threshold range, continuing to execute the step of calculating the confidence score difference value and the subsequent steps; if the number of the history matching records is smaller than the statistical sample threshold range, triggering a standby arbitration logic; the standby arbitration logic judges whether the carbon emission activity category belongs to a category list with a preset importance score higher than a set threshold value, if so, the result output by the main matching channel is adopted, and if not, the result output by the channel with a high confidence score is adopted.
  7. 7. The full cycle carbon emission statistical method based on project financial expenditures of claim 2, wherein said third level of similar material substitution matching further comprises a precision compensation step; After selecting candidate emission factors from a correlation factor library through cosine similarity, calling a material attribute influence weight library, wherein the material attribute influence weight library is pre-stored with influence coefficients of different attributes on carbon emission, calculating a compensation adjustment coefficient according to the attribute difference of the extracted material key attribute and the candidate material and the corresponding influence coefficient, wherein the range of the compensation adjustment coefficient is 0.9-1.1, and multiplying the candidate emission factor data by the compensation adjustment coefficient to obtain a final output factor after precision compensation.
  8. 8. The full cycle carbon emission statistical method based on project financial expenditures of claim 7, wherein said accuracy compensating step is performed by a physical property correction model; The physical characteristic correction model firstly searches a basic compensation coefficient from a preset correction coefficient table according to the extracted predefined physical parameters in the material key attribute, and the range of the basic compensation coefficient is 0.5-1.5; Secondly, calling an attribute influence weight library, and calculating an attribute difference correction term by combining the attribute difference between the material key attribute and the candidate material and the corresponding weight coefficient, wherein the range of the attribute difference correction term is-0.2 to +0.2; Adding the basic compensation coefficient and the attribute difference correction term to obtain a final compensation adjustment coefficient; And multiplying the candidate emission factor data by the compensation adjustment coefficient to obtain a final output factor after precision compensation.
  9. 9. The full-cycle carbon emission statistical method based on project financial expenditure according to claim 2, wherein the preset rule base is connected with a feedback learning module; The feedback learning module continuously collects accuracy evaluation of carbon emission factor data finally output by the decision unit and correlates an evaluation result to a specific rule and a material classification code triggering the decision, periodically counts the evaluation accuracy of each rule under different material classification codes, and when the statistical sample size of a certain rule under a specific material classification code reaches a preset threshold and the evaluation accuracy is lower than 70% in three continuous statistical periods, automatically sends an instruction to the rule base, and reduces the application priority of the rule under the material classification code or sets the rule to be in a disabled state.
  10. 10. The project financial expenditure based full cycle carbon emission statistical method of claim 9 wherein the rule adjustment operations performed by the feedback learning module are performed in the following order: Executing the operation of a real-time performance early warning unit, when the occurrence frequency of a negative feedback signal with inconsistent matching results under specific material classification codes exceeds a frequency threshold range of 50% -60% in a single statistical period, immediately sending an instruction to the rule base, adjusting the application priority of the rule triggering the statistical result under the classification codes to the lowest level, and starting an adjustment cooling period with 2-4 statistical periods; during the adjustment cooling period, the feedback learning module pauses the periodic statistical evaluation accuracy calculation of the rule under the classification coding; and after the cooling period is over, recovering the periodic statistics, and continuously executing the rule adjustment operation after the periodic statistics reach the threshold range.

Description

Full-period carbon emission statistical method based on project financial expenditure Technical Field The invention relates to the technical field of carbon emission accounting and environmental information. More particularly, the present invention relates to a full cycle carbon emission statistical method based on project financial expenditures. Background In prior art practice of full cycle carbon emission statistics based on project activity, it is common to rely on direct monitoring or statistics of physical quantities actually consumed by the project (e.g., kilowatt-hours of energy, kilograms of material, etc.). However, this physical quantity-based approach faces a series of challenges in practical applications. First, there is a barrier to the conversion of project expenditure data into physical consumption. The core management records of the project are usually in the form of financial expenditures and include payment amount, supplier, time and purpose classification, etc., while the carbon emissions accounting requires specific physical consumption. The prior art often lacks a systematic, automated method for efficiently and accurately converting such ubiquitous, standardized financial expenditure records into the various physical quantity data required for calculating carbon emissions. Common practice is to rely on manual review of invoices, inventory or evaluation, which is not only labor intensive, inefficient, but also prone to conversion errors or omissions due to human factors or data opacity. The root cause is that there is a lack of a uniform, dynamically associable conversion bridge between the amount of expenditure and the physical quantity, and particularly when the project involves the expenditure of multiple categories (e.g. building materials, energy sources, services), each category may require different conversion parameters (e.g. unit price, density, energy consumption coefficients), it is extremely difficult to manually determine the appropriate parameters for each expenditure and perform the conversion. Second, matching accurate carbon emission factors for a wide variety of materials is a complex and error-prone link. The accuracy of the carbon emission factor directly determines the reliability of the final accounting result. Existing methods typically rely on a fixed emission factor database for query matching. However, when dealing with thousands of materials of varying specifications that are depicted in a project purchase, simple exact name matching often fails because it is difficult for the database to overlay all of the specific material names. At this time, technicians often need to manually judge, classify or select approximation factors empirically, and this process is highly dependent on personal knowledge, has great subjectivity and uncertainty, and is difficult to perform on a large scale and standardized. When fuzzy matching is attempted through an algorithm, the difficulty of balancing the matching breadth with the result accuracy is faced, namely, how to design matching logic to find a correlation factor and quantitatively evaluate the reliability of the matching result so as to avoid using an improper factor is a technical difficulty in the field. Finally, carbon emission accounting systems lack the ability to continue self-optimization and adaptability. Even if an initial set of matching rules or threshold systems are built, problems are encountered in actual operation. For example, a fixed confidence threshold may not be able to accommodate the differing requirements of different classes of material for matching accuracy, and the rules for arbitrating conflicting results may not perform well on certain classes of material, but these problems cannot be automatically identified and corrected due to the lack of an effective feedback mechanism. Depending on the initial settings, it is difficult to learn and adjust its internal decision logic from actual use. This makes long-term maintenance and improvement of accuracy of the system require continuous manual intervention and expert analysis, increasing the operating costs and also limiting its stability and reliability in different projects or over time. The deep difficulty is how to design a mechanism, which can automatically collect effect feedback and positioning problem links and safely and progressively adjust related parameters or rules on the premise of not influencing the normal operation of the system. The three interrelated problems-efficient and accurate conversion from expenditure to physical quantity, reliable and intelligent matching of mass material emission factors, and continuous self-optimization of system decision logic-constitute the main technical bottleneck faced when project expenditure data are utilized for full-period carbon emission statistics, and restrict large-scale, automatic and high-precision application of the method. Disclosure of Invention It is an object of the present invention