Search

CN-122023087-A - Engineering project carbon emission calculation method based on financial expenditure data

CN122023087ACN 122023087 ACN122023087 ACN 122023087ACN-122023087-A

Abstract

The invention discloses a project carbon emission calculation method based on financial expenditure data, and belongs to the technical field of carbon emission calculation and project management. The method aims to solve the problems of low confidence level of an extraction result and poor parameter accuracy when the carbon emission related technical attribute parameters are automatically extracted from the object description text by utilizing a natural language processing technology. The method comprises the steps of analyzing object description texts in financial expenditure records through a natural language processing technology to extract technical attribute parameters and generate confidence scores, triggering a cross verification process if the confidence is lower than a preset threshold, acquiring verified technical attribute parameters from other systems of engineering projects according to a data association mapping rule for comparison, determining finally adopted parameters according to the credibility of a data source, and feeding back the result to a model and a professional dictionary to realize closed-loop self-optimization. The method is mainly used for carbon footprint accounting, carbon emission monitoring and emission reduction management of engineering projects, and provides reliable data support for low-carbon decisions.

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 project carbon emission calculation method based on the financial expenditure data is characterized by comprising the following steps of: S1, acquiring financial expenditure data of an engineering project, wherein the financial expenditure data comprises a plurality of expenditure records, and each expenditure record at least comprises an article category, an expenditure amount and an article description text; Analyzing the object description text by adopting a natural language processing technology, identifying and extracting technical attribute parameters related to carbon emission, and generating a confidence score for the technical attribute parameters; If the confidence score is lower than a preset threshold, triggering a cross-validation process, and acquiring at least one structured data source and corresponding validated technical attribute parameters from a system of the engineering project in a correlated manner based on a preset data correlation mapping rule according to the item category and the key identifier of the current expenditure record; Comparing the extracted technical attribute parameters with the verified technical attribute parameters, and determining the finally adopted technical attribute parameters according to a preset data source credibility evaluation rule; Taking the finally adopted technical attribute parameters and the corresponding article description text as training samples, and feeding back to a model corresponding to the natural language processing technology and a preset engineering material and process professional dictionary to realize iterative optimization of the model and the dictionary; The method comprises the steps of verifying validity of verified technical attribute parameters, screening the verified technical attribute parameters and corresponding structured data sources, wherein the verified technical attribute parameters and the corresponding structured data sources are used for screening the update time within set timeliness and the key features are matched by combining the update time of the key features and the data sources in the article description text during verification.
  2. 2. The method for calculating carbon emissions of an engineering project based on financial expenditure data of claim 1, further comprising: S2, according to a mapping relation between the predefined object type and the unit cost, converting the expenditure amount of each expenditure record into an object physical quantity; s3, acquiring carbon emission factors corresponding to each expenditure record, wherein the carbon emission factors specifically comprise: S31, a multi-level carbon emission factor library is pre-built, wherein a plurality of sub-class carbon emission factors are associated with at least one article class, and each sub-class carbon emission factor corresponds to one or more technical attribute parameters; s32, analyzing the object description text in the expenditure record, and identifying and extracting technical attribute parameters related to carbon emission through a natural language processing technology; s33, inquiring a multi-level carbon emission factor library according to the article category and the identified technical attribute parameters, and matching and obtaining a corresponding sub-class carbon emission factor; S4, multiplying the physical quantity of the object with a corresponding carbon emission factor for each expenditure record, and calculating the carbon emission quantity of the expenditure record; S5, summing the carbon emission of all expenditure records to obtain the total carbon emission of the engineering project; wherein further comprising: step S32, analyzing the object description text in the expenditure record, and identifying and extracting technical attribute parameters related to carbon emission through natural language processing technology specifically comprises the following steps: S321, predefining an engineering material and process professional dictionary which stores standard names, common aliases, abbreviations and standard technical attribute parameters mapped to the names, common aliases and abbreviations of the articles in the engineering field; s322, when analyzing the article description text, firstly, preprocessing the text and standardizing terms by using engineering materials and a process professional dictionary, and mapping nonstandard expressions in the text into canonical technical terms; S323, executing a natural language processing model to identify and extract technical attribute parameters based on the standardized text, and generating a confidence score for each extracted parameter; s324, pushing the expenditure record and the article description text thereof to a manual auditing interface if the confidence score is lower than a preset threshold, receiving correct technical attribute parameters input after manual auditing, and feeding back the parameters and the corresponding relations to an engineering material and process professional dictionary for optimizing the dictionary and a natural language processing model.
  3. 3. The method of claim 2, further comprising an automated cross-validation step prior to pushing the low confidence expense record to the manual review interface in step S324: s3241, automatically triggering a cross-validation process when the confidence score of the identified technical attribute parameter is lower than a preset threshold value; s3242, the cross verification process comprises the steps of acquiring associated structural data from other data systems of engineering projects according to the object category of the current expense record, wherein the other data systems comprise a material demand planning system or a building information model; s3243, comparing the technical attribute parameters identified by the natural language processing technology with the technical attribute parameters extracted from the associated structured data; S3244, if the comparison is consistent, automatically adopting the technical attribute parameters, taking the technical attribute parameters and the corresponding object description text as training samples to feed back and update the natural language processing model, and pushing the record to a manual auditing interface.
  4. 4. The method for calculating carbon emissions of engineering projects based on financial expenditure data as claimed in claim 3, wherein step S3242 obtains the associated structured data from other data systems of the engineering projects and the comparison process of step S3243, specifically comprises: S3242a, constructing a data association mapping table which defines one or more potential association paths between key identifications in financial expenditure records and key identifications of components or bill of materials in a material demand planning system, a building information model; S3242b, according to the data association mapping table, attempting to match one or more associated structured data sources for the current expense record, and generating an association confidence level for each matched data source; S3242c, predefining and dynamically maintaining a source credibility score for each potential data source, wherein the source credibility score is calculated and updated based on the consistency proportion of the data source and the manual auditing result in the history cross-validation; S3243a, when comparing, preferentially selecting the technical attribute parameters provided by the data sources with the associated confidence and source reliability scores higher than the respective threshold values, and comparing; S3243b, adopting the parameters if a plurality of data sources which meet the conditions and provide the same technical attribute parameters exist, and adopting the parameters from the data source with the highest source reliability score if the parameters provided by the plurality of data sources conflict.
  5. 5. The method for calculating carbon emissions of engineering projects based on financial expense data according to claim 1, wherein in step S2, converting the expense amount of each expense record into the physical quantity of the article according to the predefined mapping relation of the article category and the unit cost comprises: S21, constructing and maintaining a dynamic unit cost database, wherein the database records the market unit cost of different article types in different time intervals and/or different regions; S22, extracting transaction time information and project region information from the expenditure record to be processed; s23, inquiring a dynamic unit cost database according to the article category of the expenditure record and combining transaction time information and project region information, and matching to obtain a space-time corresponding dynamic unit cost; S24, replacing the predefined static unit cost by the dynamic unit cost, and dividing the expenditure amount by the dynamic unit cost to obtain the calibrated physical quantity of the object.
  6. 6. The method for calculating carbon emissions of an engineering project based on financial expenditure data according to claim 5, wherein the construction and data acquisition means of the dynamic unit cost database comprises: the dynamic unit cost database integrates a plurality of data sources, including at least one authoritative market quotation platform data source and at least one actual purchasing data source from the inside of the project; in step S23, the rule for matching the acquisition dynamic unit cost is: S231, preferentially inquiring an actual purchasing data source in the project, and if an actual purchasing unit price matched with the type of the object, the transaction time and the project region of the current expenditure record exists in the actual purchasing data source, adopting the actual purchasing unit price as a dynamic unit cost; S232, if the corresponding actual purchase unit price is not matched in the actual purchase data source, inquiring the authoritative market quotation platform data source, acquiring matched market unit cost, and adopting the market unit cost as dynamic unit cost.
  7. 7. The method for calculating carbon emissions of an engineering project based on financial expenditure data according to claim 1, further comprising, after step S5: s6, calculating an uncertainty evaluation index of the total carbon emission of the engineering project, wherein the uncertainty evaluation index specifically comprises the following steps: S61, defining error distribution for calculating key input parameters in the chain, wherein the key input parameters at least comprise unit cost and/or carbon emission factors; S62, repeatedly calculating the calculation process of all expenditure records of the engineering project for a plurality of times based on a Monte Carlo simulation method, and randomly sampling key input parameters according to respective error distribution in each calculation so as to obtain a series of simulation values of total carbon emission; S63, determining a confidence interval and probability distribution of the total carbon emission according to a series of simulation values of the total carbon emission; s64, when the total carbon emission of the engineering project is output, the confidence interval is output in a correlation mode.
  8. 8. The method for calculating carbon emissions of an engineering project based on financial expense data according to claim 7, wherein the calculation of the calculation process of all expense records of the engineering project based on monte carlo simulation method in step S62 is specifically optimized as: s62a, before full Monte Carlo simulation is carried out, global sensitivity analysis is carried out once to determine the contribution degree of each expenditure record to the uncertainty of the total carbon emission; s62b, dividing all expenditure records into a key record set and a non-key record set according to the contribution degree; S62c, in the follow-up Monte Carlo simulation, performing full-precision simulation on the key record set, namely randomly sampling key input parameters of the key record set in each simulation; And S62d, for the non-critical record set, adopting an approximate processing strategy, wherein the strategy is to calculate a reference carbon emission amount for the non-critical record set at the beginning of simulation, multiply the reference value by a global perturbation factor to represent the fluctuation of the reference carbon emission amount in each subsequent simulation, and generate the perturbation factor based on the overall fluctuation trend of the critical record set in the current simulation.
  9. 9. The method for calculating carbon emissions of an engineering project based on financial expenditure data according to claim 1, further comprising, after step S5: S7, carrying out attribution analysis on carbon emission of the engineering project based on the configured responsibility dimension, wherein the attribution analysis specifically comprises the following steps: S71, predefining a plurality of responsibility dimensions, wherein the responsibility dimensions at least comprise a responsibility main body dimension, an engineering structure dimension and a time stage dimension; s72, constructing a dimension mapping rule base, wherein the rule base defines the mapping relation between characteristic information in the expenditure record and each responsible dimension; S73, determining attribution of each expenditure record under each responsible dimension through a dimension mapping rule base according to the characteristic information contained in each expenditure record; And S74, distributing the carbon emission amount of the expenditure record to corresponding nodes under each responsibility dimension based on the attribution relation, and S75, aggregating the carbon emission amount according to the responsibility dimension to generate a multi-dimensional carbon emission attribution analysis report.
  10. 10. The method for calculating carbon emissions of an engineering project based on financial expenditure data according to claim 9, wherein the dimension mapping rule base in step S72 supports dynamic configuration, and the execution of steps S73 and S74 includes: S73a, when a single expense record has a plurality of possible attributions under a certain responsibility dimension, splitting the carbon emission of the expense record to a plurality of attribution nodes according to a preset weight proportion by adopting a weight distribution strategy; s73b, providing a rule conflict detection mechanism, and prompting a user to perform rule optimization when detecting that the mapping rule has conflict; S74a, supporting traceable inquiry based on attribution results, and reversely inquiring all original expense records forming the aggregation results and detailed calculation processes of the original expense records from the aggregate carbon emission of nodes in any dimension.

Description

Engineering project carbon emission calculation method based on financial expenditure data Technical Field The invention belongs to the technical field of carbon emission calculation and engineering project management, and particularly relates to an engineering project carbon emission calculation method based on financial expenditure data. Background In engineering project full life cycle management, accurate accounting of carbon emissions is a fundamental task to achieve low carbon conversion and green construction. In the prior art, the carbon emission calculation mainly depends on two types of data input, namely, activity data based on field actual measurement or an energy source and material consumption list, and theoretical calculation data based on a design drawing or a material list. However, both of these main approaches face significant challenges in practical applications. The primary difficulty is how to reliably extract the specific technical attribute parameters necessary to calculate carbon emissions from the article description text, which is abbreviated and non-standardized in the financial records. Description in financial systems is based mostly on accounting subjects or internal habits, often using acronyms, aliases or non-technical terms, e.g. "C30 commodity concrete" is denoted "business concrete" or "business mix". And the accurate matching of the carbon emission factor library is highly dependent on the specific technical specifications of materials or processes, such as the strength grade of concrete, the type of steel, and the like. When such text is parsed directly using natural language processing techniques, the technical parameters identified by the model tend to be ambiguous or erroneous due to term non-normative and contextual lack. At present, when the confidence of an automatic analysis result is low, a common processing mode is to rely on manual judgment and correction. However, for large-scale project frequently thousands of financial records, the manual auditing efficiency is low one by one, the cost is high, and the operability of large-scale application is not realized. How to effectively improve the automatic accuracy of extracting key parameters from non-standardized texts on the premise of not completely relying on manpower is a long-standing bottleneck. Secondly, after the technical parameter extraction is realized, how to ensure the accuracy of the extracted parameters faces the difficulty in verification. The correctness of the system to self-identification results is often difficult to self-check due to the lack of independent, comparable standardized data sources. Once the initial resolution deviates, a subsequent carbon emission factor matching error is caused, and the error systematically affects the final accounting result. An improvement attempt has been to introduce structured data, such as bill of materials or building information model data, into other management systems for projects as a reference. However, in actual operation, the association relationship between the financial records and other system data is often ambiguous or difficult to automatically correspond. The identifiers and classification systems among different systems are inconsistent, so that the cross-system automatic retrieval and comparison of related technical parameters become complex. In addition, there may be conflicts in the information provided by multiple data sources, and the lack of an objective set of rules to determine and select which data source is more trusted makes it difficult to reliably establish an automated cross-validation mechanism. Another key issue exists in the initial link of data conversion, namely how to reasonably convert the financial expenditure amount into the physical consumption of the item. The currently common approach is to scale the division based on a predefined, static mapping of item categories to unit costs. However, the unit cost of engineering materials and services is significantly affected by market price fluctuations, regional differences, purchasing channels and specific specifications. Scaling with a fixed, averaged unit cost introduces significant errors that result in estimated physical quantities that do not match the actual consumption of the project. While building a dynamic cost database is an obvious direction of improvement, in practice, there are difficulties in how to obtain a high quality, highly time-efficient cost data source and achieve accurate matching. The actual purchase data inside the project is most representative, but the format may not be uniform and not necessarily cover all expenditure categories, and the external authoritative market data is good in timeliness, but may have a difference with the actual purchase cost of the specific project. How to design priority and matching rules, so that the unit cost for conversion is close to project reality, and the availability of data is ensured, is a real problem to be