Carbon factor library generation method and device and carbon factor matching method and device

Abstract

The present invention provides a method and device for generating a carbon factor library, and a method and device for matching carbon factors. The carbon factor library generation method includes: obtaining multiple matching data points of material data and carbon factor data to construct a training sample set, training a first language processing model to obtain a mapping model; obtaining material attribute information used to generate an extended carbon factor library, inputting the information into the mapping model, and obtaining target carbon factor data corresponding to the material attribute information; determining the carbon emission item name of the extended carbon factor data based on the material attribute information, and determining the carbon factor of the extended carbon factor data based on the carbon factor of the target carbon factor data; and generating an extended carbon factor library using the extended carbon factor data. Through the present invention, an extended carbon factor library is generated, increasing the carbon factor data that can be matched with building material data, thereby improving the success rate of automatic carbon factor matching, reducing manual matching, and improving carbon factor matching efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of building data processing, and in particular to a carbon factor library generation method and device, and a carbon factor matching method and device. Background Art

[0002] As global warming becomes increasingly serious, reducing carbon emissions has become a global consensus, and green and low-carbon buildings have become the mainstream in China and even the world. More and more government agencies, design institutes, architectural institutes, and construction companies are demanding accurate measurement of building carbon emissions.

[0003] Among them, the carbon factor is a key indicator for measuring the relationship between energy consumption and carbon emissions. When calculating the carbon emissions of energy consumption, it is first necessary to match the carbon factor corresponding to the energy consumption. In the prior art, when matching the carbon factor of building materials, a simple search method is first used to search for the carbon factor corresponding to the building material in the carbon factor library. If the search fails, the carbon factor matching the building material is then queried and analyzed based on manual experience. In this regard, the inventors have found that in the current construction field, there are relatively few standard carbon factors related to building materials, while there are many types of building-related materials. The types of building materials actually used do not correspond one-to-one with the standard carbon factors. The data gap in between is huge, which results in a large part of the carbon factor matching usually not being able to be searched, and requiring reliance on manual query and analysis. This method is not only inefficient but also prone to errors. At the same time, when using this type of software, users of existing carbon metering software do not share data with each other, resulting in each user investing a lot of time in repeated carbon factor query and analysis work, resulting in a waste of human resources.

[0004] Therefore, how to improve the efficiency of carbon factor matching and save human resources has become a technical problem that needs to be solved urgently in this field. Summary of the Invention

[0005] The purpose of the present invention is to provide a carbon factor library generation method and device and a carbon factor matching method and device, which are used to solve the technical problems in the prior art.

[0006] On the one hand, to achieve the above-mentioned purpose, the present invention provides a method for generating a carbon factor library.

[0007] The method includes: obtaining a plurality of matching data that match material data and carbon factor data, wherein the material data includes a plurality of material attribute fields, and the carbon factor data includes a carbon emission item name and a carbon factor; constructing a training sample set according to the plurality of matching data; using the training sample set to train a first language processing model to obtain a mapping model; obtaining material attribute information for generating an extended carbon factor library; inputting the material attribute information into the mapping model to obtain target carbon factor data corresponding to the material attribute information; determining the carbon emission item name of the extended carbon factor data according to the material attribute information, and determining the carbon factor of the extended carbon factor data according to the carbon factor of the target carbon factor data; and generating an extended carbon factor library using the extended carbon factor data.

[0008] Furthermore, the material data includes a material name field and a specification field, and the step of determining the carbon emission item name of the extended carbon factor data based on the material attribute information includes: obtaining the material name and specification in the material attribute information; and using the material name and specification as the carbon emission item name of the extended carbon factor data.

[0009] Furthermore, the carbon factor data also includes emission measurement units, and the steps of determining the carbon factor of the extended carbon factor data based on the carbon factor of the target carbon factor data include: when the engineering measurement units of the material attribute information and the emission measurement units of the target carbon factor data are different, calculating the conversion coefficient of converting the emission measurement units into engineering measurement units; converting the carbon factor of the target carbon factor data into an extended carbon factor measured in engineering measurement units according to the conversion coefficient; and using the extended carbon factor as the carbon factor of the extended carbon factor data; the method also includes: using the engineering measurement units as the emission measurement units of the extended carbon factor data.

[0010] Furthermore, the carbon factor data also includes emission units, the material data also includes an engineering unit field and a unit conversion coefficient, and determining the carbon factor of the extended carbon factor data based on the carbon factor of the target carbon factor data includes: using the carbon factor of the target carbon factor data as the carbon factor of the extended carbon factor data; the method also includes: using the emission unit of the target carbon factor data as the emission unit of the extended carbon factor data; the material attribute information also includes engineering units, and the material attribute information is input into the mapping model to obtain the target emission unit and target conversion coefficient corresponding to the engineering unit information; the method also includes: generating a conversion coefficient library based on the engineering unit, the target emission unit and the target conversion coefficient.

[0011] Furthermore, the step of constructing a training sample set based on multiple matching data includes: segmenting the material data of the matching data to obtain a phrase including multiple words; determining the feature words and non-feature words of the phrase; generating a material phrase using all the feature words or all the feature words and at least one non-feature word; constructing initial sample data based on the material phrase and the carbon factor data of the matching data; generating an initial sample set using the initial sample data corresponding to the multiple matching data; and expanding the initial sample set to obtain a training sample set.

[0012] Furthermore, the step of expanding the initial sample set to obtain a training sample set includes: using the material phrases in the initial sample set to train a second language processing model to obtain a data expansion model, wherein the second language processing model includes an encoder and a decoder, the encoder is used to encode input data to generate intermediate variables, and the decoder is used to decode the intermediate variables to obtain input data; generating a number of intermediate variables that conform to preset distribution rules; inputting the several intermediate variables into the data expansion model to obtain an extended material phrase; using the initial sample set to train the first language processing model to obtain an intermediate mapping model; inputting the extended material phrase into the intermediate mapping model to obtain extended carbon factor data corresponding to the extended material phrase; constructing extended sample data based on the extended material phrase and the extended carbon factor data; and adding the extended sample data to the initial sample set to obtain a training sample set.

[0013] Furthermore, the steps of obtaining material property information for generating an extended carbon factor library include: constructing an enhanced knowledge base; obtaining material names and specifications; retrieving enhanced knowledge in the enhanced knowledge base using the material names and specifications; and generating material property information using the material names, specifications and enhanced knowledge.

[0014] On the other hand, to achieve the above object, the present invention provides a carbon factor matching method.

[0015] The carbon factor matching method includes: obtaining target material data; searching for carbon factor data matching the target material data in a standard carbon factor library; if no carbon factor data matching the target material data is found in the standard carbon factor library, searching for carbon factor data matching the target material data in an extended carbon factor library, wherein the extended carbon factor library is generated using any one of the carbon factor library generation methods provided by the present invention.

[0016] On the other hand, to achieve the above-mentioned purpose, the present invention provides a carbon factor library generation device.

[0017] The carbon factor library generation device includes: a first acquisition module, used to obtain multiple matching data that match material data and carbon factor data, wherein the material data includes several material attribute fields, and the carbon factor data includes the carbon emission item name and the carbon factor; a processing module, used to construct a training sample set according to the multiple matching data; a training module, used to train a first language processing model using the training sample set to obtain a mapping model; a second acquisition module, used to obtain material attribute information for generating an extended carbon factor library; an input module, used to input the material attribute information into the mapping model to obtain target carbon factor data corresponding to the material attribute information; a determination module, used to determine the carbon emission item name of the extended carbon factor data according to the material attribute information, and determine the carbon factor of the extended carbon factor data according to the carbon factor of the target carbon factor data; and a generation module, used to generate an extended carbon factor library using the extended carbon factor data.

[0018] On the other hand, to achieve the above-mentioned purpose, the present invention provides a carbon factor matching device.

[0019] The carbon factor matching device includes: an acquisition module for acquiring target material data; a first search module for searching for carbon factor data matching the target material data in a standard carbon factor library; and a second search module for searching for carbon factor data matching the target material data in an extended carbon factor library if the first search module cannot find carbon factor data matching the target material data in the standard carbon factor library, wherein the extended carbon factor library is generated using any one of the carbon factor library generation methods provided by the present invention.

[0020] To achieve the above objectives, the present invention also provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0021] To achieve the above object, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when executed by a processor.

[0022] The carbon factor library generation method and device and carbon factor matching method and device provided by the present invention obtain multiple matching data that match material data and carbon factor data, use the data to construct a training sample set to train a language processing model, and obtain a mapping model. Then, material attribute information used to generate an extended carbon factor library is obtained and input into the mapping model to obtain corresponding target carbon factor data. Finally, new carbon factor data is constructed using the material attribute information and the corresponding target carbon factor data to generate an extended carbon factor library. Through the present invention, more carbon factor data is expanded on the basis of the standard carbon factor library using matching data to generate an extended carbon factor library, increasing the carbon factor data that can be matched with building material data, thereby improving the success rate of automatic carbon factor matching, reducing manual matching, and improving carbon factor matching efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0024] Figure 1 Flowchart of the method for generating a carbon factor library provided in Example 1 of the present invention;

[0025] Figure 2 A flow chart of the carbon factor matching method provided in Example 2 of the present invention;

[0026] Figure 3 A block diagram of a carbon factor library generation device provided in Example 3 of the present invention;

[0027] Figure 4 A block diagram of a carbon factor matching device provided in Embodiment 4 of the present invention;

[0028] Figure 5 This is a hardware structure diagram of the computer device provided in Example 5 of the present invention. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0030] Example 1

[0031] The embodiment of the present invention provides a method for generating a carbon factor library. By this method, more carbon factor data can be further expanded on the basis of the existing standard carbon factor library to form an extended carbon factor library, thereby improving the success rate of automatic carbon factor matching, reducing manual matching, and improving carbon factor matching efficiency. Specifically, Figure 1 This is a flow chart of the method for generating a carbon factor library provided in Example 1 of the present invention, as shown in FIG. Figure 1 As shown, the carbon factor library generation method provided in this embodiment includes the following steps S101 to S107.

[0032] Step S101: Acquire a plurality of matching data pieces of material data and carbon factor data that match each other.

[0033] Among them, the material data includes several material attribute fields, and the carbon factor data includes the carbon emission item name and carbon factor.

[0034] Specifically, when calculating the carbon emissions of building materials during building design or construction, it is necessary to match the carbon factor data corresponding to the material data from the carbon factor library of national standards and local standards.

[0035] Among them, the material data includes material attribute fields, which limit multiple attributes of the material, specifically material category fields, material name fields, specification fields or model fields, etc. The material data also includes project quantity. For example, in a certain type of concrete material data A1, the material category field is cement concrete, the material name field is pumped antifreeze concrete, the specification field is C30, and the project quantity is 80 cubic meters; for another type of concrete material data A2, the material category field is cement concrete, the material name field is pumped fine stone concrete, the specification field is C20, and the project quantity is 20 cubic meters.

[0036] In a standard carbon factor library for building carbon emissions calculation, carbon factor data includes the name of the carbon emission item and the corresponding carbon factor. For example, the carbon emission item name of a carbon factor data B is C30 concrete, and the carbon factor is 295kgCO2 / m 3 .

[0037] When the user side calculates the carbon emissions of concrete material data A1, it automatically searches for the carbon factor data B corresponding to the concrete material data A1 in the carbon factor library. According to the carbon factor of carbon factor data B, 295kgCO2 / m 3 The engineering usage of concrete material data A1 is 80 cubic meters, and the carbon emissions of concrete material data A1 are calculated as 295*80=23600kgCO2.

[0038] When the user side calculates the carbon emissions of concrete material data A2, the carbon factor data corresponding to the concrete material data A2 is not automatically found in the carbon factor library. The user can manually select carbon factor data B as the carbon factor data corresponding to the concrete material data A2, and calculate the carbon emissions of the concrete material data A2 based on the carbon factor data B.

[0039] Regardless of whether the user automatically or manually matches the material data with the carbon factor data, the data that matches the material data and the carbon factor data can be used as matching data. Optionally, the user writes a historical record of the mutual matching of the material data and the carbon factor data into a log. Multiple historical records are obtained by reading the log. These historical records can originate from different user terminals. Further, after obtaining the historical records from different user terminals, the number of historical records with the same material attribute field and the same matching carbon factor data is counted for the historical records of the user manually selecting the carbon factor data. Only when this number meets a preset requirement, for example, when this number is greater than a threshold, this portion of historical records is used to construct the following training sample set. That is, if different users manually match the same carbon factor to material data with the same attribute field, it indicates that the accuracy of the matching relationship is high, that is, the accuracy of the training sample set is high. If this number does not meet the preset requirement, one case is that the user manually matched different carbon factors to material data with the same attribute field. In this case, data filtering can be performed to avoid using incorrect historical records to construct the training sample set. Another case is that there are fewer material data samples of the calibration type of the attribute field. In this case, more material data of the attribute field can be further collected for statistics.

[0040] In addition, matching data between material data and carbon factor data can also be generated through manual labeling and other methods.

[0041] Step S102: construct a training sample set based on multiple matching data.

[0042] After obtaining multiple matching data, one matching data is used as a sample data to construct a training sample set.

[0043] Optionally, when constructing a training sample set, sample expansion is performed based on the obtained sample data to increase the sample size in the training sample set, thereby avoiding the problem of a small amount of matching data resulting in a small amount of data in the training sample set, which in turn affects the generalization ability and accuracy of the subsequent training model.

[0044] Step S103: Using the training sample set to train the first language processing model to obtain a mapping model.

[0045] Optionally, use PyTorch's loading function to load multiple matching data, then use the pre-trained tokenizer in the transformers library to tokenize and encode the matching data, convert the encoded sample data into a PyTorch dataset object to obtain an intermediate dataset; then use PyTorch to build and train the VAE model to obtain a training sample set.

[0046] In this step, the first language processing model uses an LLM model. The pre-trained LLM model is loaded using the transformers library. Training parameters are then defined using the DeepSpeed ​​configuration file and the TrainingArguments function in the transformers library. The Trainer class in the transformers library is used to create a Trainer object, combining the training sample set, the LLM model, and the training parameters. The Trainer object is then started to train the LLM model using the training sample set and training parameters to obtain a mapping model. After the mapping model is trained, it can output matching carbon factor data based on material properties.

[0047] Furthermore, optionally, data related to building carbon emission measurement can be automatically crawled from open Internet resources through crawlers, new matching data can be obtained through cleaning, and inserted into the training sample set. Then, the first language processing model can be trained regularly on the incremental data to produce richer carbon factor data in quasi-real time.

[0048] Step S104: Obtaining material property information for generating an extended carbon factor library.

[0049] In this step, the material attribute information used to generate the extended carbon factor library can be obtained by reading the material attribute information from the material list of the engineering file or obtaining the material attribute information from the material dictionary. The material attribute information includes information such as material category, material name, material specification, or material model.

[0050] Step S105: inputting the material property information into the mapping model to obtain target carbon factor data corresponding to the material property information.

[0051] Step S106: determining the carbon emission item name of the extended carbon factor data according to the material attribute information, and determining the carbon factor of the extended carbon factor data according to the carbon factor of the target carbon factor data.

[0052] In step S105, the mapping model is used to obtain target carbon factor data corresponding to the target carbon factor data corresponding to the material attribute information. In step S106, the material attribute information and the corresponding target carbon factor data are used to construct new carbon factor data, i.e., extended carbon factor data. During the construction, the material attribute information is used to determine the carbon emission item name of the extended carbon factor data, for example, using part or all of the data in the material attribute information as the carbon emission item name; and the carbon factor of the extended carbon factor data is used to determine the carbon factor of the extended carbon factor data, for example, using the carbon factor of the target carbon factor data directly as the carbon factor of the extended carbon factor data, or calculating the carbon factor of the extended carbon factor data based on the carbon factor of the target carbon factor data.

[0053] Step S107: Generate an extended carbon factor library using the extended carbon factor data.

[0054] After obtaining multiple pieces of extended carbon factor data, all extended carbon factor data can be integrated to generate an extended carbon factor library.

[0055] In the carbon factor library generation method provided in this embodiment, after obtaining multiple matching data that match material data and carbon factor data, a training sample set is constructed using the data to train a language processing model to obtain a mapping model. Then, material attribute information used to generate an extended carbon factor library is obtained and input into the mapping model to obtain corresponding target carbon factor data. Finally, new carbon factor data is constructed using the material attribute information and the corresponding target carbon factor data to generate an extended carbon factor library. Using the carbon factor library generation method provided in this embodiment, more carbon factor data is expanded on the basis of the standard carbon factor library using the matching data to generate an extended carbon factor library, thereby increasing the carbon factor data that can be matched with building material data, thereby improving the success rate of automatic carbon factor matching, reducing manual matching, and improving carbon factor matching efficiency.

[0056] Optionally, in one embodiment, the material data includes a material name field and a specification field, and the step of determining the carbon emission item name of the extended carbon factor data based on the material attribute information includes: obtaining the material name and specification in the material attribute information; and using the material name and specification as the carbon emission item name of the extended carbon factor data.

[0057] Specifically, with the development of materials, the specifications of the same type of materials are becoming more and more abundant. When obtaining matching data, in addition to obtaining the material name field of the material data, the specification field is also obtained. On the other hand, when determining the carbon emission item name of the extended carbon factor data based on the material attribute information, the material name and specification are obtained accordingly as the name of the carbon emission item.

[0058] By adopting the carbon factor library generation method provided in this embodiment, after generating an extended carbon factor library, the carbon emission item names of the carbon factor data all include the material name and specifications. Therefore, when using the extended carbon factor library for carbon factor matching, the matched carbon factors are more in line with the actual categories of building materials, making the calculation of carbon emissions more accurate.

[0059] Optionally, in one embodiment, the carbon factor data also includes emission units, and the step of determining the carbon factor of the extended carbon factor data based on the carbon factor of the target carbon factor data includes: when the engineering units of the material attribute information and the emission units of the target carbon factor data are different, calculating a conversion coefficient for converting the emission units into engineering units; converting the carbon factor of the target carbon factor data into an extended carbon factor measured in engineering units according to the conversion coefficient; and using the extended carbon factor as the carbon factor of the extended carbon factor data; the method also includes: using the engineering units as the emission units of the extended carbon factor data.

[0060] Specifically, when the engineering measurement unit in the material attribute information is different from the emission measurement unit in the target carbon factor data, for example, the engineering measurement unit of steel and metal materials is cubic meters, and the emission measurement unit in the carbon factor data of such materials is measured in tons, that is, how many kilograms of CO2 are included in each ton of material, when determining the carbon factor of the extended carbon factor data based on the carbon factor of the target carbon factor data, first calculate the conversion coefficient between the emission measurement unit and the engineering measurement unit. Assume that the material density is Xkg / m 3 , we can calculate that the conversion coefficient for converting tons into cubic meters is 1 / X, and then convert the carbon factor of the target carbon factor data into an extended carbon factor measured in engineering units. For example, the carbon factor of the target carbon factor data is Y, that is, each ton of material includes Ykg of CO2, which is converted to cubic meters, that is, each cubic meter of material includes X*Ykg of CO2. The expanded carbon factor after conversion is X*Y. In the final generated extended carbon factor data, the carbon factor is X*Y, and the unit of carbon emission item is measured in cubic meters, that is, kgCO2 per cubic meter.

[0061] By adopting the carbon factor library generation method provided in this embodiment, when generating the extended carbon factor, when the emission measurement unit is inconsistent with the engineering measurement unit, the carbon factor measured in the emission measurement unit is converted into a carbon factor measured in the engineering measurement unit, so that the units are consistent when the extended carbon factor library is used for carbon factor matching, which facilitates calculation.

[0062] Optionally, in one embodiment, the carbon factor data also includes emission units, the material data also includes an engineering unit field and a unit conversion coefficient, and determining the carbon factor of the extended carbon factor data based on the carbon factor of the target carbon factor data includes: using the carbon factor of the target carbon factor data as the carbon factor of the extended carbon factor data; the method also includes: using the emission unit of the target carbon factor data as the emission unit of the extended carbon factor data; the material attribute information also includes engineering units, and the material attribute information is input into the mapping model to obtain the target emission units and target conversion coefficients corresponding to the engineering unit information; the method also includes: generating a conversion coefficient library based on the engineering units, target emission units and target conversion coefficients.

[0063] Specifically, when constructing the extended carbon factor data, this embodiment uses the carbon factor of the target carbon factor data directly as the carbon factor of the extended carbon factor data. The difference from the above embodiment is that this embodiment does not perform unit conversion in the carbon factor library, and the construction method is simpler.

[0064] In this embodiment, when obtaining matching data, the engineering unit field and unit conversion coefficient of the material data are obtained. The unit conversion coefficient is a conversion coefficient for converting engineering units into emission units. Thus, a mapping model obtained by training a language processing model using matching data can realize the mapping from engineering units to emission units and between unit conversion coefficients. Furthermore, the material property information used to generate the extended carbon factor library also includes engineering units. By inputting such material property information into the mapping model, the target emission units and target conversion coefficients corresponding to the engineering unit information can be obtained, and a conversion coefficient library can be constructed using engineering units, target emission units and target conversion coefficients.

[0065] When calculating the carbon emissions of materials, if the emission units used in the matched carbon factor data are inconsistent with the engineering units commonly used in the material production and transportation process, it is still necessary to convert the engineering units into emission units before calculation. The carbon factor library generation method provided in this embodiment is used to construct a conversion coefficient library through a mapping model. Therefore, when calculating carbon emissions, the conversion coefficient library can be queried for calculation in the case of inconsistent units, which also achieves convenience in calculation.

[0066] Optionally, in one embodiment, the step of constructing a training sample set based on multiple matching data includes: segmenting the material data of the matching data to obtain a phrase including multiple words; determining the feature words and non-feature words of the phrase; generating a material phrase using all the feature words or all the feature words and at least one non-feature word; constructing initial sample data based on the material phrase and the carbon factor data of the matching data; generating an initial sample set using the initial sample data corresponding to the multiple matching data; and expanding the initial sample set to obtain a training sample set.

[0067] Specifically, material data contains complex attribute fields, such as "pumping impermeable and frost-resistant concrete, specification model C40." The purpose of word segmentation is to break these descriptions into meaningful words or phrases. For example, after word segmentation, the resulting phrase is ["pumping," "impermeability," "frost-resistant," "concrete," "specification model," "C40"]. Word segmentation tools such as BERT and GPT-3 can be used to improve word segmentation accuracy. After word segmentation, a phrase is obtained, from which feature words and non-feature words are filtered. Feature words are words that represent key information in the material data. This can be done manually or by pre-setting a feature library of feature words. The phrase is then filtered to identify words belonging to the library. Non-feature words are defined as words other than the feature words in the phrase. For example, the feature words filtered out are "pumping" and "concrete."

[0068] Then, material phrases are generated based on the characteristic words and non-characteristic words. Specifically, all characteristic words can be used as material phrases, or all characteristic words can be superimposed with at least one non-characteristic word to form a material phrase. More material phrases can be derived from the phrases after the material data segmentation. For example, the generated material phrases include ["pumping", "concrete"], ["pumping", "anti-permeability", "concrete"], or ["pumping", "concrete", "specification model", "C40"], etc.

[0069] Initial sample data is constructed based on the material phrases and the carbon factor data of the matching data. For example, an initial sample data includes two vectors ["pumping", "impermeability", "concrete"] and ["C50 concrete", "385", "kgCO2 / kg"]. Therefore, one material data can generate multiple initial sample data. An initial sample set is generated from multiple initial sample data, and then the initial sample set is expanded to obtain a training sample set, that is, to obtain more training samples.

[0070] Using the carbon factor library generation method provided in this embodiment, when constructing the initial sample data, the material data is first segmented, and after finding the characteristic words, more material phrases are derived, thereby enriching the samples so that the samples can cover more material description methods. At the same time, more samples are also generated to help improve the model training accuracy and generalization ability. On this basis, the initial sample data set is further expanded to construct a high-quality training sample set, thereby improving the prediction ability and generalization ability of the mapping model, and thus realizing accurate mapping of material data to carbon factor data.

[0071] Optionally, in one embodiment, the step of expanding the initial sample set to obtain a training sample set includes: training a second language processing model using material phrases in the initial sample set to obtain a data expansion model, wherein the second language processing model includes an encoder and a decoder, the encoder is used to encode input data to generate intermediate variables, and the decoder is used to decode the intermediate variables to obtain input data; generating a number of intermediate variables that conform to preset distribution rules; inputting the several intermediate variables into the data expansion model to obtain an expanded material phrase; training a first language processing model using the initial sample set to obtain an intermediate mapping model; inputting the expanded material phrase into the intermediate mapping model to obtain expanded carbon factor data corresponding to the expanded material phrase; constructing expanded sample data based on the expanded material phrase and the expanded carbon factor data; and adding the expanded sample data to the initial sample set to obtain a training sample set.

[0072] Specifically, a second language processing model is constructed, comprising an encoder and a decoder. The encoder encodes the input data and outputs intermediate variables, while the decoder decodes the intermediate variables and returns them to the input data space. The second language processing model can employ deep learning models such as variational autoencoders (VAEs) or generative adversarial networks (GANs). The second language processing model is trained using material phrases from the initial sample set to produce a data expansion model, which can then generate new, similar data.

[0073] After training the data expansion model, a number of intermediate variables are generated using preset distribution rules. These intermediate variables serve as input to the data expansion model to generate new material data. A variety of distribution rules, such as normal, Gaussian, or uniform distributions, can be used to generate diverse intermediate variables. The generated intermediate variables are then input into the data expansion model, and the decoder generates new material data, i.e., the expanded material phrase.

[0074] On the other hand, the initial sample set is used to train the first language processing model to obtain an intermediate mapping model, which is used to map the material phrase to the carbon factor data. Specifically, the first language processing model can be trained using advanced models such as bidirectional LSTM and Transformer to improve mapping accuracy. The mapping accuracy of the model is improved through cross-validation and parameter tuning. After the intermediate mapping model is trained, the generated expanded material phrase is input into the intermediate mapping model to obtain the corresponding carbon factor data, i.e., the expanded carbon factor data. The expanded material phrase and the corresponding expanded carbon factor data are combined to construct new expanded sample data, which is added to the initial sample set to form the final training sample set.

[0075] By adopting the carbon factor library generation method provided in this embodiment, the initial sample set is further expanded to generate a training sample set, enriching the diversity of the samples, so that the trained mapping model can better understand the relationship between material data and carbon factors, thereby improving the prediction accuracy, and enhancing the robustness of the mapping model to different input data, thereby improving the generalization ability of the model and helping to generate a more accurate expanded carbon factor library.

[0076] Optionally, in one embodiment, the step of obtaining material property information for generating an extended carbon factor library includes: constructing an enhanced knowledge base; obtaining material names and specifications; retrieving enhanced knowledge in the enhanced knowledge base using the material names and specifications; and generating material property information using the material names, specifications and enhanced knowledge.

[0077] Specifically, a knowledge base containing a large amount of building material information is constructed by crawling data from the web or e-books on density, specifications, sizes, production processes, and carbon factors of building-related materials. Material names and specifications are extracted from the dataset. For example, material names such as "steel" and "concrete" and specifications such as "Q235" and "C30" are extracted. Using these material names and specifications as query conditions, relevant knowledge fragments containing detailed material property information are retrieved from the enhanced knowledge base. The retrieved enhanced knowledge is combined with the material names and specifications to generate complete material property information.

[0078] By adopting the carbon factor library generation method provided in this embodiment, constructing an enhanced knowledge base and obtaining material property information therefrom, the diversity, completeness and accuracy of the data can be greatly improved, the performance and accuracy of the mapping model in carbon factor prediction can be improved, and ultimately more comprehensive and accurate carbon factor data can be generated.

[0079] Example 2

[0080] The embodiment of the present invention provides a carbon factor matching method, through which a standard carbon factor library and an extended carbon factor can be used for carbon factor matching, thereby improving the success rate of automatic carbon factor matching, reducing manual matching, and improving carbon factor matching efficiency. The extended carbon factor library is generated using the carbon factor library generation method provided in the first embodiment above. Specifically, Figure 2 This is a flow chart of the carbon factor matching method provided in Example 2 of the present invention, as shown in FIG. Figure 2 As shown, the carbon factor matching method provided in this embodiment includes the following steps S201 to S203.

[0081] Step S201: Acquire target material data.

[0082] Optionally, in one scenario, when calculating the carbon emissions of building materials in a construction project based on a cost document, the matching tool reads the cost document, parses the building material data of the construction project contained in the bill of materials and machinery, and obtains the target material data.

[0083] Optionally, in another scenario, the matching tool provides the user with a page for querying material data corresponding to carbon factor data, through which the user inputs the material data to be queried, namely, the target material data.

[0084] Optionally, in another scenario, the user generates a table list including several material data as needed, uploads the list to the matching tool, and the matching tool obtains the target material data.

[0085] The above-mentioned target material data may specifically include material name, specification, engineering measurement unit and engineering data, etc.

[0086] Step S202: searching for carbon factor data matching the target material data in the standard carbon factor library.

[0087] The standard carbon factors here refer to national or local standard libraries for carbon factors in the construction sector. Examples include the Building Carbon Emission Calculation Standard GBT51366 / 2019, the Jiangsu Province Civil Building Carbon Emission Calculation Guidelines (Draft for Comment), the Xiamen Building Carbon Emission Accounting Standard DBT3502 / Z 5053-2019, and the Prefabricated Building Carbon Emission Calculation Standard and Analysis Method T / CSTM00927-2023. First, search the probabilistic standard library for carbon factor data that matches the target material data.

[0088] Optionally, the target material data is segmented, and the segmentation results are used to match the carbon factor data in the standard carbon factor library. At the same time, matching rules are set. When the segmentation results match the carbon factor data in the standard carbon factor library, the carbon emissions of the target material data are calculated based on the matched carbon factor data.

[0089] Step S203: If no carbon factor data matching the target material data is found in the standard carbon factor library, the extended carbon factor library is searched for carbon factor data matching the target material data.

[0090] When no carbon factor data is matched in the standard carbon factor library, further matching is performed in the extended carbon factor library. Specifically, the same matching method as in step S202 can be adopted, which will not be described in detail here.

[0091] Optionally, no matter in which carbon factor library the matching is performed, the matching steps are divided into unit matching and name matching, that is, first use the material name and specification of the target material data to match the carbon emission item name in the carbon factor data. If the carbon emission item name is matched, then determine whether the engineering measurement unit of the target material data is consistent with the emission measurement unit in the carbon factor data. If not, look up the engineering measurement unit and emission measurement unit in the conversion coefficient library to find the conversion coefficient between the two; if the carbon emission item name is not matched in the standard carbon factor library, then match in the extended carbon factor library.

[0092] Optionally, separate Elasticsearch indexes are constructed for the standard carbon factor library and the extended carbon factor. The indexes primarily establish mappings between different material names and standard carbon factors, mapping material names to the codes of standard carbon factors. This allows for rapid retrieval of the correct carbon factor based on Elasticsearch when matching. Rapid retrieval through Elasticsearch and querying the conversion factor library allows for a quick and accurate association between the two. This allows for automatic lookup of the conversion factor when units are inconsistent, facilitating user operations.

[0093] Optionally, you can flexibly adjust the matching conditions to ensure matching accuracy by adjusting the matching parameters of Elasticsearch: scope and minimumShouldMatch.

[0094] In the carbon factor matching method provided in this embodiment, historical data can be automatically and efficiently utilized, and through deep learning, richer material data and carbon factor data matching relationships can be inferred. This solves the problem of carbon factor matching efficiency and accuracy in different energy consumption scenarios when there are relatively few national and provincial carbon factor library data for materials with widely varying names, specifications and units. It also reduces manual intervention, improves the accuracy and speed of entering material matching during the user's carbon calculation process, provides strong technical support for carbon emission management, and can provide real-time data for some green building construction and operation carbon calculation scenarios. Therefore, the carbon factor matching method provided by this embodiment has a high degree of automation, minimizes the workload of manual query and analysis of matching carbon factors, and improves the overall carbon emission calculation efficiency; elasticsearch is used to establish an index library, with a fast matching speed, and can complete carbon factor matching of a large amount of data in a short time; high accuracy, through intelligent analysis of the deep learning model, the carbon factor library is expanded, the matching accuracy is improved, and the dilemma of few carbon factors in existing national standards, provincial standards and group standards, and a large number of building materials, most of which require manual analysis and matching, is solved. Among them, it also effectively solves the inconsistency between material units and carbon factor units, and does not require manual calculation of conversion coefficients, which greatly improves efficiency.

[0095] Example 3

[0096] Corresponding to the above-mentioned embodiment 1, embodiment 3 of the present invention provides a carbon factor library generation device. The corresponding technical feature details and corresponding technical effects can be referred to the above-mentioned embodiment 1 and will not be repeated in this embodiment. Figure 3 This is a block diagram of a carbon factor library generation device provided in Example 3 of the present invention, as shown in FIG. Figure 3 As shown, the device includes: a first acquisition module 301 , a processing module 302 , a training module 303 , a second acquisition module 304 , an input module 305 , a determination module 306 and a generation module 307 .

[0097] The first acquisition module 301 is used to obtain multiple matching data that match the material data and the carbon factor data, wherein the material data includes several material attribute fields and the carbon factor data includes the carbon emission item name and the carbon factor; the processing module 302 is used to construct a training sample set based on the multiple matching data; the training module 303 is used to train the first language processing model using the training sample set to obtain a mapping model; the second acquisition module 304 is used to obtain material attribute information for generating an extended carbon factor library; the input module 305 is used to input the material attribute information into the mapping model to obtain target carbon factor data corresponding to the material attribute information; the determination module 306 is used to determine the carbon emission item name of the extended carbon factor data based on the material attribute information, and determine the carbon factor of the extended carbon factor data based on the carbon factor of the target carbon factor data; and the generation module 307 is used to generate an extended carbon factor library using the extended carbon factor data.

[0098] Optionally, in one embodiment, the material data includes a material name field and a specification field, and the determination module includes: a first acquisition unit, used to obtain the material name and specification in the material attribute information; and a first determination unit, used to use the material name and specification as the carbon emission item name of the extended carbon factor data.

[0099] Optionally, in one embodiment, the carbon factor data also includes emission units, and the determination module also includes: a first calculation unit, used to calculate the conversion coefficient of the emission units into engineering units when the engineering units of the material property information and the emission units of the target carbon factor data are different; a conversion unit, used to convert the carbon factor of the target carbon factor data into an extended carbon factor measured in engineering units according to the conversion coefficient; a second determination unit, used to use the extended carbon factor as the carbon factor of the extended carbon factor data; the determination module is also used to use the engineering units as the emission units of the extended carbon factor data.

[0100] Optionally, in one embodiment, the carbon factor data also includes emission units, the material data also includes an engineering unit field and a unit conversion coefficient, and the determination module is further used to use the carbon factor of the target carbon factor data as the carbon factor of the extended carbon factor data, and use the emission unit of the target carbon factor data as the emission unit of the extended carbon factor data; the material attribute information also includes engineering units, and the input module is further used to input the material attribute information into the mapping model to obtain the target emission units and target conversion coefficients corresponding to the engineering unit information, and the generation module is further used to generate a conversion coefficient library based on the engineering units, target emission units and target conversion coefficients.

[0101] Optionally, in one embodiment, the processing module includes: a word segmentation unit for segmenting the material data of the matching data to obtain a phrase including multiple words; a determination unit for determining the feature words and non-feature words of the phrase; a first generation unit for generating a material phrase using all the feature words or all the feature words and at least one non-feature word; a second generation unit for constructing initial sample data based on the material phrase and the carbon factor data of the matching data, and generating an initial sample set using the initial sample data corresponding to multiple matching data; and an expansion unit for expanding the initial sample set to obtain a training sample set.

[0102] Optionally, in one embodiment, when the expansion unit executes the step of expanding the initial sample set to obtain a training sample set, the specific steps executed include: training a second language processing model using the material phrases in the initial sample set to obtain a data expansion model, wherein the second language processing model includes an encoder and a decoder, the encoder is used to encode input data to generate intermediate variables, and the decoder is used to decode the intermediate variables to obtain input data; generating a number of intermediate variables that conform to preset distribution rules; inputting the several intermediate variables into the data expansion model to obtain an extended material phrase; training a first language processing model using the initial sample set to obtain an intermediate mapping model; inputting the extended material phrase into the intermediate mapping model to obtain extended carbon factor data corresponding to the extended material phrase; constructing extended sample data based on the extended material phrase and the extended carbon factor data; and adding the extended sample data to the initial sample set to obtain a training sample set.

[0103] Optionally, in one embodiment, the first acquisition module includes: a construction unit for constructing an enhanced knowledge base; a second acquisition unit for acquiring material names and specifications; a retrieval unit for retrieving enhanced knowledge in the enhanced knowledge base using the material names and specifications; and a second generation unit for generating material attribute information using the material names, specifications and enhanced knowledge.

[0104] Example 4

[0105] Corresponding to the above-mentioned embodiment 2, embodiment 4 of the present invention provides a carbon factor matching device. The corresponding technical feature details and corresponding technical effects can be referred to the above-mentioned embodiment 2 and will not be repeated in this embodiment. Figure 4 This is a block diagram of a carbon factor matching device provided in the fourth embodiment of the present invention, as shown in FIG. Figure 4 As shown, the device includes: an acquisition module 401, a first search module 402 and a second search module 403.

[0106] An acquisition module 401 is used to acquire target material data; a first search module 402 is used to search for carbon factor data that matches the target material data in a standard carbon factor library; a second search module 403 is used to search for carbon factor data that matches the target material data in an extended carbon factor library if the first search module cannot find carbon factor data that matches the target material data in the standard carbon factor library, wherein the extended carbon factor library is generated using any one of the carbon factor library generation methods provided by the present invention.

[0107] Example 5

[0108] This embodiment also provides a computer device, such as a smart phone, tablet computer, laptop computer, desktop computer, rack server, blade server, tower server or cabinet server (including an independent server or a server cluster composed of multiple servers) that can execute programs. Figure 5 As shown, the computer device 01 of this embodiment includes at least but not limited to: a memory 012 and a processor 011 which can be interconnected via a system bus. Figure 5 It should be pointed out that Figure 5 The computer device 01 is shown only with components memory 012 and processor 011 , but it should be understood that implementation of all of the components shown is not a requirement and more or fewer components may alternatively be implemented.

[0109] In this embodiment, the memory 012 (i.e., a readable storage medium) includes a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., an SD or DX memory), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the memory 012 can be an internal storage unit of the computer device 01, such as a hard disk or memory of the computer device 01. In other embodiments, the memory 012 can also be an external storage device of the computer device 01, such as a plug-in hard disk equipped on the computer device 01, a smart memory card (SMC), a secure digital (SD) card, a flash card, etc. Of course, the memory 012 can also include both the internal storage unit of the computer device 01 and its external storage device. In this embodiment, the memory 012 is generally used to store the operating system and various application software installed on the computer device 01, such as the program code of the carbon factor library generation device and the carbon factor matching device. In addition, the memory 012 can also be used to temporarily store various types of data that have been output or are to be output.

[0110] In some embodiments, the processor 011 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor 011 is generally used to control the overall operation of the computer device 01. In this embodiment, the processor 011 is used to run program codes stored in the memory 012 or process data, such as the carbon factor library generation method and the carbon factor matching method.

[0111] Example 6

[0112] This embodiment also provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic storage device, a disk, an optical disk, a server, an App store, etc., on which a computer program is stored, and when the program is executed by a processor, a corresponding function is implemented. The computer-readable storage medium of this embodiment is used to store a carbon factor library generation device and a carbon factor matching device, and when executed by a processor, a carbon factor library generation method and a carbon factor matching method are implemented.

[0113] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0114] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0115] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method.

[0116] The above are only preferred embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for generating a carbon factor library, characterized in that: include: Acquire multiple matching data of material data and carbon factor data, wherein the material data includes a plurality of material attribute fields, and the carbon factor data includes a carbon emission item name and a carbon factor; Constructing a training sample set according to the plurality of matching data; Using the training sample set to train a first language processing model to obtain a mapping model, wherein the mapping model is used to output matching carbon factor data according to material attribute information; Obtaining material property information for generating an extended carbon factor library; Inputting the material attribute information into the mapping model to obtain target carbon factor data corresponding to the material attribute information, obtaining data related to building carbon emission measurement from open Internet resources, obtaining new matching data through cleaning, inserting the new matching data into the training sample set, and regularly training the first language processing model on the incremental data; Determining the carbon emission item name of the extended carbon factor data according to the material attribute information, and determining the carbon factor of the extended carbon factor data according to the carbon factor of the target carbon factor data; and generating an extended carbon factor library using the extended carbon factor data, Among them, the step of constructing a training sample set based on the multiple matching data includes: segmenting the material data of the matching data to obtain a phrase including multiple words; determining the feature words and non-feature words of the phrase; generating a material phrase using all the feature words or all the feature words and at least one non-feature word; constructing initial sample data based on the material phrase and the carbon factor data of the matching data; generating an initial sample set using the initial sample data corresponding to the multiple matching data; and expanding the initial sample set to obtain the training sample set.

2. The method for generating a carbon factor library according to claim 1, wherein The material data includes a material name field and a specification field, and the step of determining the carbon emission item name of the extended carbon factor data according to the material attribute information includes: Obtaining the material name and specifications from the material attribute information; The material name and the specification are used as the carbon emission item name of the extended carbon factor data.

3. The carbon factor library generation method according to claim 2, wherein The carbon factor data also includes an emission measurement unit, and the step of determining the carbon factor of the extended carbon factor data according to the carbon factor of the target carbon factor data includes: When the engineering measurement unit in the material property information is different from the emission measurement unit of the target carbon factor data, calculating a conversion coefficient from the emission measurement unit to the engineering measurement unit; converting the carbon factor of the target carbon factor data into an expanded carbon factor measured in the engineering measurement unit according to the conversion coefficient; Using the extended carbon factor as the carbon factor of the extended carbon factor data; The method further includes: using the engineering measurement unit as the emission measurement unit of the extended carbon factor data.

4. The method for generating a carbon factor library according to claim 2, wherein The carbon factor data further includes an emission measurement unit, the material data further includes an engineering measurement unit field and a unit conversion coefficient, and determining the carbon factor of the extended carbon factor data based on the carbon factor of the target carbon factor data includes: using the carbon factor of the target carbon factor data as the carbon factor of the extended carbon factor data; The method further includes: using the emission measurement unit of the target carbon factor data as the emission measurement unit of the extended carbon factor data; The material property information also includes engineering measurement units. The material property information is input into the mapping model to obtain the target emission measurement units and target conversion coefficients corresponding to the engineering measurement unit information. The method also includes: generating a conversion coefficient library based on the engineering measurement units, the target emission measurement units and the target conversion coefficients.

5. The method for generating a carbon factor library according to claim 1, wherein The step of expanding the initial sample set to obtain the training sample set includes: Training a second language processing model using the material phrases in the initial sample set to obtain a data expansion model, wherein the second language processing model includes an encoder and a decoder, the encoder is used to encode input data to generate intermediate variables, and the decoder is used to decode the intermediate variables to obtain the input data; Generate several intermediate variables that conform to preset distribution rules; Inputting the plurality of intermediate variables into the data expansion model to obtain an expanded material phrase; Using the initial sample set to train the first language processing model to obtain an intermediate mapping model; Inputting the extended material phrase into the intermediate mapping model to obtain extended carbon factor data corresponding to the extended material phrase; constructing extended sample data based on the extended material phrase and the extended carbon factor data; and The extended sample data is added to the initial sample set to obtain the training sample set.

6. The method for generating a carbon factor library according to claim 1, wherein The steps for obtaining material property information used to generate the extended carbon factor library include: Build an enhanced knowledge base; Get material name and specifications; Retrieving enhanced knowledge in the enhanced knowledge base using the material name and the specification; The material property information is generated using the material name, the specification, and the enhanced knowledge.

7. A carbon factor matching method, characterized in that: include: Obtain target material data; Searching for carbon factor data matching the target material data in a standard carbon factor library; If no carbon factor data matching the target material data can be found in the standard carbon factor library, carbon factor data matching the target material data is searched in an extended carbon factor library, wherein the extended carbon factor library is generated using the carbon factor library generation method according to any one of claims 1 to 6.

8. A carbon factor library generation device, characterized in that, include: A first acquisition module is used to acquire a plurality of matching data of material data and carbon factor data, wherein the material data includes a plurality of material attribute fields, and the carbon factor data includes a carbon emission item name and a carbon factor; A processing module, configured to construct a training sample set based on the plurality of matching data; a training module configured to train a first language processing model using the training sample set to obtain a mapping model, wherein the mapping model is configured to output matching carbon factor data based on material property information, obtain data related to building carbon emission measurement from open internet resources, obtain new matching data through cleaning, insert the data into the training sample set, and periodically train the first language processing model on incremental data; A second acquisition module is used to obtain material property information for generating an extended carbon factor library; An input module, configured to input the material property information into the mapping model to obtain target carbon factor data corresponding to the material property information; a determination module, configured to determine a carbon emission item name of the extended carbon factor data according to the material attribute information, and determine a carbon factor of the extended carbon factor data according to the carbon factor of the target carbon factor data; and A generation module, for generating an extended carbon factor library using the extended carbon factor data, In which, the processing module includes: a word segmentation unit, used to segment the material data of the matching data to obtain a phrase including multiple words; a determination unit, used to determine the feature words and non-feature words of the phrase; a first generation unit, used to generate a material phrase using all the feature words or all the feature words and at least one non-feature word; a second generation unit, used to construct initial sample data based on the material phrase and the carbon factor data of the matching data, and generate an initial sample set using the initial sample data corresponding to the multiple matching data; and an expansion unit, used to expand the initial sample set to obtain a training sample set.

9. A carbon factor matching device, characterized in that: include: An acquisition module, used for acquiring target material data; A first search module is used to search for carbon factor data matching the target material data in a standard carbon factor library; A second search module is configured to search for carbon factor data matching the target material data in an extended carbon factor library if the first search module cannot find carbon factor data matching the target material data in the standard carbon factor library, wherein the extended carbon factor library is generated using the carbon factor library generation method according to any one of claims 1 to 6.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images