Method, device, apparatus, and storage medium for intelligently calculating carbon emissions
Patent Information
- Application Number
- TW114143868
- Authority / Receiving Office
- TW · TW
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2025-11-11
- Publication Date
- 2026-11-11
- Estimated Expiration
- 2045-11-10
Smart Images

Figure TWG2TB001920916_001 
Figure TWG2TB001920916_002 
Figure TWG2TB001920916_003
Abstract
Claims
1. A method for intelligently calculating carbon emissions, characterized in that the method comprises: Obtain raw material procurement information from the target supplier. This raw material procurement information list includes material classification information, material subclass tags, and procurement measurements. Based on the material classification information, query a predefined rule base to determine the carbon emission weight calculation type for each material. The predefined rule base stores the mapping relationship between the physical property categories, measurement dimensions, and basic calculation methods of materials. This mapping relationship includes calculation methods for weight units corresponding to metals, calculation methods for carbon content based on molecular formulas for chemicals, and calculation methods for wafer area units corresponding to semiconductor wafers. Determining the carbon emission weight calculation type for each material includes: obtaining the material category attributes of each material based on the material classification information and the material subclass tags; matching the corresponding basic calculation method from the predefined rule base according to the physical property categories of the material category attributes; extracting the measurement dimension tags corresponding to the material subclass tags; and generating the carbon emission weight calculation type based on the measurement dimension tags and the matched basic calculation method. Query the target supplier's historical carbon emission intensity records. If no record exists, obtain the benchmark value of companies of similar size in the same industry and generate a historical benchmark coefficient. The weight calculation type and historical benchmark coefficient are input into a preset machine learning model to obtain emission coefficients and confidence levels; and emission coefficients with confidence levels exceeding a preset threshold and the procurement measurement are calculated to obtain the carbon emissions of the target product.
2. The intelligent carbon emission calculation method as described in claim 1, characterized in that, before obtaining the raw material procurement information of the target supplier, it further includes: Obtain product information provided by the supplier and identify whether the product information includes the carbon emissions of the target product; If it is determined that the carbon emissions are not included, the target product's bill of materials is broken down step by step to obtain the corresponding basic material composition information. The bill of materials is obtained based on the product data, and the basic material composition information includes measurement dimensions. The association between the measurement dimension markers and the procurement measurement is established to construct a matching correspondence between the material category attributes and the weight calculation type.
3. The intelligent carbon emission calculation method as described in claim 2, characterized in that, obtaining benchmark values from enterprises of the same industry and scale, and generating historical benchmark coefficients, includes: Based on the product information, determine the target supplier's scale attribute marker and industry classification marker; Based on the industry classification marker, the carbon emission intensity benchmark value for the corresponding industry is extracted from the pre-constructed industry benchmark library; and based on the scale attribute marker, the corresponding correction coefficient is matched by calling the preset scale correction rule library; the carbon emission intensity benchmark value and the correction coefficient are calculated to generate the industry scale benchmark value; and the historical carbon emission intensity record library of the target supplier is searched. If the preset condition of no continuous data is met, the industry scale benchmark value is determined as the historical benchmark coefficient; if valid historical data exists, the valid historical data and the industry scale benchmark value are weighted and fused based on preset weights to generate the historical benchmark coefficient.
4. The intelligent carbon emission calculation method as described in claim 1, characterized in that, the weight calculation type and historical benchmark coefficient are input into a preset machine learning model, and the emission coefficient and confidence level are output, including: A feature vector is constructed, including material physical property category labels, supplier industry classification codes, enterprise size classification labels, and historical carbon emission intensity volatility. This feature vector is then concatenated with historical benchmark coefficients to form the model input dataset. A pre-trained gradient boosting decision tree is used to perform multi-level feature allocation and node splitting operations on the input dataset to generate predicted emission coefficients. A confidence score is output based on the decision path coverage density. When the confidence score is lower than a preset reliability threshold, the historical benchmark coefficient is used as an alternative emission coefficient and a low-confidence warning is issued. Furthermore, the deviation rate of the output emission coefficients from the industry benchmark is calculated. If the deviation rate exceeds a preset tolerance threshold, an audit flag is added, and the coefficient output is frozen.
5. The intelligent carbon emission calculation method as described in any one of claims 1 to 4, characterized in that, after obtaining the carbon emission of the target commodity, it further includes: The carbon emission intensity of the target supplier is compared with the industry benchmark value, and a carbon emission performance label is generated for the supplier according to the preset rating rules. When the performance marker reaches the preset deterioration threshold, low-carbon candidate suppliers of the same material category are matched from the pre-built alternative supplier library; the rating markers, carbon emission intensity deviation rate and historical supply stability indicators of candidate suppliers are dynamically aggregated to generate an alternative solution push data set; and a visual rating dashboard is embedded in the procurement decision interface to simultaneously output the alternative solution push data set and Sankey diagram carbon emission contribution path.
6. The intelligent carbon emission calculation method as described in claim 5, characterized in that the generation of the Sankey diagram carbon emission contribution path includes: Analyze the bill of materials hierarchy path of the target product and extract the carbon emission contribution of each material category at each level; A multi-level Sankey topology network is constructed using material category as the source node, supplier as the relay node, and carbon emission contribution as the flow vector. The allocation ratio of each relay node to the flow vector is calculated, and nodes exceeding the preset ratio threshold are marked with high carbon emission hotspots. The multi-level Sankey topology network and the high carbon emission hotspots are mapped to the Sankey graph rendering engine to output an interactive carbon emission contribution path view and associate it with procurement optimization suggestions.
7. A smart carbon emission calculation device, characterized in that the device comprises: A procurement information acquisition module is used to acquire raw material procurement information from target suppliers. The raw material procurement information list includes material classification information, material subclass tags, and procurement measurement. A rule matching engine is used to query a predefined rule library based on the material classification information to determine the carbon emission weight calculation type corresponding to each material. The predefined rule library stores the mapping relationship between the physical property category, measurement dimension, and basic calculation method of the material. The mapping relationship includes the calculation method for weight units corresponding to metals, the calculation method for carbon content calculation based on molecular formulas corresponding to chemicals, and the calculation method for wafer area units corresponding to semiconductor wafers. The rule matching engine obtains the material category attribute of each material based on the material classification information and the material subclass tags. According to the physical property category of the material category attribute, it matches the corresponding basic calculation method from the predefined rule library, extracts the measurement dimension tag corresponding to the material subclass tag, and generates the carbon emission weight calculation type based on the measurement dimension tag and the matched basic calculation method. A coefficient generation module is used to query the historical carbon emission intensity records of the target supplier. If the record does not exist, the benchmark value of the same industry and the same size enterprise is obtained to generate the historical benchmark coefficient. The weight calculation type and the historical benchmark coefficient are input into a preset machine learning model to obtain the emission coefficient and confidence level. A carbon emission calculation module is used to calculate the emission coefficient with the confidence level exceeding the preset threshold and the procurement measurement to obtain the carbon emission of the target product.
8. An electronic device, characterized in that the electronic device comprises: One or more processors; And a storage device for storing one or more programs that, when executed by the one or more processors, cause the electronic device to implement the intelligent carbon emission calculation method as described in any one of claims 1 to 6.
9. A computer-readable storage medium, characterized in that it stores a computer program thereon, which, when executed by a computer's processor, causes the computer to perform a carbon emission intelligent calculation method as described in any one of claims 1 to 6.
Citation Information
Patent Citations
Supply chain carbon data credible management method based on block chain
CN120069336A
Fund constraint type low-carbon supply chain operation analysis method
CN120822713A
