Enterprise carbon portrait generation method and system based on multi-source data fusion and medium

Abstract

The application discloses a kind of enterprise carbon portrait generation method, system and medium based on multi-source data fusion, mainly related to multi-source data fusion technical field, to solve the problem that existing scheme lacks standardized interface between internal and external data sources at data level, at technical level, traditional method is mostly dependent on statistical model, its linear hypothesis is difficult to capture the complex nonlinear relationship between carbon emission and multiple factors.The application comprises: preprocessing the uploaded data by the edge computing gateway, and obtaining structured carbon portrait feature data from the initial enterprise carbon data;Obtain the preset compatible access protocol and configure the preset standardized interface, obtain the carbon portrait feature data transmitted by the edge computing gateway;Then, based on the weight, update the specific value of each carbon portrait feature data in the feature vector;Real-time acquisition of carbon portrait feature data, and then using the trained carbon portrait generation model, obtain the current carbon emission data.

Description

Technical Field

[0001] This application relates to the field of multi-source data fusion technology, and in particular to a method, system and medium for generating enterprise carbon profiles based on multi-source data fusion. Background Technology

[0002] Currently, mainstream methods for corporate carbon emission assessments are largely based on modeling and analysis using a single data source. For example, real-time data from energy-consuming devices such as electricity and gas meters are collected to construct linear regression models to estimate carbon emissions. While this approach was initially favored due to readily available data and simple calculations, its limitations are becoming increasingly apparent in real-world scenarios. First, the reliance on a single data source makes the assessment results highly dependent on the completeness of energy bills or production reports, neglecting indirect carbon emissions from upstream and downstream of the supply chain (such as raw material transportation and waste disposal). Second, existing technologies often employ static models, failing to dynamically respond to changes in corporate operations. For instance, when a company introduces energy-saving equipment or adjusts its production plans, traditional models require manual parameter recalibration, making it impossible to update carbon emission predictions in real time, leading to decision-making delays. Furthermore, existing research generally lacks integration of external environmental data, such as regional carbon quota policies and the impact of extreme weather on energy consumption, making the assessment results difficult to adapt to policy changes or the impact of unforeseen events.

[0003] The aforementioned limitations stem from several factors. First, at the data level, the lack of standardized interfaces between internal enterprise systems (such as ERP and MES) and external data sources makes it difficult to efficiently integrate multi-source heterogeneous data. Second, at the technical level, traditional methods rely heavily on statistical models (such as multiple linear regression and time series analysis), whose linear assumptions make it difficult to capture the complex nonlinear relationships between carbon emissions and multiple factors. Summary of the Invention

[0004] This application provides a method, system, and medium for generating enterprise carbon profiles based on multi-source data fusion, in order to solve the problems of existing solutions lacking standardized interfaces between internal and external data sources at the data level, and traditional methods relying heavily on statistical models at the technical level, whose linear assumptions make it difficult to capture the complex nonlinear relationship between carbon emissions and multiple factors.

[0005] Firstly, this application provides a method for generating enterprise carbon profiles based on multi-source data fusion, the method comprising: Initial corporate carbon data is collected using a pre-set data collection platform. The types of initial corporate carbon data include: energy consumption data, production process data, supply chain data, unstructured data such as carbon policy texts, and energy market data. Based on the preset association between the preset data collection platform and the edge computing gateway, the collected enterprise carbon data is transmitted to the corresponding edge computing gateway. The data uploaded by the edge computing gateway is preprocessed, and structured carbon profile feature data is obtained from the initial enterprise carbon data. Obtain a preset compatible access protocol and configure a preset standardized interface, and then obtain carbon profile feature data transmitted by the edge computing gateway based on the preset compatible access protocol and through the preset standardized interface; Based on the type and collection time of carbon profile feature data, the carbon profile feature data are combined into feature vectors according to a preset type order; through a preset learnable weight matrix, the weights of each type of carbon profile feature data are obtained, and then the specific values ​​of each carbon profile feature data in the feature vector are updated based on the weights. The updated feature vector and carbon emission data collected at the same time are used as training data to input into the preset carbon profile generation model to obtain the trained carbon profile generation model; carbon profile feature data is acquired in real time, and then the trained carbon profile generation model is used to obtain the current carbon emission data. Based on carbon profile feature data and carbon emission data, a carbon profile with a preset display method is generated and displayed on the preset display interface.

[0006] In one implementation of this application, initial corporate carbon data is collected using a pre-set data acquisition platform, specifically including: By using pre-set energy consumption monitoring and metering equipment, the energy consumption of each production stage of the enterprise is collected in real time to obtain energy consumption data, which is presented in a time series numerical format; the pre-set energy consumption monitoring and metering equipment includes at least: smart meters, gas meters and fuel sensors. Extract production process data from the manufacturing execution system; the production process data includes at least: equipment operating parameters, raw material input, and waste output. Obtain supply chain data through a logistics management system; the supply chain data should include at least: transportation routes, vehicle types, and supplier carbon emission coefficients. By using a pre-defined platform for publicizing policies, unstructured data such as carbon policy texts related to corporate carbon profiling are extracted; Energy market data is collected through the electricity trading platform; this data includes at least real-time electricity prices and carbon intensity indices.

[0007] In one implementation of this application, before transmitting the collected enterprise carbon data to the corresponding edge computing gateway based on a preset association between the preset data acquisition platform and the edge computing gateway, the method further includes: The preset association between the preset data acquisition platform and the edge computing gateway can be obtained through the preset interface.

[0008] In one implementation of this application, the uploaded data is preprocessed through an edge computing gateway, and structured carbon profile feature data is obtained from the initial enterprise carbon data, specifically including: Remove abnormal initial enterprise carbon data that is outside the preset reasonable range; The sliding window mean filling method is used to fill the deleted initial enterprise carbon data; When the initial carbon data of an abnormal enterprise is at the beginning of the data sequence and the window forward is less than the preset window value, the mean of the beginning part is used for filling; when it is at the end and the window backward is less than the preset window value, the mean of the end part is used for filling. By using keyword extraction technology, structured feature data is extracted from the processed data; The feature data is normalized to obtain carbon profiling feature data.

[0009] In one implementation of this application, the weights of various types of carbon profile feature data are obtained by pre-setting a learnable weight matrix, and then the specific values ​​of each carbon profile feature data in the feature vector are updated based on the weights, specifically including: Learnable weight matrix can be preset: Calculate the weights of each type of carbon profile feature data; Where i represents the i-th type, b is a preset bias term, W represents the first preset weight matrix, and h represents the second preset weight matrix. This represents the carbon profile feature data of the i-th type; and =1, where n represents the total number of types; Each carbon profile feature data point is added to its corresponding weight to obtain the specific value of the updated carbon profile feature data.

[0010] In one implementation of this application, the updated feature vector and carbon emission data collected at the same time are used as training data and input into a preset carbon profile generation model to obtain a trained carbon profile generation model, specifically including: During model training, adversarial generated samples are generated using adversarial network technology, and then these adversarial generated samples are input into the carbon profiling generation model. In terms of loss function construction, the weight values ​​of mean squared error and cross entropy are obtained, and the loss function is calculated by weighted combination of mean squared error and cross entropy. The parameters of the carbon profiling generation model are updated at preset time intervals using online learning technology until the parameters remain unchanged or the preset time length is reached.

[0011] In one implementation of this application, a carbon profile with a preset display method is generated based on carbon profile feature data and carbon emission data, and then displayed on a preset display interface, specifically including: Get carbon profile templates; Based on a preset update interval, carbon profile feature data and carbon emission data are input into the carbon profile template to update the carbon profile.

[0012] Secondly, this application provides an enterprise carbon profile generation system based on multi-source data fusion, the system comprising: The data acquisition module is used to collect initial corporate carbon data using a pre-set data acquisition platform. The types of initial corporate carbon data include: energy consumption data, production process data, supply chain data, unstructured data such as carbon policy texts, and energy market data. The acquisition module is used to transmit the collected enterprise carbon data to the corresponding edge computing gateway based on the preset association between the preset acquisition platform and the edge computing gateway, perform preprocessing on the data uploaded by the edge computing gateway, and obtain structured carbon profile feature data from the initial enterprise carbon data. The transmission module is used to acquire a preset compatible access protocol and configure a preset standardized interface, and then obtain carbon profile feature data transmitted by the edge computing gateway based on the preset compatible access protocol and through the preset standardized interface. The update module is used to combine carbon profile feature data into feature vectors according to a preset type order based on the type and collection time of the carbon profile feature data; obtain the weights of each type of carbon profile feature data through a preset learnable weight matrix, and then update the specific values ​​of each carbon profile feature data in the feature vector based on the weights. The display module is used to input the updated feature vector and carbon emission data collected at the same time as training data into the preset carbon profile generation model to obtain the trained carbon profile generation model; to acquire carbon profile feature data in real time, and then use the trained carbon profile generation model to obtain the current carbon emission data; and to generate a carbon profile in a preset display mode based on the carbon profile feature data and carbon emission data, and display it on the preset display interface.

[0013] In one implementation of this application, the acquisition module includes an acquisition unit. This device is used to collect energy consumption data in real time from each production stage of an enterprise using pre-set energy consumption monitoring and metering equipment, and the energy consumption data is presented in a time series numerical format; the pre-set energy consumption monitoring and metering equipment includes at least: smart meters, gas meters and fuel sensors. Extract production process data from the manufacturing execution system; the production process data includes at least: equipment operating parameters, raw material input, and waste output. Obtain supply chain data through a logistics management system; the supply chain data should include at least: transportation routes, vehicle types, and supplier carbon emission coefficients. By using a pre-defined platform for publicizing policies, unstructured data such as carbon policy texts related to corporate carbon profiling are extracted; Energy market data is collected through the electricity trading platform; this data includes at least real-time electricity prices and carbon intensity indices.

[0014] Thirdly, this application provides a non-volatile computer storage medium storing computer instructions thereon, which, when executed, implement a method for generating enterprise carbon profiles based on multi-source data fusion as described above.

[0015] As can be seen from the above technical solutions, this application has the following advantages: Solved the data layer interface standardization issue: By acquiring a pre-defined compatible access protocol and configuring a pre-defined standardized interface, this application solves the problem of the lack of standardized interfaces between internal and external data sources at the data layer in existing solutions. This ensures that data can be transmitted and interacted smoothly and accurately between different data sources, improving the efficiency and accuracy of data processing.

[0016] Capable of capturing complex nonlinear relationships: The method employed in this application does not rely on traditional statistical models. Instead, it captures the complex nonlinear relationship between carbon emissions and multiple factors by pre-setting a learnable weight matrix and a pre-set carbon profile generation model. This overcomes the limitations of the linear assumptions of traditional methods, making the generated carbon profile more accurate and comprehensive in reflecting the carbon emissions of enterprises.

[0017] Achieved multi-source data fusion and efficient processing: This application is capable of collecting various types of initial enterprise carbon data, including energy consumption data, production process data, supply chain data, unstructured data such as carbon policy texts, and energy market data. Through preprocessing and structured extraction via an edge computing gateway, as well as subsequent feature vector combination and weight updates, efficient fusion and processing of multi-source data are achieved, providing a rich and accurate data foundation for the generation of carbon profiles.

[0018] Furthermore, this application can acquire carbon profile feature data in real time and use a trained carbon profile generation model to obtain current carbon emission data. This makes the generated carbon profile real-time and dynamic, reflecting the latest situation of corporate carbon emissions and providing timely and effective support for corporate carbon emission management and decision-making.

[0019] Furthermore, this application can generate carbon profiles in a preset display format based on carbon profile feature data and carbon emission data, and display them on a preset display interface. This intuitive display method allows users to clearly understand a company's carbon emission situation, facilitating carbon emission analysis and decision-making. At the same time, the preset display interface also provides convenient interactive functions for easy user operation and querying. Attached Figure Description

[0020] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of a method for generating enterprise carbon profiles based on multi-source data fusion, provided in an embodiment of this application.

[0022] Figure 2 This is a schematic diagram of the internal structure of an enterprise carbon profile generation system based on multi-source data fusion provided in an embodiment of this application. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Those skilled in the art should understand that the embodiments described below are merely preferred embodiments of this disclosure and do not imply that this disclosure can only be implemented through these preferred embodiments. These preferred embodiments are merely used to explain the technical principles of this disclosure and are not intended to limit the scope of protection of this disclosure. Based on the preferred embodiments provided by this disclosure, all other embodiments obtained by those skilled in the art without creative effort should still fall within the scope of protection of this disclosure.

[0025] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0026] The technical solutions proposed in the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0027] The embodiment provides a method for generating enterprise carbon profiles based on multi-source data fusion, such as Figure 1 As shown in the embodiments of this application, the method mainly includes the following steps: Step 110: Collect initial corporate carbon data using a pre-set data collection platform.

[0028] The initial enterprise carbon data includes the following types: energy consumption data, production process data, supply chain data, unstructured data such as carbon policy texts, and energy market data. In some embodiments, energy consumption data collection involves using energy monitoring and metering devices such as smart meters, gas meters, and fuel sensors to collect real-time data on energy usage at each stage of the enterprise's production process. The collected data is presented in a time-series numerical format, detailing the specific energy consumption at each point in time, providing fundamental and crucial data support for subsequent analysis. The data covers various energy sources used in the enterprise's daily production and operations, including electricity, water, gas, and coal.

[0029] Production process data acquisition: Structured data such as equipment operating parameters, raw material input, and waste output are accurately extracted from the Manufacturing Execution System (MES). The MES system comprehensively records various information throughout the enterprise's production process. Equipment operating parameters, including runtime, power, and speed, reflect the equipment's working status; raw material input specifies the exact quantity and type of raw materials used in each batch of production; and waste output visually presents the waste generated during the production process. This data is crucial for analyzing carbon emission sources in the enterprise's production process and the correlation between production efficiency and carbon emissions.

[0030] Supply Chain Data Acquisition: Information such as transportation routes, vehicle types, and supplier carbon emission coefficients are obtained through a logistics management system (TMS), with data transmission via API interface for fast and stable acquisition. Transportation routes meticulously record the entire logistics trajectory of products from suppliers to the enterprise and from the enterprise's products to customers; vehicle types clearly identify the means of transportation used during the process, with significant differences in carbon emission characteristics among different vehicles; supplier carbon emission coefficients reflect the supplier's carbon emission levels in raw material production and other stages. Through API integration between the TMS system and the enterprise's carbon profiling system, this critical supply chain data can be obtained in real time and accurately.

[0031] Unstructured data collection of policy texts: This involves extensively gathering unstructured policy texts such as regional carbon quotas and carbon tax rates from platforms where relevant policies are publicly available. Government departments and related agencies publish various carbon emission-related policy information through official websites and specific policy release platforms. Using web crawling technology or specialized data collection software, policy texts from these platforms are collected. These unstructured texts will then be processed and analyzed in depth to extract key information valuable for building corporate carbon profiles.

[0032] Energy Market Data Acquisition: Real-time electricity prices and carbon intensity indices are collected from the electricity trading platform. The electricity trading platform updates electricity market transaction price information in real time, which is closely related to enterprises' electricity procurement costs and energy use decisions; the carbon intensity index reflects the overall carbon emission intensity trend of the energy market. By connecting with the data interface of the electricity trading platform, this data can be acquired periodically or in real-time, providing enterprises with dynamic market information for energy selection and carbon emission assessment.

[0033] Step 120: Based on the preset association between the preset data acquisition platform and the edge computing gateway, transmit the collected enterprise carbon data to the corresponding edge computing gateway, preprocess the data uploaded by the edge computing gateway, and obtain structured carbon profile feature data from the initial enterprise carbon data.

[0034] It's important to note that an edge computing gateway is deployed for initial data aggregation and encrypted transmission locally. Located close to the data source, the edge computing gateway can quickly collect data from surrounding devices and perform preliminary processing and aggregation, reducing data transmission volume. Simultaneously, encryption technology is used to encrypt the data, ensuring its security during transmission. For data storage, a distributed storage cluster is employed to achieve efficient storage of massive amounts of data. This distributed storage cluster consists of multiple storage nodes, and distributed storage algorithms distribute data across these nodes, increasing storage capacity, improving read / write speeds and reliability, and meeting the enterprise's need for storing large amounts of historical and real-time data during carbon profiling.

[0035] Before transmitting the collected enterprise carbon data to the corresponding edge computing gateway based on the preset association between the preset data acquisition platform and the edge computing gateway, the method also includes: The preset association between the preset data acquisition platform and the edge computing gateway can be obtained through the preset interface.

[0036] Specifically, the data uploaded through the edge computing gateway is preprocessed, and structured carbon profile feature data is obtained from the initial enterprise carbon data. This can be achieved through: Remove abnormal initial enterprise carbon data that is outside the preset reasonable range; The sliding window mean filling method is used to fill the deleted initial enterprise carbon data; When the initial carbon data of an abnormal enterprise is at the beginning of the data sequence and the window forward is less than the preset window value, the mean of the beginning part is used for filling; when it is at the end and the window backward is less than the preset window value, the mean of the end part is used for filling. By using keyword extraction technology, structured feature data is extracted from the processed data; The feature data is normalized to obtain carbon profiling feature data.

[0037] For example, for a device that collects data every 10 minutes, if the target point is the data point at the current 50th minute, the window can be set to 5 data points before and after (i.e., data from 40 to 60 minutes).

[0038] More specifically, remove abnormal initial enterprise carbon data that is outside the preset reasonable range: Suppose that in our collected energy consumption data, the electricity consumption on a certain day suddenly surges to several times the normal value; this is clearly an outlier. Based on a preset reasonable range (e.g., upper and lower limits of electricity consumption set based on historical data or industry standards), we can identify and remove this outlier.

[0039] The sliding window mean filling method was used to fill in the deleted initial enterprise carbon data: Suppose that data for a few days is missing from energy consumption data due to equipment failure. We can use a sliding window mean to fill in the missing data.

[0040] Set a preset window value, such as 5 days. For missing data points, we calculate the average power consumption of the 5 days before and after the missing data point (if the data exists), and use this average value to fill in the missing data points.

[0041] Mean filling in special cases: When the initial carbon data of an abnormal enterprise is at the beginning of the data sequence, and the window forward is less than the preset window value (for example, the data for the first 3 days is missing, but the preset window value is 5 days), we use the mean of the initial part (i.e. the existing data points) to fill it.

[0042] Similarly, when outlier data is at the end and the window is smaller than the preset window value, we fill it with the mean of the last part.

[0043] By using keyword extraction technology, structured feature data is extracted from the processed data: For unstructured data such as carbon policy texts, we can use keyword extraction techniques to extract key information related to carbon emissions.

[0044] For example, through natural language processing technology, we can identify keywords such as "emission reduction target", "carbon tax" and "green energy" in the text, and extract these keywords and their related contextual information as structured feature data.

[0045] Normalize the feature data to obtain carbon profiling feature data: Suppose that the feature data we extract includes power consumption in different time periods, carbon emission intensity in the production process, etc., the dimensions and range of these data may be different.

[0046] To facilitate subsequent analysis and modeling, we need to normalize these feature data. For example, we can use the min-max normalization method to scale the data to the range of [0, 1], thereby obtaining standardized carbon profiling feature data.

[0047] Step 130: Obtain the preset compatible access protocol and configure the preset standardized interface, and then obtain the carbon profile feature data transmitted by the edge computing gateway based on the preset compatible access protocol and the preset standardized interface.

[0048] It should be noted that, based on the various types of data collected, standardized interfaces are built, and a unified data access protocol (such as RESTful API) is designed to support HTTP / HTTPS communication and be compatible with multiple formats such as JSON and CSV, ensuring the compatibility of multi-source heterogeneous data.

[0049] Step 140: Based on the type and collection time of the carbon portrait feature data, combine the carbon portrait feature data into a feature vector according to the preset type order; obtain the weight of each type of carbon portrait feature data through the preset learnable weight matrix, and then update the specific value of each carbon portrait feature data in the feature vector based on the weight.

[0050] In some embodiments, the weights of various types of carbon profiling feature data are obtained by pre-setting a learnable weight matrix, and then the specific values ​​of each carbon profiling feature data in the feature vector are updated based on the weights, specifically including: Learnable weight matrix can be preset: Calculate the weights of each type of carbon profile feature data; Where i represents the i-th type, b is a preset bias term, W represents the first preset weight matrix, and h represents the second preset weight matrix. This represents the carbon profile feature data of the i-th type; and =1, where n represents the total number of types; Each carbon profile feature data point is added to its corresponding weight to obtain the specific value of the updated carbon profile feature data.

[0051] It should be noted that feature vector combination involves combining the carbon profiling feature data into feature vectors according to a preset type order, based on the data type and acquisition time. This step ensures the orderliness and structure of the feature data, facilitating subsequent processing and analysis.

[0052] Application of Preset Learnable Weight Matrix: A preset learnable weight matrix is ​​used to calculate the weights of various types of carbon profile feature data. This matrix design allows the model to automatically adjust the weights during the learning process to better capture the relationship between feature data and carbon emissions.

[0053] Those skilled in the art will understand that the weights calculated using the above formula reflect the relative importance of each feature data point in carbon emission prediction or analysis. These weights are dynamically learned, meaning they are adjusted based on changes in the training data to optimize the model's predictive performance.

[0054] Multiplying each carbon biometric feature data point by its corresponding weight results in a comprehensive sum considering all feature data points and their weights, reflecting the impact of each feature on the final output. Simplifying this from a direct operational perspective, it can be described as "adjusting" each feature data point according to its weight (reflecting its contribution to the final sum in the overall calculation). The final weighted sum of all feature data points constitutes the updated feature vector. This "addition" actually occurs during the application of weights after softmax normalization; each feature data point multiplied by its weight becomes part of the overall sum, collectively forming the new feature representation.

[0055] Step 150: Input the updated feature vector and carbon emission data collected at the same time as training data into the preset carbon profile generation model to obtain the trained carbon profile generation model; acquire carbon profile feature data in real time, and then use the trained carbon profile generation model to obtain the current carbon emission data; generate a carbon profile in the preset display mode based on the carbon profile feature data and carbon emission data, and display it on the preset display interface.

[0056] It should be noted that the process of acquiring carbon profile feature data in real time and then using a trained carbon profile generation model to obtain current carbon emission data requires the following steps: "Based on the type and acquisition time of the carbon profile feature data, the carbon profile feature data is combined into a feature vector according to a preset type order; the weights of each type of carbon profile feature data are obtained through a preset learnable weight matrix; and then the specific values ​​of each carbon profile feature data in the feature vector are updated based on the weights." This process is used to obtain the specific values ​​of the real-time data, and then the model is used to obtain the current carbon emission data.

[0057] Specifically, the updated feature vector and carbon emission data collected at the same time are used as training data and input into a preset carbon profile generation model to obtain a trained carbon profile generation model. This can be achieved as follows: During model training, adversarial generated samples are generated using adversarial network technology, and then these adversarial generated samples are input into the carbon profiling generation model. In terms of loss function construction, the weight values ​​of mean squared error and cross entropy are obtained, and the loss function is calculated by weighted combination of mean squared error and cross entropy. The parameters of the carbon profiling generation model are updated at preset time intervals using online learning technology until the parameters remain unchanged or the preset time length is reached.

[0058] Specifically, based on carbon profile feature data and carbon emission data, a carbon profile with a preset display format is generated and displayed on a preset display interface. This can be achieved through: Get carbon profile templates; Based on a preset update interval, carbon profile feature data and carbon emission data are input into the carbon profile template to update the carbon profile.

[0059] In addition, this application can carry out dynamic updates and visualization of carbon profiles, mainly including the following: Real-time data sources can be monitored through message queues (such as RabbitMQ) to trigger model recalculation and achieve data stream monitoring. It adopts a Git-like mechanism to manage model versions and supports historical state backtracking and comparative analysis. The design includes a display interface that incorporates heat maps, carbon emission trend forecasts, and emission reduction recommendations, specifically including: A heat map showing the carbon emission intensity distribution of each factory area of ​​the enterprise based on a geographic information system; The carbon emission trend prediction curve for the next 7 days is predicted by combining the ARIMA model with a confidence interval of 95%. Based on the model output, priority suggestions (such as "replace high-energy-consuming equipment" and "optimize logistics routes") are generated to provide carbon emission reduction recommendations for enterprises. It supports providing a RESTful interface for enterprise ERP systems to call, enabling collaborative linkage between carbon data and production plans.

[0060] In addition, this application Figure 2 This application provides an embodiment of an enterprise carbon profile generation system based on multi-source data fusion. For example... Figure 2 As shown in the embodiments of this application, the system mainly includes: The data acquisition module 210 is used to collect initial corporate carbon data using a preset data acquisition platform. The types of initial corporate carbon data include: energy consumption data, production process data, supply chain data, unstructured data such as carbon policy texts, and energy market data.

[0061] The acquisition module 210 includes an acquisition unit. This device is used to collect energy consumption data in real time from each production stage of an enterprise using pre-set energy consumption monitoring and metering equipment, and the energy consumption data is presented in a time series numerical format; the pre-set energy consumption monitoring and metering equipment includes at least: smart meters, gas meters and fuel sensors. Extract production process data from the manufacturing execution system; the production process data includes at least: equipment operating parameters, raw material input, and waste output. Obtain supply chain data through a logistics management system; the supply chain data should include at least: transportation routes, vehicle types, and supplier carbon emission coefficients. By using a pre-defined platform for publicizing policies, unstructured data such as carbon policy texts related to corporate carbon profiling are extracted; Energy market data is collected through the electricity trading platform; this data includes at least real-time electricity prices and carbon intensity indices.

[0062] The acquisition module 220 is used to transmit the collected enterprise carbon data to the corresponding edge computing gateway based on the preset association relationship between the preset acquisition platform and the edge computing gateway, perform preprocessing on the data uploaded by the edge computing gateway, and obtain structured carbon profile feature data from the initial enterprise carbon data.

[0063] The transmission module 230 is used to acquire a preset compatible access protocol and configure a preset standardized interface, and then obtain carbon profile feature data transmitted by the edge computing gateway based on the preset compatible access protocol and through the preset standardized interface.

[0064] The update module 240 is used to combine carbon portrait feature data into feature vectors according to a preset type order based on the type and collection time of the carbon portrait feature data; obtain the weights of each type of carbon portrait feature data through a preset learnable weight matrix, and then update the specific values ​​of each carbon portrait feature data in the feature vector based on the weights.

[0065] The display module 250 is used to input the updated feature vector and carbon emission data collected at the same time as training data into the preset carbon profile generation model to obtain the trained carbon profile generation model; to acquire carbon profile feature data in real time, and then use the trained carbon profile generation model to obtain the current carbon emission data; and to generate a carbon profile in a preset display mode based on the carbon profile feature data and carbon emission data, and display it on the preset display interface.

[0066] In addition, this application embodiment also provides a non-volatile computer storage medium storing executable instructions, which, when executed, implement the enterprise carbon profile generation method based on multi-source data fusion as described above.

[0067] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for generating enterprise carbon profiles based on multi-source data fusion, characterized in that, The method includes: Initial corporate carbon data is collected using a pre-set data collection platform. The types of initial corporate carbon data include: energy consumption data, production process data, supply chain data, unstructured data such as carbon policy texts, and energy market data. Based on the preset association between the preset data acquisition platform and the edge computing gateway, the collected enterprise carbon data is transmitted to the corresponding edge computing gateway. The data uploaded by the edge computing gateway is preprocessed, and structured carbon profile feature data is obtained from the initial enterprise carbon data. Obtain a preset compatible access protocol and configure a preset standardized interface, and then obtain carbon profile feature data transmitted by the edge computing gateway based on the preset compatible access protocol and through the preset standardized interface; Based on the type and collection time of carbon profile feature data, the carbon profile feature data are combined into feature vectors according to a preset type order; through a preset learnable weight matrix, the weights of each type of carbon profile feature data are obtained, and then the specific values ​​of each carbon profile feature data in the feature vector are updated based on the weights. The updated feature vector and carbon emission data collected at the same time are used as training data to input into the preset carbon profile generation model to obtain the trained carbon profile generation model; carbon profile feature data is acquired in real time, and then the trained carbon profile generation model is used to obtain the current carbon emission data. Based on carbon profile feature data and carbon emission data, a carbon profile with a preset display method is generated and displayed on the preset display interface; By pre-setting a learnable weight matrix, the weights of various types of carbon profiling feature data are obtained. Then, based on these weights, the specific values ​​of each carbon profiling feature data in the feature vector are updated, including: Learnable weight matrix can be preset: Calculate the weights of each type of carbon profile feature data; Where i represents the i-th type, b is a preset bias term, W represents the first preset weight matrix, and h represents the second preset weight matrix. This represents the carbon profile feature data of the i-th type; and =1, where n represents the total number of types; Each carbon profile feature data is added to its corresponding weight to obtain the specific value of the updated carbon profile feature data; The updated feature vector and carbon emission data collected at the same time are used as training data and input into the preset carbon profile generation model to obtain the trained carbon profile generation model, which specifically includes: During model training, adversarial generated samples are generated using adversarial network technology, and then these adversarial generated samples are input into the carbon profiling generation model. In terms of loss function construction, the weight values ​​of mean squared error and cross entropy are obtained, and the loss function is calculated by weighted combination of mean squared error and cross entropy. The parameters of the carbon profiling generation model are updated at preset time intervals using online learning technology until the parameters remain unchanged or the preset time length is reached.

2. The enterprise carbon profile generation method based on multi-source data fusion according to claim 1, characterized in that, Initial corporate carbon data is collected using a pre-set data acquisition platform, specifically including: By using pre-set energy consumption monitoring and metering equipment, the energy consumption of each production link of the enterprise is collected in real time to obtain energy consumption data, and the energy consumption data is presented in a time series numerical format; the pre-set energy consumption monitoring and metering equipment includes at least: smart meters, gas meters and fuel sensors. Extract production process data from the manufacturing execution system; the production process data includes at least: equipment operating parameters, raw material input, and waste output. Obtain supply chain data through a logistics management system; the supply chain data should include at least: transportation routes, vehicle types, and supplier carbon emission coefficients. By using a pre-defined platform for publicizing policies, unstructured data such as carbon policy texts related to corporate carbon profiling are extracted; Energy market data is collected through the electricity trading platform; this data includes at least real-time electricity prices and carbon intensity indices.

3. The enterprise carbon profile generation method based on multi-source data fusion according to claim 1, characterized in that, Before transmitting the collected enterprise carbon data to the corresponding edge computing gateway based on the preset association between the preset data acquisition platform and the edge computing gateway, the method further includes: The preset association between the preset data acquisition platform and the edge computing gateway can be obtained through the preset interface.

4. The enterprise carbon profile generation method based on multi-source data fusion according to claim 1, characterized in that, The data uploaded through the edge computing gateway is preprocessed, and structured carbon profile feature data is obtained from the initial enterprise carbon data, specifically including: Remove abnormal initial enterprise carbon data that is outside the preset reasonable range; The sliding window mean filling method is used to fill the deleted initial enterprise carbon data; When the initial carbon data of an abnormal enterprise is at the beginning of the data sequence and the window forward is less than the preset window value, the average value of the beginning part is used to fill the gap; when it is at the end and the window backward is less than the preset window value, the average value of the end part is used to fill the gap. By using keyword extraction technology, structured feature data is extracted from the processed data; The feature data is normalized to obtain carbon profiling feature data.

5. The enterprise carbon profile generation method based on multi-source data fusion according to claim 1, characterized in that, Based on carbon profile feature data and carbon emission data, a carbon profile with a preset display format is generated and displayed on a preset display interface, specifically including: Get carbon profile templates; Based on a preset update interval, carbon profile feature data and carbon emission data are input into the carbon profile template to update the carbon profile.

6. A corporate carbon profile generation system based on multi-source data fusion, applicable to the corporate carbon profile generation method based on multi-source data fusion as described in any one of claims 1-5, characterized in that, The system includes: The data acquisition module is used to collect initial corporate carbon data using a pre-set data acquisition platform. The types of initial corporate carbon data include: energy consumption data, production process data, supply chain data, unstructured data such as carbon policy texts, and energy market data. The acquisition module is used to transmit the collected enterprise carbon data to the corresponding edge computing gateway based on the preset association between the preset acquisition platform and the edge computing gateway, perform preprocessing on the data uploaded by the edge computing gateway, and obtain structured carbon profile feature data from the initial enterprise carbon data. The transmission module is used to acquire a preset compatible access protocol and configure a preset standardized interface, and then obtain carbon profile feature data transmitted by the edge computing gateway based on the preset compatible access protocol and through the preset standardized interface. The update module is used to combine carbon profile feature data into feature vectors according to a preset type order based on the type and collection time of the carbon profile feature data; obtain the weights of each type of carbon profile feature data through a preset learnable weight matrix, and then update the specific values ​​of each carbon profile feature data in the feature vector based on the weights. The display module is used to input the updated feature vector and carbon emission data collected at the same time as training data into the preset carbon profile generation model to obtain the trained carbon profile generation model; to acquire carbon profile feature data in real time, and then use the trained carbon profile generation model to obtain the current carbon emission data; and to generate a carbon profile in a preset display mode based on the carbon profile feature data and carbon emission data, and display it on the preset display interface.

7. The enterprise carbon profile generation system based on multi-source data fusion according to claim 6, characterized in that, The acquisition module includes acquisition units. This device is used to collect energy consumption data in real time from each production stage of an enterprise using pre-set energy consumption monitoring and metering equipment, and the energy consumption data is presented in a time series numerical format; the pre-set energy consumption monitoring and metering equipment includes at least: smart meters, gas meters and fuel sensors. Extract production process data from the manufacturing execution system; the production process data includes at least: equipment operating parameters, raw material input, and waste output. Obtain supply chain data through a logistics management system; the supply chain data should include at least: transportation routes, vehicle types, and supplier carbon emission coefficients. By using a pre-defined platform for publicizing policies, unstructured data such as carbon policy texts related to corporate carbon profiling are extracted; Energy market data is collected through the electricity trading platform; this data includes at least real-time electricity prices and carbon intensity indices.

8. A non-volatile computer storage medium, characterized in that, It stores computer instructions, which, when executed, implement a method for generating enterprise carbon profiles based on multi-source data fusion as described in any one of claims 1-6.

Images