Integrated ai agent system for carbon accounting and emission tracking

Abstract

An integrated AI agent system for carbon accounting and emission tracking is provided. The present system is an integrated AI agent system in which voice, images, text, and sensor data are processed by specialized agents, respectively, and integrated into normalized standard data via an integrated agent, then carbon emissions are calculated, and reports are automatically generated. In this system, an emission calculation engine calculates emissions for each category such as Scope 1, Scope 2, and Scope 3, and a report generation agent automatically generates reports that meet various international regulatory frameworks such as GHG, TCFD, and SBTi. In addition, the present system can continuously improve the algorithm by reflecting the user's feedback, and also includes various additional functions such as real-time analysis, administrators, auditors, customized reporting for each investor, version history management, visualization functions, and external API interworking.

Description

An integrated AI agent system for carbon accounting and emissions tracking.

[0001] The present invention relates to an integrated Artificial Intelligence (AI) agent system for carbon accounting and emissions tracking, and more particularly, to an integrated AI agent system designed to enhance carbon accounting and emissions tracking within a logistics value chain.

[0002]

[0003] Carbon emissions allowances are the right to emit greenhouse gases that contribute to and exacerbate global warming. Companies allocated emissions allowances are required to use greenhouse gases within their allocated limits. Any excess or insufficient allowances can be traded on the market.

[0004] In other words, carbon emissions allowances are the right to emit greenhouse gases such as carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), hydrofluorocarbons (HFCs), perfluorocarbons (PFCs), and sulfur hexafluoride (SF6), which are the main culprits of global warming. Carbon dioxide accounts for the largest proportion of greenhouse gases, so the purpose is to regulate carbon dioxide emissions. The Kyoto Protocol member countries agreed to reduce carbon dioxide emissions by an average of 5% compared to 1990 levels by 2012, and countries or companies that fail to comply are required to purchase carbon emissions allowances from external sources. Therefore, companies with high carbon dioxide emissions must either reduce their emissions themselves through technological development such as energy conservation, or purchase the allowances from companies with low emissions and thus surplus allowances.

[0005] Carbon credits are issued by the United Nations Framework Convention on Climate Change (UNFCCC), and can be freely traded like commodities on the market. Carbon credits are divided into two types: "allowances" and "credits." Allowances are the right to emit greenhouse gases that each entity obligated to reduce emissions can emit. When the government allocates allowances to companies, companies can trade any excess or surplus emissions in the market as credits. "Credits" are issued when greenhouse gas emissions are reduced beyond the originally projected level through greenhouse gas reduction activities.

[0006] Specifically, these types of carbon emission rights include AAUs (national quotas of countries obligated to reduce emissions under the Kyoto Protocol), EUAs (quotas set by the EU ETS (European Union Emissions Trading System)), CERs (greenhouse gas reductions through the Clean Development Mechanism (CDM)), ERUs (greenhouse gas reductions through the Joint Implementation Initiative (JI)), and RMUs (greenhouse gas absorptions through afforestation projects, etc. of countries obligated to reduce emissions under the Kyoto Protocol). Among the countries implementing a carbon emissions trading system, the one with the most active trading is the European Union (EU), which first established a carbon exchange in 2005 and implemented the system.

[0007] Meanwhile, the carbon emissions trading system is leading to the development of various new technologies aimed at reducing carbon emissions, a prime example being carbon capture and storage (CCS). This technology extracts carbon dioxide (CO2) from fossil fuels before it is released into the atmosphere, pressurizes it, and stores it in a liquid state.

[0008] As global cooperation continues to halt accelerating global warming, a system capable of continuously analyzing carbon emissions and reporting them to users is needed.

[0009]

[0010] The technical problem to be solved by the present invention is to provide an integrated AI agent system for carbon accounting and emissions tracking that receives unstructured input data and generates a report related to carbon emissions.

[0011] Another technical challenge of the present invention is to provide an integrated AI agent system for carbon accounting and emissions tracking that generates secondary reports for at least one of investor, regulatory, and internal management purposes using the primary generated reports.

[0012] Another technical challenge of the present invention is to improve a report generation algorithm using feedback data, and to provide an integrated AI agent system for carbon accounting and emissions tracking that performs report writing using the improved report generation algorithm.

[0013] Another technical challenge of the present invention is to provide an integrated AI agent system for carbon accounting and emissions tracking that automatically records audit trail information on the figures and calculation basis included in a report when generating a first report, and can prevent forgery by generating a digital signature or integrity hash value.

[0014] Another technical challenge of the present invention is to provide an integrated AI agent system for carbon accounting and emissions tracking that can include a mapping function that automatically converts the format and field composition of reports to meet the requirements of various international regulatory frameworks such as the GHG Protocol, TCFD, SBTi, CBAM, and ISO 14064.

[0015] Another technical challenge of the present invention is to provide an integrated AI agent system for carbon accounting and emissions tracking that can include visualization information such as time series analysis results, reduction potential predictions, and regional emission heat maps in reports, and automatically provide customized reports to managers, auditors, and investors.

[0016] Another technical challenge of the present invention is to provide an integrated AI agent system for carbon accounting and emissions tracking that provides an API that links with an external ERP, supply chain system, or accounting program, and can support JSON, XML, XLSX, and ESG standard templates as report output formats.

[0017] Another technical challenge of the present invention is to enable the user to schedule the report generation cycle or automate it according to the legal filing cycle, and to manage the version history while generating quarterly / annual reports.

[0018]

[0019] An integrated AI agent system for carbon accounting and emissions tracking according to one embodiment of the present invention comprises: one or more processors; and one or more memories storing instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, wherein the one or more processors are configured to receive first unstructured input data, convert voice data included in the first unstructured input data into first text data, convert image data included in the first unstructured input data into second text data, perform natural language processing on at least one of the first unstructured input data, the first text data, and the second text data to generate third text data, form carbon emission data using sensor data included in the first unstructured input data, receive the first unstructured input data, the third text data, and the carbon emission data to generate first standardized data, and perform first report writing using the first standardized data.

[0020] As one embodiment, the one or more processors may calculate carbon emissions by category using the first standardized data, generate carbon emission hotspot, carbon emission trend, and carbon emission reduction area data using the carbon emissions by category, and perform the first report creation using the carbon emission hotspot, carbon emission trend, and carbon emission reduction area data.

[0021] In one embodiment, the one or more processors may use the first report to generate a customized second report for at least one of investor, regulatory, and internal management purposes.

[0022] In one embodiment, the one or more processors may receive feedback data regarding the accuracy and relevance of the first standardized data and the first report, use the feedback data to improve the first standardized data and the first report generation algorithm, and use the improved algorithm to generate second standardized data and a third report from newly received second unstructured input data.

[0023] In one embodiment, the one or more processors, when generating second standardized data and a third report from the second unstructured input data, may receive the second unstructured input data, convert voice data included in the second unstructured input data into fourth text data, convert image data included in the second unstructured input data into fifth text data, perform natural language processing on at least one of the second unstructured input data, the fourth text data, and the fifth text data to generate sixth text data, form carbon emission data using sensor data included in the second unstructured input data, receive the second unstructured input data, the sixth text data, and the carbon emission data to generate the second standardized data, and perform the third report creation using the second standardized data.

[0024] In one embodiment, the one or more processors may include audit trail information indicating the source, calculation basis, calculation logic, and connection with input data for information included in the first report, and may add digital signature or hash value-based integrity verification information to prevent tampering with the first report.

[0025] In one embodiment, the one or more processors automate report generation cycles according to user-defined or regulatory schedules, schedule quarterly or annual report generation, and manage records by version.

[0026] In one embodiment, the one or more processors include a function for evaluating the confidence score of the learned report generation algorithm in real time and automatically triggering model retraining if it falls below a certain threshold.

[0027]

[0028] Through the above solution, the present invention provides the following effects.

[0029] First, the integrated AI agent system for carbon accounting and emissions tracking according to the present invention is configured to automatically generate a highly reliable carbon emissions report by comprehensively receiving and analyzing various forms of unstructured input data including voice, image, text, and sensor data, thereby contributing to the automation and improvement of accuracy in report creation.

[0030] Second, the present invention enables efficient data structuring and heterogeneous data integration by preprocessing individual data through multiple specialized agents (e.g., voice agents, image agents, text agents, sensor data agents) and standardizing them through an integrated agent. This maximizes efficiency in terms of time, cost, and human resources compared to traditional, manual carbon emissions estimation methods.

[0031] Third, the report generation method of the present invention is designed to support international regulatory frameworks such as the GHG Protocol, TCFD, and SBTi, ensuring compliance with global reporting standards and practical usability. Furthermore, by including audit trail information and integrity verification information (e.g., digital signatures, hash values) for the source, calculation basis, and calculation logic of each data item within the report, it prevents forgery and tampering and enhances auditability.

[0032] Fourth, the report of the present invention can be customized for stakeholders such as investors, regulators, and internal management, and is advantageous for long-term management and verification through the automatic cycle setting according to user settings or regulatory schedules and the version-by-version history management function.

[0033] Fifth, the present invention enables learning-based performance improvement of AI models by continuously improving standardization algorithms and report generation algorithms based on feedback data, and also enables implementation of an adaptive system function that automatically detects reliability scores below a certain standard and triggers relearning.

[0034]

[0035] FIG. 1 is an exemplary diagram showing the configuration of an integrated AI agent system for carbon accounting and emissions tracking according to an embodiment of the present invention.

[0036] Figure 2 is an exemplary diagram showing the configuration of a server according to an embodiment of the present invention.

[0037] Figure 3 is an exemplary diagram showing the configuration of a server according to an embodiment of the present invention.

[0038] Figure 4 is a diagram for explaining learning of a neural network according to an embodiment of the present invention.

[0039] FIG. 5 is a flowchart showing the procedure of a method for carbon accounting and emissions tracking according to an embodiment of the present invention.

[0040]

[0041] The advantages and features of the invention, and the methods for achieving them, will become clearer with reference to the embodiments described in detail below together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and may be implemented in various different forms. These embodiments are provided only to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention, and the present invention is defined only by the scope of the claims. Like reference numerals refer to like elements throughout the specification.

[0042] The terminology used herein is merely used to describe specific embodiments and is not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this specification, it should be understood that the term "comprises" and the like specify the presence of a feature, number, step, operation, component, part, or combination thereof described in this specification, but does not exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

[0043] Hereinafter, the configuration and operation of the ultra-wideband microwave-based measurement technology according to the present invention will be described with reference to the attached drawings.

[0044] FIG. 1 is an exemplary diagram showing the configuration of an integrated AI agent system for carbon accounting and emissions tracking according to an embodiment of the present invention.

[0045] As illustrated in FIG. 1, the integrated AI agent system (100) for carbon accounting and emissions tracking may include a plurality of user terminals (110-1,…,110-n), a server (120), and a database (130). In one embodiment, the database (130) is illustrated as being configured separately from the server (120), but is not limited thereto, and the database (130) may be provided within the server (120). For example, the server (120) may include a plurality of artificial intelligence models for performing machine learning algorithms. In one embodiment, the plurality of user terminals (110-1,…,110-n), the server (120), and the database (130) may be connected to each other so as to be able to communicate with each other via a network (N).

[0046] Figure 2 is an exemplary diagram showing the configuration of a server according to an embodiment of the present invention.

[0047] As illustrated in FIG. 2, the server (120) may include a central receiving agent (121), a voice agent (122), an image agent (123), a text agent (124), a sensor data agent (125), an integration agent (126), an emissions calculation engine (127), a data analysis agent (128), and a report generation agent (129). According to one embodiment, the central receiving agent (121), the voice agent (122), the image agent (123), the text agent (124), the sensor data agent (125), the integration agent (126), the emissions calculation engine (127), the data analysis agent (128), and the report generation agent (129) may be connected to communicate with each other via a system bus.

[0048] The central receiving agent (121) can receive the first unstructured input data. According to one embodiment, the central receiving agent (121) can receive the first unstructured input data, which is all unstructured data inputs, and transmit it to each specialized agent (voice agent (122), image agent (123), text agent (124), and sensor data agent (125)) and the integrated agent (126). For example, a speaker included in a user terminal (e.g., 110-1) can receive a user's voice input to form voice data, and the central receiving agent (121) can receive the voice data from the user terminal (110-1). In addition, a camera included in a user terminal (e.g., 110-n) can receive a photo taken by the user to form image data, and the central receiving agent (121) can receive the image data from the user terminal (110-n).

[0049] The voice agent (122) can convert voice data included in the first unstructured input data into first text data. According to one embodiment, the voice agent (122) can convert the voice data included in the first unstructured input data into first text data using at least one of the Whisper model, the Google Cloud Speech-to-Text application programming interface (API), and the Python library. In addition, the voice agent (122) can improve the accuracy of data conversion by using a confidence score and repeated verification when converting the first text data.

[0050] The Whisper model is OpenAI's speech-to-text model that can be used to convert audio data from audio files into text. The Whisper model is trained on a large-scale English audio and text dataset. The Whisper model is optimized for transcribing the content of audio files containing English speech into text data. It can also be used to transcribe audio files containing speech in other languages. The output of the Whisper model is English text.

[0051] Google Cloud Speech-to-Text API is Google's speech-to-text model that can be used to convert speech data into text data.

[0052] A Python library is a collection of modules and packages available for use in the Python programming language. Libraries typically provide various functions that aid in writing Python code. These functions can be easily performed by calling the library, without having to write code to perform specific tasks. The voice agent (122) can convert speech data into primary text data using the openai and google-cloud-speech Python libraries.

[0053] The image agent (123) can convert image data included in the first unstructured input data into second text data using at least one of Tesseract, Google Cloud Vision API, and a Python library. In addition, the image agent (123) can perform error correction and accuracy verification using multiple conversion paths when converting the second text data.

[0054] OCR stands for Optical Character Recognition, a technology that recognizes characters using light. It allows a computer to read and copy text from printed materials, document images, and other general images captured by a camera. The image agent (123) can convert image data into secondary text data using various OCR engines.

[0055] Tesseract is an optical character recognition engine for various operating systems. It is free software distributed under the Apache License, Version 2.0, and its development has been sponsored by Google since 2006.

[0056] Google Cloud Vision API provides advanced OCR and image analysis capabilities.

[0057] The image agent (123) can convert image data into second text data using the pytesseract and google-cloud-vision Python libraries.

[0058] The text agent (124) may perform natural language processing on at least one of the first unstructured input data, the first text data, and the second text data to generate third text data. According to one embodiment, the text agent (124) may generate the third text data using at least one of a Natural Language Toolkit, a spaCy model, and a regular expression. In addition, the text agent (124) may perform a feedback loop and heuristic check when generating the third text data to ensure consistency and accuracy.

[0059] The Natural Language Toolkit is a collection of libraries and programs for symbolic and statistical natural language processing in English, written in the Python programming language. It supports classification, tokenization, morphological analysis, tagging, parsing, and semantic inference.

[0060] Spacey Model is an open-source software library for advanced natural language processing written in the programming languages ​​Python and Cython.

[0061] Regular expressions are a formal language used to represent sets of strings with specific rules. Many text editors and programming languages ​​support regular expressions for string searching and substitution, and Perl and Tcl, in particular, have powerful regular expression implementations built into their languages. Text agents (124) can utilize regular expressions to organize and standardize data.

[0062] The sensor data agent (125) can form carbon emission data using the sensor data included in the first unstructured input data. According to one embodiment, the sensor data agent (125) can form carbon emission data using a customized IoT data processing algorithm using real-time data received from various IoT (Internet of Things) sensors attached to each of a plurality of user terminals (110-1,…,110-n). In addition, the sensor data agent (125) can continuously monitor the sensor data to detect anomalies and, if necessary, perform corrections to the IoT sensors.

[0063] The integrated agent (126) can generate first standardized data using the first unstructured input data received from the central receiving agent (121), the third text data received from the text agent (124), and the carbon emission data received from the sensor data agent (125). According to one embodiment, the integrated agent (126) can generate first standardized data by integrating and standardizing the first unstructured input data received from the central receiving agent (121), the third text data received from the text agent (124), and the carbon emission data received from the sensor data agent (125). For example, the integrated agent (126) can directly process multi-modal input. According to one embodiment, the integrated agent (126) can generate the first standardized data using at least one of GPT-4o and a Python library. When generating the first standardized data, the integrated agent (126) can verify the first standardized data using a reinforcement learning technique and perform continuous improvement of the results.

[0064] GPT-4o is OpenAI's multimodal model capable of seamlessly processing text, speech, and image input. GPT-4o can integrate and standardize primary and secondary text data to generate standardized data.

[0065] The integrated agent (126) can generate standardized data using the openai Python library.

[0066] The integrated agent (126) can use reinforcement learning techniques to verify the combined output and promote continuous improvement through repeated refinement and comparison with expected results.

[0067] In a data parsing method according to an embodiment of the present invention, a central receiving agent (121) receives first unstructured input data (voice, image, text, sensor data, etc.) which is an unstructured data input, and transmits the received first unstructured input data to an appropriate specialized agent (voice agent (122), image agent (123), text agent (124), sensor data agent (125)) and an integrated agent (126).

[0068] Each specialized agent that receives the first unstructured input data performs preprocessing of the data, and then the voice agent (122) converts the voice data into first text data, the image agent (123) converts the image data into second text data, the text agent (124) generates third text data, and the sensor data agent (125) normalizes the sensor data and ensures consistency while forming carbon emission data. Each specialized agent transmits the first to third text data and carbon emission data to the integrated agent (126).

[0069] The integrated agent (126) can generate first standardized data using the first unstructured input data, the third text, and carbon emissions data. In one embodiment, the integrated agent (126) can integrate data using the multi-modal capabilities of GPT-4o and generate the first standardized data in a standardized format. Furthermore, the integrated agent (126) can verify that the standardized data is accurate and ready for use in external systems.

[0070] The emission calculation engine (127) can calculate carbon emissions by category, such as Scope 1, Scope 2, and Scope 3, using the first standardized data. According to one embodiment, the emission calculation engine (127) can ensure compliance with international standards (e.g., GHG Protocol).

[0071] The GHG (Greenhouse Gas) Protocol is a set of guidelines for greenhouse gas accounting and reporting, developed by the World Business Council for Sustainable Development (WBCSD) and the World Resources Institute (WRI). Launched in 1998, the protocol is led by WRI, a U.S. environmental non-governmental organization (NGO), and WBCSD, a consortium of 170 multinational corporations, with governments, organizations, companies, and NGOs participating.

[0072] The GHG Protocol divides greenhouse gas emissions into three categories (Scopes): direct emissions (Scope 1), indirect emissions (Scope 2), and other indirect emissions (Scope 3), depending on the source. Scope 1 refers to carbon emissions directly generated from resources owned and managed by the company, i.e., direct emissions. Scope 2 refers to greenhouse gas emissions resulting from the purchased electricity consumed by the business, i.e., indirect emissions. Finally, Scope 3 refers to greenhouse gas emissions from sources not directly owned or controlled by the business, i.e., other indirect emissions.

[0073] Meanwhile, greenhouse gases are a general term for gases that pollute the Earth's atmosphere and cause the greenhouse effect. The six greenhouse gases targeted for reduction under the Kyoto Protocol to the United Nations Framework Convention on Climate Change are carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), hydrofluorocarbons (HFCs), perfluorocarbons (PFCs), and sulfur hexafluoride (SF6).

[0074] The data analysis agent (128) can perform an in-depth analysis of carbon emissions by category to generate data on carbon emission hotspots, carbon emission trends, and potential carbon emission reduction areas. In one embodiment, the data analysis agent (128) can identify key insights such as carbon emission hotspots, carbon emission trends, and potential carbon emission reduction areas.

[0075] The report generation agent (129) may include an automatic branching function for reports based on job functions (e.g., managers, auditors, investors), a visual dashboard conversion function, a multilingual translation and localization output function, an automatic tagging function for carbon emissions keywords in the report, a function for linking with an external audit system or ESG platform, and a report version management function to increase the marketability and practicality of the report.

[0076] The report generation agent (129) includes a first report generation algorithm that performs first report creation using data on carbon emission hotspots, carbon emission trends, and carbon emission reduction potential areas. The report generation agent (129) can generate a standardized first report suitable for various regulatory frameworks (e.g., GHG Protocol, TCFD, SBTi, etc.) using data on carbon emission hotspots, carbon emission trends, and carbon emission reduction potential areas.

[0077] The report generation agent (129) includes a second report generation algorithm that uses the generated first report to generate a customized second report for at least one of the purposes of investors, regulators, and internal management. The report generation agent (129) can use the generated first report to generate a customized second report to meet the specific needs of various stakeholders, such as investors, regulators, and internal management.

[0078] The server (120) may receive feedback data regarding the accuracy and relevance of the generated first standardized data and the first report. According to one embodiment, the server (120) may receive feedback data from one user terminal (e.g., 110-1) among multiple user terminals (110-1, ..., 110-n) via a network (N).

[0079] The server (120) may use the feedback data to improve the first standardized data and first report generation algorithm. In one embodiment, the server (120) may use the received feedback data to improve the processing algorithm for generating the first standardized data and the first report, thereby enhancing future performance.

[0080] The server (120) includes a third report generation algorithm that generates second standardized data and a third report from the newly received second unstructured input data using an improved algorithm. According to one embodiment, the server (120) can continuously update the machine learning model based on new data and feedback to adapt to changing conditions and requirements. For example, the central receiving agent (121) can receive the second unstructured input data, and the voice agent (122) can convert voice data included in the second unstructured input data into fourth text data. The image agent (123) can convert image data included in the second unstructured input data into fifth text data, and the text agent (124) can perform natural language processing on at least one of the second unstructured input data, the fourth text data, and the fifth text data to generate sixth text data. The sensor data agent (125) may form carbon emission data using the sensor data included in the second unstructured input data, and the integrated agent (126) may receive the second unstructured input data, the sixth text data, and the carbon emission data to generate second standardized data. The report generation agent (129) may create a third report using the second standardized data.

[0081] According to an embodiment of the present invention, a data parsing scenario can track emissions of transport vehicles using an integrated AI agent system (100) for carbon accounting and emissions tracking in a logistics company server. That is, IoT sensors mounted on multiple user terminals (110-1,…,110-n) each deployed in multiple transport vehicles can transmit real-time data to a central receiving agent (121). A sensor data agent (125) can process the received real-time data, and an integration agent (126) can integrate and standardize the data to generate first standardized data. An emissions calculation engine (127) can calculate carbon emissions of the transport vehicle, and a report generation agent (129) can generate a first standardized report for regulatory compliance and internal monitoring.

[0082] A data parsing scenario according to an embodiment of the present invention can input voice logs, receipt images, handwritten notes, etc. through multiple user terminals (110-1,…,110-n) of an integrated AI agent system (100) for carbon accounting and emissions tracking in a manufacturing company server. A voice agent (122) can extract text from the images, and a text agent (124) can process text data. An integrated agent (126) can integrate all data to calculate carbon emissions. A data analysis agent (128) can generate a detailed emissions report for stakeholders.

[0083] The network (N) can perform wireless or wired communication among a plurality of user terminals (110-1,…,110-n), a server (120), a database (130), etc. For example, the network (N) can perform wireless communication according to a method such as LTE (long-term evolution), LTE-A (LTE Advanced), CDMA (code division multiple access), WCDMA (wideband CDMA), WiBro (Wireless BroadBand), WiFi (wireless fidelity), Bluetooth (Bluetooth), NFC (near field communication), GPS (Global Positioning System), or GNSS (global navigation satellite system). For example, the network (N) can also perform wired communication according to a method such as USB (universal serial bus), HDMI (high definition multimedia interface), RS-232 (recommended standard 232), or POTS (plain old telephone service).

[0084] The database (130) can store various data. The data stored in the database (130) is data acquired, processed, or used by at least one component of a plurality of user terminals (110-1,…,110-n) and the server (120), and may include software (e.g., a program). The database (130) may include volatile and / or non-volatile memory. The database (130) may store first and second unstructured input data, first to sixth text data, first and second standardized data, carbon emission data, first to third reports, a voice-to-text conversion model for driving a voice agent (122), an image-to-text conversion model for driving an image agent (123), a natural language processing model for driving a text agent (124), and a standardization model for driving an integrated agent (125).

[0085] In the present invention, artificial intelligence (AI) refers to a technology that mimics human learning, reasoning, and perception abilities and implements them on a computer. It may include concepts such as machine learning and symbolic logic. Machine learning (ML) is an algorithmic technology that classifies or learns the characteristics of input data on its own. AI technology analyzes input data using machine learning algorithms, learns the results of that analysis, and makes judgments or predictions based on the results of that learning. Furthermore, technologies that utilize machine learning algorithms to mimic human brain functions such as cognition and judgment can also be understood as falling under the category of AI. For example, this may include technical fields such as linguistic understanding, visual understanding, inference / prediction, knowledge representation, and motion control.

[0086] Machine learning can refer to the process of training a neural network model using data processing experience. Through machine learning, computer software can improve its data processing capabilities. Neural network models are built by modeling correlations between data, and these correlations can be expressed by multiple parameters. Neural network models extract and analyze features from given data to derive correlations between data. This process of iteratively optimizing the parameters of a neural network model can be defined as machine learning. For example, a neural network model can learn the mapping (correlation) between inputs and outputs for data presented as input-output pairs. Alternatively, a neural network model can learn the relationships between inputs and outputs by deriving regularities between the given data, even when presented with only input data.

[0087] An artificial intelligence learning model or neural network model can be designed to implement the structure of the human brain on a computer, and can include multiple network nodes that simulate the neurons of a human neural network and have weights. The multiple network nodes can have connections with each other by simulating the synaptic activity of neurons that exchange signals through synapses. In an artificial intelligence learning model, multiple network nodes can be located at layers of different depths and exchange data according to convolutional connections. The artificial intelligence learning model can be, for example, an artificial neural network model, a convolutional neural network (CNN), etc. The artificial intelligence learning model can be machine-learned using methods such as supervised learning, unsupervised learning, and reinforcement learning. Machine learning algorithms that can be used to perform machine learning include decision trees, Bayesian networks, support vector machines, artificial neural networks, Ada-boost, perceptrons, genetic programming, and clustering.

[0088] CNNs are a type of multilayer perceptron designed to utilize minimal preprocessing. They consist of one or more convolutional layers stacked on top of regular artificial neural network layers, with additional weight and pooling layers. This structure allows CNNs to fully utilize two-dimensional input data. Compared to other deep learning architectures, CNNs demonstrate excellent performance in both image and audio domains. CNNs can also be trained using standard backpropagation. Compared to other feedforward artificial neural network techniques, CNNs are easier to train and have fewer parameters.

[0089] Convolutional networks are neural networks that contain sets of nodes with bounded parameters. The increasing availability of training data and computational power, combined with advances in algorithms such as piecewise linear units and dropout training, have led to significant improvements in many computer vision tasks. With the massive datasets available for many tasks today, overfitting is less of a concern, and increasing network size improves test accuracy. Optimal utilization of computing resources becomes a limiting factor. To address this, distributed, scalable implementations of deep neural networks can be utilized.

[0090] Figure 3 is an exemplary diagram showing the configuration of a server according to an embodiment of the present invention.

[0091] As illustrated in FIG. 3, the server (120) may include one or more processors (131), one or more memories (132), and a transceiver (133). At least one of these components of the server (120) may be omitted, or another component may be added to the server (120). Additionally or alternatively, some of the components may be implemented in an integrated manner, or may be implemented as a single or multiple entities. At least some of the components inside and outside the server (120) may be connected to each other via a system bus, a general purpose input / output (GPIO), a serial peripheral interface (SPI), or a mobile industry processor interface (MIPI), and may exchange data and / or signals.

[0092] One or more processors (131) may control at least one component of a server (120) connected to the processor (131) by running software (e.g., commands, programs, etc.). In addition, the processor (131) may perform various operations related to the present invention, such as calculations, processing, data generation, and processing. In addition, the processor (131) may load data, etc. from one or more memories (132), or store data, etc. in one or more memories (132).

[0093] One or more processors (131) may receive first non-standard input data. According to one embodiment, the processor (131) may receive first non-standard input data from a plurality of user terminals (110-1,…, 110-n) via a transceiver (133).

[0094] One or more processors (131) can convert voice data included in the first unstructured input data into first text data. The processor (131) can convert the voice data included in the unstructured input data into first text data using at least one of the Whisper model, the Google Cloud Speech-to-Text application programming interface (API), and the Python library. In addition, the processor (131) can improve the accuracy of data conversion by using a reliability score and repeated verification when converting the first text data.

[0095] One or more processors (131) can convert image data included in the first unstructured input data into second text data. The processor (131) can convert the image data included in the first unstructured input data into second text data using at least one of Tesseract, Google Cloud Vision API, and Python library. In addition, the processor (131) can verify accuracy by using error correction and multiple conversion paths when converting the second text data.

[0096] One or more processors (131) may perform natural language processing on at least one of the first unstructured input data, the first text data, and the second text data to generate third text data. The processor (131) may generate the third text data using at least one of a Natural Language Toolkit, a spaCy model, and a regular expression. In addition, the processor (131) may perform a feedback loop and a heuristic check when generating the third text data to ensure consistency and accuracy.

[0097] One or more processors (131) may form carbon emission data using sensor data included in the first non-standard input data. According to one embodiment, the processor (131) may form carbon emission data using a customized IoT data processing algorithm using real-time data received from various IoT (Internet of Things) sensors attached to each of a plurality of user terminals (110-1,…,110-n) through a transmitter / receiver (133).

[0098] One or more processors (131) may generate first standardized data using the first unstructured input data received from the central receiving agent (121), the third text data received from the text agent (124), and the carbon emission data received from the sensor data agent (125). According to one embodiment, the processor (131) may integrate and standardize the first unstructured input data received from the central receiving agent (121), the third text data received from the text agent (124), and the carbon emission data received from the sensor data agent (125) to generate the first standardized data. For example, the processor (131) may directly process multi-modal input. According to one embodiment, the processor (131) may generate the first standardized data using at least one of GPT-4o and a Python library. When generating the first standardized data, the processor (131) may use a reinforcement learning technique to verify the first standardized data and perform continuous improvement of the results.

[0099] One or more processors (131) may perform first report creation using the first standardized data. According to one embodiment, the processor (131) may calculate carbon emissions by category using the first standardized data, perform an in-depth analysis of the carbon emissions by category to generate data on carbon emission hotspots, carbon emission trends, and carbon emission reduction potential areas, and may perform first report creation using the data on carbon emission hotspots, carbon emission trends, and carbon emission reduction potential areas.

[0100] One or more memories (132) may store the first and second unstructured input data, the first to sixth text data, the first and second standardized data, carbon emission data, the first to third reports, a voice-to-text conversion model for driving a voice agent (122), an image-to-text conversion model for driving an image agent (123), a natural language processing model for driving a text agent (124), and a standardization model for driving an integrated agent (125). In addition, one or more memories (132) may store commands that, when executed by one or more processors (131), cause one or more processors (131) to perform operations.

[0101] According to one embodiment, the server (120) may further include a transceiver (133). The transceiver (133) may perform wireless or wired communication between the server (120) and various external servers (e.g., logistics company servers, manufacturing company servers, etc.), databases, client devices, and / or other devices. For example, the transceiver (133) can perform wireless communication according to a method such as eMBB (enhanced Mobile Broadband), URLLC (Ultra Reliable Low-Latency Communications), MMTC (Massive Machine Type Communications), LTE (long-term evolution), LTE-A (LTE Advance), UMTS (Universal Mobile Telecommunications System), GSM (Global System for Mobile communications), CDMA (code division multiple access), WCDMA (wideband CDMA), WiBro (Wireless Broadband), WiFi (wireless fidelity), Bluetooth (Bluetooth), NFC (near field communication), GPS (Global Positioning System), or GNSS (global navigation satellite system). For example, the transceiver (133) can also perform wired communication according to a method such as USB (universal serial bus), HDMI (high definition multimedia interface), RS-232 (recommended standard232), or POTS (plain old telephone service).

[0102] According to one embodiment, one or more processors (131) can control a transceiver (133) to obtain information from various external servers and databases (130). The information obtained from various external servers and databases (130) can be stored in one or more memories (132).

[0103] According to one embodiment, the server (120) may be a device of various forms. For example, the server (120) may be a portable communication device, a computer device, or a device according to a combination of one or more of the devices described above. The server (120) of the present invention is not limited to the devices described above.

[0104] The various embodiments of the server (120) according to the present invention can be combined with each other. Each embodiment can be combined according to a number of cases, and the combined server (120) embodiments also fall within the scope of the present invention. Furthermore, the internal / external components of the server (120) according to the present invention described above can be added, changed, replaced, or deleted depending on the embodiment. Furthermore, the internal / external components of the server (120) described above can be implemented as hardware components.

[0105] Figure 4 is a diagram for explaining learning of a neural network according to an embodiment of the present invention.

[0106] As illustrated in FIG. 4, the learning device can train a neural network (135) to extract text data from unstructured input data. In one embodiment, the learning device may be a separate entity from the server (120), but is not limited thereto.

[0107] The neural network (135) includes an input layer (134) into which training samples are input and an output layer (136) that outputs training outputs, and can be learned based on the differences between the training outputs and labels. Here, the labels can be defined based on text data corresponding to unstructured input data. The neural network (135) is connected as a group of multiple nodes and is defined by weights between the connected nodes and an activation function that activates the nodes.

[0108] The learning device can train a neural network (135) using the GD (Gradient Descent) technique or the SGD (Stochastic Gradient Descent) technique. The learning device can use a loss function designed based on the outputs and labels of the neural network.

[0109] The learning device can calculate a training error using a predefined loss function. The loss function can be predefined as input variables, including labels, outputs, and parameters, where the parameters can be set by weights within the neural network (135). For example, the loss function can be designed in the form of a Mean Square Error (MSE) or entropy, and various techniques or methods can be employed in the design of the loss function.

[0110] The learning device can use the backpropagation technique to identify weights that influence training errors. Here, the weights represent relationships between nodes within the neural network (135). The learning device can utilize the SGD technique using labels and outputs to optimize the weights identified through the backpropagation technique. For example, the learning device can update the weights of a loss function defined based on the labels, outputs, and weights using the SGD technique.

[0111] According to one embodiment, the learning device can acquire unstructured input data of a training target and extract text data of the training target. The learning device can acquire pre-labeled information (first labels) for each of the training unstructured input data, and can acquire first labels representing predefined text data for the training unstructured input data.

[0112] According to one embodiment, the learning device can generate first training feature vectors based on array features, sequence features, and pattern features of the training unstructured input data. Various methods can be employed to extract features of the training unstructured input data.

[0113] According to one embodiment, the learning device can obtain training outputs by applying the first training feature vectors to the neural network (129). The learning device can train the neural network (129) based on the training outputs and the first labels. The learning device can calculate training errors corresponding to the training outputs and train the neural network (129) by optimizing the connection relationship between nodes in the neural network (129) to minimize the training errors. The server (120) can form first to third text data from the first unstructured input data using the neural network (129) for which training has been completed. In addition, the server (120) can form fourth to sixth text data from the second unstructured input data using the neural network (129) for which training has been completed.

[0114] FIG. 5 is a flowchart illustrating a method for carbon accounting and emissions tracking according to an embodiment of the present invention. Although the process steps, method steps, and algorithms are described in a sequential order in the flowchart of FIG. 5 , such processes, methods, and algorithms may be configured to operate in any suitable order. In other words, the steps of the processes, methods, and algorithms described in various embodiments of the present invention need not be performed in the order described herein. Furthermore, even if some steps are described as being performed asynchronously, in other embodiments, such some steps may be performed concurrently. Furthermore, the illustration of a process by depiction in the drawings does not imply that the illustrated process excludes other variations and modifications thereof, nor does it imply that the illustrated process or any of its steps is essential to one or more of the various embodiments of the present invention, nor does it imply that the illustrated process is preferred.

[0115] As illustrated in FIG. 5, in step S510, first unstructured input data is received. Referring to FIGS. 1 to 4, the central receiving agent (121) of the server (120) can receive the first unstructured input data. The central receiving agent (121) can receive a user's voice input through a speaker included in one of a plurality of user terminals (110-1,...,110-n) (e.g., 110-1) to form voice data, and the central receiving agent (121) can receive the voice data from the user terminal (110-1) through the network (N). In addition, the camera included in the user terminal (e.g., 110-n) can receive a receipt photo taken by the user to form image data, and the central receiving agent (121) can receive the image data from the user terminal (110-n) through the network (N).

[0116] In step (S520), voice data is converted into first text data. Referring to FIGS. 1 to 4, the voice agent (122) of the server (120) can convert voice data included in the first unstructured input data into first text data. The voice agent (122) can convert the voice data included in the unstructured input data into first text data using at least one of the Whisper model, the Google Cloud Speech-to-Text API (application programming interface), and the Python library. In addition, the voice agent (122) can improve the accuracy of data conversion by using a confidence score and repeated verification when converting the first text data.

[0117] In step (S530), image data is converted into second text data. For example, referring to FIGS. 1 to 4, the image agent (123) of the server (120) may convert image data included in the first unstructured input data into second text data using at least one of Tesseract, Google Cloud Vision API, and a Python library. In addition, the image agent (123) may verify accuracy by using error correction and multiple conversion paths when converting the second text data.

[0118] In step (S540), third text data is generated. Referring to FIGS. 1 to 4, the text agent (124) of the server (120) may perform natural language processing on at least one of the first unstructured input data, the first text data, and the second text data to generate the third text data. The text agent (124) may generate the third text data using at least one of a Natural Language Toolkit, a spaCy model, and a regular expression. In addition, the text agent (124) may perform a feedback loop and a heuristic check when generating the third text data to ensure consistency and accuracy.

[0119] In step (S550), carbon emission data is generated. Referring to FIGS. 1 to 4, the sensor data agent (125) of the server (120) can form carbon emission data using sensor data included in the first unstructured input data. The sensor data agent (125) can form carbon emission data using a customized IoT data processing algorithm using real-time data received from various IoT (Internet of Things) sensors attached to each of a plurality of user terminals (110-1,…,110-n).

[0120] In step (S560), first standardized data is generated. Referring to FIGS. 1 to 4, the integrated agent (126) of the server (120) may generate the first standardized data using the first unstructured input data received from the central receiving agent (121), the third text data received from the text agent (124), and the carbon emission data received from the sensor data agent (125). According to one embodiment, the integrated agent (126) may generate the first standardized data by integrating and standardizing the first unstructured input data received from the central receiving agent (121), the third text data received from the text agent (124), and the carbon emission data received from the sensor data agent (125). For example, the integrated agent (126) may directly process multi-mode input. According to one embodiment, the integrated agent (126) may generate the first standardized data using at least one of GPT-4o and a Python library. The integrated agent (126) can use reinforcement learning techniques to verify the first standardized data when generating the first standardized data and perform continuous improvement of the results.

[0121] In step (S570), a first report is created. Referring to FIGS. 1 to 4, one or more processors (131) of the server (120) may create the first report using the first standardized data. The processor (131) may calculate carbon emissions by category using the first standardized data, perform an in-depth analysis of the carbon emissions by category, generate data on carbon emission hotspots, carbon emission trends, and carbon emission reduction potential areas, and create the first report using the data on carbon emission hotspots, carbon emission trends, and carbon emission reduction potential areas.

[0122] While the method has been described through specific embodiments, the method can also be implemented as computer-readable code on a computer-readable recording medium. A computer-readable recording medium includes any type of recording device that stores data that can be read by a computer system. Examples of computer-readable recording media include ROMs, RAMs, CD-ROMs, magnetic tapes, floppy disks, and optical data storage devices. Furthermore, the computer-readable recording medium can be distributed across network-connected computer systems, such that the computer-readable code can be stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the above embodiments can be readily inferred by programmers skilled in the art to which the present invention pertains.

[0123] Although the present invention has been described above with specific details such as specific components and limited embodiments and drawings, these are provided only to help a more general understanding of the present invention, and the present invention is not limited to the above embodiments, and those with ordinary skill in the art to which the present invention pertains can make various modifications and variations based on this description. Therefore, the spirit of the present invention should not be limited to the embodiments described above, and all things that are equally or equivalently modified as well as the following claims are considered to fall within the scope of the spirit of the present invention.

[0124]

[0125] The present invention has a structure that automatically generates a quantitative carbon emission report by processing unstructured environmental data, so it can be used in the manufacturing industry that requires calculation of supply chain emissions and improvement reporting, the logistics and distribution industry that requires calculation of carbon emissions based on fuel usage / mileage by means of transportation, the energy and power generation industry that requires real-time tracking of carbon emissions at the power plant level and creation of cumulative reports, the construction and infrastructure field that requires data based on ESG evaluation of national / local government construction projects, and financial and insurance institutions that require collection of emission indicators for ESG investment evaluation and responsible investment (PRI) response and can design carbon emission-based risk insurance products. It is a technology that can satisfy a wide range of practical and industrial needs, such as responding to domestic and international carbon regulations, internalizing ESG management, and transitioning to an eco-friendly supply chain, beyond a simple emissions measurement tool, and can contribute to the establishment of a carbon accounting infrastructure encompassing the government, private sector, and international organizations.

Claims

1. One or more processors; and When executed by the one or more processors, it includes one or more memories in which instructions for causing the one or more processors to perform operations are stored, One or more of the above processors, Receive the first unstructured input data, Converting voice data included in the first unstructured input data into first text data, converting image data included in the first unstructured input data into second text data, and performing natural language processing on at least one of the first unstructured input data, the first text data, and the second text data to generate third text data, Forming carbon emission data using sensor data included in the above first unstructured input data, Receiving the first unstructured input data, the third text data, and the carbon emission data to generate first standardized data, Using the first standardized data above, carbon emissions are calculated by category, and using the carbon emissions by category, data on carbon emission hotspots, carbon emission trends, and carbon emission reduction potential areas are generated, and using the data on carbon emission hotspots, carbon emission trends, and carbon emission reduction potential areas, a first report is prepared. An integrated AI agent system for carbon accounting and emissions tracking.

2. In paragraph 1, One or more of the above processors, Using the above first report, a second report tailored to the purpose of at least one of investors, regulators, and internal management is generated. An integrated AI agent system for carbon accounting and emissions tracking.

3. In paragraph 2, One or more of the above processors, Receive feedback data on the accuracy and relevance of the first standardized data and the first report, improve the first standardized data and the first report generation algorithm using the feedback data, and generate second standardized data and a third report from the newly received second unstructured input data using the improved algorithm. An integrated AI agent system for carbon accounting and emissions tracking.

4. In paragraph 3, One or more of the above processors, When generating second standardized data and a third report from the second unstructured input data, Receiving the second non-standard input data, Converting voice data including the second non-standard input data into fourth text data, Converting image data including the second non-standard input data into fifth text data, Generating sixth text data by performing natural language processing on at least one of the second unstructured input data, the fourth text data, and the fifth text data, Forming carbon emission data using sensor data including the second non-standard input data, Receiving the second unstructured input data, the sixth text data, and the carbon emission data to generate the second standardized data, and performing the third report creation using the second standardized data. An integrated AI agent system for carbon accounting and emissions tracking.

5. In paragraph 1, One or more of the above processors, Include audit trail information indicating the source, basis for calculation, calculation logic and connection with input data for the information included in the above first report; In order to prevent falsification of the above first report, digital signature or hash value-based integrity verification information is added. An integrated AI agent system for carbon accounting and emissions tracking.

6. In paragraph 1, One or more of the above processors, Automate report generation cycles based on user-defined or regulatory schedules, schedule quarterly or unit report generation, and manage records by version. An integrated AI agent system for carbon accounting and emissions tracking.

7. In paragraph 1, One or more of the above processors, Includes the ability to evaluate the confidence score of the learned report generation algorithm in real time and automatically trigger model retraining if it falls below a certain standard. An integrated AI agent system for carbon accounting and emissions tracking.

Images