Artificial intelligence-based carbon emission prediction method and system
By deploying sensors and a real-time data processing platform in the steel production process, and combining deep learning and modular design, the problems of data dependence and real-time performance in carbon emission prediction for steel enterprises have been solved, achieving high-precision and flexible carbon emission prediction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HEFEI UNIV OF TECH
- Filing Date
- 2024-12-12
- Publication Date
- 2026-06-09
Smart Images
Figure CN119647696B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of artificial intelligence technology, specifically to an artificial intelligence-based carbon emission prediction method and system. Background Technology
[0002] In the steel production process, energy consumption and carbon emissions mainly originate from key stages such as ironmaking, steelmaking, and steel rolling. These stages involve substantial energy use, including electricity, coal, and natural gas, with electricity consumption being particularly critical as it is directly related to carbon emissions. The purpose of this invention is to develop an artificial intelligence-based carbon emission prediction technology that, by analyzing equipment power consumption data, can predict carbon emissions at each production stage, providing decision support for enterprises.
[0003] In related technologies, the prediction of carbon emissions from steel enterprises mainly relies on traditional machine learning algorithms and deep learning algorithms. Specifically, firstly, machine learning-based carbon emission prediction methods typically use algorithms such as Support Vector Machines (SVM), Random Forests, or Gradient Boosting Machines (GBM) to build predictive models by combining historical energy consumption data and carbon emission data. Secondly, deep learning algorithms, such as Convolutional Neural Networks (CNN) or Recurrent Neural Networks (RNN), can be used to predict carbon emissions. Deep learning algorithms can handle more input variables and capture the nonlinear relationships between variables.
[0004] However, machine learning-based carbon emission prediction methods are limited by their requirement for large amounts of labeled data and their limited ability to process real-time data, making them unsuitable for real-time prediction. Deep learning algorithms have improved prediction accuracy, but they typically require large amounts of labeled data and have limited ability to process real-time data, making them unsuitable for real-time prediction; they also cannot effectively handle real-time data streams, resulting in insufficient system flexibility and scalability. Summary of the Invention
[0005] To address the shortcomings of existing technologies, this application provides a carbon emission prediction method and system based on artificial intelligence, which solves the problem of inaccurate prediction accuracy caused by excessive limitations in carbon emission prediction for steel enterprises.
[0006] To achieve the above objectives, this application provides the following technical solution:
[0007] In a first aspect, embodiments of this application provide an artificial intelligence-based carbon emission prediction method, which includes: deploying multiple sensors in the steel production process, integrating sensor data acquisition, data synchronization, data cleaning, and feature extraction methods to obtain first target stream data information; building a real-time data processing platform, and formatting the data based on the first target stream data information to obtain second target stream data information; based on the second target stream data information, integrating deep learning model selection, transfer learning application, model training strategies, and model compression technology, obtaining multiple target models through model training and optimization to predict carbon emissions from steel production; merging real-time prediction granularity, user interface design, and report generation technologies to provide hourly carbon emission prediction results for intuitive display; and periodically updating the target models based on acquired user feedback data and actual carbon emission data to improve the accuracy of carbon emission prediction.
[0008] Secondly, embodiments of this application provide an artificial intelligence-based carbon emission prediction system, which includes: a data acquisition and preprocessing module, a real-time data processing module, a model training and optimization module, a prediction output module, and an iterative improvement module.
[0009] Specifically, the data acquisition and preprocessing module is used to deploy multiple sensors based on the steel production process, integrating sensor data acquisition, data synchronization, data cleaning, and feature extraction methods to obtain the first target stream data information; the real-time data processing module is used to build a real-time data processing platform and format the data based on the first target stream data information to obtain the second target stream data information; the model training and optimization module is used to integrate deep learning model selection, transfer learning application, model training strategies, and model compression technology based on the second target stream data information, and obtain multiple target models through model training and optimization to predict carbon emissions from steel production; the prediction output module is used to combine real-time prediction granularity, user interface design, and report generation technology to provide hourly carbon emission prediction results for intuitive display; and the iterative improvement module is used to periodically update the target models based on the acquired user feedback data and actual carbon emission data to improve the accuracy of carbon emission prediction.
[0010] Thirdly, embodiments of this application provide an electronic device, which includes: a processor, a memory, and a program stored in the memory and executable on the processor. When the program is executed by the processor, it implements the artificial intelligence-based carbon emission prediction method described in the first aspect above.
[0011] Fourthly, embodiments of this application provide a computer-readable storage medium storing a program or instructions that, when executed by a processor, implement the artificial intelligence-based carbon emission prediction method described in the first aspect above.
[0012] This application provides a carbon emission prediction method and system based on artificial intelligence. Compared with existing technologies, it has the following advantages:
[0013] This application deploys multiple sensors to collect and process data during the steel production process to obtain the first target flow data information. A real-time data processing platform is built to meet the steel company's need for timely response, enhancing the ability to predict production changes in a timely manner. This application integrates deep learning model selection, transfer learning application, model training strategies, and model compression technology to improve the prediction accuracy of carbon emissions at each stage of steel production, reducing errors caused by data dependence and insufficient model generalization ability. Through modular design and transfer learning technology, the applicability and accuracy of the model in different production environments and processes are improved, making the model not limited to specific production conditions. This application also combines real-time prediction granularity, user interface design, and report generation technology, enabling intuitive display of carbon emission prediction results. It can also periodically update the target model based on user feedback data and actual carbon emission data, supporting online learning and allowing the model to be dynamically updated based on the latest data. Attached Figure Description
[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.
[0015] Figure 1 This is a flowchart illustrating an artificial intelligence-based carbon emission prediction method provided in an embodiment of this application.
[0016] Figure 2 yes Figure 1 An exemplary process diagram of S130;
[0017] Figure 3 This is a schematic diagram of the structure of an artificial intelligence-based carbon emission prediction system provided in an embodiment of this application;
[0018] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are described clearly and completely. Obviously, the described embodiments are only some embodiments of the present invention, 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.
[0020] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, 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 a process, method, article, or apparatus. Without further limitations, 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 the element.
[0021] This application provides an artificial intelligence-based carbon emission prediction method and system, which solves the problem of inaccurate prediction accuracy caused by excessive limitations in carbon emission prediction for steel enterprises.
[0022] The technical solution in this application is to solve the above-mentioned technical problems, and the general idea is as follows:
[0023] Against the backdrop of global climate change and sustainable development, the carbon emissions of the steel manufacturing industry, as an energy-intensive sector, have received widespread attention. The technology involved in this invention is a method for predicting carbon emissions from steel enterprises. It aims to help companies optimize production processes, reduce environmental impact, and comply with increasingly stringent environmental regulations by accurately predicting and managing carbon emissions.
[0024] In the steel production process, energy consumption and carbon emissions mainly originate from key stages such as ironmaking, steelmaking, and steel rolling. These stages involve substantial energy use, including electricity, coal, and natural gas, with electricity consumption being particularly critical as it is directly related to carbon emissions. The purpose of this invention is to develop an artificial intelligence-based carbon emission prediction technology that, by analyzing equipment power consumption data, can predict carbon emissions at each production stage, providing decision support for enterprises.
[0025] In related technologies, the prediction of carbon emissions from steel enterprises mainly relies on traditional machine learning algorithms and deep learning algorithms. Specifically, firstly, machine learning-based carbon emission prediction methods typically use algorithms such as Support Vector Machines (SVM), Random Forests, or Gradient Boosting Machines (GBM) to build predictive models by combining historical energy consumption data and carbon emission data. Secondly, deep learning algorithms, such as Convolutional Neural Networks (CNN) or Recurrent Neural Networks (RNN), can be used to predict carbon emissions. Deep learning algorithms can handle more input variables and capture the nonlinear relationships between variables.
[0026] However, machine learning-based carbon emission prediction methods are limited by their requirement for large amounts of labeled data and their limited ability to process real-time data, making them unsuitable for real-time prediction. Deep learning algorithms have improved prediction accuracy, but they typically require large amounts of labeled data and have limited ability to process real-time data, making them unsuitable for real-time prediction; they also cannot effectively handle real-time data streams, resulting in insufficient system flexibility and scalability.
[0027] Specifically, the relevant technology has the following drawbacks:
[0028] (1) High data dependence: Many existing carbon emission prediction models, especially those based on traditional machine learning, rely heavily on large amounts of labeled data. In the steel production sector, obtaining sufficient labeled data is often impractical because the collection and labeling of such data is both time-consuming and expensive.
[0029] (2) Insufficient real-time processing capability: Existing models are often unable to effectively process real-time data streams. In the steel production process, it is crucial to be able to respond to production changes and predict carbon emissions in real time, but existing technologies often cannot meet this requirement.
[0030] (3) Limited generalization ability: Many models perform well under specific conditions, but their predictive accuracy drops significantly when applied to different production environments or processes. This limits the practicality and scalability of the models.
[0031] (4) Difficulty in updating models: Traditional models need to be retrained when faced with changes in production conditions or shifts in data distribution. This is not only time-consuming, but may also require additional labeled data, which is not feasible in practice.
[0032] (5) Lack of modular design: Existing forecasting systems often do not follow the specific process of steel production in a modular design, resulting in insufficient system flexibility and scalability.
[0033] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.
[0034] The following section first introduces an artificial intelligence-based carbon emission prediction method provided in the embodiments of this application.
[0035] This application provides a flowchart illustrating an artificial intelligence-based carbon emission prediction method, as shown in the embodiments below. Figure 1 As shown, the artificial intelligence-based carbon emission prediction method may include the following steps S110-S150.
[0036] S110. Based on the deployment of multiple sensors in the steel production process, integrate sensor data acquisition, data synchronization, data cleaning and feature extraction methods to obtain the first target flow data information;
[0037] S120. Build a real-time data processing platform and format the data based on the first target stream data information to obtain the second target stream data information;
[0038] S130. Based on the second target stream data information, integrate deep learning model selection, transfer learning application, model training strategy and model compression technology, and obtain multiple target models through model training and optimization to predict carbon emissions from steel production.
[0039] S140 integrates real-time forecast granularity, user interface design, and report generation technology to provide hourly carbon emission forecasts for intuitive visualization;
[0040] S150. Update the target model regularly based on the obtained user feedback data and actual carbon emission data to improve the accuracy of carbon emission prediction.
[0041] The above describes a specific implementation of an artificial intelligence-based carbon emission prediction method provided in this application. It is understood that this application deploys multiple sensors in the steel production process to collect and process data to obtain first target flow data information, and builds a real-time data processing platform to meet the steel company's need for timely response, thereby enhancing the ability to predict production changes in a timely manner. This application integrates deep learning model selection, transfer learning application, model training strategies, and model compression technology to improve the prediction accuracy of carbon emissions at each stage of the steel production process, reduce errors caused by data dependence and insufficient model generalization ability, and improve the applicability and accuracy of the model in different production environments and processes through modular design and transfer learning technology, making the model not limited to specific production conditions.
[0042] Furthermore, this application reduces the reliance on large amounts of labeled data through unsupervised and self-supervised learning, and enhances the model's learning ability by utilizing unlabeled data; it also employs streaming data processing technology to achieve efficient processing of real-time data streams, meeting the needs of real-time prediction.
[0043] Furthermore, this application integrates real-time prediction granularity, user interface design, and report generation technologies, enabling intuitive display of carbon emission prediction results. It also allows for regular updates to the target model based on user feedback data and actual carbon emission data, supporting online learning and enabling the model to dynamically update according to the latest data. This application overcomes the shortcomings of existing technologies in terms of data processing capabilities, prediction accuracy, real-time performance, and generalization ability.
[0044] In some embodiments, the aforementioned deployment of multiple sensors based on the steel production process integrates sensor data acquisition, data synchronization, data cleaning, and feature extraction methods to obtain first target stream data information. Specifically, S110 may include the following steps:
[0045] S210, deploy multiple sensors based on key equipment and processes in steel production; the processes include: raw material preparation stage, coking stage, blast furnace ironmaking stage, steelmaking stage, continuous casting stage, hot rolling stage, cold rolling stage, post-processing stage, energy management stage, and waste treatment and recycling stage;
[0046] S220: Data is collected in real time through multiple sensors, and the data from different sensors and devices are time-stamped and synchronized to obtain initial flow data information, which includes temperature data, pressure data, flow rate data and energy consumption data.
[0047] S230. Introduce an unsupervised learning algorithm and use automated scripts to identify and correct missing and outlier values in the initial streaming data and remove noise.
[0048] S240. Extract key features related to carbon emissions from the data to obtain the first target flow data information.
[0049] In the embodiments of this application, it is understood that after obtaining the initial streaming data information, unsupervised learning algorithms, such as K-means clustering, can be introduced to automatically identify and correct outliers, thereby enhancing the robustness of the data. Statistical analysis and domain knowledge can be applied in the process of extracting key features related to carbon emissions.
[0050] In some embodiments, the aforementioned real-time data processing platform is built, and data is formatted based on the first target stream data information to obtain the second target stream data information. That is, the aforementioned S120 may specifically include the following steps:
[0051] S310. Build a real-time data processing platform using a pre-defined stream processing framework; the stream processing framework includes at least one of Apache Kafka and Apache Storm.
[0052] S320. Apply sliding window time window technology to process the first target stream data information, and perform normalization and encoding conversion to complete the data format conversion, thereby obtaining the second target stream data information.
[0053] In the embodiments of this application, it is understood that this application can utilize streaming processing frameworks such as Apache Kafka and Apache Storm to build a real-time data processing platform to process high-speed data streams; and apply time window techniques such as sliding windows to process streaming data in order to provide data at fixed time intervals required by the model. Through data format conversion, the data can be converted into an input format acceptable to the model.
[0054] In some embodiments, such as Figure 2 As shown, based on the second target stream data information, the aforementioned method integrates deep learning model selection, transfer learning application, model training strategy, and model compression technology. Through model training and optimization, multiple target models are obtained for predicting carbon emissions from steel production. Specifically, S130 may include the following steps:
[0055] S410. Select a suitable deep learning model architecture to build multiple pre-trained models, wherein each pre-trained model corresponds one-to-one with a stage in the process flow.
[0056] S420. Define input and output interfaces for each pre-trained model to determine compatibility and data flow between the various pre-trained models;
[0057] S430. Transfer each pre-trained model to the carbon emission prediction task at each stage of the process flow, and fine-tune it using the data corresponding to each stage in the second target flow data information.
[0058] S440. The mini-batch gradient descent optimization algorithm is used to train the model and adjust the hyperparameters. Model pruning and quantization techniques are applied to reduce the model size to improve inference speed, resulting in multiple target models.
[0059] The S450 employs an online learning mechanism to update the parameters of the target model in real time based on new data collected by the sensors.
[0060] In the embodiments of this application, it can be understood that this application is based on deep learning algorithms to improve the prediction accuracy of carbon emissions at each stage of steel production and reduce errors caused by data dependence and insufficient model generalization ability.
[0061] It should be noted that this application adopts a modular design for the steel enterprise production process, decomposing the production process into independent processes and customizing pre-trained models for each independent process to enhance the system's generalization ability and adaptability. In the model training strategy, this application uses optimization algorithms such as mini-batch gradient descent to train the model, and applies model compression techniques such as model pruning and quantization to reduce the model size and improve inference speed.
[0062] It is important to emphasize that this application improves the applicability and accuracy of the model in different production environments and processes through modular design and transfer learning techniques, making the model not limited to specific production conditions. By utilizing transfer learning techniques, the model pre-trained on a large-scale general dataset can be transferred to a specific module's carbon emission prediction task, reducing the dependence on large amounts of labeled data.
[0063] Furthermore, this application introduces an online learning mechanism, enabling the model to update parameters in real time based on newly arriving data, using algorithms such as online gradient descent. It also continuously monitors the predictive performance of the monitoring model, such as accuracy and recall, and dynamically adjusts the learning rate and model parameters based on performance feedback. This application supports online learning and model fine-tuning, allowing for rapid adaptation without retraining the entire model when production conditions change or data distribution shifts.
[0064] In one example, the aforementioned pre-trained models include:
[0065] The raw material preparation model is based on a convolutional neural network (CNN) setup. The raw material preparation model is used to process image data, which includes at least one of ore images and coke images.
[0066] The coking model, based on a Long Short-Term Memory (LSTM) network or a Gated Recurrent Unit (GRU) setup, is used to process time-series data and predict temperature and energy consumption changes during the coking process.
[0067] Blast furnace ironmaking model, based on deep belief network (DBN) or variational autoencoder (VAE) settings, is used to capture the complex physicochemical reactions inside the blast furnace.
[0068] The steelmaking model, based on a recurrent neural network (RNN) or Transformer model, is used to process complex time-series data in the steelmaking process.
[0069] The continuous casting model, based on a convolutional LSTM network and combining the spatial feature extraction capability of a convolutional neural network (CNN) with the time series analysis capability of a long short-term memory network (LSTM), is used to predict carbon emissions during the continuous casting process.
[0070] A hot rolling model, based on a one-dimensional convolutional neural network, is used to analyze stress and strain data during the hot rolling process.
[0071] The cold rolling model is a hybrid model combining convolutional neural networks (CNN) and long short-term memory networks (LSTM). The cold rolling model is used to process multidimensional data in the cold rolling process, including pressure, speed, and temperature.
[0072] The post-processing model is a fully connected deep neural network (DNN) used to learn complex nonlinear relationships in the post-processing process.
[0073] An energy management model, an attention-enhanced Transformer model, is used to analyze and predict energy consumption and carbon emissions across modules; and
[0074] A waste treatment and recycling model, based on a graph neural network (GNN) setup, is used to simulate the flow and transformation in waste treatment and recycling networks.
[0075] It should be noted that this application achieves accurate prediction of carbon emissions at each stage of the steel production process through modular design and customized deep learning models, thereby providing steel enterprises with a more effective carbon emission management tool. Addressing the complexity of steel production, this application proposes a modular solution, enabling targeted carbon emission prediction for each production stage, thus improving the system's flexibility and scalability.
[0076] In some embodiments, the aforementioned integration of real-time prediction granularity, user interface design, and report generation technologies provides hourly carbon emission prediction results for intuitive display; that is, the aforementioned S140 may specifically include the following steps:
[0077] S510. Configure the target model to achieve hourly carbon emission prediction;
[0078] S520: Develop an intuitive user interface to display hourly carbon emission forecasts in real time and present them visually through charts and dashboards;
[0079] S530 automatically generates hourly carbon emission reports, which include forecast results, actual emissions, energy consumption data, and support for corporate decision-making and environmental responsibility reporting.
[0080] In some embodiments, the aforementioned target model is periodically updated based on acquired user feedback data and actual carbon emission data to improve the accuracy of carbon emission prediction. Specifically, S150 may include the following steps:
[0081] S610. Collect user feedback data and actual carbon emission data, and establish a feedback mechanism to evaluate the performance of the target model.
[0082] S620. Based on feedback and performance evaluation results, the target model is updated regularly to adapt to changes in the production process.
[0083] In some embodiments, this application provides an artificial intelligence-based carbon emission prediction system 700, such as... Figure 3 As shown, the carbon emission prediction system 700 may include the following modules:
[0084] The data acquisition and preprocessing module 710 is used to deploy multiple sensors based on the steel production process, and integrate sensor data acquisition, data synchronization, data cleaning and feature extraction methods to obtain the first target stream data information.
[0085] The real-time data processing module 720 is used to build a real-time data processing platform and to format data based on the first target stream data information to obtain the second target stream data information.
[0086] The model training and optimization module 730 is used to integrate deep learning model selection, transfer learning application, model training strategy and model compression technology based on the second target stream data information, and obtain multiple target models through model training and optimization to predict carbon emissions from steel production.
[0087] The prediction output module 740 is used to combine real-time prediction granularity, user interface design and report generation technology to provide hourly carbon emission prediction results for intuitive display.
[0088] The iterative improvement module 750 is used to periodically update the target model based on acquired user feedback data and actual carbon emission data to improve the accuracy of carbon emission prediction.
[0089] According to embodiments of this application, any multiple modules among the data acquisition and preprocessing module 710, real-time data processing module 720, model training and optimization module 730, prediction output module 740, and iterative improvement module 750 can be combined into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module.
[0090] In some embodiments, the data acquisition and preprocessing module 710 may specifically be used for:
[0091] Multiple sensors are deployed based on key equipment and processes in steel production; the processes include: raw material preparation stage, coking stage, blast furnace ironmaking stage, steelmaking stage, continuous casting stage, hot rolling stage, cold rolling stage, post-processing stage, energy management stage, and waste treatment and recycling stage;
[0092] Data is collected in real time by multiple sensors, and the data from different sensors and devices are time-stamped and synchronized to obtain initial flow data information, which includes temperature data, pressure data, flow rate data and energy consumption data.
[0093] An unsupervised learning algorithm is introduced, and an automated script is used to identify and correct missing and outlier values in the initial streaming data, and to remove noise.
[0094] Key features related to carbon emissions are extracted from the data to obtain the first target flow data information.
[0095] In some embodiments, the real-time data processing module 720 may specifically be used for:
[0096] Build a real-time data processing platform using a pre-defined stream processing framework; the stream processing framework includes at least one of Apache Kafka and Apache Storm.
[0097] The sliding window time window technique is applied to process the first target stream data information, and normalization and encoding conversion are performed to complete the data format conversion to obtain the second target stream data information.
[0098] In some embodiments, the model training and optimization module 730 may specifically be used for:
[0099] Choose a suitable deep learning model architecture to build multiple pre-trained models, where each pre-trained model corresponds one-to-one with a stage in the process flow.
[0100] Define input and output interfaces for each pre-trained model to determine compatibility and data flow between the various pre-trained models;
[0101] Each pre-trained model is transferred to the carbon emission prediction task at each stage of the process flow and fine-tuned using the data corresponding to each stage in the second target flow data information.
[0102] The mini-batch gradient descent optimization algorithm was used to train the model and adjust the hyperparameters. Model pruning and quantization techniques were applied to reduce the model size to improve inference speed, resulting in multiple target models.
[0103] An online learning mechanism is adopted to update the parameters of the target model in real time based on new data collected by the sensors.
[0104] In some embodiments, the prediction output module 740 may specifically be used for:
[0105] Configure the target model to achieve hourly carbon emission predictions;
[0106] Develop an intuitive user interface to display hourly carbon emission forecasts in real time and present them visually through charts and dashboards;
[0107] It automatically generates hourly carbon emission reports, which include forecast results, actual emissions, energy consumption data, and reports to support corporate decision-making and environmental responsibility.
[0108] In some embodiments, the iterative improvement module 750 can specifically be used for:
[0109] Collect user feedback data and actual carbon emission data, and establish a feedback mechanism to evaluate the performance of the target model;
[0110] Based on feedback and performance evaluation results, the target model is updated regularly to adapt to changes in the production process.
[0111] Figure 3 Each module in the system shown has the function of implementing each step in the aforementioned artificial intelligence-based carbon emission prediction method and can achieve its corresponding technical effect. For the sake of brevity, it will not be described in detail here.
[0112] In some embodiments, this application provides an electronic device, the structural schematic of which is shown below. Figure 4 As shown.
[0113] The electronic device may include a processor 810 and a memory 820 storing computer program instructions.
[0114] Specifically, the processor 810 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.
[0115] Memory 820 may include mass storage for data or instructions. For example, and not limitingly, memory 820 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 820 may include removable or non-removable (or fixed) media. Where appropriate, memory 820 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 820 is non-volatile solid-state memory.
[0116] Memory 820 may include read-only memory (ROM), random access memory (RAM), disk storage media device, optical storage media device, flash memory device, electrical, optical, or other physical / tangible memory storage device. Therefore, typically, memory 820 includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it can perform the operations described in any of the AI-based carbon emission prediction methods in the above embodiments.
[0117] The processor 810 reads and executes computer program instructions stored in the memory 820 to implement any of the artificial intelligence-based carbon emission prediction methods in the above embodiments.
[0118] In one example, the electronic device may also include a communication interface 830 and a bus 800. For example, Figure 4 As shown, the processor 810, memory 820, and communication interface 830 are connected via bus 800 and communicate with each other.
[0119] The communication interface 830 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.
[0120] Bus 800 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 800 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.
[0121] Furthermore, in conjunction with the AI-based carbon emission prediction method in the above embodiments, this application embodiment can provide a computer storage medium for implementation. This computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the AI-based carbon emission prediction methods in the above embodiments.
[0122] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.
[0123] The functional blocks shown in the above block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.
[0124] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.
[0125] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.
[0126] In summary, compared with the prior art, this application has the following beneficial effects:
[0127] 1. This application integrates deep learning model selection, transfer learning application, model training strategy and model compression technology to improve the prediction accuracy of carbon emissions at each stage of steel production, reduce errors caused by data dependence and insufficient model generalization ability, and improve the applicability and accuracy of the model in different production environments and processes through modular design and transfer learning technology, so that the model is not limited to specific production conditions.
[0128] 2. This application combines real-time prediction granularity, user interface design, and report generation technology, which can intuitively display carbon emission prediction results; it can also regularly update the target model based on user feedback data and actual carbon emission data, support online learning, and enable the model to be dynamically updated based on the latest data.
[0129] 3. This application can process data streams in real time and quickly output prediction results; by utilizing transfer learning and unsupervised learning techniques, it reduces the need for large amounts of labeled data. This application employs deep learning technology to customize models for each production process, enabling it to more accurately capture complex patterns of carbon emissions and improve prediction accuracy.
[0130] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A carbon emission prediction method based on artificial intelligence, characterized in that, include: Based on the deployment of multiple sensors in the steel production process, the sensor data acquisition, data synchronization, data cleaning and feature extraction methods are integrated to obtain the first target stream data information; A real-time data processing platform is built, and data is formatted based on the first target stream data information to obtain the second target stream data information; Based on the second target stream data information, deep learning model selection, transfer learning application, model training strategy and model compression technology are integrated to obtain multiple target models through model training and optimization for carbon emission prediction of steel production. By combining real-time forecast granularity, user interface design, and report generation technology, hourly carbon emission forecasts are provided for intuitive visualization. The target model is updated regularly based on user feedback data and actual carbon emission data to improve the accuracy of carbon emission prediction. Based on the second target stream data information, the process integrates deep learning model selection, transfer learning application, model training strategies, and model compression techniques to obtain multiple target models through model training and optimization for predicting carbon emissions from steel production, including: Select a suitable deep learning model architecture to build multiple pre-trained models, wherein the multiple pre-trained models correspond one-to-one with each stage in the process flow; Define input / output interfaces for each pre-trained model to determine compatibility and data flow between the pre-trained models; Each of the pre-trained models is transferred to the carbon emission prediction task at each stage of the process flow, and fine-tuned using the data corresponding to each stage in the second target flow data information; The mini-batch gradient descent optimization algorithm was used to train the model and adjust the hyperparameters. Model pruning and quantization techniques were applied to reduce the model size to improve inference speed, resulting in multiple target models. An online learning mechanism is adopted to update the parameters of the target model in real time based on new data collected by the sensors; The multiple pre-trained models include: A raw material preparation model, based on a convolutional neural network (CNN), is used to process image data, which includes at least one of ore images and coke images. A coking model, based on a Long Short-Term Memory (LSTM) network or a Gated Recurrent Unit (GRU) setup, is used to process time-series data and predict temperature and energy consumption changes during the coking process. A blast furnace ironmaking model, based on a deep belief network (DBN) or variational autoencoder (VAE) setup, is used to capture the complex physicochemical reactions inside the blast furnace. The steelmaking model, based on a recurrent neural network (RNN) or a Transformer model, is used to process complex time-series data during the steelmaking process. A continuous casting model, based on a convolutional LSTM network and combining the spatial feature extraction capability of a convolutional neural network (CNN) with the time series analysis capability of a long short-term memory network (LSTM), is used to predict carbon emissions during the continuous casting process. A hot rolling model, based on a one-dimensional convolutional neural network, is used to analyze stress and strain data during the hot rolling process. The cold rolling model is a hybrid model combining a convolutional neural network (CNN) and a long short-term memory network (LSTM). The cold rolling model is used to process multidimensional data during the cold rolling process, including pressure, speed, and temperature. The post-processing model is a fully connected deep neural network (DNN) used to learn complex nonlinear relationships in the post-processing process. An energy management model, an attention-enhanced Transformer model, is used to analyze and predict energy consumption and carbon emissions across modules; and A waste treatment and recycling model, based on a graph neural network (GNN) setup, is used to simulate the flow and transformation in waste treatment and recycling networks.
2. The artificial intelligence-based carbon emission prediction method as described in claim 1, characterized in that, The method involves deploying multiple sensors based on the steel production process, integrating sensor data acquisition, data synchronization, data cleaning, and feature extraction methods to obtain the first target stream data information, including: Multiple sensors are deployed based on key equipment and processes in steel production; the processes include: raw material preparation stage, coking stage, blast furnace ironmaking stage, steelmaking stage, continuous casting stage, hot rolling stage, cold rolling stage, post-processing stage, energy management stage, and waste treatment and recycling stage; Data is collected in real time by the multiple sensors, and the data from different sensors and devices are time-stamped and synchronized to obtain initial flow data information, which includes temperature data, pressure data, flow rate data and energy consumption data. An unsupervised learning algorithm is introduced, and an automated script is used to identify and correct missing and outlier values in the initial streaming data, and to remove noise. Key features related to carbon emissions are extracted from the data to obtain the first target flow data information.
3. The artificial intelligence-based carbon emission prediction method as described in claim 2, characterized in that, The process of building a real-time data processing platform and formatting data based on the first target stream data information to obtain the second target stream data information includes: A real-time data processing platform is built using a pre-defined stream processing framework; the stream processing framework includes at least one of Apache Kafka and Apache Storm. The sliding window time window technique is applied to process the first target stream data information, and normalization and encoding conversion are performed to complete the data format conversion to obtain the second target stream data information.
4. The artificial intelligence-based carbon emission prediction method as described in claim 1, characterized in that, The combined real-time forecast granularity, user interface design, and report generation technology provide hourly carbon emission forecast results for intuitive display, including: The target model is configured to predict carbon emissions every hour; Develop an intuitive user interface to display hourly carbon emission forecasts in real time and present them visually through charts and dashboards; Automatically generate hourly carbon emission reports, which include forecast results, actual emissions, energy consumption data, and reports supporting corporate decision-making and environmental responsibility.
5. The artificial intelligence-based carbon emission prediction method as described in claim 1, characterized in that, The step of periodically updating the target model based on acquired user feedback data and actual carbon emission data to improve the accuracy of carbon emission prediction includes: Collect user feedback data and actual carbon emission data, and establish a feedback mechanism to evaluate the performance of the target model. Based on feedback and performance evaluation results, the target model is updated regularly to adapt to changes in the production process.
6. A carbon emission prediction system based on artificial intelligence, characterized in that, include: The data acquisition and preprocessing module is used to deploy multiple sensors based on the steel production process, and integrates sensor data acquisition, data synchronization, data cleaning and feature extraction methods to obtain the first target stream data information. The real-time data processing module is used to build a real-time data processing platform and to format the data based on the first target stream data information to obtain the second target stream data information. The model training and optimization module is used to integrate deep learning model selection, transfer learning application, model training strategy and model compression technology based on the second target stream data information, and obtain multiple target models through model training and optimization to predict carbon emissions from steel production. The forecast output module combines real-time forecast granularity, user interface design, and report generation technology to provide hourly carbon emission forecasts for intuitive visualization. An iterative improvement module is used to periodically update the target model based on the acquired user feedback data and actual carbon emission data in order to improve the accuracy of carbon emission prediction. Based on the second target stream data information, the process integrates deep learning model selection, transfer learning application, model training strategies, and model compression techniques to obtain multiple target models through model training and optimization for predicting carbon emissions from steel production, including: Select a suitable deep learning model architecture to build multiple pre-trained models, wherein the multiple pre-trained models correspond one-to-one with each stage in the process flow; Define input / output interfaces for each pre-trained model to determine compatibility and data flow between the pre-trained models; Each of the pre-trained models is transferred to the carbon emission prediction task at each stage of the process flow, and fine-tuned using the data corresponding to each stage in the second target flow data information; The mini-batch gradient descent optimization algorithm was used to train the model and adjust the hyperparameters. Model pruning and quantization techniques were applied to reduce the model size to improve inference speed, resulting in multiple target models. An online learning mechanism is adopted to update the parameters of the target model in real time based on new data collected by the sensors; The multiple pre-trained models include: A raw material preparation model, based on a convolutional neural network (CNN), is used to process image data, which includes at least one of ore images and coke images. A coking model, based on a Long Short-Term Memory (LSTM) network or a Gated Recurrent Unit (GRU) setup, is used to process time-series data and predict temperature and energy consumption changes during the coking process. A blast furnace ironmaking model, based on a deep belief network (DBN) or variational autoencoder (VAE) setup, is used to capture the complex physicochemical reactions inside the blast furnace. The steelmaking model, based on a recurrent neural network (RNN) or a Transformer model, is used to process complex time-series data during the steelmaking process. A continuous casting model, based on a convolutional LSTM network and combining the spatial feature extraction capability of a convolutional neural network (CNN) with the time series analysis capability of a long short-term memory network (LSTM), is used to predict carbon emissions during the continuous casting process. A hot rolling model, based on a one-dimensional convolutional neural network, is used to analyze stress and strain data during the hot rolling process. The cold rolling model is a hybrid model combining a convolutional neural network (CNN) and a long short-term memory network (LSTM). The cold rolling model is used to process multidimensional data during the cold rolling process, including pressure, speed, and temperature. The post-processing model is a fully connected deep neural network (DNN) used to learn complex nonlinear relationships in the post-processing process. An energy management model, an attention-enhanced Transformer model, is used to analyze and predict energy consumption and carbon emissions across modules; and A waste treatment and recycling model, based on a graph neural network (GNN) setup, is used to simulate the flow and transformation in waste treatment and recycling networks.
7. An electronic device, characterized in that, include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the artificial intelligence-based carbon emission prediction method as described in any one of claims 1 to 4.
8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program or instructions that, when executed by a processor, implement the artificial intelligence-based carbon emission prediction method as described in any one of claims 1 to 4.