Blast furnace carbon emission anomaly prediction method based on digital twinning

Abstract

The invention provides a blast furnace carbon emission anomaly prediction method based on digital twinning, relates to the technical field of blast furnace carbon emission, and provides a blast furnace top abnormal carbon emission detection method based on deep learning and mechanism fusion. Firstly, key features of carbon traces in the blast furnace production process are collected, and gas components at the top end of a blast furnace and peripheral production data are collected through a sensor. Feature data which cannot be acquired by some sensors in real time are iteratively solved through a Navier-Stokes NS equation. Then, cross-correlation analysis is carried out to find the hysteresis relation of carbon traces in production, and a carbon emission hysteresis relation model is established; abnormal operating conditions are detected by a residual neural network. And finally, testing and evaluating the performance of the model by using actual production data. The result shows that the provided method can accurately detect the abnormal carbon emission at the top end of the blast furnace, early warning can be given out in time, and environmental pollution and energy waste caused by the abnormal carbon emission are effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of blast furnace carbon emission, and in particular to a method for predicting blast furnace carbon emission anomalies based on digital twins. Background Art

[0002] Carbon emissions from the blast furnace ironmaking process are an important source of greenhouse gas emissions in steel production. Effectively detecting and controlling abnormal carbon emissions from the top of the blast furnace is of great significance for reducing carbon emissions, improving energy efficiency, and achieving green steel production. Current blast furnace gas data-driven methods have high data dependence and low interpretability.

[0003] As the core equipment in the steel production process, the blast furnace is responsible for the key task of converting iron ore into molten iron. The blast furnace reactor is large in scale, and the smelting process is carried out under high temperature and high pressure conditions. This is a core challenge in steel smelting technology. The stable operation of the blast furnace is not only crucial to the continuity and efficiency of steel production, but also an important indicator of the technical level and competitiveness of steel enterprises. In this case, the importance of blast furnace abnormality prediction becomes particularly obvious. As a key technical means, abnormality prediction can timely identify and warn potential abnormal hazards through real-time monitoring and comprehensive analysis of the blast furnace operation status, thereby providing a scientific basis for production management and maintenance decisions. This not only helps to avoid production interruptions and safety accidents, reduce economic losses and casualties, but also optimizes blast furnace operating parameters, improves production efficiency and product quality, and promotes technological progress and industrial upgrading in the steel industry. Therefore, the in-depth research and application of blast furnace abnormality prediction technology has great production value and practical significance for ensuring the stability, safety and efficiency of steel production.

[0004] During the blast furnace ironmaking process, various types of faults may occur. According to the location and nature of the fault, it can be divided into abnormal furnace condition faults, equipment faults and raw material quality faults. Among them, abnormal furnace condition faults pose the greatest threat to production. This abnormality is mainly manifested in uneven airflow distribution in the blast furnace, large temperature fluctuations and poor charge descent. Specifically, it includes suspended materials, material collapse, pipeline accumulation, and furnace accumulation. Due to the complexity of multiphase fluid mechanics in the blast furnace and the uncertainty of internal chemical reactions, it is difficult to accurately and quantitatively analyze physical variables such as velocity and temperature of components and materials at various positions in the blast furnace. Therefore, it is difficult to quickly establish an accurate mechanism model to describe the production process of blast furnace ironmaking. In blast furnace production, when abnormal conditions occur, it will cause serious fluctuations in the operation of the blast furnace, which will seriously affect the blast furnace output, molten iron quality and production energy consumption.

[0005] After years of unremitting efforts by researchers, many methods for blast furnace anomaly detection have been developed. Early research methods rely more on expert experience and sensors, such as the analytic hierarchy process (AHP), the top-ranking preference technique based on ideal solution similarity (TOPSIS), and Bayesian networks to establish evaluation models and obtain better evaluation results. However, relying on expert experience will bring high economic costs and have certain limitations. With the improvement of computing power, data-driven models have become an important method for blast furnace anomaly detection. For example, support vector machine (SVM), kernel correlation filter (KCF), and other machine learning methods. Zhou et al. proposed a novel integrated PCA-ICA method to monitor and diagnose blast furnace anomalies. Despite extensive research, there are still some unresolved issues in the establishment of blast furnace ironmaking process monitoring models. Most of the above models have relatively simple structures and work well when processing simple data structures. When the data structure is complex, shallow models are insufficient in capturing complex nonlinear relationships between data. Wu et al. used deep learning models to predict blast furnace anomalies and proposed an anomaly prediction method based on multi-channel dynamic GCN (MDGCN). Assign weights to different nodes in the graph to extract more useful information about process dynamics for fault diagnosis. This method has attracted widespread attention and has been applied in many fields. Feature extraction plays a vital role in deep learning models. Raw data often contains a lot of redundant information. Directly processing these high-dimensional data is not only computationally intensive, but may also lead to model overfitting. Feature extraction extracts key information from the data and converts it into low-dimensional but more representative feature vectors, thereby reducing computational costs and improving the generalization ability of the model. Efficient feature extraction methods can improve the accuracy of anomaly detection. Lu Siwei et al. proposed a local dynamic generalized kernel stationary subspace analysis method (local-dbkssa), which has higher robustness. Feature extraction can also utilize encoders, convolutional neural networks (CNNs), recurrent neural networks and other technologies. Huang Hang et al. proposed an anomaly prediction and recognition method based on a multidimensional gated recurrent unit (GRU) network, which has a simpler structure and fewer parameters, but still provides good prediction results. Although these methods show high accuracy in detecting abnormalities in blast furnace states, relying solely on data-driven models for fault identification has disadvantages, such as poor generalization ability and long training time, which cannot meet the speed requirements of blast furnace production. From the perspective of blast furnace carbon emissions, the main sources of carbon dioxide gas emissions are coke and sinter. Coke and sinter enter the blast furnace through the distribution bell at the top, and after complex chemical reactions, they are converted into blast furnace gas and then discharged. This reaction process takes a long time. When using data-driven models for abnormality prediction, the significant lag in the production process and the impact of internal reactions on abnormality prediction are often ignored. Relying solely on the nonlinear mapping ability of deep learning may slightly reduce accuracy.At present, fluid mechanics is developing rapidly. By adopting computational fluid dynamics technology, the velocity field and temperature field in the blast furnace can be quantitatively analyzed to extract additional features, which is helpful for the analysis of blast furnace anomalies.

[0006] Digital twin technology is a key advancement in digital transformation and intelligent upgrading. Driven by data and models, digital twin technology simulates the dynamic response of physical entities to internal and external mechanisms, and is able to perform monitoring, simulation, correlation, prediction, optimization and other operations. As a core component of steelmaking, the chemical reactions inside the blast furnace are both complex and critical. Among these reactions, carbon dissolution reactions, direct reduction reactions, indirect reduction reactions, and internal reduction reactions are crucial. These reactions not only affect the quality and output of blast furnace molten iron, but also reveal the reaction state inside the blast furnace through the composition of blast furnace gas discharged from the top of the blast furnace.

[0007] The carbon dioxide content in blast furnace gas is a key indicator of the reactor condition, which is affected by many factors, including raw material quality, operating parameter adjustments, and changes in equipment status. Any abnormal fluctuations in carbon monoxide content will have an adverse effect on the operating efficiency of the blast furnace and may endanger the normal operation of the blast furnace and the normal operation of subsequent process equipment, thereby affecting the economy and operational stability of the entire steel production line. Given the importance of this issue, it is crucial to diagnose abnormal carbon monoxide content in blast furnace gas in a timely and accurate manner and take appropriate measures to adjust and control it. This process not only ensures the smooth operation of the blast furnace, but also significantly improves production efficiency, reduces energy consumption, and reduces the emission of harmful gases.

[0008] In the blast furnace production process, the carbon sources mainly include ore and coke added at the top, and coal powder injected through the nozzle at the bottom. These carbon sources experience high temperature and pressure in the furnace, and undergo a series of reduction reactions with iron ore, eventually producing molten iron and coal gas. However, there is a significant delay from the furnace top to the completion of the reaction and the discharge of blast furnace gas. This delay poses a major challenge to analyzing the reaction status in the furnace. By detecting and analyzing the proportion of each gas component in the blast furnace gas, it is possible to more accurately assess whether the reduction reaction in the blast furnace is sufficient and monitor changes in key parameters such as temperature and pressure. Summary of the invention

[0009] In view of the shortcomings of the prior art, the present invention provides a method for predicting abnormal carbon emissions from blast furnaces based on digital twins; blast furnace operating parameters, such as ore addition and coke addition rate, are used to predict carbon emissions from the top of the blast furnace. By precisely controlling these parameters and combining the expertise of on-site staff, we can achieve real-time monitoring and dynamic adjustment of the blast furnace operating status. Once any abnormal reaction or signs of failure in the blast furnace are detected, appropriate intervention measures can be quickly taken to prevent failures and ensure stable operation and efficient production of the blast furnace.

[0010] A method for predicting abnormal carbon emissions from a blast furnace based on digital twins includes the following steps:

[0011] Step 1: Construct a digital twin mechanism model to simulate and monitor the changes in velocity field, temperature field and pressure field in real time; specifically use digital twin technology to solve the Navier-Stokes NS equations of two-dimensional incompressible fluid; determine the velocity field, temperature field and pressure field by solving the Navier-Stokes NS equations and heat conduction differential equations;

[0012] Step 1.1: Generate the blast furnace modeling area, specifically using elliptic partial differential equations to generate the grid of the modeling area;

[0013] The arrangement of the grids is consistent with the longitudinal direction of the blast furnace;

[0014] Step 1.2: Assuming that the fluid in the blast furnace is incompressible, simplify the NS equations into a two-dimensional form; obtain the velocity field and pressure field inside the blast furnace by iteratively solving the discretized two-dimensional NS equations; at the same time, use the temperature values ​​obtained by the infrared thermocouples in the blast furnace tuyere whirlpool zone to solve the thermal differential equation to determine the temperature field; the temperature field, pressure field and velocity field constitute the mechanism model.

[0015] Step 1.3: Build a digital twin data-driven model;

[0016] Step 1.3.1: Take edge sampling points from the mechanism model and fuse them with real-time sensors to form input data for the data-driven model;

[0017] Step 1.3.2: Use cross-correlation analysis to determine the carbon hysteresis time lag value under the current operating conditions;

[0018] The cross-correlation analysis is specifically as follows:

[0019] Through cross-correlation analysis, the lag relationship between the blast furnace top distribution data and the blast furnace top gas composition data is determined, and the formula is as follows:

[0020]

[0021] Where x(i) is the carbon input, y(i) is the carbon output, N is the time span of the data, and R xy (k) is the result of cross-correlation analysis;

[0022] The carbon-related data were converted from the time domain to the frequency domain and then back to the time domain using fast Fourier transform;

[0023] R xy(k) =ifft(fft(x).*conj(fft(y))) (2)

[0024] In the formula, fft is the fast Fourier transform, ifft is the inverse fast Fourier transform, and conj is the conjugate.

[0025] Through cross-correlation analysis, the carbon lag time of the blast furnace under the current working condition is obtained; the furnace top pressure, temperature and edge detection data are used as the input data of the cross-correlation, and the historical data of the furnace top gas composition is used as the target data, that is, x and y in the cross-correlation analysis; based on the lag time under the current working condition obtained through the cross-correlation analysis, the input data and output data are matched, the PCA dimensionality reduction method is used, and the order after dimensionality reduction is used as the input of the data-driven model;

[0026] Step 1.3.3: Build a data-driven model, namely a residual neural network;

[0027] After obtaining the carbon lag time under the current working conditions, deep learning is used to predict carbon emission anomalies; specifically, an optimized convolutional neural network (CNN) is used to diagnose carbon emission anomalies under various working conditions; the optimized convolutional neural network (CNN) takes feature input as the starting point, first through two sets of parallel convolution operations, each set contains 32 convolution kernels, and then accesses the ReLU activation function. The feature map enters the next stage after addition fusion, and then the feature map is reduced in size through the pooling layer, and the size adjustment operation is completed by the method of filling 0 to complete the preprocessing; in the next stage, the number of convolutional layers is expanded to 64, and the activation function is continued to be used for processing. After fusion, the features enter the final fully connected layer to complete feature extraction and mapping; the entire network realizes layer-by-layer feature extraction and information aggregation through multi-level convolution, pooling and fully connected operations, and the output vector y is calculated as follows:

[0028] y=H(x)=F(x,{W i})+x (3)

[0029] Where F(x, {Wi}) represents the learned residual mapping, and x is the input vector;

[0030] Step 1.4: Preprocess the data of the data-driven model to obtain the digital twin mechanism model; specifically normalize the data; apply the normalization formula to scale the data to the range of [-1,1]:

[0031] x norm =2*(xx min ) / (x max -x min )-1 (4)

[0032] In the formula, x represents the input data that needs to be preprocessed, x norm represents the normalized result, and x min and x maxRespectively represent the minimum and maximum values ​​of a variable set;

[0033] Step 2: After obtaining the sensor data that can be obtained in real time at the current moment, construct the velocity field, pressure field, and temperature field models, and fuse them with the sensor data as the input data of the data-driven model, and obtain the lag time in short-term working conditions through cross-correlation analysis. After inputting the data into the data-driven model, the amount of carbon emissions from the furnace top is obtained, and combined with the lag time, the time of the predicted carbon emissions is determined.

[0034] The beneficial effects of adopting the above technical solution are:

[0035] The present invention provides a method for predicting abnormal carbon emissions from blast furnaces based on digital twins. In blast furnace ironmaking, the detection of abnormal conditions is crucial to reducing accident risks, improving economic benefits, and achieving safe and sustainable production. The present invention proposes a digital twin mechanism model, that is, variables that cannot be obtained in real time, and then combines these variables with real-time sensor data. Cross-correlation analysis is used to determine the time lag of carbon emissions under various operating conditions. A deep neural network is then constructed using residual mapping to detect abnormal carbon emissions from the furnace top. The results of verification using production data from a steel plant in this embodiment show that this method can accurately detect insufficient reduction conditions in the ironmaking process. In order to improve the detection accuracy, we will further improve the mechanism part, establish the internal composition field of the blast furnace, and improve the accuracy of abnormal prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a thermal image of the top of the blast furnace of the present invention;

[0037] Figure 2 This is an overall flow chart of the method for predicting abnormal carbon emissions from a blast furnace according to the present invention;

[0038] Figure 3 A flow chart of a method for generating a modeling region of the present invention;

[0039] Figure 4 This is a diagram showing the infrared temperature measurement results of the tuyere raceway area of ​​the present invention;

[0040] Figure 5 is a hysteresis relationship curve diagram of carbon emission of the present invention;

[0041] Figure 6 This is a diagram of the residual neural network architecture of the present invention;

[0042] Figure 7 This is a schematic diagram of the jump connection of the present invention;

[0043] Figure 8 A schematic diagram showing the comparison of the performance of the algorithms used in the prediction of the present invention;

[0044] Fig. 9 A scatter plot of the regression relationship of the present invention;

[0045] Fig.10 A histogram showing the distribution of errors in the data for the present invention;

[0046] Fig.11 A confusion matrix diagram for anomaly detection of the present invention;

[0047] Fig.12 This is a cumulative error curve diagram of the present invention. DETAILED DESCRIPTION

[0048] The specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0049] A method for predicting abnormal carbon emissions from blast furnaces based on digital twins, such as Figure 2 As shown, the following steps are included:

[0050] Step 1: Build a digital twin mechanism model to simulate and monitor the changes in velocity field, temperature field and pressure field in real time; specifically use digital twin technology to solve the Navier-Stokes NS equations of two-dimensional incompressible fluid; digital twin technology not only improves the accuracy of simulation, but also helps us better understand and predict the operation of the system. By solving the Navier-Stokes NS equations and heat conduction differential equations, the velocity field, temperature field and pressure field are determined;

[0051] Step 1.1: Generate the blast furnace modeling area, specifically using elliptic partial differential equations to generate the grid of the modeling area;

[0052] In this embodiment, the blast furnace is 27.1 meters high and the radius of the maximum cross section is 6.6 meters. Figure 1 As shown, Figure 3 The iterative process of meshing the modeling area is shown. This example chooses to divide the modeling area radially into 60 steps and vertically into 271 steps, creating a total of 16,260 meshes. The average step length of all meshes is about 100 mm, which can accommodate multiple particles, reduce calculation delays, and ensure the accuracy of the model. The arrangement of the grid is consistent with the longitudinal direction of the blast furnace; according to experience, the gas in the blast furnace flows from the bottom to the top, and its flow trajectory is consistent with the grid arrangement, which helps the model converge.

[0053] The NS equations are a basic mathematical tool for describing fluid motion. These equations describe the motion of fluids and are crucial in various fields of engineering and science. They reveal the essence of fluid motion: the fluid velocity changes under the influence of external forces (such as pressure gradients, volume forces) and internal forces (such as viscous stress). This change is not only related to the current velocity and pressure distribution, but also to the physical properties of the fluid (such as density, viscosity) and the flow history (such as convection terms). Therefore, the NS equations provide a powerful tool for studying the laws of fluid flow. The three conservation laws contained in the NS equations are: Figure 3 shown.

[0054] Step 1.2: Currently, there is no analytical solution to the three-dimensional compressible fluid NS equation. The process of iterative calculation using discrete methods to find approximate solutions is extremely slow and cannot meet the requirements of industrial computing speed. Assuming that the fluid in the blast furnace is incompressible, the NS equation is simplified to a two-dimensional form; the velocity field and pressure field inside the blast furnace are obtained by iteratively solving the discretized two-dimensional NS equation; at the same time, the temperature value obtained by the infrared thermocouple in the blast furnace tuyere whirlpool zone is used to solve the thermal differential equation to determine the temperature field; at this stage, the temperature field, pressure field, and velocity field that cannot be measured in real time constitute the mechanism model.

[0055] Step 1.3: Build a digital twin data-driven model;

[0056] Step 1.3.1: Take edge sampling points from the mechanism model and fuse them with real-time sensors to form the input data of the data-driven model; including sensors such as furnace top pressure sensor and infrared temperature measuring instrument in the tuyere whirlpool area; the infrared temperature measurement results of the tuyere whirlpool area are shown in the figure below. Figure 4 As shown;

[0057] Step 1.3.2: In order to solve the carbon hysteresis problem, cross-correlation analysis is used to determine the carbon hysteresis time lag value under the current working conditions;

[0058] The cross-correlation analysis is specifically as follows:

[0059] Because there is a long lag between the addition and emission of carbon in blast furnace production, there will be a time lag in the sampling and analysis of blast furnace gas. Figure 5 As shown, it is crucial to determine the time required for carbon elements to be added to emissions. From a steel plant, blast furnace top distribution data and blast furnace top gas composition data were obtained. Through cross-correlation analysis, the lag relationship between blast furnace top distribution data and blast furnace top gas composition data was determined, and the formula is as follows:

[0060]

[0061] Where x(i) is the carbon input, y(i) is the carbon output, N is the time span of the data, and R xy(k) is the result of cross-correlation analysis.

[0062] Since all the time difference points need to be nested in two layers of for loops, the computational load is particularly large and the scale is large. Therefore, the carbon-related data is converted from the time domain to the frequency domain using fast Fourier transform, and then converted back to the time domain to simplify the operation.

[0063] R xy(k) =ifft(fft(x).*conj(fft(y))) (2)

[0064] In the formula, fft is the fast Fourier transform, ifft is the inverse fast Fourier transform, and conj is the conjugate.

[0065] Through cross-correlation analysis, the carbon lag time of the blast furnace under the current working condition is obtained; the furnace top pressure, temperature and edge detection data are used as the input data of the cross-correlation, and the historical data of the furnace top gas composition is used as the target data, that is, x and y in the cross-correlation analysis; based on the lag time under the current working condition obtained through cross-correlation analysis, the input data and output data are matched. In order to meet the industry's requirements for speed, reduce training components, and make the model converge faster, the PCA dimensionality reduction method is used, and the order after dimensionality reduction is used as the input of the data-driven model;

[0066] Step 1.3.3: Build a data-driven model, namely a residual neural network, such as Figure 6 , Figure 7 As shown;

[0067] After obtaining the carbon lag time under the current working conditions, deep learning is used to predict carbon emission anomalies; inspired by the residual neural network, in the residual block, the input can be transmitted faster through the cross-layer data line. In addition, through the jump connection, the signal can be transmitted without attenuation during the back propagation process, alleviating the gradient vanishing or gradient exploding problem caused by deepening the number of layers. It can also reduce the occurrence of overfitting in deep networks. Specifically, an optimized convolutional neural network (CNN) is used to diagnose abnormal carbon emissions under various working conditions. The optimized convolutional neural network (CNN) starts with feature input, first passes through two sets of parallel convolution operations, each set contains 32 convolution kernels, and then accesses the ReLU activation function. The feature map enters the next stage after addition fusion. Subsequently, the feature map is reduced in size through the pooling layer, and the size adjustment operation is completed by filling 0 to complete the preprocessing. In the next stage, the number of convolutional layers is expanded to 64, and the activation function is continued to be used for processing. After fusion, the features enter the final fully connected layer to complete feature extraction and mapping. The entire network realizes layer-by-layer feature extraction and information aggregation through multi-level convolution, pooling and fully connected operations, and has efficient feature learning capabilities. The training accuracy of the model with too many convolutional layers is lower than that of the three-layer convolutional model; increasing the number of layers will increase the training error. Compared with using multiple network layers to fit the hidden nonlinear mapping, the network is easier to learn its residual. This premise constitutes the basic concept of the residual block. The basic structure of the residual block is shown in the figure. The model based on the residual learning method reverses this calculation process, and the output vector y is calculated as follows:

[0068] y=H(x)=F(x,{W i})+x (3)

[0069] Where F(x, {W i}) represents the learned residual mapping, x is the input vector; Figure 7 In the two-layer structure shown, the dimensions of x and F must be equal when performing the summation. When the input or output channels change, the resize module size step is adjusted to adjust the dimensions of the data to match their sizes, such as Figure 7 In this example, a uniform kernel size of 3×1 is used, and the number of channels and network structure are shown in Figure 6 shown.

[0070] Step 1.4: Data-driven model data preprocessing; industrial data is often affected by noise and outliers, and data preprocessing is required to mitigate these effects. In order to improve the accuracy of the model and accelerate the convergence speed, the data must be preprocessed. Specifically, the data is normalized; the normalization formula is applied to scale the data to the range of [-1,1]:

[0071] x norm =2*(xx min ) / (xmax -x min )-1 (4)

[0072] In the formula, x represents the input data that needs to be preprocessed, x norm represents the normalized result, and x min and x max Respectively represent the minimum and maximum values ​​of a variable set. In blast furnace production, different variables have different effects on operating conditions. Using all variables may lead to slow model convergence and increased risk of overfitting. Therefore, in this embodiment, principal component analysis PCA is used for feature selection;

[0073] Step 2: Since the input and output of carbon in blast furnace production are lagging, after obtaining the sensor data that can be obtained in real time at the current moment, construct the velocity field, pressure field, and temperature field models, and fuse them with the sensor data as the input data of the data-driven model, and obtain the lag time in short-term working conditions through cross-correlation analysis. After inputting the data into the data-driven model, the amount of carbon emissions from the furnace top is obtained, and combined with the lag time, the time of the predicted carbon emissions is determined.

[0074] In this embodiment, more than one thousand sets of production data were selected from an experimental blast furnace of a steel plant, with a sampling frequency of once per hour. After preprocessing, the data of the first one thousand hours were selected and divided into a training set and a test set according to a ratio of 17:3. The test set includes 6 small anomalies, 4 medium anomalies and 6 large anomalies. In order to speed up the convergence of the model, the network structure only designs two residual mappings. The Xavier initializer generates weights with normal distribution, zero mean and specific variance. The activation function used is ReLU, and the output is generated through a fully connected layer. The gradient descent algorithm used is Adam, with a maximum training round of 500 times, a minimum batch size of 300, and a learning rate adjusted every 300 rounds. The learning rate adjustment coefficient is 0.2.

[0075] To demonstrate the effectiveness of the proposed method, we conducted comparative experiments with GRU, CNN, SVM, random forest, and resnet18. The GRU algorithm performs single-step prediction, while CNN uses two convolutional layers. All methods are implemented in MATLAB 2023a and run on a laboratory computer equipped with an Intel Core i7-13700KF CPU, RTX 4060Ti GPU with a maximum frequency of 5.4 GHz, and 32 GB of memory. The accuracy is measured by averaging the results of five experiments.

[0076] Figure 8,Tables 1 and 2 compare the performance of different algorithms to evaluate their effectiveness in ,the prediction task.,The results show that the time series prediction algorithm performs poorly on chaotic ,industrial production data and can only achieve good results in the prediction of the next ,step.

[0077] But it is still insufficient compared to convolutional neural network algorithms. Signal decomposition algorithms, such as empirical mode decomposition (EMD), show good results, but in actual production scenarios, it is challenging to obtain enough data for such decomposition. The accuracy of convolutional neural networks decreases as the network structure deepens. Fewer convolutional layers can lead to insufficient nonlinear mapping capabilities, resulting in inaccurate fault point regression. Resnet addresses the challenges of training deep neural networks by introducing residual mapping. It exhibits strong nonlinear mapping capabilities and performs well under some stable working conditions. However, short-term faults cannot be accurately predicted when production inputs change significantly. Experiments on existing ResNet-50 and ResNet-18 models show poor performance and long training time. ResNet-50 takes ten times longer to train than other models, and ResNet-18 also suffers from extended training time and poor performance. In the network structure described in this article, increasing or decreasing the number of network layers will lead to a decrease in accuracy. Fig. 9 A scatter plot showing the regression relationship is shown.

[0078] Table 1: Accuracy and calculation of different methods

[0079]

[0080] Table 2: Accuracy of anomaly predictions

[0081]

[0082] It shows the correlation between the predicted variables and the actual values. Fig.10 A histogram of the error distribution is displayed to help analyze the error frequency and its distribution. Fig.11 A confusion matrix for anomaly detection is provided, which clearly shows the classification accuracy and performance of the model on different categories. Fig.12 A cumulative error curve is drawn to show the error accumulation and stability as the amount of data increases.

[0083] From the above analysis, it can be seen that the residual mapping neural network based on cross-correlation analysis is significantly better than other methods. Cross-correlation analysis establishes a connection between carbon input and output, enhances the consistency between model input and output, improves the interpretability of the neural network, and reduces the nonlinearity of input and output data compared with direct input of different working conditions. This makes the network regression more accurate. In addition, the introduction of residual mapping enhances the feature extraction ability of the model, making the model more accurate compared with other algorithms.

[0084] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the embodiments of the present disclosure (but not limited to) to form a technical solution.

Claims

1. A method for predicting abnormal carbon emissions from blast furnaces based on digital twins, characterized in that: The following steps are involved: Step 1: Build a digital twin mechanism model to simulate and monitor the changes in velocity field, temperature field and pressure field in real time; Step 2: After acquiring the real-time sensor data, construct the velocity field, pressure field, and temperature field models and fuse them with the sensor data as the input data of the data-driven model. Then, obtain the lag time in short-term working conditions through cross-correlation analysis. After inputting the data into the data-driven model, obtain the amount of carbon emissions from the furnace top, combine it with the lag time, and determine the time of the predicted carbon emissions, thereby realizing the prediction of abnormal carbon emissions.

2. The method for predicting abnormal carbon emissions from blast furnaces based on digital twins according to claim 1 is characterized in that: The step 1 specifically utilizes digital twin technology to solve the Navier-Stokes NS equations of a two-dimensional incompressible fluid; by solving the Navier-Stokes NS equations and the heat conduction differential equation, the velocity field, temperature field and pressure field are determined.

3. The method for predicting abnormal carbon emissions from blast furnaces based on digital twins according to claim 1 is characterized in that: The step 1 specifically comprises the following steps: Step 1.1: Generate the blast furnace modeling area, specifically using elliptic partial differential equations to generate the grid of the modeling area; The arrangement of the grids is consistent with the longitudinal direction of the blast furnace; Step 1.2: Assuming that the fluid in the blast furnace is incompressible, simplify the NS equation into a two-dimensional form; obtain the velocity field and pressure field inside the blast furnace by iteratively solving the discretized two-dimensional NS equation; at the same time, use the temperature value obtained by the infrared thermocouple in the blast furnace tuyere raceway to solve the thermal differential equation to determine the temperature field; the temperature field, pressure field and velocity field constitute the mechanism model; Step 1.3: Build a digital twin data-driven model; Step 1.4: Preprocess the data driven model to obtain the final digital twin mechanism model.

4. The method for predicting abnormal carbon emissions from blast furnaces based on digital twins according to claim 1 is characterized in that: The step 1.3 comprises the following steps: Step 1.3.1: Take edge sampling points from the mechanism model and fuse them with real-time sensors to form input data for the data-driven model; Step 1.3.2: Use cross-correlation analysis to determine the carbon hysteresis time lag value under the current operating conditions; Step 1.3.3: Build a data-driven model, namely a residual neural network.

5. The method for predicting abnormal carbon emissions from blast furnaces based on digital twins according to claim 4 is characterized in that: The cross-correlation analysis described in step 1.3.2 is as follows: Through cross-correlation analysis, the lag relationship between the blast furnace top distribution data and the blast furnace top gas composition data is determined, and the formula is as follows: Where x(i) is the carbon input, y(i) is the carbon output, N is the time span of the data, and R xy (k) is the result of cross-correlation analysis; The carbon-related data were converted from the time domain to the frequency domain and then back to the time domain using fast Fourier transform; R xy(k) =ifft(fft(x).*conj(fft(y))) (2) In the formula, fft is the fast Fourier transform, ifft is the inverse fast Fourier transform, and conj is the conjugate. Through cross-correlation analysis, the carbon lag time of the blast furnace under the current operating conditions is obtained; the furnace top pressure, temperature and edge detection data are used as the input data of the cross-correlation, and the historical data of the furnace top gas composition is used as the target data, i.e., x and y in the cross-correlation analysis; based on the lag time under the current operating conditions obtained through cross-correlation analysis, the input data and output data are matched, the PCA dimensionality reduction method is adopted, and the order after dimensionality reduction is used as the input of the data-driven model.

6. The method for predicting abnormal carbon emissions from blast furnaces based on digital twins according to claim 4 is characterized in that: Step 1.3.3 is specifically as follows: after obtaining the carbon lag time under the current working conditions, deep learning is used to predict carbon emission anomalies; specifically, an optimized convolutional neural network CNN is used to diagnose carbon emission anomalies under various working conditions; the optimized convolutional neural network CNN takes feature input as the starting point, first through two sets of parallel convolution operations, each set contains 32 convolution kernels, and then accesses the ReLU activation function, and the feature map enters the next stage after addition fusion, and then the feature map is reduced in size through the pooling layer, and the size adjustment operation is completed by filling 0 to complete the preprocessing; in the next stage, the number of convolution layers is expanded to 64, and the activation function is continued to be used for processing, and the features are fused and enter the final fully connected layer to complete feature extraction and mapping; the entire network realizes layer-by-layer feature extraction and information aggregation through multi-level convolution, pooling and fully connected operations, and the output vector y is calculated as follows: y=H(x)=F(x,{W i })+x (3) Where F(x, {W i }) represents the learned residual mapping, and x is the input vector.

7. The method for predicting abnormal carbon emissions from blast furnaces based on digital twins according to claim 3 is characterized in that: The preprocessing described in step 1.4 specifically normalizes the data and applies the normalization formula to scale the data to the range of [-1,1]: x norm =2*(x-x min ) / (x max -x min )-1 (4) In the formula, x represents the input data that needs to be preprocessed, x norm represents the normalized result, and x min and x max Represent the minimum and maximum values ​​of a variable set respectively.

Images