Cement decomposing furnace temperature field simulation method based on digital twinning

Abstract

The invention belongs to the technical field of cement production, and particularly discloses a cement decomposing furnace temperature field simulation method based on digital twinning, and the method comprises the steps: firstly, carrying out the preliminary selection of input parameters according to a cement clinker sintering technology, and then building a time sequence data set from a decomposing furnace; secondly, data preprocessing is carried out through mean filtering and homogenization, and a time sequence data set is divided into a training set, a verification set and a test set; establishing a deep learning model for training to obtain a digital twin virtual model layer, and establishing a decomposition furnace geometric model for training to obtain a prediction model; and carrying out anomaly detection and prediction simulation on the input decomposition furnace variables by using the two trained models. According to the method, the decomposition furnace digital twinborn model is established through the decomposition furnace mechanism model and the data driving model, so that the method has the advantages of prediction capability, operation optimization, decision support and the like, temperature distribution and dynamic change in the decomposition furnace can be reflected in real time, and abnormal conditions in operation can be found.

Description

Technical Field

[0001] The present invention belongs to the technical field of cement production, and in particular, relates to a temperature field simulation method of a cement decomposition furnace based on digital twin. Background Art

[0002] As a basic raw material, the production quality of cement directly determines the safety and stability of subsequent products. In cement production, the cement decomposition furnace mainly completes tasks such as coal powder combustion and raw material decomposition.

[0003] At present, the research on the decomposition furnace technology is relatively mature, but the decomposition of raw materials and the combustion of coal powder in the decomposition furnace are all based on the experience and judgment of the operators on the production site.

[0004] Existing technologies such as the Chinese invention patent application with application number 202411358309.8 discloses a rotary kiln operating condition prediction method using coupled simulation and deep learning neural network. The method simulates the rotary kiln to obtain sample data, combines the POD method with the neural network, and develops an efficient prediction model to achieve rapid prediction of the flow field and temperature field of the rotary kiln, significantly reducing the huge overhead of traditional CFD simulation in computing resources and time. At the same time, it solves the limitations of previous simulation technology in real-time, and adjusts the predicted data through a correction algorithm, thereby improving the practicality and reliability of the model, realizing the deployment of rotary kiln reduction and digital twin models in industrial sites, providing real-time guidance for production processes, and ensuring improved production efficiency and quality.

[0005] Prior art, such as the digital twin system of a cement decomposition furnace based on POD and deep learning disclosed in the Chinese invention patent application with application number 202210256544.9, includes a data acquisition and processing module, a database retrieval module, a model reduction prediction module, a comprehensive display module, and a remote control module connected in sequence. By combining computational fluid dynamics with POD and deep learning, it realizes the rapid calculation of the entire temperature field and flow field in the entire cement decomposition furnace based on a finite number of discrete operating points, and displays it in a variety of ways. The present invention enables cement decomposition furnace production and operation personnel to obtain the real-time changes of the temperature field and flow field in the cement decomposition furnace more intuitively and quickly, so as to timely regulate the coal powder combustion and raw material decomposition in the furnace according to it.

[0006] With regard to the above technical solutions, it is obvious that the current measurement of the temperature field of the cement decomposition furnace still has the following deficiencies: 1. Although it involves calculating the real-time changes of the entire temperature field and flow field in the entire cement decomposition furnace, there is still a certain lack of precision in describing the dynamic characteristics of the decomposition furnace temperature field, and the parameters are not selected from the most basic physical and chemical process of cement clinker burning. There are certain limitations in the understanding and characterization of the internal mechanism of complex physical processes.

[0007] 2. The depth and pertinence of data processing are insufficient, resulting in insufficient support for model training from data quality. At present, the role and relationship of different data sets are not considered for fine-tuning of data division, and the strategies for adjusting model parameters and preventing overfitting are not perfect and detailed enough. The simulation purpose and application scenarios are single, and currently they mainly focus on the prediction and display of flow field and temperature field, which leads to certain deficiencies in the comprehensiveness and effectiveness of using simulation results to guide production and solve practical problems. Summary of the invention

[0008] In view of this, in order to solve the problems raised in the above background technology, a temperature field simulation method of a cement decomposition furnace based on digital twin is proposed.

[0009] The purpose of the present invention can be achieved through the following technical scheme: The present invention provides a temperature field simulation method of a cement decomposition furnace based on digital twin, the method comprising: selecting input parameters according to the cement clinker burning process, and constructing a decomposition furnace time series data set.

[0010] The decomposition furnace time series data set is preprocessed and divided into training set, validation set and test set, which are used as model input.

[0011] A decomposition furnace geometric model is trained to obtain a prediction model, and a deep learning model based on CNN-GRU-Attention is trained to obtain a digital twin virtual model layer.

[0012] Based on the trained digital twin virtual model layer and prediction model, anomaly detection and temperature field simulation of the decomposition furnace are performed, and the predicted value of the decomposition furnace temperature field is output.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) By combining the cement clinker burning process mechanism, data-driven technology and digital twin technology, the present invention realizes high-precision simulation and optimization of the temperature field of the cement decomposition furnace, which can reflect the temperature distribution and dynamic changes inside the decomposition furnace in real time, promptly discover abnormal conditions during operation, optimize operating parameters, and improve production efficiency and clinker quality.

[0014] (2) The present invention starts from the basic physical and chemical processes of cement clinker burning to ensure that the model can accurately reflect the complex physical processes inside the decomposition furnace. At the same time, the input parameters are comprehensively selected and combined with the mechanism model and the data-driven model to improve the understanding and characterization of the complex physical processes inside the decomposition furnace.

[0015] (3) This invention improves the support of data quality for model training by performing deep data preprocessing and refined data set division, ensuring that the strategies for model training, parameter adjustment and prevention of overfitting are more complete. At the same time, it expands the simulation purpose and application scenarios from single prediction simulation to anomaly detection, operation optimization and decision support.

[0016] (4) The present invention achieves high-precision simulation of the temperature field of the decomposition furnace by establishing a digital twin virtual model layer and a decomposition furnace geometric model, combined with deep learning technology. And through digital twin technology, the temperature field distribution and dynamic changes inside the decomposition furnace can be simulated in real time, significantly improving the precision of the description of the dynamic characteristics of the temperature field.

[0017] (5) The present invention utilizes the trained digital twin virtual model and prediction model to not only realize the predictive simulation of the temperature field, but also supports anomaly detection and operation optimization. Through the simulation results, the abnormal conditions in the operation of the decomposition furnace can be discovered in time, and optimization suggestions can be provided to effectively guide actual production. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0019] Figure 1 The present invention is a schematic flow chart of the steps for implementing the method.

[0020] Figure 2 This is a simplified diagram of the process of building a digital twin layer according to the present invention.

[0021] Figure 3 A schematic diagram of the process of constructing a virtual model layer according to the present invention.

[0022] Figure 4 This is an auxiliary diagram of the CNN-GRU-Attention process described in the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] See also Figure 1As shown, the present invention provides a cement decomposition furnace temperature field simulation method based on digital twin, the method comprising: step 1, selecting input parameters according to the cement clinker burning process, and constructing a decomposition furnace time series data set.

[0025] Step 2: Preprocess and divide the decomposition furnace time series data set to obtain the training set, validation set, and test set, which are used as model input.

[0026] The embodiment of the present invention improves the support of data quality for model training by performing deep data preprocessing and refined data set division, ensuring that the strategies for model training, parameter adjustment and prevention of overfitting are more complete. At the same time, it expands the simulation purpose and application scenarios from single prediction simulation to anomaly detection, operation optimization and decision support.

[0027] Step 3: Establish a decomposition furnace geometric model to train and obtain a prediction model, and establish a deep learning model based on CNN-GRU-Attention to train and obtain a digital twin virtual model layer.

[0028] Step 4: Use the trained digital twin virtual model layer and prediction model to perform anomaly detection and temperature field simulation on the decomposition furnace, and output the predicted value of the decomposition furnace temperature field.

[0029] The embodiment of the present invention combines the cement clinker burning process mechanism, data-driven technology and digital twin technology to achieve high-precision simulation and optimization of the temperature field of the cement decomposition furnace. It can reflect the temperature distribution and dynamic changes inside the decomposition furnace in real time, promptly discover abnormal conditions during operation, optimize operating parameters, and improve production efficiency and clinker quality.

[0030] Regarding step 1, it should be noted that the selection of input parameters includes selecting raw material feeding amount, coal injection amount, kiln tail smoke chamber temperature, tertiary air temperature, C4A cone outlet temperature, C4A cone outlet temperature and decomposition furnace outlet temperature offline collection data as input parameters.

[0031] Understandably, temperature sensors are installed at key locations of the decomposition furnace, such as different heights and radial positions inside the furnace body, various inlets and outlets such as raw material discharge port, coal injection port, tertiary air inlet, kiln tail smoke chamber outlet, decomposition furnace outlet, etc. These sensors can collect temperature data in real time. For example, thermocouple sensors can accurately obtain temperature information by measuring the thermoelectric potential of different materials, and thermal resistor sensors can measure temperature by using the change characteristics of resistance with temperature. And the sensors can communicate and interconnect with the DCS system.

[0032] Understandably, in the cement production scenario, the DCS system can comprehensively collect, transmit and store various key information in the production process, such as temperature data, pressure data, flow data, equipment operation data, operation data and environmental data, etc. Among them, temperature data are like tertiary air temperature, kiln tail smoke chamber temperature, decomposition furnace outlet temperature and each preheater cone outlet temperature, such as C4A cone outlet temperature, etc., and pressure data include decomposition furnace internal pressure and each cone pressure, such as C4A cone pressure, C4B cone pressure, etc. Flow data includes raw material feeding amount, coal injection amount, tertiary air volume and other flow data, equipment data such as decomposition furnace diameter, height, etc., motor start and stop status, valve opening, etc., operation data such as primary air volume, secondary air volume, kiln tail flue gas temperature, decomposition furnace temperature setting value, decomposition furnace pressure setting value and residence time, etc., and environmental data such as external ambient temperature and atmospheric pressure.

[0033] Among them, temperature data can reflect the thermal state in the decomposition furnace. If the temperature of the smoke chamber at the tail of the kiln rises abnormally, it may be due to poor ventilation in the kiln or incomplete combustion of coal powder. If the outlet temperature of the decomposition furnace is too low, it may lead to insufficient decomposition of raw materials and affect the quality of clinker. The pressure change can intuitively reflect the ventilation and material flow state of the system. If the pressure inside the decomposition furnace fluctuates greatly, it may indicate that there is a risk of blockage in the system, affecting the continuity of production. Flow data is directly related to the chemical reaction and heat exchange efficiency in the decomposition furnace. Unstable raw material feeding will affect the decomposition effect of raw materials. Inappropriate coal injection will cause temperature fluctuations, which will affect the burning quality of clinker. The operating state of the motor determines whether the equipment is operating normally, and the valve opening affects the flow and direction of materials and gases. These data are crucial to ensure the smooth operation of the entire production process. Therefore, comprehensive collection is carried out through the DCS system to ensure the operability and effectiveness of subsequent simulation. The primary air volume is used to transport pulverized coal and provide the oxygen required for combustion, which affects the combustion efficiency and temperature distribution. The secondary air volume is used to supplement the oxygen required for combustion and further adjust the combustion state in the furnace. The flue gas temperature at the kiln tail refers to the flue gas temperature entering the decomposition furnace from the rotary kiln, which affects the heat input of the decomposition furnace. The decomposition furnace temperature setting value refers to the target temperature of the decomposition furnace. The residence time refers to the residence time of the raw material in the decomposition furnace.

[0034] It should be noted that step 2 includes preprocessing including mean filtering and normalization, and the normalization can be performed by minimum-maximum normalization.

[0035] What needs to be explained about step 3 is that the geometric model of the decomposition furnace is established, including: R1. According to the design drawing of the decomposition furnace of the cement plant, a 3D model of the decomposition furnace is established using three-dimensional modeling software.

[0036] R2. According to the cement burning process, the 3D model of the decomposition furnace is set to 4 raw material feeding inlets, 4 tail coal injection inlets, 2 tertiary air inlets, 1 kiln tail smoke chamber outlet, 1 decomposition furnace outlet and the decomposition furnace inner wall inlet and wall.

[0037] R3. Use hybrid meshing technology to mesh the internal area of ​​the decomposition furnace, the boundary layer and the pre-set analysis area, and perform mesh quality detection by calculating the orthogonal quality index of the mesh; among them, the hybrid mesh includes layered polyhedral mesh, pure polyhedral mesh and hexahedral mesh.

[0038] Furthermore, hexahedral meshes are used to decompose the internal area of ​​the furnace. Hexahedral meshes are less in number at the same size, suitable for simple geometries, and can reduce the amount of calculation. Layered polyhedral meshes are used to decompose the transition area from the outer wall to the interior of the furnace, that is, the boundary layer. Layered polyhedral meshes can better capture the physical phenomena of the boundary layer. Pure polyhedral meshes are used to decompose the mixing area of ​​raw materials, coal powder and wind at the bottom of the furnace. Pure polyhedral meshes are suitable for complex geometries and can improve the accuracy of the simulation.

[0039] The meshing quality is evaluated by calculating the orthogonal quality index of the mesh. The orthogonal quality formula is: ; It is a grid orthogonality quality index, which is used to evaluate the relationship between the shape of the grid unit and its actual volume. The closer the orthogonality quality is to 1, the closer the shape of the grid unit is to orthogonality, and the higher the grid quality is. represents the determinant of the Jacobian matrix, which is usually expressed as , the Jacobian determinant is used to describe the coordinate transformation from the original coordinate system Change to the new coordinate system The proportional relationship of the change of volume microelement, , , They represent the partial derivatives of the coordinates of the grid cells in the reference space, , and are the coordinates of the grid cell in physical space, is the volume of the mesh unit. This formula is mainly used to evaluate the quality of mesh partitioning, especially in finite element analysis (FEA) or computational fluid dynamics (CFD) simulation. By calculating the orthogonal quality index of the mesh, the quality and rationality of the mesh partitioning can be evaluated, thus providing a reliable basis for subsequent numerical calculations. Table 1 shows the orthogonal quality of the partitioned mesh.

[0040] Table 1 Grid orthogonality quality

[0041]

[0042] R4. Set the boundary conditions for the raw material feeding inlet, tail coal injection inlet, tertiary air inlet, kiln tail smoke chamber outlet, decomposition furnace outlet, decomposition furnace inner wall and wall.

[0043] R5. The nearest neighbor algorithm based on Euclidean distance is mapped into a data model, and a physical simulation inside the decomposition furnace is performed. According to the simulation results, the model parameters are adjusted to output the decomposition furnace geometric model.

[0044] The embodiment of the present invention starts from the basic physical and chemical processes of cement clinker burning to ensure that the model can accurately reflect the complex physical processes inside the decomposition furnace. At the same time, it comprehensively selects input parameters and combines mechanism models and data-driven models to improve the understanding and characterization of the complex physical processes inside the decomposition furnace.

[0045] Furthermore, the physical simulation inside the calciner can be combined with the mass conservation equation, momentum conservation equation and energy conservation equation, and the model parameters can be adjusted according to the simulation results to achieve the best combustion state simulation. Specifically, it includes: adjusting the raw material feeding amount, coal injection amount, tertiary air temperature and other operating parameters, as well as optimizing the grid division and boundary condition settings to improve the simulation accuracy.

[0046] Specifically, the model parameters can be adjusted by combining optimization algorithms such as genetic algorithms, particle swarm optimization algorithms, etc. to perform automated parameter search, and at the same time, combined with grid optimization such as grid encryption and type selection, boundary condition adjustment, physical model selection, initial condition setting, solver configuration such as time step and convergence criteria, parameter sensitivity analysis such as single factor and multi-factor analysis, and experimental data comparison and verification such as error analysis and model calibration, etc., to comprehensively optimize the model parameters and ensure the accuracy and reliability of the simulation model, thereby providing a solid foundation for the simulation of the temperature field of the decomposition furnace. Existing common technologies are used in the process, which will not be repeated.

[0047] Understandably, boundary condition setting is a key step in the construction of the geometric model of the decomposition furnace, which is used to define the distribution of physical quantities in areas such as the inlet, outlet and wall during the simulation process. Specifically, it includes setting the mass flow rate, temperature and chemical composition of the raw material feeding inlet, the tail coal injection inlet and the tertiary air inlet. Set the temperature, pressure and composition of the kiln tail smoke chamber outlet and the decomposition furnace outlet. And set the temperature, thermal conductivity and roughness of the inner wall and wall of the decomposition furnace. In addition, it is necessary to set the initial temperature field, pressure field and flow rate field at the beginning of the simulation, and set boundary conditions such as thermal radiation and chemical reaction according to the simulation requirements. The setting of these boundary conditions directly affects the accuracy and reliability of the simulation, ensuring that the model can truly reflect the actual working conditions of the decomposition furnace.

[0048] What needs to be explained about step 3 is that the specific training process of the prediction model is as follows: G1. After completing the construction of the decomposition furnace geometric model, set its initial parameters. Based on the mass conservation equation, momentum conservation equation and energy conservation equation, in the geometric model, first do not consider the inlet and outlet temperatures, and iterate according to the set number of iterations until the decomposition furnace outlet wind speed reaches the set maximum value.

[0049] G2. Import pulverized coal chemical analysis and raw material chemical analysis to start simulating the chemical reaction behaviors of pulverized coal combustion and raw material decomposition.

[0050] G3. After completing the chemical reaction simulation, analyze the simulation results and compare them with the actual data. If the difference between the simulation results and the corresponding actual data is greater than the set difference, adjust the model parameters and output the final prediction model.

[0051] Specifically, the initial parameters of the geometric model of the decomposition furnace are set comprehensively based on the professional knowledge in the field of cement production, the statistical analysis results of a large amount of historical production data, and the commonly used empirical values ​​in the industry. These parameters include chemical reaction kinetic parameters, such as the rate constants of different coal powder combustion reactions and raw meal decomposition reactions, which are preliminarily determined based on the reaction characteristics of common coal powder and raw meal components, heat transfer and mass transfer parameters such as heat transfer coefficient, which are estimated based on the heat transfer research data of similar industrial equipment and combined with the actual structure and material of the decomposition furnace. The mass transfer coefficient is estimated based on the principles of fluid mechanics and the flow characteristics of materials in the decomposition furnace, as well as fluid physical property parameters, for example, the initial values ​​of fluid parameters such as density and viscosity are determined based on the manual of physical properties of coal powder, raw meal and gas at different temperatures and pressures.

[0052] Understandably, the data of coal powder chemical analysis and raw meal chemical analysis record in detail the contents of volatile matter, fixed carbon, ash, moisture and various trace elements in coal powder, the proportions of main components such as calcium carbonate, silicon oxide, aluminum oxide, iron oxide in raw meal, and other impurity components that may affect chemical reactions.

[0053] Understandably, the data of coal powder chemical analysis and raw meal chemical analysis record in detail the contents of volatile matter, fixed carbon, ash, moisture and various trace elements in coal powder, the proportions of main components such as calcium carbonate, silicon oxide, aluminum oxide, iron oxide in raw meal, and other impurity components that may affect chemical reactions.

[0054] It is also understandable that the physical simulation results based on the three major conservation laws provide a large amount of valuable data for model training. These data reflect the physical and chemical changes of the decomposition furnace under different working conditions. The model improves its prediction ability by learning the laws in these data. During the training process, the results under different parameter settings are simulated according to the conservation laws, and the differences between the model prediction values ​​and the actual simulation values ​​calculated based on the conservation laws are observed. These differences are used to adjust the model parameters such as the weights and thresholds of the neural network to make the model prediction more accurate. For example, if there is a deviation between the temperature field predicted by the model and the temperature field calculated according to the law of conservation of energy, the model parameters can be adjusted through the optimization algorithm to make the model prediction closer to the actual situation, thereby continuously optimizing the prediction model and improving its prediction accuracy and reliability for the operation status of the decomposition furnace.

[0055] It should be noted that step 3 is that the digital twin virtual model layer consists of a digital twin layer and a virtual model layer, where Figure 2 As shown, the specific acquisition process of the digital twin layer is as follows: C1. Filter out the tertiary air temperature, kiln tail smoke chamber temperature, tail coal chemical analysis and cement raw material chemical analysis from the input parameters.

[0056] C2. Use the geometric model of the decomposition furnace as a physical model and generate temperature field data through physical model simulation.

[0057] C3. Preprocess the tertiary air temperature, kiln tail smoke chamber temperature, tail coal chemical analysis and cement raw material chemical analysis by mean filtering and normalization to obtain preprocessed data.

[0058] C4. Use the preprocessed data as input and build a temperature field prediction model based on the K-nearest neighbor algorithm.

[0059] C5. Integrate the temperature field data generated by the physical model and the temperature field prediction data based on the K-nearest neighbor algorithm to build a data model.

[0060] C6. Through the data model, the collected and predicted temperature data are mapped to the actual spatial structure of the decomposition furnace to form a temperature field distribution.

[0061] C7. Based on the mapped temperature field, combined with the real-time operation data and historical data of the decomposition furnace, a digital twin layer model is constructed.

[0062] Understandably, by calculating the distance between the target data and the historical data, the K nearest neighbor data are screened out based on the K-nearest neighbor algorithm. Based on the temperature field information of these neighbor data, the temperature field under the current working conditions is predicted, providing more timely prediction data for subsequent models.

[0063] Understandably, the temperature field data generated by the physical model and the temperature field prediction data based on the K-nearest neighbor algorithm are integrated to construct a data model. In this process, data mining and machine learning techniques are used to deeply analyze the correlation and potential laws between the data, so that the data model can more accurately describe the characteristics of the decomposition furnace temperature field.

[0064] It can also be understood that, through the data model, the collected and predicted temperature data are mapped to the actual spatial structure of the decomposition furnace to form a temperature field distribution. During the mapping process, factors such as the geometric shape and internal components of the decomposition furnace are fully considered to ensure the accuracy of the temperature field mapping.

[0065] See also Figure 3 and Figure 4 As shown, further, the specific acquisition process of the virtual model layer is: H1. Generate a model file through the digital twin in the digital twin layer, and organize it according to the time series. According to the needs of model training, the continuous time series data is divided into windows of fixed length, and the data in each window is used as an independent input sample.

[0066] H2. Perform feature enhancement and normalization on the divided time series data, input the preprocessed input sequence into the CNN layer of the deep learning model based on CNN-GRU-Attention, and set convolution kernels of different sizes and steps in the convolution layer.

[0067] Preferably, the input sequence passes through the convolution layer of the CNN layer, and the convolution operation is performed by sliding the convolution layer kernel on the data to extract features from the input sequence. The convolution operation formula is: ,in, is the data in the input sequence, is the convolution kernel weight, is the bias, and is the convolution kernel size, The coordinates of the output feature map. The data processed by the convolution layer is then reduced in dimension by the pooling layer. The maximum pooling method is used, the pooling window size is 2×2, and the maximum value in each window is taken as the pooling output.

[0068] Understandably, the data in each window is used as an independent input sample for subsequent model training and prediction. The selection of window length needs to comprehensively consider the frequency of change of temperature field data and the model's ability to capture time series information. The optimal value is generally determined through experiments and verification. For example, for temperature field data that changes more frequently, a shorter window length such as 5-10 time steps can be selected. For data that changes relatively slowly, a longer window length such as 20-30 time steps can be selected.

[0069] The GRU layer contains a reset gate and an update gate. The reset gate controls the degree of forgetting or resetting the previous hidden state before calculating the candidate hidden state, and the update gate is used to determine the amount of previous information that should be passed to the future.

[0070] The reset gate controls the degree to which the previous hidden state is forgotten or reset before calculating the candidate hidden state. The update gate is used to determine the amount of previous information that should be passed to the future. The reset gate calculation formula is , the update gate calculation formula is , the candidate hidden state calculation formula is , the final hidden state calculation formula is , Sigmoid activation function is used to compress the input value to between 0 and 1. It is often used in gating mechanisms to control the flow of information. , and is the weight matrix, and Respectively represent the calculation of reset gate and update gate for input vector and the hidden state at the previous moment The weight matrix for linear transformation, Indicates the candidate hidden state calculation, used for the input vector and the previous hidden state after the reset gate The weight matrix for linear transformation, , and Respectively represent the bias vectors in the reset gate calculation, the update gate calculation, and the candidate hidden state calculation, Represents the input vector and the hidden state at the previous moment The concatenation of forms a new vector as the input for subsequent calculations. It is a hyperbolic tangent activation function that compresses the input value to between -1 and 1 and is used to calculate the candidate hidden state. , Represents the element-by-element multiplication (Hadamard product) operator, which multiplies the elements of corresponding positions of two vectors.

[0071] H3. Through the sliding convolution operation of the convolution kernel on the input data, the local features in the data are extracted. The data processed by the convolution layer is then reduced in dimension through the pooling layer using the maximum pooling or average pooling method.

[0072] H4. Convert the input sequence processed by the CNN layer into a feature vector and send it to the GRU layer. Then input the feature sequence output by the GRU layer into the Attention layer for attention mechanism calculation.

[0073] H5. The output of the Attention layer passes through an output layer for the final output. The output layer adopts a fully connected layer structure. The calculation formula of the output layer is: , is the weight matrix of the output layer, Represents the predicted output of the model, Represents the feature vector output by the Attention layer, A learnable constant term is introduced into the calculation result to increase the expressive power of the model. The overall function of this formula is to map the feature vector obtained by the previous layer processing to the required output dimension through the operation of the weight matrix and the bias vector to obtain the final prediction value.

[0074] Preferably, the Attention layer is divided into road, Luhe For the input matrix , each path is first mapped through a linear layer to obtain the corresponding query vector , key vector Sum value vector , where the formulas are , , ,in, , and They are road, Luhe The weight matrix of the path is calculated and The dot product of , when performing scaling operations, is , is the key vector The dimension of represents the transpose symbol, Represents the score in Attention, Use the Softmax function to normalize and get the attention weight , and finally the attention weights and value vector Multiply and sum to get the output of the Attention layer. Through this process, the Attention layer can calculate the similarity or correlation between the feature sequence and the feature value, highlight the weight of important elements, and filter out more valuable information for temperature field prediction.

[0075] Specifically, Figure 3 middle Represents the absolute value of the difference between the predicted value and the actual value. Indicates the difference threshold between the predicted value and the actual value that is allowed.

[0076] Furthermore, the training process of the digital twin virtual model layer also includes: 1) using the training set to train the constructed CNN-GRU-Attention-based model, and using a preset loss function to calculate the difference between the model prediction result and the true value, and using the back propagation algorithm to calculate the gradient of the loss function to the model parameters during the training process, and using the optimizer to update the model parameters according to the gradient so that the loss function reaches the minimum value.

[0077] 2) Adopt a learning rate decay strategy to gradually reduce the learning rate as the number of training rounds increases, and use regularization techniques to constrain the virtual model layer.

[0078] Understandably, the loss function can specifically select mean square error loss function MSE, mean absolute error loss function MAE, etc. to measure the difference between the model prediction result and the true value. The optimizer can select Adam, SGD, etc. to update the model parameters according to the gradient. Regularization technology can select L1 regularization, L2 regularization, Dropout, etc. to constrain the model to prevent overfitting of the model and improve the generalization ability of the model. By continuously adjusting the model parameters and training strategies, the model can learn the laws in the temperature field data more accurately, improve the prediction ability of the decomposition furnace temperature field, and finally train to obtain the digital twin virtual model layer.

[0079] It should be noted that in order to further improve the performance of the Attention layer, the basic attention mechanism can be improved and expanded. For example, a multi-head attention mechanism can be used to obtain multiple sets of query vectors, key vectors, and value vectors through multiple different linear transformations, and the attention weights can be calculated and fused respectively, so that different aspects of the input features can be paid attention to at the same time, and the model's ability to capture complex features can be enhanced. Alternatively, position encoding information can be introduced to incorporate the position information of the input sequence into the attention calculation, so that the model can better understand the relative position relationship of the elements in the sequence and improve the ability to process time series data.

[0080] It is also understandable that the prediction results need to be adjusted and post-processed according to actual application requirements, such as limiting the predicted temperature value within a reasonable range, or smoothing the prediction results to improve the stability and reliability of the prediction.

[0081] The embodiment of the present invention achieves high-precision simulation of the temperature field of the decomposition furnace by establishing a digital twin virtual model layer and a decomposition furnace geometric model, combined with deep learning technology. And through digital twin technology, the temperature field distribution and dynamic changes inside the decomposition furnace can be simulated in real time, significantly improving the precision of the description of the dynamic characteristics of the temperature field.

[0082] What needs to be added about step 4 is that abnormality detection and temperature field simulation of the decomposition furnace are performed, including: obtaining real-time data from the actual operation of the decomposition furnace as current input data, and performing mean filtering and normalization processing on the current input data.

[0083] The preprocessed data is input into the trained digital twin virtual model layer for temperature field simulation, and the digital twin virtual model layer outputs the temperature field data set inside the decomposition furnace, including temperature distribution, pressure distribution, and flow velocity distribution.

[0084] Determine the steady-state operating point of the virtual model layer, establish a hybrid model using a series-parallel hybrid connection method, and use the hybrid model to determine whether the predicted value of the decomposition furnace input parameter deviates from the normal value. If it deviates, it is determined as an abnormal operating condition, and an early warning message is output, and the next step is executed. If it does not deviate, the next step is executed directly.

[0085] The temperature field is simulated based on the hybrid model and the predicted value of the temperature field of the decomposition furnace is output.

[0086] Furthermore, temperature field simulation is performed, including: Geometric model simulation: based on the 3D geometric model and mesh division of the decomposition furnace, flow, heat transfer and chemical reaction simulation are performed.

[0087] Physical model simulation: Based on the laws of conservation of mass, momentum and energy, the temperature field distribution inside the decomposition furnace is calculated.

[0088] Chemical reaction simulation: Import coal powder chemical analysis and raw material chemical analysis to simulate the chemical reaction behavior of coal powder combustion and raw material decomposition.

[0089] Furthermore, a hybrid model is established, including: connecting a CNN-GRU-Attention-based data-driven model to the temperature field data model of the digital twin layer.

[0090] The series-connected model is connected in parallel with another data-driven model based on CNN-GRU-Attention to form a series-parallel hybrid model.

[0091] In a specific embodiment, if the predicted value deviates from the normal value, the system will output abnormal warning information and store it in the database of the digital twin layer to provide reference data for subsequent analysis and prompt the operator to make corresponding adjustments. At this time, the temperature field simulation results can help operators better understand the causes and effects of the abnormality. According to the abnormal detection results, the operator can adjust the raw material feeding amount, coal injection amount, tertiary air temperature and other parameters to optimize the operation status of the decomposition furnace.

[0092] The embodiment of the present invention not only realizes the predictive simulation of the temperature field by utilizing the trained digital twin virtual model and the prediction model, but also supports anomaly detection and operation optimization. Through the simulation results, it is possible to timely discover abnormal conditions in the operation of the decomposition furnace, and provide optimization suggestions to effectively guide actual production.

[0093] In another embodiment of the present invention, the effect of the cement decomposition furnace temperature field simulation method of the present invention is also verified.

[0094] When the raw meal feeding amount and tail coal data are stable and the raw meal feeding amount and coal injection amount are relatively stable, the input variables are simplified, and the kiln tail smoke chamber temperature, tertiary air temperature, C4A cone outlet temperature and C4A cone outlet temperature are selected as the input of the CNN-GRU-Attention model. The input of the geometric model in the data twin layer is the actual operation data of the factory.

[0095] In this way, 103 sets of cement decomposition furnace temperature field samples are generated. The 103 sets of data of kiln tail smoke chamber temperature, tertiary air temperature, C4A cone outlet temperature and C4B cone outlet temperature are the actual operation data of the factory. The outlet temperature of the decomposition furnace of the first 103 sets of samples is replaced by the calculation results of the data twin layer. The 103 sets of temperature field data sets of the digital twin layer are used as training sets to train the steady-state working points of the virtual model layer. The outlet temperature of the decomposition furnace predicted by the steady-state working point of the virtual model layer is the actual operation data of the factory. Finally, the outlet temperature of the decomposition furnace predicted by the steady-state working point of the virtual model layer is output and compared with the actual operation data of the factory. With the increase of the number of iterations, the error curve of the training set gradually tends to be stable, indicating that the training results are good and it has good prediction ability of the outlet temperature of the decomposition furnace. For example, after 250 iterations, the error curve of the test set gradually decreases and finally falls below 0.1. After the iteration of the model is completed, the test set of the outlet temperature of the decomposition furnace is normalized and input into the established CNN-GRU-Attention model. The prediction results of the test set and the actual comparison results are shown in Table 2.

[0096] Table 2 Comparison of predicted values ​​and actual values ​​of digital twin layer

[0097]

[0098] As shown in Table 2, except for the raw material feeding amount and tail coal, which remain normal, the other abnormal variables that are abnormal from the test set are the internal pressure of the decomposition furnace, the C5A cone pressure, the C5B cone pressure, the temperature of the kiln tail smoke chamber, the decomposition furnace outlet temperature, the internal temperature of the decomposition furnace, and the decomposition furnace outlet temperature. The test set data of the above variables exceeds the threshold given by the validation set, and the model judges it as abnormal. The deviation of the predicted values ​​of each variable is shown in Table 3.

[0099] Table 3 Comparison of predicted values ​​and actual values

[0100]

[0101] From the table above, we can see that when an abnormal working condition occurs, CNN-GRU-Attention can determine that the predicted value has deviated from the normal value. Since the temperature of the kiln tail smoke chamber drops most significantly, other temperatures have dropped to varying degrees, and the internal pressure of the decomposition furnace, the C5A cone pressure, and the C5B cone pressure first deviate from the normal threshold. Therefore, it is judged that the data anomaly is caused by the on-site operator in the cement plant opening the kiln tail door to clean the kiln skin.

[0102] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they shall all fall within the protection scope of the present invention.

Claims

1. A cement decomposition furnace temperature field simulation method based on digital twin, characterized in that: The method includes: Select input parameters according to the cement clinker burning process and construct a time series data set for the calciner; Preprocess and divide the decomposition furnace time series data set into training set, validation set and test set, which are used as model input; Establish a geometric model of the decomposition furnace to obtain a prediction model, and establish a deep learning model based on CNN-GRU-Attention to obtain a digital twin virtual model layer; Based on the trained digital twin virtual model layer and prediction model, anomaly detection and temperature field simulation of the decomposition furnace are performed, and the predicted value of the decomposition furnace temperature field is output.

2. According to a method for simulating temperature field of cement decomposition furnace based on digital twinning according to claim 1, it is characterized in that: The method of establishing a calciner geometric model comprises: According to the design drawing of the cement plant calciner, a 3D model of the calciner is established using 3D modeling software; According to the cement burning process, the 3D model of the calciner is set to have 4 raw material feeding inlets, 4 tail coal injection inlets, 2 tertiary air inlets, 1 kiln tail smoke chamber outlet, 1 calciner outlet and the calciner inner wall inlet and wall; The hybrid meshing technology is used to mesh the internal area of ​​the calciner, the boundary layer and the pre-set analysis area, and the mesh quality is tested by calculating the orthogonal quality index of the mesh. The hybrid mesh includes layered polyhedral mesh, pure polyhedral mesh and hexahedral mesh. Set the boundary conditions of raw material feeding inlet, tail coal injection inlet, tertiary air inlet, kiln tail smoke chamber outlet, calciner outlet, calciner inner wall and wall; The nearest neighbor algorithm based on Euclidean distance is mapped into a data model, and a physical simulation inside the decomposition furnace is performed. According to the simulation results, the model parameters are adjusted, thereby outputting a geometric model of the decomposition furnace.

3. The method for simulating temperature field of cement decomposition furnace based on digital twin according to claim 1, characterized in that: The specific training process of the prediction model is as follows: After the construction of the calciner geometric model is completed, its initial parameters are set. Based on the mass conservation equation, momentum conservation equation and energy conservation equation, in the geometric model, the inlet and outlet temperatures are not considered first, and the iteration is performed according to the set number of iterations until the outlet wind speed of the calciner reaches the set maximum value; Import pulverized coal chemical analysis and raw material chemical analysis to start simulating the chemical reaction behaviors of pulverized coal combustion and raw material decomposition; After completing the chemical reaction simulation, analyze the simulation results and compare them with the actual data. If the difference between the simulation results and the corresponding actual data is greater than the set difference, adjust the model parameters and output the final prediction model.

4. The method for simulating temperature field of cement decomposition furnace based on digital twin according to claim 1, characterized in that: The digital twin virtual model layer consists of a digital twin layer and a virtual model layer, wherein the specific acquisition process of the digital twin layer is as follows: From the input parameters, the tertiary air temperature, the temperature of the kiln tail smoke chamber, the chemical analysis of the tail coal, and the chemical analysis of the cement raw material were selected; The geometric model of the calciner is used as a physical model, and temperature field data is generated through physical model simulation; Pre-process the tertiary air temperature, kiln tail smoke chamber temperature, tail coal chemical analysis and cement raw material chemical analysis to obtain pre-processed data; The preprocessed data is used as input to build a temperature field prediction model based on the K-nearest neighbor algorithm; Integrate the temperature field data generated by the physical model and the temperature field prediction data based on the K-nearest neighbor algorithm to build a data model; Through the data model, the collected and predicted temperature data are mapped to the actual spatial structure of the decomposition furnace to form the temperature field distribution; Based on the mapped temperature field, combined with the real-time operation data and historical data of the decomposition furnace, a digital twin layer model is constructed.

5. The method for simulating temperature field of cement decomposition furnace based on digital twin according to claim 4 is characterized in that: The specific acquisition process of the virtual model layer is as follows: Generate model files through the digital twins in the digital twin layer and organize them according to time series. According to the needs of model training, divide the continuous time series data into windows of fixed length, and use the data in each window as an independent input sample; Perform feature enhancement and normalization on the divided time series data, input the preprocessed input sequence into the CNN layer of the deep learning model based on CNN-GRU-Attention, and set convolution kernels of different sizes and steps in the convolution layer; The local features in the data are extracted by sliding convolution operation of the convolution kernel on the input data. The data processed by the convolution layer is then reduced in dimension by the pooling layer using the maximum pooling or average pooling method. The input sequence processed by the CNN layer is converted into a feature vector and sent to the GRU layer. The feature sequence output by the GRU layer is input into the Attention layer for attention mechanism calculation. The output of the Attention layer passes through an output layer for final output. The output layer adopts a fully connected layer structure. The calculation formula of the output layer is: , is the weight matrix of the output layer, Represents the predicted output of the model, Represents the feature vector output by the Attention layer, is a constant term.

6. The method for simulating the temperature field of a cement decomposition furnace based on digital twin according to claim 5, characterized in that: The Attention layer has three inputs: road, Luhe road; For the input matrix , each path is first mapped through a linear layer to obtain the corresponding query vector , key vector Sum value vector ; calculate and The dot product of ,right Use the Softmax function to normalize and get the attention weight, and add the attention weight to the value vector Multiply and sum to get the output of the Attention layer.

7. The method for simulating temperature field of cement decomposition furnace based on digital twin according to claim 5, characterized in that: The training process of the digital twin virtual model layer also includes: Use the training set to train the constructed CNN-GRU-Attention-based model, and use the preset loss function to calculate the difference between the model prediction result and the true value. In the training process, the back propagation algorithm is used to calculate the gradient of the loss function to the model parameters, and the optimizer is used to update the model parameters according to the gradient to minimize the loss function. A learning rate decay strategy is adopted to gradually reduce the learning rate as the number of training rounds increases, and regularization techniques are used to constrain the virtual model layer.

8. The method for simulating temperature field of cement decomposition furnace based on digital twin according to claim 1, characterized in that: The abnormality detection and temperature field simulation of the decomposition furnace include: Real-time data is obtained from the actual operation of the decomposition furnace as current input data, and mean filtering and normalization processing are performed on the current input data; The preprocessed data is input into the trained digital twin virtual model layer for temperature field simulation, and the temperature field data set inside the decomposition furnace is output through the digital twin virtual model layer; Determine the steady-state working point of the virtual model layer, establish a hybrid model using a series-parallel hybrid connection method, and use the hybrid model to determine whether the predicted value of the decomposition furnace input parameter deviates from the normal value. If it deviates, it is determined as an abnormal operating condition, and an early warning message is output, and the next step is executed. If it does not deviate, the next step is directly executed; The temperature field is simulated based on the hybrid model and the predicted value of the temperature field of the decomposition furnace is output.

9. The method for simulating temperature field of cement decomposition furnace based on digital twin according to claim 8, characterized in that: The temperature field simulation comprises: Geometric model simulation: Based on the 3D geometric model and meshing of the calciner, flow, heat transfer and chemical reaction simulation are performed; Physical model simulation: Based on the laws of conservation of mass, momentum and energy, calculate the temperature field distribution inside the decomposition furnace; Chemical reaction simulation: Import coal powder chemical analysis and raw material chemical analysis to simulate the chemical reaction behavior of coal powder combustion and raw material decomposition.

10. The method for simulating temperature field of cement decomposition furnace based on digital twin according to claim 8, characterized in that: The hybrid model is established, comprising: Connect a CNN-GRU-Attention-based data-driven model to the temperature field data model of the digital twin layer; The series-connected model is connected in parallel with another data-driven model based on CNN-GRU-Attention to form a series-parallel hybrid model.

Images