A method and system for predicting the permeability of a sintering mix

Abstract

The application provides a sintering mixture air permeability prediction method and system, real-time parameter data affecting sintering mixture air permeability is acquired, the real-time parameter data is input into a neural network model established in advance, and a real-time air permeability label output by the neural network model is accepted, and the real-time air permeability label is output as an air permeability prediction value. The technical scheme of the application is stable, high in accuracy, good in robustness, and advances the sintering air permeability analysis node, so that the sintering air permeability condition in production can be grasped earlier, the particle size distribution condition is optimized by adjusting water addition, speed adjustment and the like in the granulation process, the air permeability is optimized, the yield and quality of subsequent sintered ore are effectively improved, the main exhaust fan energy consumption is reduced, resource utilization is improved, and production cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of sintering mixture permeability technology, and in particular to a method and system for predicting the permeability of sintering mixtures. Background Technology

[0002] Sintering is a crucial step in iron and steel smelting. Sintering involves mixing various powdered iron-containing raw materials with appropriate amounts of fuel, flux, and water. After mixing and pelletizing, the materials undergo a series of physicochemical changes on sintering equipment, eventually melting into lumps. Finally, after crushing, screening, and granulation, honeycomb-shaped particles meeting the particle size requirements of blast furnaces are formed, which is called sintered ore. Sintered ore is commonly used in blast furnace ironmaking. High-performance sintered ore can effectively improve iron production and quality, while also conserving resources and reducing energy consumption.

[0003] The permeability of the mixture refers to the airflow per minute per unit sintering area of ​​the mixture laid on the sintering machine under certain layer thickness and negative pressure. Poor permeability of the mixture layer indicates high bed resistance, leading to increased suction intensity, high air box pressure, insufficient air supply, slower vertical sintering speed, and a poorer sintering and melting reaction process. It also causes a shift in the burn-through point, high residual sulfur, and poor agglomeration, affecting the yield and quality of sintered ore. Good permeability of the mixture layer improves the thermal conductivity of the sintered material, enhancing heat exchange conditions within the layer. This helps control the sintering zone within a narrower range, stabilizing the production process and maintaining a stable burn-through point, thus improving the yield and quality of sintered ore and increasing fuel utilization. Therefore, maintaining good permeability of the mixture layer is crucial for improving the yield and quality of sintered ore and reducing energy consumption.

[0004] With the widespread application of sintering technology, many companies have conducted in-depth research on the permeability of the sintering process in order to improve the production efficiency of the sintering process. In the sintering process operation guidance system developed by Kawasaki Steel Corporation of Japan, permeability is comprehensively evaluated using a two-dimensional matrix based on the optimal exhaust gas flow rate of the main exhaust fan, the highest temperature at the air box, and the sintering endpoint. This method mainly uses parameters at the end of sintering to judge the permeability of the entire process, with a time lag of about one hour. At this point, there is no room for further adjustment to optimize the sintering process.

[0005] Therefore, it is necessary to propose a method and system for predicting the permeability of sintered mixtures in order to solve or at least alleviate the above-mentioned defects. Summary of the Invention

[0006] The main objective of this invention is to provide a method and system for predicting the permeability of sintered mixtures, in order to solve the problem that the existing technology uses parameters at the end of sintering to evaluate the permeability of the entire process, which has a long time lag and cannot meet the needs of actual production.

[0007] To achieve the above objectives, the present invention provides a method for predicting the permeability of sintered mixtures, comprising the following steps:

[0008] S1, acquire real-time parameter data affecting the permeability of the sintering mixture; wherein, the real-time parameter data includes the real-time particle size distribution matrix of the sintering mixture, the real-time moisture content of the sintering mixture, the real-time solid fuel ratio, the real-time average thickness of the sintering layer, and the real-time temperature of the sintering mixture.

[0009] S2, input the real-time parameter data into a pre-established neural network model, and accept the real-time breathability label output by the neural network model, and output the real-time breathability label as a breathability prediction value; wherein, the neural network model contains the mapping relationship between the real-time parameter data and the breathability label.

[0010] Preferably, step S2 further includes the following step before "outputting the real-time breathability label as a breathability prediction value":

[0011] S201, determine the deviation value between the real-time breathability label and the breathability calculation value corresponding to the real-time breathability label, and determine whether the deviation value is within a preset range;

[0012] S202, when the deviation value is within the preset range, proceed to the step of "outputting the real-time breathability label as a breathability prediction value"; when the deviation value is not within the preset range, retrain the neural network model.

[0013] Preferably, the "real-time particle size distribution matrix of the sintering mixture" in step S1 is obtained through the following steps:

[0014] S11, acquire images of the sintering mixture surface on the sintering machine trolley at preset time intervals, and extract effective surface images from the sintering mixture surface images;

[0015] S12, count the total number n of effective sintered mixture particles in the effective material surface image, and determine the proportion η1 of particle size d1 in the sintered mixture material surface image to the total number n, the proportion η2 of particle size d2 in the sintered mixture material surface image to the total number n, the proportion η3 of particle size d3 in the sintered mixture material surface image to the total number n, the proportion η4 of particle size d4 in the sintered mixture material surface image to the total number n, the proportion η5 of particle size d5 in the sintered mixture material surface image to the total number n, and the proportion η6 of particle size d6 in the sintered mixture material surface image to the total number n; wherein, d1 < 3mm, 3mm ≤ d2 < 5mm, 5mm ≤ d3 < 8mm, 8mm ≤ d4 < 10mm, 10mm ≤ d5 < 15mm, and d6 ≥ 15mm;

[0016] S13, aggregate η1, η2, η3, η4, and η5 to obtain the real-time particle size distribution matrix of the sintered mixture. Here... Therefore, it can be excluded from the real-time matrix.

[0017] Preferably, the "air permeability calculation value" in step S201 is obtained through the following steps:

[0018] According to the formula

[0019] Where Q is the real-time gas flow rate entering the ignition furnace, A is the furnace area, H is the real-time average thickness of the sintering material layer, and ΔP is the real-time pressure difference between the ignition furnace furnace and the wind box; wherein the real-time pressure difference between the ignition furnace furnace and the wind box is obtained through the following steps:

[0020] The real-time values ​​of the ignition furnace pressure and the bellows pressure are obtained, and the difference between the real-time values ​​of the ignition furnace pressure and the bellows pressure is taken as the real-time pressure difference between the ignition furnace furnace and the bellows.

[0021] Preferably, the "neural network model" in step S2 is obtained through the following steps:

[0022] S21, Obtain N sets of historical sample data that meet the needs of training and testing; and divide the N sets of historical sample data into a training set and a test set according to a preset ratio, wherein the number of historical sample data in the training set is greater than the number of historical sample data in the test set.

[0023] S22, the training set includes historical sample input data and historical sample output data; wherein, the historical sample input data includes historical values ​​of sintering mixture particle size distribution matrix, historical values ​​of sintering mixture moisture content, historical values ​​of solid fuel ratio, historical values ​​of average thickness of sintering material layer, and historical values ​​of sintering mixture temperature.

[0024] S23, preprocess the historical sample input data, initialize the weights and biases of the preliminary neural network model, and use the preprocessed historical sample input data in the training set as the input to the preliminary neural network model to obtain the corresponding air permeability label network value; the preprocessing includes normalization processing and smoothing processing.

[0025] S24. Determine the calculated value of the loss function operation based on the breathability label network value and the historical values ​​of the breathability labels in the preprocessed training set, and determine whether the calculated value of the loss function operation is less than a preset value. If the calculated value of the loss function operation is greater than the preset value, proceed to step S25. If the calculated value of the loss function operation is less than the preset value, proceed to step S26.

[0026] S25, backpropagate the calculated values ​​according to the loss function to update the network weight values ​​and biases, and then repeat steps S23 to S24.

[0027] S26, input the test set into the preliminary neural network model, and when the test set feedback results meet the preset conditions, use the preliminary neural network as the neural network model.

[0028] Preferably, the neural network model is an ANN neural network model, specifically:

[0029]

[0030] Where Y0 is the target output, X it For the current input, WH ij Here, WO represents the weighted neuron connecting the i-th input and the j-th hidden, where m is the number of input neurons. j f is the connection weight between the j-th hidden neuron and the output neuron. h f is the activation function for the hidden neuron. o For the output neuron activation function, b j For the j-th bias hidden neuron, b o H is the bias of the output neuron, and HN is the number of hidden neurons in the output neuron.

[0031] Preferably, step S2 is followed by the step:

[0032] When the predicted air permeability value is greater than the first preset threshold, a first alarm command is sent to the target object;

[0033] When the predicted air permeability value is less than the second preset threshold, a second alarm command is sent to the target object; wherein the first preset threshold is greater than the second preset threshold.

[0034] Preferably, the normalization process in step S23 specifically employs a 0-1 normalization model to normalize the historical sample input data. The 0-1 normalization model is as follows:

[0035]

[0036] Where x represents the historical sample input data, max represents the maximum value of the historical sample input data, and min represents the minimum value of the historical sample input data.

[0037] Preferably, in step S23, the normalization process uses a mean smoothing model for smoothing, and the mean smoothing model is:

[0038]

[0039] Where n is the nth data point of the historical sample input data, and y is the moving smoother setting value.

[0040] The present invention also provides a prediction system for the permeability of sintered mixtures, for performing the above-described prediction method for the permeability of sintered mixtures, including a real-time data acquisition module, a permeability label generation module, and a permeability prediction value determination module;

[0041] The real-time data acquisition module is used to acquire real-time parameter data that affects the permeability of the sintering mixture; wherein, the real-time parameter data includes the real-time particle size distribution matrix of the sintering mixture, the real-time moisture content of the sintering mixture, the real-time solid fuel ratio, the real-time average thickness of the sintering layer, and the real-time temperature of the sintering mixture.

[0042] The breathability label generation module is used to input the real-time parameter data into a pre-established neural network model and receive the breathability label output by the neural network model; wherein, the neural network model contains a mapping relationship between the real-time parameter data and the breathability label;

[0043] The air permeability prediction value determination module is used to determine the deviation value between the air permeability label and the calculated air permeability value, and to determine whether the deviation value is within a preset range; when the deviation value is within the preset range, the air permeability label is used as the predicted air permeability value of the sintering mixture; when the deviation value is not within the preset range, the neural network model is retrained.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] This invention provides a method and system for predicting the permeability of sintered mixtures. The method acquires real-time parameter data affecting the permeability of the sintered mixture, including real-time particle size distribution matrix, real-time moisture content, real-time solid fuel ratio, real-time average thickness of the sintered material layer, and real-time temperature. This real-time parameter data is input into a pre-established neural network model, which outputs real-time permeability labels as predicted values. This invention is a prediction system based on a neural network model. By acquiring relevant parameters affecting sintered permeability, the system uses a trained neural network model to predict sintered permeability. This method is stable, accurate, robust, and advances the analysis of sintered permeability, providing better timeliness. This allows factories to understand the sintered permeability situation during production earlier, effectively helping them improve the output and quality of sintered ore, increase resource utilization, and reduce costs. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0047] Figure 1 This is a flow chart of the sintering process in one embodiment of the present invention;

[0048] Figure 2 This is a schematic diagram of a process in one embodiment of the present invention;

[0049] Figure 3 This is a schematic diagram of the process for obtaining the "real-time particle size distribution matrix of sintered mixture" in step S1 of an embodiment of the present invention.

[0050] Figure 4 This is a flowchart illustrating the process of obtaining the neural network model in step S2 of one embodiment of the present invention;

[0051] Figure 5 This is a schematic diagram of a neural network model in one embodiment of the present invention.

[0052] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0053] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0055] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0056] Furthermore, the use of terms such as "first" and "second" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. If the combination of technical solutions is contradictory or impossible to implement, such a combination should be considered non-existent and not within the scope of protection claimed by this invention.

[0057] Please see the appendix Figures 1-5 The present invention provides a method for predicting the permeability of sintering mixtures in one embodiment, comprising the following steps:

[0058] S1, acquire real-time parameter data affecting the permeability of the sintering mixture; wherein, the real-time parameter data includes the real-time particle size distribution matrix of the sintering mixture, the real-time moisture content of the sintering mixture, the real-time solid fuel ratio, the real-time average thickness of the sintering layer, and the real-time temperature of the sintering mixture.

[0059] It is important to note that this application aims to advance the analysis of sinter permeability, improving timeliness and enabling factories to understand the sinter permeability during production earlier. This can effectively help factories increase the yield and quality of sinter, improve resource utilization, and reduce costs. It should also be noted that in the actual analysis process, the material in the predicted mixture can be selected from a single point area, a surface area, or a mixture over a specific period, depending on the actual needs.

[0060] The real-time particle size distribution matrix of the sintering mixture refers to the distribution ratio of the material particle size. Specifically, considering that the mixed particles in the actual production process generally need to fall into a suitable range (generally 3mm to 15mm in this field), materials with too small a particle size will seriously affect the permeability of the material, while materials with too large a particle size will easily lead to an increase in the unevenness of the sintered ore formed later. Therefore, this application conducts particle size distribution matrix analysis on the material to be ignited. That is, the real-time particle size distribution matrix of the sintering mixture will affect the permeability of the sintering mixture, and there is a corresponding relationship between the two.

[0061] Preferably, the "real-time matrix of particle size distribution of sintering mixture" in step S1 is obtained through the following steps: S11, acquiring the surface image of the sintering mixture located on the sintering machine trolley at a preset time interval, and extracting the effective surface image from the surface image of the sintering mixture; specifically, the surface image of the sintering mixture can be acquired by an image acquisition device, and then the surface image of the sintering mixture within the effective range can be cropped as the effective surface image.

[0062] S12, count the total number n of effective sintered mixture particles within the effective material surface image, and determine the proportion η1 of particle size d1 in the sintered mixture material surface image to the total number n, the proportion η2 of particle size d2 in the sintered mixture material surface image to the total number n, the proportion η3 of particle size d3 in the sintered mixture material surface image to the total number n, the proportion η4 of particle size d4 in the sintered mixture material surface image to the total number n, the proportion η5 of particle size d5 in the sintered mixture material surface image to the total number n, and the proportion η6 of particle size d6 in the sintered mixture material surface image to the total number n; wherein, d1 < 3mm, 3mm ≤ d2 < 5mm, 5mm ≤ d3 < 8mm, 8mm ≤ d4 < 10mm, 10mm ≤ d5 < 15mm, d6 ≥ 15mm; S13, aggregate η1, η2, η3, η4, and η5 to obtain the real-time particle size distribution matrix of the sintered mixture. Here... Therefore, it can be excluded from the real-time matrix. The real-time particle size distribution matrix of the sintered mixture obtained in this way can accurately reflect the particle size distribution of the sintered mixture, and converting the particle size of the material into a proportion facilitates analysis and calculation. In another embodiment, the number of proportions can be appropriately increased or decreased according to the actual situation.

[0063] In other embodiments, considering that the data are relatively stable under stable production conditions, those skilled in the art can also determine the real-time particle size distribution matrix of the sintered mixture by analyzing the data in each image and then averaging all the data based on multiple images of the sintered mixture surface taken within a certain period of time. For example, the real-time particle size distribution matrix of the sintered mixture obtained from all image data within 3 seconds can be used as the real-time matrix within these 3 seconds. The data obtained in this way has higher accuracy.

[0064] The real-time moisture content of the sintering mixture can be collected using a moisture sampler. The real-time solid fuel ratio is obtained through data tracking. It's important to note the spatial relative position between the pellet mill and the ignition furnace; specifically, the ignition furnace is downstream of the pellet mill. Therefore, it's understandable that there's a difference of over 10 minutes between the real-time data collected at the pellet mill and the data collected at the ignition furnace. This means that data on the particle size distribution, fuel ratio, and moisture content of the same batch of material must be collected, and then the layer thickness and temperature must be obtained at the material distribution point after another 10 minutes. Finally, the permeability data must be obtained from the ignition furnace after another 3-5 minutes. The material ratio includes iron-containing mixed ore p1, fuel p2, limestone p3, lightly calcined dolomite p4, quicklime p5, and cold-returned ore p6, all expressed as percentages. The real-time average thickness of the sintering layer can be measured by a layer thickness gauge, and the real-time temperature of the sintering mixture can be obtained by a temperature sensor. Based on this basic concept of the relationship, it is understood that those skilled in the art can also select other parameters that can affect permeability as needed.

[0065] Furthermore, it should be noted that the real-time parameter data of this application can be correlated with each other using data tracking technology. Specifically, in actual production, there is a time difference, for example, from the time the material enters the pellet mill to the time the corresponding sintered mixture is generated and enters the ignition furnace. Therefore, this invention uses data tracking technology, for example, according to the formula s = vt or using the integral formula of speed over time. During the stable operation of the equipment, v can be regarded as constant. The aforementioned data is tracked and matched by associating the moving speed of the batch of materials with the moving speed of the equipment in the process. When the equipment stops, the data moves; when the equipment moves fast, the data moves fast. When the material in the process moves to the particle size analysis device, temperature measurement device, moisture detection device, sintered material layer thickness detection device, and ignition furnace, the data is also synchronized to these devices. In this way, the data corresponding to the same batch of materials at different times and locations can be accurately obtained.

[0066] S2, input the real-time parameter data into a pre-established neural network model, and accept the real-time breathability label output by the neural network model, and output the real-time breathability label as a breathability prediction value; wherein, the neural network model contains the mapping relationship between the real-time parameter data and the breathability label.

[0067] It is worth noting for those skilled in the art that the pre-established neural network model can be configured in a specific way, such as using ANN, CNN, or other algorithm models, depending on the actual data structure.

[0068] Preferably, the "neural network model" in step S2 is obtained through the following steps:

[0069] S21, acquire N sets of historical sample data suitable for training and testing; and divide the N sets of historical sample data into a training set and a test set according to a preset ratio, wherein the number of historical sample data in the training set is greater than the number of historical sample data in the test set; the sample size for acquiring samples suitable for training and testing can be [10000, 100000]. The ratio of the training set to the test set can be set to 9:1.

[0070] S22, the training set includes historical sample input data and historical sample output data; wherein, the historical sample input data includes historical values ​​of sintering mixture particle size distribution matrix, historical values ​​of sintering mixture moisture content, historical values ​​of solid fuel ratio, historical values ​​of average thickness of sintering material layer, and historical values ​​of sintering mixture temperature.

[0071] It should be noted that since the permeability of the sintered mixture cannot be directly measured, this application uses a soft-measurement formula. Determine the calculated value of the air permeability;

[0072] Where Q is the real-time gas flow rate entering the ignition furnace, A is the furnace area, H is the real-time average thickness of the sintering material layer, and ΔP is the real-time pressure difference between the ignition furnace furnace and the wind box; wherein the real-time pressure difference between the ignition furnace furnace and the wind box is obtained through the following steps:

[0073] The real-time values ​​of the ignition furnace pressure and the bellows pressure are obtained, and the difference between the real-time values ​​of the ignition furnace pressure and the bellows pressure is taken as the real-time pressure difference between the ignition furnace furnace and the bellows.

[0074] Furthermore, it's important to note that since there's a time difference between the real-time gas flow rate and furnace negative pressure values ​​entering the ignition furnace and the historical sample input data, the aforementioned data tracking technology is also necessary. For example, if the historical sample input data is obtained at time k, then the historical sample output data will be the data corresponding to time k+t, ensuring that the historical sample input data and historical sample output data represent the same material. Based on this, historical sample data is obtained by acquiring materials from different time periods.

[0075] S23, preprocess the historical sample input data, initialize the weights and biases of the preliminary neural network model, and use the preprocessed historical sample input data in the training set as the input to the preliminary neural network model to obtain the corresponding air permeability label network value; the preprocessing includes normalization processing and smoothing processing.

[0076] S24. Determine the calculated value of the loss function operation based on the breathability label network value and the historical values ​​of the breathability labels in the preprocessed training set, and determine whether the calculated value of the loss function operation is less than a preset value. If the calculated value of the loss function operation is greater than the preset value, proceed to step S25. If the calculated value of the loss function operation is less than the preset value, proceed to step S26.

[0077] S25, based on the calculated values ​​from the loss function, backpropagate to update the network weights and biases, then repeat steps S23-S24; through continuous iterative training, update the model's weight parameters, bias parameters, and learning rate, and calculate the mean squared error (MSE). Training stops when the MSE is within the allowable range or the number of iterations exceeds a threshold. For example, model training ends when the MSE is less than 0.001 or the maximum number of iterations exceeds 50.

[0078] S26, the test set is input into the preliminary neural network model. When the test set feedback results meet the preset conditions, the preliminary neural network is used as the neural network model. The remaining 10% of the test set is used to test the accuracy of the previously trained model. The testing method is to input the sample data into the trained model to obtain the predicted value of the permeability of the sintered mixture, compare it with the actual measured reference value of the sintered permeability index, and calculate the mean square error of both to determine whether the mean square error is within the allowable range. If the error is within the allowable range, it indicates that the model training is successful and can be applied to the prediction system; otherwise, the relevant parameters of the model are adjusted and the model is retrained.

[0079] Preferably, the neural network model is an ANN neural network model, specifically:

[0080]

[0081] The weighted neurons of the links between them, where m is the number of input neurons, WO j f is the connection weight between the j-th hidden neuron and the output neuron. h f is the activation function for the hidden neuron. o For the output neuron activation function, b j For the j-th bias hidden neuron, b o H is the bias of the output neuron, and HN is the number of hidden neurons in the output neuron.

[0082] A schematic diagram of the constructed neural network model is shown below. Figure 5As shown. The neural network consists of three layers: an input layer, a hidden layer, and an output layer. The preset input layer has nine input nodes: x1…x9 correspond to the proportions η1, η2, η3, η4, and η5 of the sintered mixture with particle sizes d1, d2, d3, d4, and d5, respectively, as well as the moisture content MI, solid fuel ratio P, sintering layer thickness h, and sintering temperature t of the sintered mixture. An additional bias node (marked with a circle of 1) is also added. The number of input nodes can be increased or decreased as needed. After receiving the sample input data, the input layer correlates x1…x9 with the weight matrix WH. ij Combined, with WH ij *x i The input to the hidden layer is in the form of +b (bias b = +1), and then passes through the activation function f of the hidden neurons. h The processing yields output results h1……h9, which are then combined with the corresponding weights and biases and used as inputs to the output layer. After processing by the activation function of the output neurons, the final result is output in the output layer, which is in the form of Y, i.e., the predicted value of sintering permeability index.

[0083] Furthermore, the predicted sintering permeability value is output to the management interface and logs are recorded according to the strategy.

[0084] Preferably, step S2 further includes the following step before "outputting the real-time breathability label as a breathability prediction value":

[0085] S201, determine the deviation value between the real-time breathability label and the breathability calculation value corresponding to the real-time breathability label, and determine whether the deviation value is within a preset range;

[0086] S202, when the deviation value is within the preset range, proceed to the step of "outputting the real-time breathability label as a breathability prediction value"; when the deviation value is not within the preset range, retrain the neural network model.

[0087] Preferably, the "air permeability calculation value" in step S201 is obtained through the following steps:

[0088] According to the formula Determine the calculated value of the air permeability;

[0089] Where Q is the real-time gas flow rate entering the ignition furnace, A is the furnace area, H is the real-time average thickness of the sintering material layer, and ΔP is the real-time pressure difference between the ignition furnace furnace and the wind box; wherein the real-time pressure difference between the ignition furnace furnace and the wind box is obtained through the following steps:

[0090] The real-time values ​​of the ignition furnace pressure and the bellows pressure are obtained, and the difference between the real-time values ​​of the ignition furnace pressure and the bellows pressure is taken as the real-time pressure difference between the ignition furnace furnace and the bellows.

[0091] It is important to note that, to further verify the accuracy of the current neural network model, this embodiment also determines the deviation between the real-time breathability label and the calculated breathability value corresponding to the real-time breathability label, and judges whether the deviation value is within a preset range. Since the calculated breathability value corresponding to the real-time breathability label cannot be directly measured by instruments, the predicted value of the neural network is compared with the actual value reflected by the soft measurement method. When the deviation value is within the preset range, it indicates that the accuracy of the current neural network model is sufficient. When the deviation value is not within the preset range, the neural network model is retrained, and the weights and / or biases are readjusted to improve the prediction accuracy of the neural network model. It is important to note that there is a correspondence between the real-time breathability label and the calculated breathability value corresponding to the real-time breathability label, ensuring that they are the predicted and calculated values ​​for the same material, thus guaranteeing data accuracy.

[0092] This invention uses a soft measurement formula to calculate the air permeability of materials. By detecting relevant parameters through a detection device, the corresponding air permeability index is calculated. Data tracking technology is also used here. Based on the relationship between measurable variables and the unmeasurable variable (air permeability), a mathematical model is established with measurable variables as input and the unmeasurable variable (air permeability) as output. Soft measurement technology is used to achieve complex (advanced) control of the process.

[0093] Preferably, step S2 is followed by the step:

[0094] When the predicted air permeability value is greater than a first preset threshold, a first alarm command is sent to the target object; when the predicted air permeability value is less than a second preset threshold, a second alarm command is sent to the target object; wherein, the first preset threshold is greater than the second preset threshold.

[0095] This embodiment introduces alarm commands, enabling staff / maintenance personnel to quickly understand and grasp abnormal situations and make timely judgments and decisions. In other words, an alarm command will be triggered when the permeability of the sintering mixture is too high or too low.

[0096] Preferably, the normalization process in step S23 specifically employs a 0-1 normalization model to normalize the historical sample input data. The 0-1 normalization model is as follows:

[0097]

[0098] Where x represents the historical sample input data, max represents the maximum value of the historical sample input data, and min represents the minimum value of the historical sample input data. This calculation ensures that each sample data falls within the range [0,1], eliminating the influence of units on the data sample. Specifically, the average thickness of the sintering layer and the temperature of the sintering mixture can be normalized.

[0099] Preferably, in step S23, the normalization process uses a mean smoothing model for smoothing, and the mean smoothing model is:

[0100]

[0101] Where n is the nth data point of the historical sample input data, and y is the moving smoother setting value.

[0102] The present invention also provides a prediction system for the permeability of sintered mixtures, for performing the above-described prediction method for the permeability of sintered mixtures, including a real-time data acquisition module, a permeability label generation module, and a permeability prediction value determination module;

[0103] The real-time data acquisition module is used to acquire real-time parameter data that affects the permeability of the sintering mixture; wherein, the real-time parameter data includes the real-time particle size distribution matrix of the sintering mixture, the real-time moisture content of the sintering mixture, the real-time solid fuel ratio, the real-time average thickness of the sintering layer, and the real-time temperature of the sintering mixture.

[0104] The breathability label generation module is used to input the real-time parameter data into a pre-established neural network model and receive the breathability label output by the neural network model; wherein, the neural network model contains a mapping relationship between the real-time parameter data and the breathability label;

[0105] The air permeability prediction value determination module is used to determine the deviation value between the air permeability label and the calculated air permeability value, and to determine whether the deviation value is within a preset range; when the deviation value is within the preset range, the air permeability label is used as the predicted air permeability value of the sintering mixture; when the deviation value is not within the preset range, the neural network model is retrained.

[0106] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for predicting the permeability of sintered mixtures, characterized in that, Including the following steps: S1, acquire real-time parameter data affecting the permeability of the sintering mixture; wherein, the real-time parameter data includes the real-time particle size distribution matrix of the sintering mixture, the real-time moisture content of the sintering mixture, the real-time solid fuel ratio, the real-time average thickness of the sintering layer, and the real-time temperature of the sintering mixture. S2, input the real-time parameter data into a pre-established neural network model, and receive the real-time breathability label output by the neural network model, and output the real-time breathability label as a breathability prediction value; wherein, the neural network model contains the mapping relationship between the real-time parameter data and the breathability label; Real-time parameter data are correlated with each other using data tracking technology. When materials in the process move to the particle size analysis device, temperature measurement device, moisture detection device, sintering material layer thickness detection device and ignition furnace, the data is also synchronized here to obtain the data corresponding to the same batch of materials at different times and locations.

2. The method for predicting the permeability of sintered mixtures according to claim 1, characterized in that, Before step S2, which involves "outputting the real-time breathability label as a predicted breathability value", the following step is also included: S201, determine the deviation value between the real-time breathability label and the breathability calculation value corresponding to the real-time breathability label, and determine whether the deviation value is within a preset range; S202, when the deviation value is within the preset range, proceed to the step of "outputting the real-time breathability label as a breathability prediction value"; when the deviation value is not within the preset range, retrain the neural network model.

3. The method for predicting the permeability of sintered mixtures according to claim 2, characterized in that, The "real-time particle size distribution matrix of the sintered mixture" in step S1 is obtained through the following steps: S11, according to a preset time interval, acquire the mixed material taken out from the rear belt of the pellet mill to form a sintered mixed material surface image, and extract the effective material surface image from the sintered mixed material surface image; S12, count the total number n of effective sintered mixture particles in the effective material surface image, and determine the proportion η1 of particle size d1 in the sintered mixture material surface image to the total number n, the proportion η2 of particle size d2 in the sintered mixture material surface image to the total number n, the proportion η3 of particle size d3 in the sintered mixture material surface image to the total number n, the proportion η4 of particle size d4 in the sintered mixture material surface image to the total number n, the proportion η5 of particle size d5 in the sintered mixture material surface image to the total number n, and the proportion η6 of particle size d6 in the sintered mixture material surface image to the total number n; wherein, d1 < 3mm, 3mm ≤ d2 < 5mm, 5mm ≤ d3 < 8mm, 8mm ≤ d4 < 10mm, 10mm ≤ d5 < 15mm, and d6 ≥ 15mm; S13, aggregate η1, η2, η3, η4, η5 to obtain the real-time particle size distribution matrix of the sintered mixture.

4. The method for predicting the permeability of sintered mixtures according to claim 2, characterized in that, The "air permeability calculation value" in step S201 is obtained through the following steps: According to the formula Determine the calculated value of the air permeability; Where Q is the real-time gas flow rate entering the ignition furnace, A is the furnace area, H is the real-time average thickness of the sintering material layer, and ΔP is the real-time pressure difference between the ignition furnace furnace and the wind box; wherein the real-time pressure difference between the ignition furnace furnace and the wind box is obtained through the following steps: The real-time values ​​of the ignition furnace pressure and the bellows pressure are obtained, and the difference between the real-time values ​​of the ignition furnace pressure and the bellows pressure is taken as the real-time pressure difference between the ignition furnace furnace and the bellows.

5. The method for predicting the permeability of sintered mixtures according to claim 1, characterized in that, The "neural network model" in step S2 is obtained through the following steps: S21, Obtain N sets of historical sample data that meet the needs of training and testing; and divide the N sets of historical sample data into a training set and a test set according to a preset ratio, wherein the number of historical sample data in the training set is greater than the number of historical sample data in the test set. S22, the training set includes historical sample input data and historical sample output data; wherein, the historical sample input data includes historical values ​​of sintering mixture particle size distribution matrix, historical values ​​of sintering mixture moisture content, historical values ​​of solid fuel ratio, historical values ​​of average thickness of sintering material layer, and historical values ​​of sintering mixture temperature. S23, preprocess the historical sample input data, initialize the weights and biases of the preliminary neural network model, and use the preprocessed historical sample input data in the training set as the input to the preliminary neural network model to obtain the corresponding air permeability label network value; the preprocessing includes normalization processing and smoothing processing. S24. Determine the calculated value of the loss function operation based on the breathability label network value and the historical values ​​of the breathability labels in the preprocessed training set, and determine whether the calculated value of the loss function operation is less than a preset value. If the calculated value of the loss function operation is greater than the preset value, proceed to step S25. If the calculated value of the loss function operation is less than the preset value, proceed to step S26. S25, backpropagate the calculated values ​​according to the loss function to update the network weight values ​​and biases, and then repeat steps S23~S24; S26, input the test set into the preliminary neural network model, and when the test set feedback results meet the preset conditions, use the preliminary neural network as the neural network model.

6. The method for predicting the permeability of sintered mixtures according to claim 5, characterized in that, The neural network model is an ANN neural network model, specifically: ； Where Y0 is the target output, X it WH represents the current input quantity. ij Here, WO represents the weighted neuron connecting the i-th input and the j-th hidden, where m is the number of input neurons. j f is the connection weight between the j-th hidden neuron and the output neuron. h f is the activation function for the hidden neuron. o For the output neuron activation function, b j For the j-th bias hidden neuron, b o H is the bias of the output neuron, and HN is the number of hidden neurons in the output neuron.

7. The method for predicting the permeability of sintered mixtures according to claim 1, characterized in that, The step S2 is followed by the following step: When the predicted air permeability value is greater than the first preset threshold, a first alarm command is sent to the target object; When the predicted air permeability value is less than the second preset threshold, a second alarm command is sent to the target object; wherein the first preset threshold is greater than the second preset threshold.

8. The method for predicting the permeability of sintered mixtures according to claim 5, characterized in that, The normalization process in step S23 specifically employs a 0-1 normalization model to normalize the historical sample input data. Normalization processing, the 0-1 normalized model is: ； Where x represents the historical sample input data, max represents the maximum value of the historical sample input data, and min represents the minimum value of the historical sample input data.

9. The method for predicting the permeability of sintered mixtures according to claim 5, characterized in that, In step S23, the normalization process uses a mean smoothing model for smoothing. The mean smoothing model is as follows: ； Where n is the nth data point of the historical sample input data, and y is the moving smoother setting value.

10. A system for predicting the permeability of sintered mixtures, used to execute the method for predicting the permeability of sintered mixtures according to any one of claims 1-9, characterized in that, It includes a real-time data acquisition module, a breathability label generation module, and a breathability prediction value determination module; The real-time data acquisition module is used to acquire real-time parameter data that affects the permeability of the sintering mixture; wherein, the real-time parameter data includes the real-time particle size distribution matrix of the sintering mixture, the real-time moisture content of the sintering mixture, the real-time solid fuel ratio, the real-time average thickness of the sintering layer, and the real-time temperature of the sintering mixture. The breathability label generation module is used to input the real-time parameter data into a pre-established neural network model and receive the breathability label output by the neural network model; wherein, the neural network model contains a mapping relationship between the real-time parameter data and the breathability label; The air permeability prediction value determination module is used to determine the deviation value between the air permeability label and the calculated air permeability value, and to determine whether the deviation value is within a preset range; when the deviation value is within the preset range, the air permeability label is used as the predicted air permeability value of the sintering mixture; when the deviation value is not within the preset range, the neural network model is retrained.

Images