Chemical production method and system for metal smelting

Abstract

The invention discloses a chemical production method and system for metal smelting. A sensor and a reference electrode are arranged to monitor the generation position and reduction parameters of reduction parameters of blast furnace equipment, a flue gas denitrification device reducing agent storage tank influenced by the ammonia gas concentration is predicted, the ammonia gas concentration of the flue gas denitrification device reducing agent storage tank is obtained based on a monitoring node, and correction parameters are generated on the reducing agent storage tank through an injection device; the method is used for inhibiting the flue gas denitrification device reducing agent storage tank leakage effect caused by reduction parameters. According to the management method, the safety and stability of the reducing agent storage tank of the flue gas denitrification device can be effectively improved, the adverse effect of the reduction parameters of the blast furnace equipment on the reducing agent storage tank of the flue gas denitrification device is reduced, and the safety of metal smelting chemical production is improved.

Description

Technical Field

[0001] The present invention relates to the field of chemical production control, and particularly to a chemical production method and system for metal smelting. Background Art

[0002] Metal smelting is a crucial part of the industrialization process. However, traditional metal smelting methods generally have problems such as high energy consumption and high pollution. Especially during the high-temperature smelting process, a large amount of flue gas is generated, and the nitrogen oxides (NOx) in it cause serious environmental pollution. In addition, the reduction reactions involved in the smelting process require a large amount of reducing agents, such as liquid ammonia and urea solutions, and the efficiency of their use and management directly affects the cost and environmental protection effect of metal smelting. In the modern metal smelting field, in order to reduce environmental pollution during the production process and improve resource utilization rate, flue gas denitrification technology has gradually been widely applied. By introducing an efficient monitoring and control system into the smelting equipment, reducing agents can be accurately injected into the flue gas denitrification device to remove nitrogen oxides.

[0003] Due to the lack of high-precision real-time monitoring and dynamic response capabilities, it is difficult to accurately control the injection amount of reducing agents by the technical methods for monitoring nitrogen oxides in metal smelting, which easily causes waste or insufficient injection of reducing agents and affects the denitrification effect. In addition, in existing smelting equipment, the layout of sensors is relatively single, and it is difficult to comprehensively reflect the concentration gradient, temperature distribution, and pressure changes in the reaction environment. The incompleteness of data collection limits further intelligent analysis and response mechanisms. In the prior art,

[0004] CN110221030B provides an exhaust gas concentration detection structure, an exhaust gas concentration detection method, and an exhaust gas monitoring device. This technical solution can achieve precise monitoring of metal smelting exhaust gas, but it cannot automatically detect the leakage source and is difficult to automatically control the leakage situation. In addition, the method provided in CN118425437B can control the liquid gasification ratio of the scrubbing tower in chemical production and has good exhaust gas monitoring effects, but it also cannot identify the leakage source and cannot perform automatic management according to the parameters of the leakage data. Therefore, the prior art urgently needs to be further improved. Summary of the Invention

[0005] In view of the above problems, the present invention provides a chemical production method and system for metal smelting. It realizes the monitoring of the generation of reduction parameters of a blast furnace device through a three-layer structure of the blast furnace device, the grounding end of the blast furnace device, and the tuyere, generates a monitoring value of the reduction parameters of the blast furnace device, obtains the reduction parameters, and based on the reduction parameters, determines the reducing agent storage tank of the affected flue gas denitrification device. The monitoring nodes on the reducing agent storage tank of the flue gas denitrification device monitor the current leakage characteristics and generate correction parameters to offset the leakage characteristics.

[0006] The object of the invention of the present application can be achieved by the following technical means:

[0007] A chemical production method for metal smelting, comprising the following steps:

[0008] Step 1: A plurality of first sensors are arranged equidistantly on the blast furnace equipment, at least one second sensor is arranged at the grounding end of the blast furnace equipment, and a plurality of reference electrodes are distributed on the tuyere;

[0009] Step 2: n monitoring nodes are arranged on n reductant storage tanks of the flue gas denitrification device, and any one monitoring node controls a spraying device;

[0010] Step 3: In a communication cycle, n monitoring nodes access the node coordinator, and the node coordinator generates a first response table based on the corrected reduction parameters;

[0011] Step 4: The first sensor periodically collects the first data K, the second sensor periodically collects the second data P, and sends the first data K and the second data P to the data processing unit;

[0012] Step 5: If the first data K > Tk and the second data P > Tp, then go to Step 6, otherwise, return to Step 4, where Tk is the first data threshold and Tp is the second data threshold;

[0013] Step 6: The data processing unit generates reduction parameters and sends the reduction parameters to the node coordinator, and the node coordinator generates a second response table and sends it to n monitoring nodes;

[0014] Step 7: n monitoring nodes form a network, elect m monitoring nodes, and m monitoring nodes send the ammonia concentration U to the node coordinator, where m < n;

[0015] Step 8: If the ammonia concentration U > Tu, then go to Step 9, otherwise, return to Step 4, where m < n and Tu is the leakage threshold, which is the minimum reduction parameter for the reductant storage tank to generate leakage;

[0016] Step 9: m monitoring nodes send the ammonia concentration U to m spraying devices, and m spraying devices generate correction parameters;

[0017] Step 10: The spraying device sends the corrected reduction parameters to the node coordinator and returns to Step 7.

[0018] In the present invention, the first data K is the ammonia concentration on the blast furnace equipment, and the second data P is the potential difference between the grounding end of the blast furnace equipment and the reference electrode.

[0019] In the present invention, the reduction parameters include the intensity, frequency, and waveform characteristics of the reduction parameters.

[0020] In the present invention, the calibration reduction parameter is a three-dimensional vector, and any component of the three-dimensional vector corresponds to the intensity and direction of the temperature in one direction.

[0021] In the present invention, the first response table contains the position codes of each monitoring node in the flue gas denitrification device, and any one of the position codes corresponds to a calibration reduction parameter respectively.

[0022] In the present invention, the first response table and the calibration reduction parameter form a training set. The node coordinator inputs the first data K and the second data P into the training set based on the neural network, and predicts the position codes of each monitoring node affected by the reduction parameter in the second response table.

[0023] In the present invention, the second response table contains the position codes of each monitoring node affected by the reduction parameter, and any one of the position codes corresponds to a reduction parameter respectively.

[0024] In the present invention, in step 7, n monitoring nodes receive the second response table, establish an Ad-Hoc network, and perform a primary network formation. Any one monitoring node sends a confirmation message to its neighbor nodes. Set a time threshold. If any one monitoring node fails to receive the feedback message from its neighbor node within the time limit, it disconnects from that neighbor node and performs a secondary network formation with the neighbor nodes that have received the feedback message. All the monitoring nodes included in the secondary network formation are the m monitoring nodes recommended.

[0025] A production system for implementing a chemical production method for metal smelting, including a first sensor, a second sensor, a reference electrode, a data processing unit, monitoring nodes, a node coordinator, and a spraying device, wherein,

[0026] The first sensor is a temperature sensor, and the second sensor is a potentiometer;

[0027] The reference electrode is a metal material with a potential of 0 and electrical conductivity. The reference electrode is connected to the first sensor and the second sensor through a measurement terminal;

[0028] The monitoring node is a Hall current sensor with a communication function, and is configured to periodically collect the ammonia concentration on the reductant storage tank of the flue gas denitrification device and communicate with the node coordinator;

[0029] The data processing unit is configured to obtain the data generated by the first sensor and the second sensor, and perform data preprocessing and send it to the node coordinator. The data preprocessing includes data cleaning, removing outliers, and normalization processing.

[0030] The node coordinator is configured to receive the ammonia concentration sent by the monitoring node, and the node coordinator also has a classifier;

[0031] The classifier generates prediction data based on the first data K and the second data P;

[0032] The injection device is configured to generate correction parameters with controllable temperature direction and intensity.

[0033] Implementing a chemical production method and system for metal smelting according to the present invention, the beneficial effects are as follows: The technical solution of the present invention combines the collaborative work of the data processing unit and the node coordinator. The system can generate accurate reduction parameters based on the data collected by the sensors, realize dynamic regulation, ensure the efficient progress of the reduction reaction, and can trigger an alarm in time and correct relevant parameters when leakage occurs to prevent the further expansion of danger. In addition, for the problem of flue gas denitrification during the smelting process, the present invention precisely controls the injection amount of the reducing agent by using the injection device, and optimizes the concentration gradient, temperature distribution and pressure gradient in the reaction environment by correcting the reduction parameters, effectively reducing the emission of nitrogen oxides and significantly improving the flue gas denitrification efficiency. In addition, the present invention can accurately locate the key nodes in the flue gas denitrification device, reasonably allocate the usage amount of the reducing agent, and avoid waste of resources. At the same time, the real-time monitoring and correction of the concentrations of liquid ammonia and urea solution further improve the utilization efficiency of the reducing agent, thereby reducing the smelting cost. Description of the Drawings

[0034] Figure 1 It is a flowchart of the chemical production method for metal smelting according to the present invention;

[0035] Figure 2 It is a schematic diagram of the arrangement of the reduction parameter monitoring sensor and the reference electrode according to the present invention;

[0036] Figure 3 It is a schematic diagram showing that the coverage range of the reduction parameters in the flue gas denitrification device according to the present invention involves a first region and a second region;

[0037] Figure 4 It is a flowchart of the method for the node coordinator to judge the first region and the second region according to the present invention;

[0038] Figure 5 It is a hardware block diagram of the production system for implementing the chemical production method for metal smelting according to the present invention. Detailed Embodiments

[0039] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention.

[0040] Embodiment 1

[0041] The blast furnace smelting is carried out in a high-temperature environment, and a large amount of reducing agents (such as liquid ammonia, urea solution) are used for denitrification reactions. Due to inaccurate process control, the emissions of nitrogen oxides (NOx) often exceed the standards. Considering that parameters such as ammonia concentration, potential difference, and temperature gradient in the smelting reaction process have an important impact on the efficiency of the reduction reaction. It should be understood that the flue gas denitrification device and the monitoring system involved in this embodiment together form a separate one-pole circuit, and the load of the flue gas denitrification device is greater than that of the monitoring system. The monitoring system can be understood as the minimum system for maintaining the operation of the blast furnace equipment in this embodiment. This embodiment details a chemical production method for metal smelting, referring to Figure 1 , including the following steps:

[0042] Step 1: Referring to Figure 2 , a plurality of first sensors are arranged equidistantly on the blast furnace equipment, at least one second sensor is arranged at the grounding end of the blast furnace equipment, and a plurality of reference electrodes are distributed on the tuyeres. In this embodiment, the first sensor cannot directly obtain the first data K. The ammonia concentration on the blast furnace equipment is collected in real time by the first sensor, and after filtering and feature extraction, the features of the reduction parameters are retained. The second sensor cannot directly obtain the origin position of the reduction parameters. By comparing with a plurality of reference electrodes distributed on the tuyeres, the potential difference between the reference electrode and the reduction parameter is used to identify the interval where the reduction parameter originates. The tuyere described in this embodiment can be understood as any material made of metal below the blast furnace equipment.

[0043] Step 2: n monitoring nodes are arranged on the n reductant storage tanks of the flue gas denitrification device, and any one of the monitoring nodes controls a spraying device. In this embodiment, the spraying device is arranged on the reductant storage tank. The monitoring node can communicate with the spraying device through a wireless sensor network and call the spraying device to work through a control instruction. Specifically, if the distance between the reductant storage tanks is too far to meet the coverage condition of the wireless sensor network, preferably, the CAN bus communication method can be used for control.

[0044] Step 3: In a communication cycle, the n monitoring nodes access the node coordinator, and the node coordinator generates a first response table based on the calibrated reduction parameters. Among them, the calibrated reduction parameter is a three-dimensional vector, and any one component of the three-dimensional vector corresponds to the intensity and direction of the temperature in one direction. The calibrated reduction parameter = (Cx, Cy, Cz), where Cx represents the calibrated parameter component in the x direction, Cy represents the calibrated parameter component in the y direction, and Cz represents the calibrated parameter component in the z direction. In this embodiment, the first response table contains the position codes of each monitoring node in the flue gas denitrification device, and any one of the position codes corresponds to a calibrated reduction parameter respectively.

[0045] Step 4: The first sensor periodically collects the first data K, and the second sensor periodically collects the second data P, and sends the first data K and the second data P to the data processing unit. Herein, the first data K is the ammonia concentration on the blast furnace equipment, and the second data P is the potential difference between the grounding end of the blast furnace equipment and the reference electrode. In this embodiment, the potential between the reference electrode and the ground is measured in advance by a potential measuring device to obtain a potential reference value, and taking this as a benchmark, the potential difference between the second data P and the potential reference value is compared, and this potential difference constitutes the second data P.

[0046] Step 5: If the first data K > Tk and the second data P > Tp, then enter Step 6; otherwise, return to Step 4. Herein, Tk is the first data threshold, and Tp is the second data threshold. In this embodiment, the first data threshold Tk and the second data threshold Tp are the results of prior calculation. During the prior calculation process, Tk = aP 3 + bcosP + ce -p , where a, b, and c are undetermined coefficients determined by prior measurement.

[0047] Step 6: The data processing unit generates reduction parameters and sends the reduction parameters to the node coordinator. The node coordinator generates a second response table and sends it to n monitoring nodes. Herein, the reduction parameters include the intensity, frequency, and waveform characteristics of the reduction parameters. In this embodiment, the second response table contains the position codes of each monitoring node affected by the reduction parameters, and any one of the position codes corresponds to a reduction parameter respectively.

[0048] In this embodiment, the first response table and the calibration reduction parameters form a training set. The node coordinator inputs the first data K and the second data P into the training set based on a neural network, and predicts the position codes of each monitoring node affected by the reduction parameters in the second response table.

[0049] Step 7: The n monitoring nodes form a network and elect m monitoring nodes. The m monitoring nodes send the ammonia concentration U to the node coordinator, where m < n. In this embodiment, the n monitoring nodes receive the second response table and establish an Ad-Hoc network for one-time networking. Any monitoring node and its neighbor nodes send confirmation messages to each other. A time threshold is set. If any monitoring node does not receive the feedback message from its neighbor node within the time limit, then it disconnects from that neighbor node and conducts secondary networking with the neighbor nodes that have received the feedback message. All the monitoring nodes included in the secondary networking are the elected m monitoring nodes.

[0050] Step 8: If the ammonia concentration U > Tu, proceed to Step 9; otherwise, return to Step 4, where m < n, Tu is the leakage threshold, which is the minimum reduction parameter for the leakage of the reductant storage tank. In this embodiment, the calculation of the leakage threshold Tu is determined through prior measurement, combined with the working state of the actual smelting equipment and the usage conditions of the reductant storage tank, and optimized with specific monitoring data. In different reductant storage tank devices, due to differences in operating conditions such as volume, storage medium, temperature, and pressure, the leakage threshold Tu will also vary. Therefore, during the initial equipment commissioning stage, it is necessary to calibrate the leakage threshold for each device through experimental testing. To achieve this process, the present invention adopts a dynamic calibration method. Specifically, a set of initial thresholds Tu is set, and the change in ammonia concentration under different operating conditions is simulated through experiments. By comparing the relationship between the ammonia concentration and the leakage situation of the reductant storage tank under different conditions, the data acquisition and processing unit is used to perform correlation analysis on the ammonia concentration and the actual data of leakage occurrence under different working conditions, and a machine learning algorithm is used to model the data, thereby calculating the optimal leakage threshold Tu for each condition.

[0051] Step 9: m monitoring nodes send the ammonia concentration U to m injection devices, and the m injection devices generate correction parameters. In this embodiment, after the ammonia concentration U is injected into the reductant storage tank, the distribution of the leakage temperature in terms of position and direction in space is fixed. The injection devices need to generate a temperature with a correction effect based on the relative relationship between the position and the leakage temperature. It should be understood that the reductant storage tanks of the flue gas denitrification device are scattered in multiple regions, and the correction parameters generated by the m injection devices will not be superimposed.

[0052] Step 10: The injection devices send the corrected reduction parameters to the node coordinator, and return to Step 7. In this embodiment, the node coordinator can generate multiple sets of prediction data, where the prediction data is the temperature direction and temperature magnitude of the leakage temperature of the reductant storage tank after the ammonia concentration U is injected into the reductant storage tank of the flue gas denitrification device.

[0053] Embodiment 2

[0054] In a specific embodiment of the method for monitoring the reduction parameters of the blast furnace equipment of the present invention and managing the leakage of the reductant storage tank of the flue gas denitrification device caused thereby, after the blast furnace equipment generates reduction parameters, the first sensor, the second sensor, and the reference electrode arranged can be used to extract the ammonia concentration and the source point of the reduction parameter generation. Refer to Figure 3 , based on the said source point, the coverage range of the reduction parameter in the flue gas denitrification device involves the first region, and the coverage range of the influence leakage effect of the reduction parameter on the reductant storage tank of the flue gas denitrification device involves the second region. The node coordinator receives the ammonia concentration and is used to judge the first region and the second region. In this embodiment, a method for the node coordinator to judge the first region and the second region is described in detail. Refer toFigure 4 , including the following steps:

[0055] Step 101: The node coordinator receives multiple groups of calibration and restoration parameters, converts the multiple groups of calibration and restoration parameters into a matrix, where the rows of the matrix are a group of calibration and restoration parameters, and the columns of the matrix are the features corresponding to the calibration and restoration parameters.

[0056] Step 102: Input the position information of all reductant storage tanks in the flue gas denitrification device to form a reductant storage tank position database.

[0057] Step 103: The node coordinator extracts the first data K and the second data P corresponding to multiple groups of calibration and restoration parameters, and obtains the position information of the reductant storage tanks that generate leaked reductant corresponding to multiple groups of temperature parameters.

[0058] Step 104: The multiple groups of position information of the reductant storage tanks that generate leaked reductant form a tensor, and the multiple groups of the first data K and the second data P form a multi-dimensional array, jointly forming a three-dimensional tensor, where the first dimension is the number of samples, the second dimension is the number of features of the first data K, the second data P, and the position information, and the third dimension is the time step. Specifically, if time series data is involved, a fourth dimension can be added to represent the time step.

[0059] Step 105: Construct a neural network model through a multi-layer perceptron, and form a data set with the three-dimensional tensor and the matrix obtained in the above steps, where the data set includes a training set and a test set. Preferably, the train_test_split function is used to divide the data set.

[0060] Step 106: The trained neural network model performs weight optimization based on gradient descent. Preferably, cross-validation is used to optimize the model. Preferably, an appropriate loss function can be selected during the training process to perform quadratic fitting on the training data.

[0061] Step 107: Input the ammonia concentration monitored during implementation, output the position information of the reductant storage tanks affected by the restoration parameters, and generate the coverage range of the first area; convert the ammonia concentration into temperature parameters according to a priori rules, input the temperature parameters into the neural network model, and output the position information of the reductant storage tanks that generate leaked reductant, and generate the coverage range of the second area.

[0062] In this embodiment, in step 104, multiple groups of position information are imported into the Numpy library for numerical and matrix calculations. The generated variable stores the position information of the reference electrode. Based on the array function, the multiple groups of position information and the position information of the reference electrode are output as a tensor, preferably a tensor of (3, num_samples), where 3 represents the number of features and num_samples represents the number of samples. The array function further organizes the tensor to obtain (U, num_samples), where U represents the position information of the leakage-producing reducing agent storage tank covered by the reduction parameters involved.

[0063] Embodiment III

[0064] A production system for a chemical production method for metal smelting, referring to Figure 5 , includes a first sensor, a second sensor, a reference electrode, a monitoring node, a node coordinator, and a spraying device. Among them, the first sensor is a temperature sensor, and the second sensor is a potentiometer. The reference electrode is a metal material with a potential of 0 and electrical conductivity. The reference electrode is connected to the first sensor and the second sensor through a measurement terminal. The monitoring node is a Hall current sensor with communication functions, which is configured to periodically collect the ammonia concentration on the reducing agent storage tank of the flue gas denitrification device and communicate with the node coordinator;

[0065] The data processing unit is configured to obtain the data generated by the first sensor and the second sensor, and perform data preprocessing and send it to the node coordinator. The data preprocessing includes data cleaning, outlier removal, and normalization processing.

[0066] The node coordinator is configured to receive the ammonia concentration sent by the monitoring node. The node coordinator also has a classifier. The classifier generates prediction data based on the first data K and the second data P.

[0067] The spraying device is configured to generate correction parameters with controllable temperature direction and intensity.

[0068] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A chemical production method for metal smelting, characterized in that, It includes the following steps: Step 1: A plurality of first sensors are equidistantly arranged on the blast furnace equipment, at least one second sensor is arranged at the grounding end of the blast furnace equipment, and a plurality of reference electrodes are distributed on the tuyere; Step 2: n monitoring nodes are arranged on n reductant storage tanks of the flue gas denitrification device, and any one of the monitoring nodes controls a spraying device; Step 3: In a communication cycle, n monitoring nodes access the node coordinator, and the node coordinator generates a first response table based on the corrected reduction parameters; Step 4: The first sensor periodically collects the first data K, the second sensor periodically collects the second data P, and sends the first data K and the second data P to the data processing unit; Step 5: If the first data K>Tk and the second data P>Tp, then go to Step 6; otherwise, return to Step 4, where Tk is the first data threshold and Tp is the second data threshold; Step 6: The data processing unit generates reduction parameters and sends the reduction parameters to the node coordinator. The node coordinator generates a second response table and sends it to n monitoring nodes; Step 7: n monitoring nodes form a network, elect m monitoring nodes, and the m monitoring nodes send the ammonia concentration U to the node coordinator, where m<n; Step 8: If the ammonia concentration U>Tu, then go to Step 9; otherwise, return to Step 4, where m<n, Tu is the leakage threshold, which is the minimum reduction parameter for the reductant storage tank to generate leakage; Step 9: The m monitoring nodes send the ammonia concentration U to the m spraying devices, and the m spraying devices generate correction parameters; Step 10: The spraying devices send the corrected reduction parameters to the node coordinator and return to Step 7.

2. The management method for monitoring the reduction parameters of a blast furnace equipment and the leakage of a reducing agent storage tank in a flue gas denitrification device caused thereby, characterized in that, The first data K is the ammonia concentration on the blast furnace equipment, and the second data P is the potential difference between the grounding end of the blast furnace equipment and the reference electrode.

3. The chemical production method for metal smelting according to claim 1, characterized in that, The reduction parameters include the concentrations of liquid ammonia and urea solution in the reductant storage tank.

4. The chemical production method for metal smelting according to claim 1, characterized in that, The corrected reduction parameter is a three-dimensional vector, and any one of the components of the three-dimensional vector corresponds to a key parameter of the reaction environment. The key parameters are respectively the concentration gradient of the reducing atmosphere, the temperature distribution gradient, and the pressure distribution gradient.

5. The chemical production method for metal smelting according to claim 1, characterized in that, The first response table contains the position codes of each monitoring node in the flue gas denitrification device, and any one of the position codes corresponds to a corrected reduction parameter respectively.

6. The chemical production method for metal smelting according to claim 5, characterized in that, The first response table and the corrected reduction parameters form a training set. The node coordinator inputs the first data K and the second data P into the training set based on the neural network, and predicts the position codes of each monitoring node affected by the reduction parameters in the second response table.

7. The chemical production method for metal smelting according to claim 1, characterized in that, The second response table contains the position codes of each monitoring node affected by the reduction parameters, and any one of the position codes corresponds to a reduction parameter respectively.

8. The chemical production method for metal smelting according to claim 1, characterized in that, In Step 7, n monitoring nodes receive the second response table, establish an Ad-Hoc network, and conduct a primary network formation. Any one of the monitoring nodes sends a confirmation message to its neighbor nodes. Set a time threshold. If any one of the monitoring nodes fails to receive the feedback message from the neighbor node within the timeout period, then disconnect from the neighbor node and conduct a secondary network formation with the neighbor nodes that receive the feedback message. All the monitoring nodes included in the secondary network formation are the elected m monitoring nodes.

9. A production system for a chemical production method of metal smelting according to claim 1, characterized in that, It includes a first sensor, a second sensor, a reference electrode, a monitoring node, a node coordinator, and an injection device. Among them, the first sensor is a temperature sensor and the second sensor is a potentiometer; the reference electrode is a metal material with a potential of 0 and electrical conductivity, and the reference electrode connects the first sensor and the second sensor through a measurement terminal; the monitoring node is a Hall current sensor with communication function, which is configured to periodically collect the ammonia concentration on the reductant storage tank of the flue gas denitrification device and communicate with the node coordinator; The data processing unit is configured to obtain the data generated by the first sensor and the second sensor, and perform data preprocessing and send it to the node coordinator. The data preprocessing includes data cleaning, outlier removal, and normalization processing. The node coordinator is configured to receive the ammonia concentration sent by the monitoring node, and the node coordinator also has a classifier; The classifier generates prediction data based on the first data K and the second data P; The injection device is configured to generate correction parameters with controllable temperature direction and intensity.

Images