Intelligent control method and system for a porous media burner

Abstract

The application discloses an intelligent control method and system of a porous medium combustor, and belongs to the technical field of automatic control. Firstly, operation and state parameters and experimental data of the porous medium combustor during normal work are collected as reference historical data; the operation and state parameters are collected; an importance analysis model is established, the correlation degree between power demand and the number of combustion modules is analyzed, and the importance factor of each combustion module is determined; the correlation degree between the state parameters and the operation parameters of each combustion module is analyzed, and the importance factor of each operation parameter is determined. When the state parameters exceed a preset threshold value, the current operation and state parameters and the reference historical data are analyzed, the state parameters at the next time step are predicted, when the deviation between the prediction result and the reference historical data exceeds a preset threshold value, the operation parameter with the largest importance factor is adjusted, and the adjustment amount is calculated according to the optimal regression principle. The application can maximize the stable combustion of the porous medium combustor, and reduce the emission of NOx and CO.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of automatic control, and particularly relates to an intelligent control method and system for a porous medium combustor. BACKGROUND

[0002] Efficient and clean use of energy has been the goal pursued by all countries. The combined heat and power technology can simultaneously produce "electricity" and "heat", realizes the cascade utilization of energy, improves the energy utilization efficiency, and becomes one of the main directions of the development of coal-fired thermal systems. The unit realizing the combined heat and power is called a heat supply unit.

[0003] In order to reduce carbon emissions, renewable energy is connected to the grid on a large scale, and its intermittency causes great threat to the safety of the power grid. In order to accommodate renewable energy, the heat supply unit must participate in flexible peak regulation. However, due to the difference in thermal inertia between the boiler and the steam turbine, the rapid load change causes strong fluctuation of the unit parameters, seriously affecting the safety and efficiency of the unit operation, and therefore it is urgent to improve the flexible peak regulation capacity of the heat supply unit. SUMMARY

[0004] In order to solve the defects existing in the prior art, the purpose of the present application is to provide an intelligent control method and system for a porous medium combustor, which can maximize the stable combustion of the porous medium combustor and reduce the emission of NOx and CO.

[0005] The present application is realized by the following technical solutions:

[0006] An intelligent control method for a porous medium combustor, comprising:

[0007] S1: collecting the operating parameters and state parameters obtained from the historical experiments under the normal operating state of the porous medium combustor as reference historical data;

[0008] S2: collecting the operating parameters and state parameters of each combustion module inside the porous medium combustor as real-time data;

[0009] S3: based on a machine learning algorithm, establishing an importance analysis model of the power demand and the number of combustion modules, and analyzing the influence of the operating parameters of each combustion module on the porous medium combustor;

[0010] S4: when the state parameters of the combustion module are monitored to exceed the preset threshold value, a long-short time prediction neural network algorithm is used to retrieve the reference historical data of the porous medium combustor and compare and analyze the real-time data, predict the state parameters at the next time step, and when the deviation between the prediction result and the reference historical data is greater than the preset threshold value, adjust the number of operating modules and the operating parameters of the porous medium combustor;

[0011] S5: According to the optimal regression principle, the adjustment amount of the operating parameter is calculated, and the operating parameter of the combustion module is adjusted, and after the adjustment, it returns to S2 to start the next cycle.

[0012] Preferably, in S1, the operating parameters include operating power, enabled module number, inlet gas temperature, fuel flow and air flow; and the state parameters include combustion temperature, oxygen content at the upper part of the combustion module, NOx content at the upper part of the combustion module, CO content at the upper part of the combustion module and noise.

[0013] Preferably, in S2, the collected operating parameters and state parameters of each combustion module inside the porous medium burner automatically form an engineering file and are saved to a reference history database; and in S3, a working log file generated by executing a machine learning algorithm is saved to the reference history database.

[0014] Preferably, S3 specifically comprises: applying a KNN algorithm to analyze the correlation between power demand and the number of combustion modules, finding the closest rated power and the number of combustion modules matched with the rated power under the specified power; and applying a random forest algorithm to analyze the correlation between the state parameters and the operating parameters, determining the importance factor of each operating parameter and finding the operating parameter with the largest importance factor by judging the influence of the operating parameters of each combustion module on the state parameters of the combustion module corresponding to each tree in the random forest.

[0015] The application discloses an intelligent control system of a porous medium burner.

[0016] A first collection unit collects operating parameters and state parameters obtained in a normal operating state of the porous medium burner and in historical experiments as reference historical data;

[0017] A second collection unit collects operating parameters and state parameters of each combustion module inside the porous medium burner as real-time data;

[0018] A calculation unit establishes an importance analysis model of power demand and the number of combustion modules based on a machine learning algorithm, and analyzes the influence of the operating parameters of each combustion module on the porous medium burner;

[0019] An analysis unit adopts a long-short term prediction neural network algorithm to analyze the reference historical data and real-time data of the porous medium burner when the state parameters of the combustion module exceed a preset threshold value, predicts the state parameters at the next time step, and adjusts the operating parameters when the deviation between the prediction result and the reference historical data is greater than a preset threshold value, and transmits the analysis result to an adjustment unit;

[0020] An adjustment unit calculates the adjustment amount of the operating parameter according to the optimal regression principle, and adjusts the operating parameters of the combustion adjustment unit and the delivery adjustment unit of the combustion module;

[0021] a combustion unit, which realizes premixed combustion of fuel and air;

[0022] a conveying unit, which adjusts and conveys fuel, air and flue gas.

[0023] Preferably, the first collecting unit comprises a memory card used with a high-speed single-chip microcomputer; the second collecting unit comprises a thermocouple, an infrared temperature detector, an electromagnetic flowmeter, a flue gas analyzer and a noise detector; the computing unit, the analyzing unit and the adjusting unit share devices, which comprise a high-speed information collecting card, a high-speed single-chip microcomputer, a computer and a display; the combustion unit comprises an air inlet pipe, a spring electromagnetic valve, a premixing cavity, a burner shell, an electronic ignition hole, heat preservation cotton, a heat insulation plate and a combustion module; the conveying unit comprises a pipeline, a flow controller, an oxygen amount controller, a spring electromagnetic valve and a pressure reducing valve.

[0024] Further preferably, the combustion unit comprises a plurality of combustion modules, and the number of combustion modules enabled according to power matching.

[0025] Further preferably, the number of combustion modules is 2-4.

[0026] Further preferably, the combustion module is of a foam ceramic type, a honeycomb type or a metal fiber type; the material is silicon carbide, silicon nitride, zirconia, quartz ceramic, mullite, nickel-based alloy, nickel-chromium-iron alloy or porous tungsten.

[0027] Further preferably, when the combustion module is of a foam ceramic type, the pore density is 30-60 PPI, and the porosity is 0.7-0.9.

[0028] Compared with the prior art, the present application has the following beneficial technical effects:

[0029] The intelligent control method of the porous medium combustor disclosed by the application collects the operation and state parameters of the porous medium combustor when it is working normally and experimental data as reference historical data; collects the operation and state parameters of the combustion module of the porous medium combustor; establishes an importance analysis model, analyzes the correlation degree of power demand and the number of combustion modules, and determines the importance factor of each combustion module; analyzes the correlation degree of the state parameters and the operation parameters of each combustion module, determines the importance factor of each operation parameter, and selects the maximum operation parameter. When the state parameter exceeds the preset threshold value, the current operation and state parameters and the reference historical data are analyzed by using a long-short time prediction neural network to predict the next time step state parameter, and when the deviation of the prediction result and the reference historical data exceeds the preset threshold value, the operation parameter with the largest importance factor is adjusted, and the adjustment amount is calculated according to the optimal regression principle. The application is based on the historical experimental data and the field working condition data of the porous medium combustor, continuously updates the demand under different working conditions during operation, avoids the error estimation caused by the pre-set related parameters and reference values, and sets the combustion temperature, NOx content, CO content, oxygen content and noise of each combustion module according to the actual working condition demand. At the same time, different parameter input scales are pre-set for parameter collection under different time scales; the long-short time sequence neural network algorithm can only ensure the stability of the time sequence tracking trend when the parameters produce large fluctuations, and cannot realize zero-error time sequence tracking system; when the parameters produce large fluctuations, the error caused by the jitter is balanced by the weighted method, and when the running trend is consistent with the actual working condition state, the difference between the actual value and the predicted value is reduced according to the corresponding weight. The application can maximize the stable combustion of the porous medium combustor and reduce the emission of NOx and CO.

[0030] The intelligent control system of the porous medium combustor disclosed by the application has simple structure, high automation degree, compatibility with existing system hardware, low cost and wide application range. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 It is a flow chart of the method according to the application;

[0032] Figure 2 It is a principle diagram of the KNN algorithm;

[0033] Figure 3 It is a principle diagram of the random forest algorithm;

[0034] Figure 4 It is a principle diagram of the long-short time prediction neural network algorithm;

[0035] Figure 5 It is a logic diagram of the intelligent control method of the porous medium combustor in the embodiment;

[0036] Figure 6System structure diagram of the intelligent control system of the porous medium burner in the embodiment;

[0037] Figure 7 Structure schematic diagram of the combustion unit of the intelligent control system of the porous medium burner.

[0038] In the figure: 1 is an air inlet pipe, 2 is a spring electromagnetic valve, 3 is a premixing cavity, 4 is a burner shell, 5 is an electronic ignition hole, 6 is thermal insulation cotton, 7 is a heat insulation plate, and 8 is a combustion module. DETAILED DESCRIPTION

[0039] The present application will be further described in detail below in combination with the accompanying drawings and specific embodiments, which are an explanation of the present application rather than a limitation.

[0040] Comprise:

[0041] S1: Collect the operating parameters and state parameters obtained from historical experiments under the normal operating state of the porous medium burner as reference historical data; in S1, the operating parameters include operating power, enabled module number, air inlet temperature, fuel flow and air flow; and the state parameters include combustion temperature, oxygen content at the upper part of the combustion module, NOx content at the upper part of the combustion module, CO content at the upper part of the combustion module and noise.

[0042] S2: Collect the operating parameters and state parameters of each combustion module inside the porous medium burner as real-time data respectively; in S2, the collected operating parameters and state parameters of each combustion module inside the porous medium burner automatically form an engineering file and are saved into the reference historical database; and in S3, the working log file generated by executing the machine learning algorithm is saved into the reference historical database.

[0043] S3: Based on the machine learning algorithm, an importance analysis model of power demand and combustion module number is established, and the influence of the operating parameters of each combustion module on the porous medium burner is analyzed; S3 specifically comprises: applying the KNN algorithm to analyze the correlation degree of power demand and combustion module number, finding out the closest rated power and the combustion module number matched with the same under the specified power; and applying the random forest algorithm to analyze the correlation degree of the state parameters and the operating parameters, determining the importance factor of each operating parameter and finding out the operating parameter with the largest importance factor by judging the influence of the operating parameters of each combustion module on the combustion module state parameters in each tree of the random forest.

[0044] S4: When it is monitored that the state parameters of the combustion module exceed the preset threshold value, the long-short term prediction neural network algorithm is adopted to retrieve the reference historical data of the porous medium burner and compare and analyze the real-time data, predict the state parameters at the next time step, and adjust the operating module number and operating parameters of the porous medium burner when the deviation of the prediction result from the reference historical data is greater than the preset threshold value.

[0045] S5: Calculate the adjustment amount of the operating parameter according to the optimal regression principle and adjust the operating parameter of the combustion module, and return to S2 to start the next cycle after adjustment.

[0046] The intelligent control system of the porous medium combustor comprises:

[0047] The first acquisition unit acquires the operating parameters and state parameters obtained in historical experiments under the normal operating state of the porous medium combustor as reference historical data.

[0048] The second acquisition unit acquires the operating parameters and state parameters of each combustion module in the porous medium combustor as real-time data.

[0049] The calculation unit establishes an importance analysis model of power demand and the number of combustion modules based on a machine learning algorithm and analyzes the influence of the operating parameters of each combustion module on the porous medium combustor.

[0050] When the state parameter of the combustion module exceeds the preset threshold, the analysis unit analyzes the reference historical data and real-time data of the porous medium combustor by using a long short-term prediction neural network algorithm to predict the state parameter at the next time step, and adjusts the operating parameter when the deviation between the prediction result and the reference historical data is greater than the preset threshold.

[0051] The adjustment unit calculates the adjustment amount of the operating parameter according to the optimal regression principle and adjusts the operating parameters of the combustion adjustment unit and the delivery adjustment unit of the combustion module.

[0052] The combustion unit realizes premixed combustion of fuel and air.

[0053] The delivery unit adjusts and delivers fuel, air and flue gas.

[0054] The first acquisition unit comprises a memory card used in conjunction with a high-speed single-chip microcomputer; the second acquisition unit comprises a thermocouple, an infrared temperature detector, an electromagnetic flowmeter, a flue gas analyzer and a noise detector; the calculation unit, the analysis unit and the adjustment unit share equipment, which comprises a high-speed information acquisition card, a high-speed single-chip microcomputer, a computer and a display; the combustion unit comprises an air inlet pipe 1, a spring electromagnetic valve 2, a premixing cavity 3, a combustor shell 4, an electronic ignition hole 5, thermal insulation cotton 6, a heat insulation plate 7 and a combustion module 8; and the delivery unit comprises a pipeline, a flow controller, an oxygen amount controller, a spring electromagnetic valve and a pressure reducing valve.

[0055] The combustion unit comprises a plurality of combustion modules 8, and the number of combustion modules 8 enabled according to power matching. Preferably, the combustion unit comprises combustion modules, and the number of combustion modules can be 2-4; the form can be a ceramic foam type, a honeycomb type, a metal fiber type; and the material can be silicon carbide, silicon nitride, zirconia, quartz ceramic, mullite, nickel-based alloy, nickel-chromium-iron alloy, and porous tungsten.

[0056] If the porous medium material is a ceramic foam type, the pore density is in the range of 30-60 PPI, and the porosity is in the range of 0.7-0.9.

[0057] The application will be further explained and described below with reference to a specific embodiment:

[0058] As shown in the figure, the intelligent control method of the porous medium burner of the application comprises: Figure 1

[0059] Step S1: The first acquisition unit acquires the operation and state parameters obtained in the normal running state of the porous medium burner and historical experiments, and takes the operation and state parameters as reference historical data; the operation parameters include operation power, inlet air temperature, fuel flow, and air flow; the state parameters include combustion temperature, oxygen content, NOx content, and CO content at the outlet of the burner; in this embodiment, the historical experimental data of the porous medium burner experimental table of the Key Laboratory of Thermal Science and Engineering of the Ministry of Education of Xi'an Jiaotong University are taken as the reference historical data, and preset threshold values are set, including a combustion temperature of 900℃, residual oxygen content of 2%, NOx content of 30mg / (kW·h), CO content of 0.02%(α=1), and noise of 48dB(A).

[0060] Step S2: Based on the intelligent control system of the porous medium burner, the second acquisition unit acquires the operation and state parameters of each combustion module, and the parameter structure is the same as that in S1, and the operation parameters further include the number of modules, and the state parameters further include noise. The data of the same parameters of each combustion module are classified to generate an operation power database, an enabled module database, a temperature database, a flow database, a flue gas composition database, and a noise database. In this embodiment, the number of combustion modules is 4, that is, a maximum of 4 porous medium combustion modules can run at the same time, and each combustion module generates a separate operation and state parameter database.

[0061] Step S3: The calculation unit establishes an importance analysis model between the power demand and the number of combustion modules and the state parameters and operation parameters of the combustion modules based on a machine learning algorithm; the KNN algorithm is applied to analyze the correlation between the power demand and the number of combustion modules, and the closest rated power and the number of combustion modules matched therewith under a specified power are found out. The principle of the KNN algorithm is as follows: Figure 2 ​As shown, the horizontal axis is the number of combustion modules, the vertical axis is the power, K is the number of adjacent points (3 in this example), and the coordinate points in the quadrant are historical experimental data; taking summer and winter operation as examples, the summer burner power demand is low, about 800kW·m -2 , the power parameter is substituted into the KNN algorithm, the nearest three points power and combustion module number are found out, and the required combustion module number is determined to be 2 by voting calculation; the winter burner power demand is high, about 1645kW·m -2 , the calculation method is the same, and the required combustion module number is determined to be 4.

[0062] The random forest algorithm is applied to analyze the correlation degree of the state parameters and the operation parameters, the importance factor of each operation parameter is determined and the operation parameter with the largest importance factor is found out by judging the influence of the operation parameter of each combustion module on the state parameter of the combustion module corresponding to each tree in the random forest. In the random forest algorithm, there are several decision trees, and the number of samples and the number of features are randomly selected for each decision tree, so as to ensure randomness. In this example, the features are selected based on the GINI coefficient, and the selection standard of the GINI coefficient is that each sub-node reaches the highest purity, that is, all observations falling in the sub-node belong to one category, at this time, the GINI coefficient is the smallest, the purity is the highest, and the uncertainty is the smallest. For a general decision tree, assuming that there are N categories in total, and the probability of a sample belonging to the nth category is p n , then the GINI coefficient of the probability distribution is formula (1):

[0063]

[0064] The larger the GINI coefficient, the smaller the difference between the samples, the lower the uncertainty, the more complete the parameter segmentation, and the cleaner. The decision tree diagram constructed in this embodiment is shown in Figure 3 , there are two decision trees of air flow decision tree and fuel flow decision tree, N=4, and the selected features include temperature, CO content, NOx content and noise. Figure 3 This is only a schematic diagram of the method used in this embodiment, and the intelligent control system of the porous medium burner actually includes but is not limited to the decision tree described in Figure 3 .

[0065] In addition, when the external computer and the display are connected, the importance function of the operation parameter with respect to the state parameter is constructed according to the random forest algorithm in machine learning, and the operation parameter importance analysis function of each combustion module is also constructed. The importance column chart is formed and output, and the influence of the operation parameters of each module on the system is analyzed.

[0066] Step S4: In the analysis unit, when it is monitored that the state parameter of the combustion module exceeds the preset threshold value, the running, state parameter of the combustion module is analyzed with the reference historical data by using a long-short time prediction neural network to predict the next time step state parameter, and when the deviation of the predicted result of the combustion module from the reference historical data is greater than a preset threshold value, the running parameter is adjusted. The principle of the long-short time sequence prediction neural network algorithm is as shown in Figure 4 The analysis logic is as shown in Figure 5 X i is a grid input value, which is a state parameter (temperature, NOx content, etc.) obtained by the second acquisition unit at runtime in the present case; Function is a prediction algorithm; P i is the output predicted value, which is also a state parameter. A certain state parameter X t at time t is transmitted to the analysis unit, and the long-short time sequence prediction neural network predicts the predicted value P t of the state parameter at the next time step (the time step length is taken as 0.5 s in the present case) by using the historical data (X0, X1, X2…) of the parameter, and compares the predicted value with the set threshold value to determine whether the running parameter needs to be adjusted. When the parameter does not need to be adjusted, the predicted curve is compared with the real running condition in real time to observe whether the change trend of the predicted value is similar to that of the actual value, and if so, the predicted result is processed by weighting according to the corresponding deviation amplitude (the weight is obtained from the optimal regression algorithm), and if not, the grid search method is used to adjust the hyperparameters, update the historical data, and re-analyze. In addition, when the external computer and display are connected, the analysis unit also establishes the running state parameter time sequence curve of the control system. At the same time, according to the analysis result obtained by the importance function, an important influence running parameter and system state parameter time sequence comparison analysis diagram is established to intuitively compare and evaluate the analysis result of the importance influence.

[0067] Step S5: The running parameter with the largest importance factor is adjusted, the adjustment amount is calculated according to the optimal regression principle, and the state parameter of the combustion module is continuously monitored by the monitoring unit after the adjustment. The present application takes the running state parameter reference historical data set as the standard, uses the optimal regression algorithm in machine learning, draws the running state trend curve of the porous medium burner, and outputs the corresponding weight parameters and intercept parameters (r1,...rn, b).

[0068] Figure 6 is a structural diagram of the intelligent control system of the porous medium burner of the present application, and the system comprises:

[0069] The first acquisition unit is used for acquiring operation and state parameters obtained in a normal running state of the porous medium burner and historical experiments, and the operation and state parameters are taken as reference historical data; preferably, the operation parameters acquired by the first acquisition unit include operation power, inlet air temperature, fuel flow and air flow; and the state parameters include combustion temperature, oxygen content, NOx content and CO content at the outlet of the burner.

[0070] The second acquisition unit is used for acquiring real-time operation and state parameters of each combustion module in actual operation; preferably, the second acquisition unit acquires operation and state parameters of each combustion module, the operation parameters include the number of combustion modules, operation power, inlet air temperature, fuel flow and air flow; and the state parameters include combustion temperature, oxygen content, NOx content, CO content and noise.

[0071] The first acquisition unit and the second acquisition unit in the intelligent control system of the porous medium burner classify data of the same type of parameters acquired, generate a database of operation and state parameters of the intelligent control system of the porous medium burner, and deliver the classified parameters to the calculation unit, including an operation power database, an enabled module database, a temperature database, a flow database, a flue gas composition database and a noise database.

[0072] The equipment of the first acquisition unit and the second acquisition unit includes a thermocouple, an infrared temperature detector, an electromagnetic flowmeter, a flue gas analyzer, a memory card and a noise detector.

[0073] The calculation unit is used for establishing an importance analysis model between power demand and the number of combustion modules and state parameters and operation parameters of the combustion modules based on a machine learning algorithm; a KNN algorithm is applied to analyze the correlation degree between power demand and the number of combustion modules, to find out the closest rated power and the number of combustion modules matched with the closest rated power under a specified power; a random forest algorithm is applied to analyze the correlation degree between state parameters and operation parameters, to determine the importance factor of each operation parameter and find out the operation parameter with the largest importance factor by judging the influence of the operation parameter of each combustion module on the state parameters of the combustion module corresponding to each tree in the random forest, and to deliver the calculation result to the analysis unit.

[0074] Preferably, after each operation, the calculation unit forms an engineering file of operation and state parameters of each combustion module acquired by the first acquisition unit and the second acquisition unit, saves the engineering file to the reference historical data and stores the engineering file in the single-chip microcomputer, and the engineering file further includes a work log file generated by executing the machine learning algorithm during operation of each combustion module.

[0075] The analysis unit analyzes the operation and state parameters of the combustion module and historical reference historical data using a long-short time prediction neural network when it is monitored that the state parameters of the combustion module exceed a preset threshold, and predicts the next time step state parameters, and when the deviation between the prediction result of the combustion module and the reference historical data is greater than the deviation exceeds a preset threshold, the analysis result is transmitted to the adjustment unit;

[0076] The adjustment unit adjusts the operation parameter with the largest importance factor, calculates an adjustment amount according to the optimal regression principle, and transmits the adjustment amount to the devices included in the conveying unit and the combustion unit.

[0077] The devices commonly used by the calculation unit, the analysis unit and the adjustment unit include a high-speed information acquisition card, a high-speed single-chip microcomputer, a computer, a display and the like.

[0078] The combustion unit is used to realize fuel and air premixed combustion, and includes, for example, an air inlet pipe 1, a spring electromagnetic valve 2, a premixing cavity 3, a burner shell 4, an electronic ignition hole 5, heat preservation cotton 6, a heat insulation plate 7, a combustion module 8 and the like. Figure 7

[0079] Preferably, the combustion unit includes four combustion modules, and the combustion modules are in the form of foam ceramic porous media, the material can be silicon carbide, the pore density is 60 PPI, and the porosity is 0.8.

[0080] The conveying unit is used to adjust and convey fuel, air and flue gas, and includes, for example, a pipeline, a flow controller, an oxygen amount controller, a spring electromagnetic valve, a pressure reducing valve and the like.

[0081] It should be noted that the above only describes part of the embodiments of the present application, and equivalent changes made to the system described in the present application are included in the protection scope of the present application. Those skilled in the art can make similar substitutions to the described specific examples, as long as they do not deviate from the structure of the present application or exceed the scope defined by the present claims, and they are within the protection scope of the present application.​

Claims

1. A smart control method for a porous media burner, characterized in that, include: S1: Collect operating parameters and status parameters obtained from historical experiments under normal operating conditions of the porous media burner as reference historical data; S2: Collect the operating parameters and status parameters of each combustion module inside the porous media burner as real-time data; S3: Based on machine learning algorithms, establish an importance analysis model of power demand and the number of combustion modules, and analyze the impact of the operating parameters of each combustion module on the porous media burner; S3 specifically involves: applying the KNN algorithm to analyze the correlation between power demand and the number of combustion modules, finding the closest rated power and the number of combustion modules that match it under a given power; applying the random forest algorithm to analyze the correlation between state parameters and operating parameters, determining the importance factor of each operating parameter by judging the influence of the operating parameters of each combustion module on the state parameters of the corresponding combustion module on each tree in the random forest, and finding the operating parameter with the largest importance factor; S4: When the state parameters of the combustion module exceed the preset threshold, a long-short-time prediction neural network algorithm is used to retrieve the reference historical data of the porous media burner and compare it with the real-time data to predict the state parameters for the next time step. When the deviation between the prediction result and the reference historical data is greater than the preset threshold, the number of operating modules and operating parameters of the porous media burner are adjusted. Specifically, the long-short-time prediction neural network uses the historical data of this parameter to make predictions and gives the predicted value P of the state parameter for the next time step. t The predicted curve is compared with the set threshold to determine whether the operating parameters need to be adjusted. If the operating parameters do not need to be adjusted, the predicted curve is compared with the actual operating conditions in real time to observe whether the trend of the predicted value is similar to the actual value. If they are similar, the prediction results are weighted according to the corresponding deviation. If they are not similar, the grid search method is used to adjust the hyperparameters, update the historical data, and re-analyze. S5: Calculate the adjustment amount of the operating parameters based on the optimal regression principle and adjust the operating parameters of the combustion module. After adjustment, return to S2 to start the next cycle.

2. The intelligent control method for a porous media burner as described in claim 1, characterized in that, In S1, the operating parameters include operating power, number of activated modules, intake air temperature, fuel flow rate, and air flow rate; the status parameters include combustion temperature, oxygen content at the top of the combustion module, NOx content at the top of the combustion module, CO content at the top of the combustion module, and noise.

3. The intelligent control method for a porous media burner as described in claim 1, characterized in that, In S2, the operating parameters and status parameters of each combustion module inside the porous media burner are automatically generated into an engineering file and saved to the reference history database; in S3, the work log file generated by the machine learning algorithm is saved to the reference history database.

4. An intelligent control system for a porous media burner, characterized in that, The intelligent control method for the porous media burner according to any one of claims 1-3 includes: The first data acquisition unit collects the operating parameters and status parameters of the porous media burner under normal operating conditions and historical experiments, which are used as reference historical data. The second acquisition unit: collects the operating parameters and status parameters of each combustion module inside the porous media burner as real-time data; The computing unit, based on machine learning algorithms, establishes an importance analysis model of power demand and the number of combustion modules, and analyzes the impact of the operating parameters of each combustion module on the porous media burner. When the state parameters of the combustion module exceed the preset threshold, the analysis unit uses a long-short-time prediction neural network algorithm to analyze the reference historical data and real-time data of the porous media burner, predicts the state parameters of the next time step, and adjusts the operating parameters when the deviation between the prediction result and the reference historical data is greater than the preset threshold, and transmits the analysis results to the adjustment unit. The adjustment unit calculates the adjustment amount of the operating parameters based on the optimal regression principle and adjusts the operating parameters of the combustion adjustment unit and the delivery adjustment unit of the combustion module. The combustion unit enables the premixed combustion of fuel and air; The conveying unit regulates and conveys fuel, air, and flue gas.

5. The intelligent control system for the porous media burner as described in claim 4, characterized in that, The first acquisition unit includes a memory card used in conjunction with a high-speed microcontroller; the second acquisition unit includes a thermocouple, an infrared thermometer, an electromagnetic flowmeter, a flue gas analyzer, and a noise detector; the calculation unit, analysis unit, and adjustment unit share equipment, including: a high-speed information acquisition card, a high-speed microcontroller, a computer, and a monitor; the combustion unit includes an intake pipe (1), a spring solenoid valve (2), a premixing chamber (3), a burner housing (4), an electronic ignition hole (5), insulation cotton (6), a heat insulation plate (7), and a combustion module (8); the delivery unit includes pipelines, a flow controller, an oxygen controller, a spring solenoid valve, and a pressure reducing valve.

6. The intelligent control system for the porous media burner as described in claim 5, characterized in that, The combustion unit includes multiple combustion modules (8), and the number of combustion modules (8) activated is matched according to the power.

7. The intelligent control system for the porous media burner as described in claim 6, characterized in that, The number of combustion modules is 2 to 4.

8. The intelligent control system for the porous media burner as described in claim 5, characterized in that, The combustion module is of foam ceramic type, honeycomb type or metal fiber type; the material is silicon carbide, silicon nitride, zirconium oxide, quartz ceramic, mullite, nickel-based alloy, nickel-chromium-iron alloy or porous tungsten.

9. The intelligent control system for the porous media burner as described in claim 8, characterized in that, When the combustion module is made of foam ceramic, the pore density is 30-60 PPI and the porosity is 0.7-0.9.

Images