Partitioning and real-time control method of transverse non-uniform temperature field in circular tunnel under high temperature

Abstract

The present invention discloses a method for zoning and real-time control of a transverse non-uniform temperature field under high temperature in a circular tunnel, which includes two steps: non-uniform temperature field zoning and non-uniform temperature field active control. The non-uniform temperature field zoning includes top burner zoning and side burner zoning, which aims to divide the high-temperature area that each burner can control, so as to determine the range that each burner can affect, and provide a basis for active control. The active control of the non-uniform temperature field includes determining the temperature monitoring time interval, monitoring the actual temperature, calculating the temperature difference, predicting the burner power, and adjusting the burner power. The method can quantitatively define the control areas of different burners in an energy-efficient manner, facilitate the generation and control of the temperature field, and effectively predict the power changes of the burner through the BP neural network, thereby efficiently controlling the generation of the temperature field, and providing great convenience for subsequent fire resistance research.

Description

Technical Field

[0001] The present invention relates to the field of tunnel fire resistance in structural engineering, and in particular to a method for partitioning and real-time controlling a transverse non-uniform temperature field in a circular tunnel under high temperature. Background Art

[0002] According to relevant statistics, since the 21st century, the average annual growth rate of China's highway tunnels has reached 20%. By the end of 2021, a total of 118 large-diameter shield tunnel engineering projects have been started in the country, and 61 ultra-large diameter shield tunnel projects with a diameter of over 14m have been launched.

[0003] As the number and size of tunnels continue to expand and traffic volume increases, the probability of fires in tunnels is also increasing. When a fire occurs, due to the shortcomings of the tunnel structure being long and narrow and the relatively poor ventilation conditions, a large amount of heat from the fire will be transferred to the tunnel lining structure and the surrounding strata, resulting in uneven temperature distribution. In order to reduce the damage caused by tunnel fires to structures, personnel, vehicles, etc., researchers have conducted many studies on tunnel fire resistance. However, physical tunnel fire tests are expensive, test conditions are difficult to control, and the test scale is limited. Conventional fire test furnaces can only produce a uniform temperature field and cannot restore the non-uniform temperature field of tunnel fires. If the lining structure is placed in an existing combustion furnace for fire resistance research, it cannot fully reflect the performance of the lining structure in an actual fire, and the research results are quite limited.

[0004] Currently, studying tunnel fire temperature fields through numerical simulation is more convenient and less expensive than conducting full-scale tests. Furthermore, compared to scaled-scale tests, numerical simulations eliminate the need to consider scale effects, making them highly effective for studying issues related to non-uniform temperature fields in tunnel fires. Research on the transverse temperature distribution patterns of tunnel fires has primarily focused on factors such as ventilation velocity, fire source size, fire source location, and tunnel slope. It can be seen that the factors influencing tunnel cross-sectional temperature distribution are complex, and existing research on tunnel cross-sectional temperature distribution has primarily focused on the temperature field distribution patterns, with limited research on the regional division of the cross-sectional temperature field.

[0005] In addition, current research on furnace temperature control technology is limited, and conventional fire resistance test furnaces mostly use computer monitoring + PLC field control equipment. PID (proportional, integral, differential) controllers have the characteristics of simple structure and strong robustness, and are widely used in the industrial field, covering more than 90% of industrial processes. However, the temperature field generated by the fire resistance test furnace of building components is relatively uniform, and the temperature of one area in its control system is controlled by one burner, while the non-uniform temperature field may involve one burner controlling the temperature of multiple areas at the same time, which cannot be achieved by the PID algorithm. Due to the complexity of furnace temperature changes, many phenomena cannot be directly described based on data expressions. In summary, the control of non-uniform temperature fields in tunnels is characterized by nonlinearity, high timeliness requirements, and high precision requirements. Traditional furnace temperature control methods often find it difficult to achieve ideal control effects. Summary of the Invention

[0006] In response to the above-mentioned technical deficiencies, the purpose of the present invention is to provide a method for zoning and real-time control of the transverse non-uniform temperature field in a circular tunnel under high temperature, which can solve the problem of difficulty in real-time generation and control of non-uniform temperature fields in current research in the field of tunnel fire resistance, and provide support for research in the fields of tunnel fire resistance and fire safety.

[0007] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0008] The present invention provides a method for zoning and real-time controlling the transverse non-uniform temperature field in a circular tunnel under high temperature, including two steps: zoning the non-uniform temperature field and actively controlling the non-uniform temperature field;

[0009] In the non-uniform temperature field partitioning step, the burners are divided into top burners and side burners;

[0010] The top burner partitioning method is as follows: the burner is established in the Fluent simulation, and the control area is determined by the angle between the burner centerlines, which includes the left and right sides of the burner. The principle for determining the angle is that the area formed by the angle must completely surround the high-temperature temperature field generated by the burner.

[0011] The side burner partitioning method is as follows: Based on Fluent simulation, the side burners need to be simulated based on two factors: the angle between the burner and the horizontal line and the number of burners. The optimal burner partitioning is found, and the optimal angle is used as the quantitative indicator of the final partitioning result.

[0012] The non-uniform temperature field active control step is performed after the non-uniform temperature field is partitioned. The process of using a small number of burners to control the temperature of multiple zones is as follows:

[0013] (1) Determine the temperature monitoring time interval; starting from 0 seconds, monitor the temperature change every t seconds, that is, Δ t =Δt-1 =t seconds;

[0014] (2) Monitoring actual temperature: monitoring the actual temperature values ​​of the five areas within a time period of t seconds, and then transmitting the temperature signal to the control system, so that the temperature of the five areas in each time period can be observed at the control end;

[0015] (3) Calculate the temperature difference; based on the given target temperature curve and the actual temperature monitored, calculate the temperature value Δ that needs to be adjusted for each area within the time period T , Δ T The calculation method of is as shown in formula 2;

[0016] (4) Predicting burner power: Based on the temperature values ​​that need to be adjusted in each area during the time period, the power adjustment value Δ of each burner is predicted through the neural network. W ;

[0017] (5) Adjusting the burner power; converting the burner power adjustment value into a signal and outputting it to the control system, which makes adjustments based on the burner power at the previous moment, and ultimately achieving the target temperature curve for each area by adjusting the burner power;

[0018] Δ t =t2-t1=Δ t-1 =t1-t0=10s (1)

[0019] Δ T =T2-T1 (2)

[0020] Δ w =W2-W1 (3)

[0021] Where, Δ T is Δ t The temperature value that each area needs to rise or fall within a certain period of time; T1 is Δ t The actual temperature value at the beginning of time is t1; T2 is Δ t The target temperature at the end of time, i.e., at time t2; W1 is Δ t The burner power at the beginning of time, i.e., at time t1, remains unchanged; W2 is Δ t The burner power after the change at the beginning of time, i.e., at time t1; Δ W To change the power of the burner.

[0022] Preferably, when the side burners are zoned, the principle for selecting the optimal angle is that the angle between adjacent control lines of adjacent areas is less than 10 degrees to prevent the two areas from being too close together, which will narrow the range of the common control area of ​​the two burners and make it easier to achieve temperature uniformity.

[0023] At the same time, the angle between adjacent control lines in adjacent areas is greater than 40 degrees, which prevents the top and side burners from jointly controlling a wide range of areas, increases the complexity of operation, and makes control more difficult in actual operation, which is not conducive to precise temperature control.

[0024] Preferably, in the step of determining the temperature monitoring time interval, since the shortest time for the current temperature sensor to transmit test data is 10 seconds, the temperature change is monitored every 10 seconds starting from 0 seconds.

[0025] Preferably, in the non-uniform temperature field zoning step, the burner nozzles that are at 90° to the horizontal line are set as top burners, and the remaining burners are set as side burners.

[0026] Preferably, the burner power is predicted by a BP neural network; specifically, by inputting the temperature difference of each area in the tunnel, i.e., the difference between the actual temperature measured by the thermocouple and the target curve temperature, the initial power and initial temperature of each burner, and outputting the power change value required for the burner to achieve the target temperature, the burner power is cyclically adjusted to realize the burner power prediction demand based on the BP neural network.

[0027] The beneficial effects of the present invention are:

[0028] 1. The present invention proposes a non-uniform temperature field zoning method for circular tunnels, which can effectively and quantitatively define the control areas of different burners and facilitate the generation and control of the temperature field.

[0029] 2. The present invention also proposes a real-time active control method for non-uniform temperature fields. Through the BP neural network, it can effectively predict the power changes of the burner, and then efficiently control the generation of the temperature field, providing great convenience for subsequent fire resistance research.

[0030] 3. The real-time active control method for non-uniform temperature fields proposed in this invention can optimize burner energy usage and avoid unnecessary energy consumption, thereby achieving energy conservation and emission reduction, meeting the needs of sustainable development. Furthermore, implementation of this invention will help researchers gain a deeper understanding of the development mechanisms of tunnel fires and provide a theoretical basis and experimental support for fire prevention and control technologies in non-uniform temperature fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0032] Figure 1 A diagram of a partitioning example provided in an embodiment of the present invention;

[0033] Figure 2 A schematic diagram of a neural network prediction approach provided by an embodiment of the present invention;

[0034] Figure 3 The dimensions of the small tower combustion vehicle provided in the embodiment of the present invention.

[0035] Figure 4 A schematic diagram of a tower combustion vehicle and a tunnel structure provided by an embodiment of the present invention;

[0036] Figure 5 A schematic diagram of the two-dimensional dimensions of a model provided in an embodiment of the present invention;

[0037] Figure 6 A schematic diagram of the final mesh division of the tunnel model provided by an embodiment of the present invention;

[0038] Figure 7 A schematic diagram of the mesh division of a tunnel model provided by an embodiment of the present invention;

[0039] Figure 8 A schematic diagram of the temperature monitoring line provided in an embodiment of the present invention;

[0040] Figure 9 A schematic diagram of the temperatures of different monitoring lines of a top burner provided in an embodiment of the present invention;

[0041] Figure 10 A schematic diagram of temperature distribution under top burner control provided by an embodiment of the present invention;

[0042] Figure 11 A schematic diagram of the main control area when the nozzle angle of the side burner provided in an embodiment of the present invention is 30°;

[0043] Figure 12 A schematic diagram of the main control area when the nozzle angle of the side burner provided in an embodiment of the present invention is 45°;

[0044] Figure 13 A schematic diagram of the main control area when the nozzle angle of the side burner provided in an embodiment of the present invention is 60°;

[0045] Figure 14 A schematic diagram of the temperature field regional distribution provided by an embodiment of the present invention;

[0046] Figure 15 The temperature distribution diagram when all burners are turned on according to the embodiment of the present invention;

[0047] Figure 16 A schematic diagram of a data-driven model control performance simulation verification provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0049] like Figures 1 to 3 As shown, this embodiment provides a method for zoning and real-time control of the transverse non-uniform temperature field in a circular tunnel under high temperature, specifically:

[0050] Step 1: Determine the tunnel and burning car model

[0051] A combustion vehicle model and a tunnel model are set up in advance. This embodiment shows a small tower combustion vehicle with a three-dimensional body as shown in FIG. Figure 3 As shown. The top width of the small tower burner vehicle body is 525mm, the bottom width is 1150mm, the body height is 840mm, and the longitudinal length is 1100mm. This size ensures that the flame of the burner can fully cover the tunnel cross-section, and enables the burner vehicle to flexibly adapt to tunnel sections of different lengths to achieve effective temperature control. The angle between the plane direction of the burners on both sides and the plane direction of the top burner is set to 45°, and the direction of each burner nozzle is perpendicular to the plane. However, the direction of the burner flame is not fixed. The flame direction can be adjusted by connecting an external refractory conduit so that it is not limited to the vertical plane direction.

[0052] According to the specification of GB / T 51438-2021, the design standard for shield tunnel engineering, the thickness of the concrete segments is 300mm. Based on the above dimensional information, a model diagram of the tower burner and the tunnel is constructed, as shown in the figure below. Figure 4 shown.

[0053] In addition, when designing a tower-type combustion vehicle, several holes need to be left on one side of the vehicle body for the connection of the burner air intake duct and air intake duct inside the vehicle body to the external air intake duct and air duct. Similarly, a hole needs to be opened at one end of the tunnel blockage to allow the gas pipe and air duct to pass through. After the pipe connection is completed, the hole is tightly closed to prevent the internal temperature and heat from being lost, ensuring safety during the experiment. However, due to the complexity of the pipeline connection, it is simplified during modeling and the details are not clearly drawn in the model. At the same time, in the design, the wheel height at the bottom of the vehicle body is about 135mm, and the track height and its bottom support are about 160mm. Therefore, a height of 300mm is left from the bottom of the combustion vehicle body to the tunnel ground for the use of wheels and tracks. However, when building the model, the model is simplified, the wheels and tracks are not drawn, and the gap at the bottom of the vehicle body is left vacant and regarded as a fluid domain. The two-dimensional size diagram of the model is as follows Figure 5 shown.

[0054] The simulation software used is Fluent, and the SIMPLE solution algorithm is used. The initial temperature is set to room temperature, that is, T = 300K, and the relative atmospheric pressure is 0Pa. These parameters indicate the starting state of the simulation. In order to ensure the effectiveness of the simulation, a grid sensitivity analysis is performed to compare the longitudinal temperature of the tunnel under five different grid numbers: 60,000, 135,000, 200,000, 260,000, and 400,000. The longitudinal temperature of the tunnel is as follows: Figure 6 As shown in the figure, it can be found that except for the 60,000 grids, the calculation accuracy of the remaining grids is similar. Therefore, the overall model grid size is selected as 80mm, and the grid is encrypted at the burner nozzle and exhaust port, and the grid size is taken as 40mm. The final grid division of the tunnel model is shown in the figure. Figure 7 shown.

[0055] Step 2: Non-uniform temperature field partitioning method

[0056] In order to accurately perform quantitative zoning, it is necessary to design the layout of temperature monitoring points. The model established in this paper is symmetrical on the left and right sides with the center line as the axis, so the temperature measurement points are only arranged on one side of the center line to improve the convenience and efficiency of the analysis. On the section where the burner nozzle is located, starting from the center line, a straight line connecting to the center of the combustion vehicle is established every 5° on the left side of the center line, such as Figure 8 As shown, there is no temperature monitoring line on the right. The temperature monitoring line is named "line". For example, the temperature monitoring line with an angle of 10° based on the center line is represented as "line10".

[0057] Temperature monitoring points are set on each temperature monitoring line, and three temperature monitoring points are set on each straight line. The outermost temperature monitoring point is 200mm from the tunnel wall, the innermost temperature monitoring point is 500mm from the combustion vehicle wall, and the middle temperature monitoring point is placed at the midpoint of the line connecting the outermost and innermost temperature monitoring points. The average temperature of the monitoring line is the average value of all the measuring points on the line.

[0058] Top burner control area division

[0059] For the top burner, the 0° to 30° monitoring line is selected to compare the temperature change (the burner injection rate is 17m / s). Figure 9 As shown in the figure, it can be found that the temperatures of the three monitoring lines at 0°, 5°, and 10° are significantly higher than those of the other monitoring lines. The temperature distribution cloud diagram under the control of the top burner is shown in the figure. Figure 10 As shown in the figure, the angle between the black line and the center line is 10°. According to the above-mentioned angle selection principle, the temperature of the black line at this time includes the high-temperature temperature field area. Therefore, the range of 0 to 10° to the left and right of the center line is defined as the control area of ​​the top burner.

[0060] Side burner control area division

[0061] For side burners, it is necessary to comprehensively consider the side burner nozzle angle and the number of openings, among which the angles considered are 30°, 45° and 60°, and the number of openings considered is single-sided opening of the burner and simultaneous opening of both sides. Figure 11 、 12 , 13 show the maximum control area when the two burner opening numbers are 30°, 45° and 60° respectively.

[0062] It can be found that when the burner nozzle angle is 30°, the area controlled by the top and side burners is relatively wide, which is not conducive to precise temperature control. When the burner nozzle angle is 60°, the area directly controlled by the non-top and side burners is smaller, which is more likely to form a uniform temperature field, which is not conducive to achieving the specific requirements of the non-uniform temperature field in the tunnel. Both angles do not meet the selection principle of the optimal angle for the side burners. Therefore, the burner distribution with a nozzle angle of 45° can better balance the control range of the top and side burners.

[0063] In summary, the final division of the non-uniform temperature field is as follows Figure 14 shown.

[0064] Turn on all burners to verify the effectiveness of the proposed temperature field division method. Figure 15 It can be seen that the temperature within the direct control area of ​​the three groups of burners is significantly higher than that in the other two areas, and the darker colors are concentrated within the edge lines of the direct control area of ​​the burners, proving that the division of the areas is reasonable.

[0065] Step 3: Real-time active control method for non-uniform temperature field

[0066] The idea behind using a small number of burners to control the temperature of multiple zones is as follows:

[0067] (1) Determine the temperature monitoring time interval. Since the shortest time for the temperature sensor currently used in the laboratory to transmit test data is 10s, the temperature change is monitored every 10s, that is, Δ t =Δ t-1 =10s;

[0068] (2) Monitor actual temperature. Monitor the actual temperature values ​​of the five areas within a 10-second period, and then transmit the temperature signal to the control system. The temperature of the five areas in each time period can be observed at the control end.

[0069] (3) Calculate the temperature difference. Based on the given target temperature curve and the actual temperature monitored, calculate the temperature value Δ that needs to be adjusted for each area during the time period. T , Δ T The calculation method is as shown in Formula 4-11.

[0070] (4) Predict burner power. Based on the temperature value that needs to be adjusted in each area during this time period, the power adjustment value of each burner Δ is predicted through the neural network. W .

[0071] (5) Adjust the burner power. The burner power adjustment value is converted into a signal and output to the control system. The control system makes adjustments based on the burner power at the previous moment. Ultimately, the target temperature curve of each area is achieved by adjusting the power of each burner.

[0072] Δ t =t2-t1=Δ t1 =t1-t0=10s (4)

[0073] Δ T =T2-T1 (5)

[0074] Δ w =W2-W1 (6)

[0075] Where, Δ T is Δ t The temperature value that each area needs to rise or fall within a certain period of time; T1 is Δ t The actual temperature value at the beginning of time (i.e., time t1); T2 is Δ t The target temperature at the end of time (i.e., time t2); W1 is Δ t The burner power does not change at the beginning of time (i.e., time t1); W2 is Δ t Burner power after change at the beginning of time (i.e. time t1); Δ W To change the power of the burner.

[0076] In the above step 4, the burner power needs to be predicted. The modeling process of the BP neural network used in the prediction will be introduced in detail below.

[0077] First, the input and output of the neural network are determined. According to the active control idea, the network input and output are as follows: the input layer is the initial temperature, the initial power of burner 1, the initial power of burner 2, the initial power of burner 3, the changing temperature of area 1, the changing temperature of area 2, the changing temperature of area 3, the changing temperature of area 4 and the changing temperature of area 5; the output layer parameters are determined to be the changing power of burner 1, the changing power of burner 2 and the changing power of burner 3.

[0078] The neural network training samples were obtained through simulations, covering the entire range of possible input and output conditions. The specific settings were as follows: the initial temperature range was set to 300–1500 K, with 300 K representing normal temperature and 1500 K representing the maximum temperature setting. The initial temperature was the average temperature of the five zones, with 200 K intervals used as the temperature range. The goal was to ensure that the temperatures of the five zones and their average temperatures fell within the same range. During neural network training, temperature ranges were numbered; for example, the 300–500 K range was numbered 1, and so on, for a total of six temperature ranges. The initial burner power values ​​were set to 0 kW, 84.98 kW, 158.21 kW, 231.44 kW, 304.68 kW, 377.91 kW, 451.14 kW, 524.36 kW, and 597.59 kW, a total of nine levels covering both the on and off states of the burner. Furthermore, the post-change burner power levels remained the same as the initial burner power.

[0079] The number of hidden layer nodes is typically determined using an empirical formula to determine a range of values. The neural network is then trained using different values, and the final value is determined based on the training results. The empirical formula estimates that the number of hidden layer nodes ranges from 5 to 13.

[0080]

[0081] h=log2m (8)

[0082] Where α is a natural number between 1 and 10, and m, h, and n are the number of nodes in the input layer, hidden layer, and output layer.

[0083] The calculations were performed for different numbers of hidden layer nodes, and the average of the five results was taken, as shown in Table 1. As can be seen from the results in the table, when the number of hidden layer nodes is 11, the average error of the model is small, so the number of hidden layer neurons in this model is 11.

[0084] Table 1 Calculation results of different hidden layer node numbers

[0085]

[0086] The sample set capacity of the neural network is estimated based on the following two empirical methods:

[0087] (1) Estimation based on the relationship between the sample set capacity and the neurons in each layer of the network

[0088] P=(5~10)×(mh+hn) (9)

[0089] (2) Estimation based on the relationship between sample set capacity and training error ε

[0090]

[0091] For a three-layer BP neural network, n w The calculation method is as follows:

[0092] n w =m(h+1)+h(n+1) (11)

[0093] After calculation, the sample capacity range was preliminarily determined to be 300 to 1720.

[0094] The sample design method adopts the orthogonal design method. According to this method, the initial power and the changed power of the three burners are orthogonally designed, and the orthogonal design table (Table 2) L81.9.6 is obtained. L81.9.6 means that 81 samples are designed based on the orthogonal design method under the 6-factor 9-level working condition, and then a comprehensive design is carried out with the initial temperature. Finally, the sample set capacity is determined to be 486.

[0095] Table 2 Factors and their levels

[0096]

[0097]

[0098] The normalization method for neural network training data is the maximum value method, and its formula is as follows:

[0099]

[0100] The model sample set is divided into a ratio of 7:1.5:1.5, that is, the number of sample sets in the training set, validation set, and test set are 340, 73, and 73, respectively. In addition, the trial calculation results of different sample set division ratios (Table 3) verify the effectiveness of the division method.

[0101] Table 3 Calculation results of different sample set division ratios

[0102]

[0103] The initialization method, activation function, learning algorithm, error function, learning rate, etc. in the BP network were calculated in a similar way, and the final hyperparameters were determined as shown in Table 4.

[0104] Table 4 Model parameter configuration

[0105]

[0106] In addition, this model uses MAE as the model error evaluation indicator, and the calculation of the model accuracy is shown in Formula 13.

[0107]

[0108] The model was trained using the training and validation sets according to the above model parameter configurations, and the results are shown in Table 5. The results show that the prediction accuracy differences among the training, validation, and test sets are small, with no overfitting. All power variations can be accurately predicted, verifying that the method has high accuracy and, therefore, good control capabilities.

[0109] Table 5 Prediction accuracy of training set, validation set and test set

[0110]

[0111] Finally, in Fluent, the neural network is used to predict data and control the formation of the temperature field in the tunnel, preliminarily verifying the application performance of the neural network.

[0112] Compare and analyze the temperature change curves of the five areas in the figure with the HC fire curve ( Figure 16 ), it can be found that the overall deviation is small, especially at high temperature, the temperature curves of the five regions gradually tend to be consistent with the HC curve, the degree of agreement is high, and the error does not exceed 10%, which shows the accuracy of the neural network model in control.

[0113] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for partitioning and real-time control of the horizontal non-uniform temperature field in a circular tunnel under high temperature, characterized in that: It includes two steps: non-uniform temperature field partitioning and non-uniform temperature field active control; In the non-uniform temperature field partitioning step, the burners are divided into top burners and side burners; The top burner partitioning method is as follows: the burner is established in the Fluent simulation, and the control area is determined by the angle between the burner centerlines, which includes the left and right sides of the burner. The principle for determining the angle is that the area formed by the angle must completely surround the high-temperature temperature field generated by the burner. The side burner partitioning method is as follows: Based on Fluent simulation, the side burners need to be simulated based on two factors: the angle between the burner and the horizontal line and the number of burners. The optimal burner partitioning is found, and the optimal angle is used as the quantitative indicator of the final partitioning result. The non-uniform temperature field active control step is performed after the non-uniform temperature field is partitioned. The process of using a small number of burners to control the temperature of multiple zones is as follows: (1) Determine the temperature monitoring time interval; starting from 0 seconds, monitor the temperature change every t seconds, that is, Δ t =Δ t-1 =t seconds; (2) Monitoring actual temperature: monitoring the actual temperature values ​​of the five areas within a time period of t seconds, and then transmitting the temperature signal to the control system, so that the temperature of the five areas in each time period can be observed at the control end; (3) Calculate the temperature difference; based on the given target temperature curve and the actual temperature monitored, calculate the temperature value Δ that needs to be adjusted for each area within the time period T , Δ T The calculation method of is as shown in formula 2; (4) Predicting burner power: Based on the temperature values ​​that need to be adjusted in each area during the time period, the power adjustment value Δ of each burner is predicted through the neural network. W ; (5) Adjusting the burner power; converting the burner power adjustment value into a signal and outputting it to the control system, which makes adjustments based on the burner power at the previous moment, and ultimately achieving the target temperature curve for each area by adjusting the burner power; D t =t2-t1=Δ t-1 =t1-t0=10s (1) D T =T2-T1 (2) Δ w =W2-W1 (3) Where, Δ T is Δ t The temperature value that each area needs to rise or fall within a certain period of time; T1 is Δ t The actual temperature value at the beginning of time, i.e., at time t1; T2 is Δ t The target temperature at the end of time, i.e., at time t2; W1 is Δ t The burner power at the beginning of time, i.e., at time t1, remains unchanged; W2 is Δ t The burner power after the change at the beginning of time, i.e., at time t1; Δ W To change the power of the burner.

2. The method for zoning and real-time control of the horizontal non-uniform temperature field in a circular tunnel under high temperature as claimed in claim 1, characterized in that: When partitioning the side burners, the principle for selecting the optimal angle is that the angle between adjacent control lines of adjacent areas is less than 10 degrees to prevent the two areas from being too close together, which will narrow the range of the common control area of ​​the two burners and make it easier to achieve uniform temperature. At the same time, the angle between adjacent control lines in adjacent areas is greater than 40 degrees, which prevents the top and side burners from jointly controlling a wide range of areas, increases the complexity of operation, and makes control more difficult in actual operation, which is not conducive to precise temperature control.

3. The method for zoning and real-time control of the horizontal non-uniform temperature field in a circular tunnel under high temperature as claimed in claim 1, characterized in that: In the step of determining the temperature monitoring time interval, since the shortest time for the current temperature sensor to transmit test data is 10 seconds, the temperature change is monitored every 10 seconds starting from 0 seconds.

4. The method for zoning and real-time control of the horizontal non-uniform temperature field in a circular tunnel under high temperature as claimed in claim 1, characterized in that: In the non-uniform temperature field zoning step, the burner nozzles with an angle of 90° to the horizontal line are set as top burners, and the remaining burners are set as side burners.

5. The method for zoning and real-time control of the horizontal non-uniform temperature field in a circular tunnel under high temperature as claimed in claim 1, characterized in that: The burner power is predicted through the BP neural network. Specifically, the temperature difference of each area in the tunnel is input, that is, the difference between the actual temperature measured by the thermocouple and the target curve temperature, the initial power and initial temperature of each burner, and the power change value required for the burner to achieve the target temperature is output. The burner power is cyclically adjusted to realize the burner power prediction demand based on the BP neural network.

Images