Transverse non-uniform temperature field partitioning and active control method for rectangular immersed tunnel

Abstract

The invention discloses a rectangular immersed tube tunnel transverse non-uniform temperature field partitioning and active control method. The method specifically comprises the three steps of tunnel region internal temperature field partitioning, region temperature active control and temperature prediction error compensation. The division of the temperature field in the tunnel area comprises preliminary determination of a combustor control area and quantitative determination of the control area; the temperature active control of each area adopts a machine learning technology, and a neural network model is trained to control a combustor; according to temperature prediction error compensation, the average temperature error of the five temperature control areas is counted once every 10 seconds, the prediction error of the combustor power is reduced, and the reduction precision of temperature rise curves of the five temperature control areas is improved. According to the method, the problems of partition and active control of the transverse non-uniform temperature field of the rectangular tunnel section when the combustor is adopted can be solved, and technical convenience is provided for research related to fire resistance of the immersed tunnel.

Description

Technical Field

[0001] The invention relates to the field of fire resistance of immersed tube tunnel structures, and in particular to a transverse non-uniform temperature field partitioning and active control method for a rectangular immersed tube tunnel. Background Art

[0002] As one of the important forms of underwater tunnel construction, immersed tube tunnels are widely used in practical engineering. When a fire occurs in a tunnel, due to the tunnel's semi-enclosed space structure and poor ventilation conditions, a large amount of heat will gather in the tunnel ceiling area and transfer to the tunnel lining structure, forming a non-uniform temperature field. Immersed tube tunnels are geographically located underwater. Once a fire occurs, smoke exhaust and heat dissipation conditions are worse, personnel evacuation and rescue are more difficult, and tunnel structure damage is more serious. In order to reduce tunnel damage caused by fire and reduce post-disaster maintenance costs, it is very necessary to study the fire resistance of immersed tube tunnels. However, it is difficult to carry out destructive tunnel fire resistance tests under existing conditions, so there are limitations in studying tunnel fire resistance.

[0003] At present, tunnel fire tests are usually conducted in real in-situ tunnels and laboratory scaled tunnels, mainly focusing on the study of tunnel fire dynamics, such as the distribution law of tunnel fire temperature field and smoke spread. Considering the safety of the test process and reducing the test cost, few scholars have conducted tunnel fire resistance tests in real tunnels to study tunnel damage mechanisms and destruction characteristics. At present, the fire resistance test of tunnel lining structure is carried out in a fire resistance test furnace. It is modified on the basis of the building fire resistance test furnace. The loading method is changed, but it cannot achieve the non-uniform temperature field of the tunnel fire, and it is difficult to reflect the performance of the tunnel lining structure in an actual fire. The test results

[0004] At the same time, a few scholars have carried out in-situ test research on immersed tube tunnels, mainly focusing on the study of temperature field and deformation, and rarely involving the fire resistance of immersed tube tunnel lining structures under high temperature; most scholars use numerical simulation or laboratory scale test methods to study the stress state of immersed tube tunnel lining structures under high temperature, but lack universality.

[0005] Since the experimental research on immersed tube tunnel fire is costly and dangerous, the numerical simulation method is more suitable for the study of the transverse temperature field of immersed tube tunnel fire. By comparing the results of the test with the same parameters and the numerical simulation, the numerical simulation can accurately restore the transverse temperature field of the tunnel under the real fire. When a fire occurs in the tunnel, the temperature distribution of the transverse temperature field is related to the distance from the fire source, the duration of combustion and the cross-sectional form. The factors affecting the cross-sectional temperature distribution of the tunnel are complex. At present, scholars' research mainly focuses on the qualitative study of the distribution law of the temperature field of tunnel fire, and there is little research on the quantitative division of the cross-sectional temperature field. Therefore, there is an urgent need for a method for the division and active control of the non-uniform temperature field of the transverse section. Summary of the invention

[0006] In view of the above-mentioned technical deficiencies, the purpose of the present invention is to provide a method for zoning and actively controlling the transverse non-uniform temperature field of a rectangular immersed tube tunnel, which can solve the problem of zoning and actively controlling the transverse non-uniform temperature field of a rectangular tunnel cross-section when a burner is used, thereby providing technical convenience for research related to fire resistance of immersed tube tunnels.

[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 partitioning and actively controlling a transverse non-uniform temperature field in a rectangular immersed tube tunnel, comprising: S1, partitioning a temperature field in a tunnel area; the specific steps include:

[0009] S11, preliminarily determining the burner control area, that is, roughly determining the non-uniform temperature field area controlled by the burner according to the location of the burner;

[0010] S12, quantitatively determine the control area, that is, determine the specific size of the control area;

[0011] S2. Active control of the temperature in each zone; using machine learning technology to train a neural network model to control the burner; the specific steps include:

[0012] S21. Through Fluent numerical simulation, the corresponding relationship between the power of three rows of burners on the small tower combustion vehicle and the temperature and initial power of five temperature control areas is constructed to obtain a sample set;

[0013] S22. Use the sample set to train the machine learning algorithm and construct a prediction model for the lateral non-uniform temperature field of the immersed tube tunnel. Before training the model, the sample set capacity needs to be estimated, and the sample scheme needs to be designed using the orthogonal design method. In addition, reasonable training, verification and test sets need to be generated through sample data analysis and standardized sample set division.

[0014] S3, temperature prediction error compensation: the average temperature error of the five temperature control areas is calculated every 10 seconds, the purpose of which is to reduce the prediction error of the burner power and improve the restoration accuracy of the temperature rise curves of the five temperature control areas.

[0015] Preferably, in step S12, the burners are divided into top burners with a flame outlet facing the ground at an angle of 90° and side burners with other side flame outlets according to the nozzle angle, and the two burners control the top and side temperature fields respectively;

[0016] The top temperature field determines the range of the controlled high temperature area according to the power of the top burner. At this time, the temperature control area of ​​the top burner is divided according to the temperature area with high sensitivity, that is, the temperature area that starts to heat up earlier is the temperature control area of ​​the top burner;

[0017] The side temperature field is determined by the burner nozzle scheduling, that is, the angle between the burner and the horizontal line to determine the range of the high-temperature area controlled by it; the principle of dividing the side temperature field is that the control range of the side burner should be focused on the tunnel side wall as much as possible to avoid providing a large amount of heat to the top and bottom; the entire process is simulated by Fluent software to perform quantitative zoning of the non-uniform temperature field.

[0018] Preferably, in step S22, taking a single burner controlling a single temperature control area as an example, the control process of a single burner controlling a single temperature control area is mapped to three rows of burners controlling five temperature control areas, which is the active control process to be implemented. The specific process is as follows:

[0019] Step 1: Divide the temperature rise curve; divide the known temperature rise curves of the five temperature control areas ① to ⑤ into time periods of 10s, and obtain the start and end temperatures of the five temperature control areas in several time periods;

[0020] Step 2: Predict the burner power; through the neural network algorithm, the power required to be set for the left, top and right burners in the first 10s time period is predicted based on the start and end temperatures of the five zones in the first 10s time period and the initial power of the three rows of burners, which is 0kW; the power required to be set for the left, top and right burners in the second 10s time period is predicted based on the start and end temperatures of the five zones in the second 10s time period and the predicted power of the three rows of burners in the first 10s time period as the initial power; and so on, the power required to be set for the burners in all time periods is obtained;

[0021] Step 3: Input the burner power: Before the test begins, input the power required by the three rows of burners in each time period into the burner control system;

[0022] Step 4: Monitor the temperature of the temperature control area; monitor the temperature changes of the five temperature control areas respectively, and calculate the average deviation between the actual temperature and the target temperature of the five temperature control areas in each time period;

[0023] Step 5: Burner power prediction error compensation: During the test, all burner powers predicted by the neural network are uniformly increased or decreased according to the cumulative temperature difference change trend and the average temperature difference in each time period.

[0024] Preferably, the specific process of the temperature prediction error compensation step is as follows:

[0025] Step 1: Calculate the average temperature error; count the average temperature error between the actual temperature and the target temperature of the five temperature control areas in the first 10s period;

[0026] Step 2: Determine the cumulative trend of temperature error, i.e., make qualitative adjustments; when the cumulative trend of temperature error increases, the actual temperature is greater than the target temperature, and all burner power values ​​predicted by the neural network need to be reduced; when the cumulative trend of temperature error decreases, the actual temperature is less than the target temperature, and all neural network prediction values ​​need to be increased;

[0027] Step 3: Adjust the predicted value of the neural network according to the average temperature error, that is, quantitative adjustment; when the average temperature error is 0-50°C, the predicted value is increased by 1.1 times, when the average temperature error is -50-0°C, the predicted value is decreased by 0.9 times; when the average temperature error is 50-100°C, the predicted value is increased by 1.2 times, when the average temperature error is -100--50°C, the predicted value is increased by 0.8 times, and so on;

[0028] Step 4: Repeat steps 1 to 3 until the test is stopped.

[0029] The beneficial effects of the present invention are:

[0030] 1. This method can be applied to a variety of rectangular tunnels. The proposed algorithm for quantitative partitioning and active control of the transverse non-uniform temperature field can realize the active control of the transverse non-uniform temperature field in the immersed tube tunnel. This method can be extended to any rectangular tunnel.

[0031] 2. Due to the limited space of the tunnel and the semi-enclosed environment, and the influence of the ceiling jet, the high temperature area is mainly concentrated on the top of the tunnel, resulting in a non-uniform lateral temperature field in the immersed tube tunnel. The quantitative division strategy proposed in this method can reasonably divide the lateral non-uniform temperature field of the immersed tube tunnel, so that each burner can directly correspond to its high temperature influence range.

[0032] 3. The burner power prediction model proposed in this method has undergone rigorous optimization training. It only needs to pre-set the target temperature curve of each area and input it into the model to predict the power required by the burner in each area and realize the generation of non-uniform temperature field.

[0033] 4. In order to avoid the prediction error of the neural network, this method proposes an error compensation strategy to compensate the error according to the actual error size. In our verification model, the errors in all areas have been effectively compensated, which is closer to the target curve and effectively realizes the active control of the non-uniform temperature field. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. 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 creative work.

[0035] Figure 1 A schematic diagram of a single temperature control area provided in an embodiment of the present invention;

[0036] Figure 2 It is a schematic diagram of five temperature control zones and three rows of burners;

[0037] Figure 3 This is a schematic diagram of a small tower combustion vehicle;

[0038] Figure 4 This is a schematic diagram of an immersed tube tunnel with a smaller width;

[0039] Figure 5 Schematic diagram of the numerical model of a small tower combustion vehicle and immersed tube tunnel;

[0040] Figure 6 It is the preliminary zoning map of the transverse non-uniform temperature field;

[0041] Figure 7 This is a schematic diagram of the simulation results of the grid sensitivity analysis;

[0042] Figure 8 State diagram for top burner on;

[0043] Fig. 9 This is the layout diagram of the top partition survey line of the immersed tube tunnel;

[0044] Fig.10 This is the temperature rise curve of the top burner at 170kW;

[0045] Fig.11 This is the temperature rise curve of the top burner at 340kW;

[0046] Fig.12 Diagram of different angles between the side burner and the horizontal plane;

[0047] Fig.13 This is the layout of the survey lines for the side sections of the immersed tube tunnel;

[0048] Fig.14 Temperature cloud diagram of the side burner at different angles at 360s;

[0049] Fig.15 This is the temperature rise curve of the side burner at 30°;

[0050] Fig.16 Temperature cloud diagram of different powers at an angle of 30° for 360s;

[0051] Fig.17 It is a schematic diagram of quantitative partitioning results;

[0052] Fig.18 This is a schematic diagram of the maximum temperature difference in five areas of 480 groups of samples;

[0053] Fig.19 It is a schematic diagram of the basic structure of BP neural network;

[0054] Fig. 20 Schematic diagram of temperature comparison with the addition of error compensation mechanism. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] In order to verify the effectiveness of the active control technology of the non-uniform temperature field of the immersed tube tunnel, the following is based on a simulated combustion vehicle to run the entire process and verify the methodological nature of the model. Since the implementation and verification process of the method is relatively complicated, the implementation method includes Fluent numerical simulation of rectangular cross-section tunnels, building a tower combustion vehicle to simulate combustion conditions and other preliminary preparations. All work will be rigorously simulated to ensure the rationality of the verification effect.

[0057] This embodiment provides a method for zoning and actively controlling the transverse non-uniform temperature field of a rectangular immersed tube tunnel, specifically:

[0058] Step 1: Determine the basic simulation verification parameters such as the burning car, flamethrower, tunnel size, etc. (This step is to better demonstrate and verify the specific process of the invention);

[0059] A small tower combustion vehicle is used to carry a burner to simulate fire sources at different locations to meet the requirements of tunnels of different sizes and types. The cross-section of the small tower combustion vehicle is designed with a bottom width of 1150mm, a top width of 525mm, and a height of 840mm. Figure 3 (a) is shown. In addition, the length of the combustion vehicle is set to 1.1m. In order to ensure that the small tower combustion vehicle forms an approximately uniform temperature field in the longitudinal direction of the immersed tube tunnel and to ensure that the distance between the two burners is less than or equal to 1200mm, the small tower combustion vehicle adopts two rows of burners in the longitudinal arrangement, and the distance is set to 550mm, as shown in FIG. Figure 3 (b) as shown.

[0060] A model of immersed tube tunnel that needs to be equipped with a small tower combustion vehicle was selected. According to the specification of "Highway Tunnel Design Detailed Rules", the size of the 1:2 scale immersed tube tunnel model was determined, with the length, width and height dimensions of 1.5m, 4m and 3.5m respectively. The cross section is as follows: Figure 4 shown.

[0061] Considering the actual needs of the project, two smoke exhaust ports with a diameter of 281mm are arranged on both sides of the immersed tube tunnel. The burner nozzle diameter is 160mm and the maximum power is 600kW. The small tower burner is located in the center of the tunnel and is 300mm away from the tunnel ground. Combined with the aforementioned dimensions of the small tower burner, the numerical model of the small tower burner and the immersed tube tunnel was finally established through Space Claim. The specific model is as follows: Figure 5 a- Figure 5 As shown in c.

[0062] Step 2: Division of burner control area in non-uniform temperature field.

[0063] Initial division process:

[0064] When a fire occurs in a tunnel, there is a certain difference in temperature between the middle area and the two sides of the tunnel, and the temperature field is non-uniform. According to the distribution law of the tunnel temperature field and the range of the high-temperature control area of ​​the burner mentioned above, the horizontal non-uniform temperature field of the immersed tube tunnel is initially partitioned. The initial partition is based on the approximate control range of the burner in different directions, and the size of the control area is not quantitatively determined. The partition results are as follows: Figure 6 As shown. The five temperature control areas are numbered ①, ②, ③, ④ and ⑤. Among them, temperature control areas ①, ③ and ⑤ are areas directly controlled by the burner; temperature control areas ② and ④ are areas indirectly controlled by the side and upper burners. After the preliminary zoning of the transverse non-uniform temperature field of the immersed tube tunnel, it is necessary to consider the impact of different burner powers on the high-temperature control area of ​​the top burner, and finally determine the specific range of temperature control area ③; it is necessary to consider the impact of different burner angles on the high-temperature control area of ​​the side burner, and finally determine the specific location and range of temperature control areas ① and ⑤; after determining the range of temperature control areas ①, ③ and ⑤, the remaining area is the range of temperature control areas ② and ④. The following quantitatively determines the range of each area through parameter analysis.

[0065] First, during the entire operation and verification process, a simulated grid sensitivity analysis was performed before quantitative partitioning. Based on the established numerical model of the transverse non-uniform temperature field of the immersed tube tunnel, the influence of the grid size on the calculation accuracy was analyzed. Four grid sizes were selected for numerical simulation and comparative analysis. The grid sizes were 0.06m, 0.08m, 0.1m and 0.2m, and the corresponding grid numbers were 320,000, 220,000, 140,000 and 70,000, respectively. Ten measuring lines were evenly arranged at 1 cm below the ceiling in the transverse center section of the immersed tube tunnel. Three measuring points were arranged on each measuring line, and the average temperature of these three measuring points was taken as the measuring line temperature. With the number of grids as the only variable, all burners were turned on for numerical simulation, and the simulation calculation time was set to 400s.

[0066] Table 1 Average temperature error of measurement lines with different grid numbers

[0067]

[0068] according to Figure 7 As shown in Table 1, when the grid size is 0.1m and 0.08m, the average temperature error of the measurement line is within 5%, which is 3.33% and 0.91% respectively. Considering the calculation speed and calculation accuracy, the final grid size is 0.1m.

[0069] Before quantitative zoning of the high temperature control area of ​​the top burner, according to the above burner layout plan, it has been determined that the top burners of the small tower burner are arranged in two rows, one in each row, and are located on the longitudinal center line of the burner and perpendicular to the top of the burner, spraying flames directly above the immersed tube tunnel. Figure 8 shown.

[0070] Quantitative division process:

[0071] Next, the high temperature control area of ​​the top burner on the top of the immersed tunnel cross section is further determined, and the factor considered is the influence of the power of the burner on the high temperature area it controls. The two top burners are started at the same time, with the power of a single burner as the only variable, and the burner power is designed to be 170kW (low fire) and 340kW (high fire) for two simulation conditions, and the calculation time is set to 360s.

[0072] In order to reflect the temperature changes in various areas near the inner wall of the immersed tube tunnel, measurement lines are evenly arranged at intervals of 200mm around it, and 3 measurement points are arranged on each measurement line. The average temperature of the 3 measurement points is taken as the measurement line temperature. By comparing and analyzing the heating speed and temperature of each measurement line, the location of the high-temperature area measurement line is determined. In order to quantitatively divide the high-temperature control area of ​​the top burner, a total of 16 measurement lines are arranged in the top area of ​​the immersed tube tunnel. The measurement lines are numbered 16 to 31. The specific arrangement of the measurement lines is as follows: Fig. 9 shown.

[0073] The results of the two top burners with a power of 170kW are as follows Fig.10 As shown. It can be found that when the simulation time reaches 360s, the temperature of the 22-25th measurement line is significantly higher than that of the other measurement lines, and the temperature difference of the other measurement lines is small, all within 10°C, and clustered in the figure; 25s before the temperature rise curve, the temperature of the 22-25th measurement line is always slightly higher than that of the other measurement lines, indicating that the 22-25th measurement line starts to heat up before the other measurement lines, and has a high sensitivity to the heat released by the two top burners. Therefore, when the power of the two top burners is 170kW, the area where the 22-25th measurement line is located is determined to be the high temperature control area of ​​the top burner.

[0074] According to the same method, the simulation results of the top two burners with a power of 340kW can be seen Fig.11Compared with the 170kW simulation results, as the burner power increases, the temperature of the 22-25th measurement line is still significantly higher than that of the other measurement lines, and it starts to heat up 25s before the temperature rise curve. Therefore, the area where the 22-25th measurement line is located is finally selected as the high temperature control area of ​​the top burner on the top of the immersed tube tunnel, and it is quantitatively divided with the 21st and 26th measurement lines as the boundaries, with a control range of 1m.

[0075] Next, the control area of ​​the side burner is quantitatively determined. Considering the limitations of actual engineering technology, three conventional flame injection angles are set for the nozzle of the side burner, namely 0°, 30° and 45°. Fig.12 As shown in the figure, since the immersed tube tunnel and the small tower combustion vehicle model are symmetrically distributed in the transverse direction, it is only necessary to study the high temperature control area of ​​the left wall of the immersed tube tunnel by the burner on the left side of the small tower combustion vehicle, and the high temperature control area of ​​the burner on the right side can be obtained by the mirroring method.

[0076] First, the burner angle is determined, and the power of the left burner at a certain angle is taken as the only control variable. Two simulation working conditions (170 kW and 340 kW) are also designed to study the influence of the side burner power change on the high-temperature control area on the side of the immersed tube tunnel. The calculation time is set to 360 s. The specific information of the working conditions is shown in Table 2.

[0077] In addition, in order to verify the rationality of the side high-temperature control area obtained by this method, the left and right burners were turned on at the same time for numerical simulation. The purpose of condition No. 5 in Table 2 is to compare whether the high-temperature control areas on both sides are symmetrically distributed.

[0078] Table 2 Simulated working conditions of the left burner

[0079]

[0080]

[0081] The measurement line arrangement scheme of the side burner is the same as that in Section 3.2. In order to study the temperature influence range of the side burner of the small tower combustion vehicle on the side areas of the immersed tube tunnel, 15 measurement lines are arranged on the left and right sides of the immersed tube tunnel. The measurement lines in the left area are numbered 1 to 15, and the measurement lines in the right area are numbered 32 to 46. The specific arrangement of the measurement lines is as follows: Fig.13 shown.

[0082] Fig.14Temperature cloud maps at different angles are displayed. When the angle of the burner nozzle on the left is 0°, the cloud map shows that the lower left side of the tunnel is darker in color, and the high temperature area is located in the lower middle part of the left side of the tunnel, close to the bottom of the left side of the tunnel; when the angle of the burner nozzle on the left is 30°, the middle left side of the tunnel in the cloud map is the darkest, followed by the upper and top parts, the high temperature area is located in the middle, and begins to have a certain impact on the temperature of the top area; when the angle is 45°, the upper left side of the tunnel in the cloud map is the darkest, and the burner flame almost touches the top of the left side of the tunnel, which has a more significant impact on the temperature of the upper middle and top areas of the left side of the immersed tube tunnel, and there is a certain overlap with the effect of the burner on the top of the small tower combustion vehicle.

[0083] Further by Fig.15 It can be seen that when the angle between the left burner nozzle and the horizontal plane is 30°, the temperature of the measuring line is divided into three gradients, among which the measuring lines 7 to 10 are located in the first gradient, and the temperature is slightly higher than that of the other measuring lines; therefore, the high temperature control area of ​​the burner is located near the measuring lines 7 to 10, in the middle of the left side of the tunnel. This shows that under the 30° temperature rise curve of the side burner, its control area can be quantitatively taken as the measuring lines 7 to 10.

[0084] In addition, when the simulation time is 360s, the temperature cloud diagram of the left burner with a power of 170kW and 340kW is shown in Figure 2. Fig.16 As shown. It can be observed that, regardless of the burner power, after the left burner is turned on for 360s, the contours of the temperature control areas formed on the left side of the immersed tube tunnel are roughly the same, and the temperature color in the same contour becomes darker as the power increases, indicating that the change in the left burner power will not affect the range of the various temperature control areas formed in the immersed tube tunnel, but will only affect the temperature level in each temperature control area. Therefore, in the factor analysis of the side burner, only the burner opening angle needs to be considered.

[0085] In summary, in this case, the high temperature control area of ​​the top burner is quantitatively divided with the No. 21 and No. 26 measuring lines at the top of the immersed tube tunnel as the boundary, and the area width is 1m. Fig.17 Temperature control area ③.

[0086] The high temperature control area of ​​the side burner is quantitatively divided with the No. 6 and No. 11 measuring lines on the left side of the immersed tube tunnel as the boundary, and the area width is 1m. Fig.17 The control area of ​​the burner on the right is symmetrically divided quantitatively with the 36th and 41th measuring lines on the right side of the immersed tube tunnel as the boundary, and the area width is 1m. Fig.17 Temperature control area No. ⑤.

[0087] After quantitatively dividing the specific ranges of temperature control areas ① and ③, the remaining area in the middle is temperature control area ②, which is the indirect control area jointly controlled by the left burner and the top burner; similarly, temperature control area ④ can be obtained, which is the indirect control area jointly controlled by the right burner and the top burner.

[0088] Step 3: Establishment of machine learning active control model for non-uniform temperature field.

[0089] Before the immersed tunnel fire test, the temperature rise curves of the five temperature control areas and the initial rate (0kW) of the first time period of the three-row burner were known. During the test, in order to ensure that the five temperature control areas are heated strictly according to the known temperature rise curves, it is necessary to accurately predict the power required for the three-row burners on the small tower burner vehicle based on the known temperature rise curves of the five temperature control areas and the initial power of the three-row burners with the help of a neural network algorithm.

[0090] Take a single burner controlling a single temperature control zone as an example ( Figure 1 ), firstly, before the test, the target temperature rise curve of the temperature control area is known. Taking into account the time interval between the temperature acquisition instrument and the burner regulating valve, the target temperature rise curve is divided into several sections with a time period of 10s, and the start and end temperatures of each time period are obtained; by using numerical simulation, sample data of the start and end temperatures of the area, the initial power of the three rows of burners and the required setting power of the burner are randomly constructed; the rules of the samples are learned by a neural network, so that before the test, the power required by the burner in each time period is predicted according to the start and end temperatures of each time period of the known temperature rise curve and the initial power of the three rows of burners, and these powers are input into the burner control system; finally, a neural network prediction error compensation algorithm is proposed. In the experiment, the prediction value of the neural network is adjusted in real time according to the deviation between the actual temperature of the previous time period and the target temperature.

[0091] The control process of a single burner controlling a single temperature control area is mapped to three rows of burners controlling five temperature control areas, which is the active control process that needs to be implemented in this chapter, such as Figure 2 The specific process is as follows:

[0092] Step 1: Divide the temperature rise curve. Divide the known temperature rise curves of temperature control areas ① to ⑤ into time periods of 10 seconds to obtain the start and end temperatures of the five temperature control areas in several time periods;

[0093] Step 2: Predict the burner power. Through the neural network algorithm, the power required for the left, top and right burners in the first 10s period is predicted based on the start and end temperatures of the five zones in the first 10s period and the initial power of the three rows of burners, which is 0kW; the power required for the left, top and right burners in the second 10s period is predicted based on the start and end temperatures of the five zones in the second 10s period and the predicted power of the three rows of burners in the first 10s period as the initial power; and so on, the power required for the burners in all time periods is obtained;

[0094] Step 3: Input the burner power. Before the test begins, input the power required by the three rows of burners in each time period into the burner control system;

[0095] Step 4: Monitor the temperature of the temperature control area. Monitor the temperature changes of the five temperature control areas separately, and calculate the average deviation between the actual temperature and the target temperature of the five temperature control areas in each time period;

[0096] Step 5: Burner power prediction error compensation: During the test, all burner powers predicted by the neural network are uniformly increased or decreased according to the cumulative temperature difference change trend and the average temperature difference in each time period.

[0097] In the above step 2, it is necessary to establish a neural network model to predict the burner power. The model establishment process is described in detail below.

[0098] First, estimate the sample size using the following empirical formula:

[0099] (1) Based on the network training error estimation formula

[0100]

[0101] n w =l(m+1)+m(n+1) (2)

[0102] Where: N is the sample set capacity; ε is the training error; l is the number of input layer nodes; m is the number of hidden layer nodes; n is the number of output layer nodes.

[0103] (2) Estimation formula based on the number of neurons in each layer of the neural network

[0104] N=(2~10)×(lm+mn) (3)

[0105] The estimation of the sample set capacity requires the determination of the number of nodes in the hidden layer of the neural network, and the number of nodes can only be determined after a large number of trials. Before the neural network trial calculation begins, the number of nodes in the hidden layer can be roughly estimated by the estimation formula based on the number of nodes in the input layer and the number of nodes in the output layer. The formula is as follows:

[0106]

[0107] m=log2l (5)

[0108]

[0109] Wherein: a is a natural number between 1 and 10.

[0110] The number of nodes in the input layer of this model is 13, and the number of nodes in the output layer is 3. According to the estimation of formulas (4), (5) and (6), the number of nodes in the hidden layer of this model ranges from 4 to 14. Assuming that the training error is 10%, the sample set capacity of this model ranges from 128 to 2510 through the estimation of formulas (1), (2) and (3).

[0111] Next, the orthogonal design was used to design the experiment, which is an efficient experimental design method that aims to minimize the number of experiments while maximizing the acquisition of information. In the active control process of the transverse non-uniform temperature field of the immersed tube tunnel, the power required for the three rows of burners on the small tower combustion vehicle during the time period (3 data) was predicted by the neural network algorithm based on the start and end temperatures of the five temperature control zones and the initial power of the three rows of burners in the 10s time period (13 data). Therefore, in the neural network, 13 data are used as input parameters (the start and end temperatures of the five temperature control zones and the initial power of the three rows of burners), and 3 data are used as output parameters (the required power of the three rows of burners). In the numerical simulation, it is necessary to set the power of the three rows of burners and the start temperature of the five temperature control zones, and obtain the end temperature of the five temperature control zones after 10s of combustion, so as to obtain the corresponding samples of input and output.

[0112] In Fluent software, the burner on the small tower combustion vehicle can be simulated through the velocity inlet in the boundary conditions. The power of the burner mainly depends on two parameters: the velocity and temperature of the velocity inlet. Since the flame temperature of the laboratory burner nozzle is about 1500℃, the inlet temperature is set to 1800K (about 1500℃) in the Fluent simulation to be close to the actual situation.

[0113] In the sample design, the power of the burner is changed by changing the speed of the speed inlet; by setting different starting heating rates, the burner can change based on different power levels; by setting different change rates, the burner can achieve different power increases within 10 seconds. This design ensures that the burner undergoes different power changes based on different starting powers within 10 seconds, forming different temperature fields in the five temperature control areas, thereby reflecting the corresponding relationship between the burner power change on the small tower combustion vehicle and the temperature change in the five areas.

[0114] It is known that the maximum power of the burner on the small tower combustion vehicle is 600kW, corresponding to a velocity inlet velocity of 19m / s and a temperature of 1800K (about 1500℃). Therefore, the maximum inlet velocity that the burner can achieve in the sample design is 19m / s. The factors and levels of the orthogonal design of the sample are shown in Table 3.

[0115] Table 3 Number and value of factor levels

[0116]

[0117] According to the orthogonal design method, the orthogonal design is carried out based on the 4 factors and 6 levels in the table, and the sample size is L 48 (6 4 )=48. Considering that the same power will have different temperature rise performance based on different ambient temperature levels, each sample is calculated at 10 different ambient temperature levels, which are 300K, 400K, 500K, 600K, 700K, 800K, 900K, 1000K, 1100K and 1200K (27℃, 127℃, 227℃, 327℃, 427℃, 527℃, 627℃, 727℃, 827℃, 927℃). Therefore, the number of samples finally obtained is L 48 (6 4 )×10=480, which is within the range of sample set capacity.

[0118] Next, the overall sample analysis was carried out to ensure the non-uniformity of the temperature in the temperature control area obtained by the implementation method. According to the sample plan, a total of 480 samples were obtained by numerical simulation, and the maximum temperature difference (highest end temperature - lowest end temperature) of the end temperature of the five temperature control areas of each sample was calculated, as shown in Fig.18 Analysis shows that the maximum difference in the end temperature of the five temperature zones of most samples is about 50°C, and the maximum difference in the end temperature of the five temperature zones reaches 218°C, indicating that the end temperatures of the five temperature control zones obtained by numerical simulation have certain non-uniformity.

[0119] The normalization method of the neural network is the maximum value method to normalize the data values ​​to [0,1]. The formula is as follows:

[0120]

[0121] In order to enhance the training effect of the model, this model uses a ratio of 7:1.5:1.5 to divide the training set, validation set, and test set. The number of samples in the training set is 336, the number of samples in the validation set is 72, and the number of samples in the test set is 72.

[0122] The BP neural network is used as the training model, and its structure diagram can be seen Fig.19, the input layer nodes are the start and end temperatures of the five temperature control zones and the initial powers of the three rows of burners, a total of 13; the output layer nodes are the required powers of the three rows of burners, a total of 3. After trial calculation, the number of hidden layer nodes is 11 (Table 4).

[0123] Table 4 BP neural network hidden layer node trial calculation results

[0124]

[0125] According to the same trial calculation method, the initialization method, activation function, training function, error function, training accuracy, and training rounds of the BP network are tested. Finally, the BP network hyperparameter selection is shown in Table 5:

[0126] Table 5 BP neural network model parameter configuration

[0127]

[0128]

[0129] In order to test the real prediction ability, i.e., generalization ability, of the trained BP neural network model, performance evaluation indicators are usually used for direct response. This model uses mean relative error (MRE) as the performance evaluation indicator, as shown in formula (8).

[0130]

[0131] Where: y i is the predicted value; is the true value.

[0132] The test set is predicted by the trained BP neural network, and the average relative error between the predicted result and the target result is calculated. At the same time, the training results of the radial basis model (RBF) and the radial basis model after particle swarm optimization (PSO-RBF) are used as comparison to verify the superiority of the method, and the results are shown in Table 6.

[0133] It can be seen from the table that the average relative error of the BP neural network test set is within 10%. The model performs well in the test set and has excellent generalization ability, meeting the expected accuracy requirements. Compared with RBF and PSO-RBF neural networks, the performance of the BP neural network is more superior. Therefore, the BP neural network is finally selected as the prediction model for the lateral non-uniform temperature field of the immersed tube tunnel.

[0134] Table 6 Average relative error of BP neural network test set

[0135]

[0136] Step 4: Neural network error compensation mechanism.

[0137] In order to make the actual temperature closer to the target temperature during the test, a prediction error compensation mechanism is proposed. This mechanism counts the average temperature error of the five temperature control areas every 10 seconds. The specific process of this mechanism is as follows:

[0138] Step 1: Calculate the average temperature error. Count the average temperature error between the actual temperature and the target temperature in the five temperature control areas in the first 10s period;

[0139] Step 2: Determine the cumulative trend of temperature error (qualitative adjustment). When the cumulative trend of temperature error increases, the actual temperature is greater than the target temperature, and all burner power values ​​predicted by the neural network need to be adjusted down; when the cumulative trend of temperature error decreases, the actual temperature is less than the target temperature, and all neural network prediction values ​​need to be adjusted up;

[0140] Step 3: Adjust the neural network prediction value according to the average temperature error (quantitative adjustment). According to experience, when the average temperature error is 0-50℃, the prediction value is increased by 1.1 times, when the average temperature error is -50-0℃, the prediction value is decreased by 0.9 times; when the average temperature error is 50-100℃, the prediction value is increased by 1.2 times, when the average temperature error is -100--50℃, the prediction value is increased by 0.8 times, and so on;

[0141] Step 4: Repeat steps 1 to 3 until the test is stopped.

[0142] Based on the above process, assuming that the five temperature control areas all need to achieve the temperature of the first 60s of the HC curve, according to the 10s division principle, the start and end temperatures of the five temperature control areas in the six time periods and the initial power (0kW) of the three-row burners are obtained. The power of the three-row burners at the six time nodes is predicted by the neural network, and the power is input into Fluent for numerical simulation to obtain the temperature rise curves of the five areas. Tables 7 and 8 show the power prediction values ​​of the burners at the time nodes before and after the error compensation mechanism is added.

[0143] Table 7 Prediction values ​​of six time nodes for three rows of burners

[0144]

[0145] Table 8 Adjustment values ​​of six time nodes for three-row burners

[0146]

[0147] It can be seen from Tables 7 and 8 that compared with the unadjusted neural network prediction value, since the actual temperature is lower than the target temperature in the first 10s time period and the deviation is within 50°C, the burner power of the 10s node is amplified by 1.1 times; in the remaining time periods, the actual temperature is lower than the target temperature, so the burner power is reduced.

[0148] The data is adjusted using the aforementioned error compensation method. Fig. 20 The target temperature, temperature before adjustment, and temperature after adjustment of the five areas are shown. It can be found that after adding the prediction error compensation mechanism, the errors between the actual temperature and the target temperature of the five temperature control areas at 60s are reduced. At 60s, the errors of the temperature control areas ①, ②, ③, ④, and ⑤ are 15%, 4%, -5%, -2%, and 2%, respectively, indicating that the mechanism has a certain compensation effect on the error of the predicted power of the three rows of burners; area ① is still the area with the largest error, indicating that the compensation mechanism can play a role in overall adjustment.

[0149] At this point, the operation and verification process of the entire invention has been demonstrated.

[0150] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for partitioning and actively controlling the transverse non-uniform temperature field of a rectangular immersed tube tunnel, characterized in that: include: S1. Temperature field division in tunnel area; The specific steps include: S11, preliminarily determining the burner control area, that is, roughly determining the non-uniform temperature field area controlled by the burner according to the location of the burner; S12, quantitatively determine the control area, that is, determine the specific size of the control area; S2. Active control of the temperature in each zone; using machine learning technology to train a neural network model to control the burner; the specific steps include: S21. Through Fluent numerical simulation, the corresponding relationship between the power of three rows of burners on the small tower combustion vehicle and the temperature and initial power of five temperature control areas is constructed to obtain a sample set; S22. Use the sample set to train the machine learning algorithm and construct a prediction model for the lateral non-uniform temperature field of the immersed tube tunnel. Before training the model, the sample set capacity needs to be estimated, and the sample scheme needs to be designed using the orthogonal design method. In addition, reasonable training, verification and test sets need to be generated through sample data analysis and standardized sample set division. S3, temperature prediction error compensation: the average temperature error of the five temperature control areas is calculated every 10 seconds, the purpose of which is to reduce the prediction error of the burner power and improve the restoration accuracy of the temperature rise curves of the five temperature control areas.

2. The method for zoning and actively controlling the transverse non-uniform temperature field of a rectangular immersed tube tunnel according to claim 1, characterized in that: In step S12, the burners are divided into top burners with a flame outlet facing the ground at an angle of 90° and side burners with other side flame outlets according to the nozzle angle, and the two burners control the top and side temperature fields respectively; The top temperature field determines the range of the controlled high temperature area according to the power of the top burner. At this time, the temperature control area of ​​the top burner is divided according to the temperature area with high sensitivity, that is, the temperature area that starts to heat up earlier is the temperature control area of ​​the top burner; The side temperature field is determined by the burner nozzle scheduling, that is, the angle between the burner and the horizontal line to determine the range of the high-temperature area controlled by it; the principle of dividing the side temperature field is that the control range of the side burner should be focused on the tunnel side wall as much as possible to avoid providing a large amount of heat to the top and bottom; the entire process is simulated by Fluent software to perform quantitative zoning of the non-uniform temperature field.

3. The method for zoning and actively controlling the transverse non-uniform temperature field of a rectangular immersed tube tunnel according to claim 1, characterized in that: In step S22, taking a single burner controlling a single temperature control area as an example, the control process of a single burner controlling a single temperature control area is mapped to three rows of burners controlling five temperature control areas, which is the active control process to be implemented. The specific process is as follows: Step 1: Divide the temperature rise curve; divide the known temperature rise curves of the five temperature control areas ① to ⑤ into time periods of 10s, and obtain the start and end temperatures of the five temperature control areas in several time periods; Step 2: Predict the burner power; through the neural network algorithm, the power required to be set for the left, top and right burners in the first 10s time period is predicted based on the start and end temperatures of the five zones in the first 10s time period and the initial power of the three rows of burners, which is 0kW; the power required to be set for the left, top and right burners in the second 10s time period is predicted based on the start and end temperatures of the five zones in the second 10s time period and the predicted power of the three rows of burners in the first 10s time period as the initial power; and so on, the power required to be set for the burners in all time periods is obtained; Step 3: Input the burner power: Before the test begins, input the power required by the three rows of burners in each time period into the burner control system; Step 4: Monitor the temperature of the temperature control area; Monitor the temperature changes of the five temperature control areas respectively, and calculate the average deviation between the actual temperature and the target temperature of the five temperature control areas in each time period; Step 5: Burner power prediction error compensation: During the test, all burner powers predicted by the neural network are uniformly increased or decreased according to the cumulative temperature difference change trend and the average temperature difference in each time period.

4. The method for zoning and actively controlling the transverse non-uniform temperature field of a rectangular immersed tube tunnel according to claim 1, characterized in that: The specific process of the temperature prediction error compensation step is as follows: Step 1: Calculate the average temperature error; count the average temperature error between the actual temperature and the target temperature of the five temperature control areas in the first 10s period; Step 2: Determine the cumulative change trend of temperature error and make qualitative adjustments; When the cumulative trend of temperature error increases, the actual temperature is greater than the target temperature, and all the burner power values ​​predicted by the neural network need to be reduced; when the cumulative trend of temperature error decreases, the actual temperature is less than the target temperature, and all the predicted values ​​of the neural network need to be increased; Step 3: Adjust the predicted value of the neural network according to the average temperature error, that is, quantitative adjustment; when the average temperature error is 0-50°C, the predicted value is increased by 1.1 times, when the average temperature error is -50-0°C, the predicted value is decreased by 0.9 times; when the average temperature error is 50-100°C, the predicted value is increased by 1.2 times, when the average temperature error is -100--50°C, the predicted value is increased by 0.8 times, and so on; Step 4: Repeat steps 1 to 3 until the test is stopped.

Images