Tunnel fire temperature field prediction and safety area division method and system

Abstract

The present invention relates to the field of data processing technology, in particular to a method and system for predicting the temperature field of a tunnel fire and dividing a safe area, the method comprising the following steps: preprocessing historical data of a tunnel fire to obtain standardized data; establishing a combustion equation for a tunnel fire based on the standardized data; constructing a combustion model for a tunnel fire based on the combustion equation; training the combustion model using the standardized data; predicting the temperature field of a tunnel fire based on the trained combustion model, and visualizing the prediction results for dividing safe areas. The present invention constructs a combustion model for a tunnel fire based on a neural network with physical constraints, selects the most suitable combustion model according to different tunnel fire conditions, predicts the temperature field of a tunnel fire in real time, and visualizes the prediction results, effectively improving the accuracy and real-time performance of tunnel fire predictions, and dividing safe areas based on the visualized results, providing strong support for emergency rescue.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, in particular to a method and system for predicting a tunnel fire temperature field and dividing a safe area. Background Art

[0002] Tunnel fires are often highly dangerous and difficult to predict due to their enclosed environment and complex flow fields. Their long and narrow geometry leads to complex thermodynamic phenomena such as turbulence, thermal convection, radiation, and chemical reactions. The strong coupling of these phenomena complicates the temperature field, smoke diffusion, and flame propagation. Therefore, developing accurate numerical simulation and prediction models is crucial for temperature field prediction, real-time monitoring, emergency rescue, and fire control in tunnel fires.

[0003] Traditional numerical simulation techniques primarily include Reynolds averaging, large eddy simulation (LES), and direct numerical simulation (DNS). The Reynolds averaging method simplifies computation by averaging the transient behavior of large and small eddies. However, it struggles to capture transient turbulence details and local vortex structure, leading to underestimation of the impact of turbulence on flame propagation and smoke diffusion. In contrast, LES and DNS can more accurately simulate the complex physical processes involved in tunnel fires. However, DNS requires extremely high computational resources and is generally only suitable for small-scale or simplified scenarios. While LES excels in reducing computational costs, its computational efficiency in large, complex scenarios remains to be improved. With the rapid advancement of computer technology, data-driven prediction techniques have become a hot topic of research. These methods can provide rapid predictions. However, purely data-driven models, such as long-short-term memory networks (LSTMs) and artificial neural networks (ANNs), lack physical constraints and can produce predictions inconsistent with the physical field. To address this issue, physics-based neural network algorithms have been applied in engineering practice. By predicting flow field parameters and incorporating them into and solving partial differential equations, they enhance prediction accuracy and ensure physical consistency. However, their computational speed still cannot fully meet the real-time requirements of tunnel fire rescue.

[0004] Therefore, considering the advantages of numerical simulation technology and artificial intelligence algorithms, the tunnel fire combustion equation established based on the large eddy simulation numerical model is used as the physical governing equation. The residual of this physical equation is embedded in the loss function of the neural network. This not only improves the physical consistency of the model but also significantly increases computational efficiency. This method is applicable to complex tunnel fire environments and can predict the tunnel fire temperature field and demarcate safe areas, providing a basis for emergency rescue. Summary of the Invention

[0005] In view of the defects in the prior art, the present invention provides a method and system for predicting the temperature field of a tunnel fire and dividing the safe area.

[0006] To achieve the above objectives, in a first aspect, the present invention provides a method for predicting the temperature field of a tunnel fire and dividing a safe area. The method comprises the following steps: preprocessing historical data of tunnel fires to obtain standardized data; establishing a combustion equation for tunnel fires based on the standardized data; constructing a combustion model for tunnel fires based on the combustion equation; training the combustion model using the standardized data; predicting the temperature field of a tunnel fire based on the trained combustion model, and visualizing the prediction results for use in dividing a safe area. The present invention introduces physical constraints into a neural network to construct a combustion model, trains the combustion model based on historical data, and adapts it to different tunnel fire conditions. This allows for real-time prediction of the temperature field of various complex tunnel fires, visualizing the prediction results, and then dividing a safe area based on the image data. In practical applications, this method can provide a scientific reference for emergency rescue decisions in a timely manner, and has important practical value.

[0007] Optionally, preprocessing the historical data of tunnel fires to obtain standardized data includes: obtaining the historical data of tunnel fires; performing interval mapping on the spatial temperature field of the historical data; and standardizing the historical data. The present invention utilizes historical data from past tunnel fires as data support and standardizes the historical data, providing a rigorous and reliable data source for the subsequent determination and training of combustion models. This effectively improves the prediction accuracy and generalization capabilities of the combustion model, significantly enhancing the feasibility and practicality of the present invention.

[0008] Optionally, establishing a combustion equation for tunnel fires based on the standardized data includes: obtaining a mass conservation expression during the combustion process; obtaining a momentum conservation expression that accounts for the influence of wind speed during the combustion process; constructing an energy equation expression that accounts for the influence of thermal radiation during the combustion process; and obtaining a mass fraction equation expression that accounts for the combustion rate during the combustion process. The present invention introduces physical constraints by establishing mass, momentum, energy, and mass fraction conservation equations for tunnel fires, which effectively improves the physical consistency of the combustion model of the present invention while significantly increasing computational efficiency. Combined with the constraints of physical conditions, the present invention can better adapt to complex tunnel fire environments, thereby enhancing the flexibility and practical value of the present invention.

[0009] Optionally, the combustion equation includes:

[0010] The mass conservation expression is:

[0011]

[0012] Momentum conservation expression:

[0013]

[0014] Energy equation expression:

[0015]

[0016] The mass fraction equation expression is:

[0017]

[0018] in, is the fluid density, For time, is the divergence operator, For speed, is the fluid velocity vector, is the gradient operator, For pressure, is the molecular viscosity coefficient, is the subgrid viscosity coefficient, is the acceleration due to gravity, is the subgrid stress tensor, The wind force factor is is the velocity vector of the external wind, is the velocity vector of the wind inside the tunnel, is enthalpy, is the molecular thermal conductivity, is the subgrid eddy thermal conductivity, is the temperature, is the volume heat source term, is the radiation heat flux vector, is the quality score, is the molecular diffusion coefficient, is the subgrid turbulent diffusion coefficient, is the generation rate. This method describes the physical processes involved in combustion based on fluid dynamics. It constructs conservation expressions for mass, momentum, energy, and mass fraction based on standardized historical tunnel fire data. This digitizes complex tunnel fire conditions, facilitating practical processing and analysis. This provides physical constraints for the subsequent construction of a universal tunnel fire combustion model, enhancing the applicability and physical consistency of the combustion model, and effectively improving the rigor and reliability of the method.

[0019] Optionally, constructing a combustion model for tunnel fires based on the combustion equation includes: obtaining model parameters of the combustion model based on the historical data; obtaining residuals of the mass, momentum, energy, and mass fraction equations respectively; and constructing a loss function to obtain the combustion model. The present invention establishes model parameters for the combustion model based on historical data of tunnel fires, while simultaneously establishing physical conservation equations and introducing physical constraints. This allows the constructed combustion model to adapt to various complex tunnel fire environments, while enhancing the accuracy of the combustion model and providing more accurate and reliable prediction results.

[0020] Optionally, constructing the loss function to obtain the combustion model includes:

[0021]

[0022] in, is the loss function, is the mean square error between the predicted value and the true value, is the predicted value, is the true value, is the weight coefficient, 、 、 and are the residuals of the mass, momentum, energy and mass fraction equations, respectively, 、 、 and are the weights of each physical residual term respectively. The present invention obtains the residuals based on the obtained mass, momentum, energy and mass fraction equations, embeds the physical residuals into the loss function of the neural network with corresponding weights, and uses the physical residuals as a regularization term to prevent the constructed combustion model from overfitting. Introducing physical constraints into the combustion model can effectively capture the process of vortex shedding during turbulent combustion, making the constructed combustion model more consistent with the actual tunnel fire situation, and improving the generalization ability of the combustion model, so that the combustion model can better adapt to the complexity of the actual tunnel fire scene, and provide strong support for the practical application of the combustion model.

[0023] Optionally, the use of the standardized data to train the combustion model includes: using an optimizer to set training parameters for the combustion model; importing the standardized data into the combustion model; and training the combustion model using the optimizer. The present invention imports a large amount of standardized historical data on tunnel fires into the combustion model, uses an optimizer to train the constructed combustion model, and optimizes the combustion model, which can effectively improve the prediction accuracy of the combustion model and enhance the reliability of the combustion model. This allows the combustion model constructed by the present invention to be flexibly applied to various complex tunnel fire scenarios in practical applications, greatly enhancing the practicality of the present invention.

[0024] Optionally, the temperature field of the tunnel fire is predicted based on the trained combustion model, and the prediction results are visualized for dividing the safe area, including: selecting a suitable combustion model based on the monitoring data of the detection points in the tunnel; importing the monitoring data into the combustion model to predict the temperature field of the tunnel fire in real time; visualizing the prediction results to obtain a three-dimensional visualization image; and dividing the safe area according to the three-dimensional visualization image. The present invention selects the most suitable combustion model based on the monitoring data in the tunnel, which significantly improves the rigor of the present invention. Based on the real-time monitoring data, the combustion model is used to predict the temperature field of the tunnel fire in real time, effectively improving the real-time nature of the prediction, and further visualizing the prediction results. The safe area is divided according to the obtained image data, which enhances the practical value of the method of the present invention and has important reference value for emergency safety rescue decision-making.

[0025] Optionally, the performing image processing on the prediction result to obtain a three-dimensional visualization image includes:

[0026]

[0027] in, For the three-dimensional data visualization process, Indicates three-dimensional image rendering. It is a three-dimensional grid structure. is the constructed continuous temperature field function, is the interpolation function, is a three-dimensional data point set, is the predicted temperature value, are the three-dimensional point coordinates, and is the total number of traversed points. The present invention constructs a continuous temperature field function through an interpolation function and renders a three-dimensional image to obtain a three-dimensional temperature field visualization image. This can reflect the real-time situation of a tunnel fire, allowing emergency rescue personnel to quickly grasp the temperature field distribution, providing a reference for demarcating safe areas, effectively improving the emergency response capability to tunnel fires, significantly enhancing the practicality of the method, and providing a scientific basis for timely safety rescue decisions.

[0028] In a second aspect, the present invention provides a system for predicting the temperature field of a tunnel fire and dividing it into safe areas. The system implements the method for predicting the temperature field of a tunnel fire and dividing it into safe areas provided by the present invention. The system includes an input device, an output device, a processor, and a memory. Its benefits lie in: the hardware facilities integrated by the present invention have excellent performance, the input device, output device, processor, and memory are interconnected, and information transmission between the various components is smooth. Through the interaction of multiple hardware facilities, an efficient information processing system is constructed. The system of the present invention introduces the physical constraints of tunnel fires into a neural network, further constructs a combustion model for tunnel fires, improves the physical consistency of the combustion model, and makes the constructed combustion model applicable to various complex tunnel fire scenarios, thereby performing real-time predictions of the temperature field of tunnel fires and further visualizing the prediction results, thereby providing a reference for dividing safe areas and providing a reference for emergency safety and rescue decisions for tunnel fires. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a flow chart of a method for predicting the temperature field and dividing safe areas in a tunnel fire according to an embodiment of the present invention;

[0030] Figure 2 This is a front view comparison diagram of the actual value and predicted value of the tunnel fire temperature field according to an embodiment of the present invention;

[0031] Figure 3 This is a top-down comparison diagram of the actual and predicted values ​​of the tunnel fire temperature field according to an embodiment of the present invention;

[0032] Figure 4 This is a comparison diagram of the actual value and predicted value of the tunnel fire temperature field according to an embodiment of the present invention;

[0033] Figure 5 This is a framework diagram of the tunnel fire temperature field prediction and safety area division system according to an embodiment of the present invention. Specific embodiments

[0034] Specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the present invention. In the following description, numerous specific details are set forth to provide a thorough understanding of the present invention. However, it will be apparent to one of ordinary skill in the art that these specific details are not necessarily required to practice the present invention. In other instances, well-known circuits, software, or methods are not specifically described to avoid obscuring the present invention.

[0035] Throughout this specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the present invention. Therefore, appearances of the phrases "in one embodiment," "in an embodiment," "an example," or "an example" in various places throughout this specification are not necessarily all referring to the same embodiment or example. Furthermore, the particular features, structures, or characteristics may be combined in any suitable combinations and / or subcombinations in one or more embodiments or examples. Furthermore, those of ordinary skill in the art will appreciate that the figures provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0036] See Figure 1 One embodiment of the present invention provides a method for predicting a tunnel fire temperature field and dividing a safe area, the method comprising the following steps:

[0037] S1. Preprocess the historical data of tunnel fire to obtain standardized data.

[0038] S1 specifically includes the following steps:

[0039] S11. Obtain historical data on tunnel fires.

[0040] Specifically, a processor is used to obtain historical data of tunnel fires through numerical simulation software, and the historical data is stored in a memory, thereby ensuring the reliability of the data source of the present invention and providing accurate data support for the present invention.

[0041] S12. Perform interval mapping on the spatial temperature field of the historical data.

[0042] Specifically, using a processor to establish a temperature interval mapping function and storing the obtained temperature interval mapping function in a memory includes:

[0043]

[0044] in, is the temperature interval mapping function, is the temperature, Indicates rounding down.

[0045] S13: Standardize the historical data.

[0046] Specifically, based on the mean and standard deviation of the historical data, the historical data of the tunnel fire is standardized by a processor, and the standardized data obtained after the standardization is stored in a memory, including:

[0047]

[0048] in, To standardize the data, For historical data, is the mean, is the standard deviation.

[0049] S2. Establishing a combustion equation for tunnel fire based on the standardized data.

[0050] Specifically, a combustion equation is obtained using a processor based on the standardized data, and the obtained combustion equation expression is stored in a memory, including:

[0051] The mass conservation expression is:

[0052]

[0053] Momentum conservation expression:

[0054]

[0055] Energy equation expression:

[0056]

[0057] The mass fraction equation expression is:

[0058]

[0059] in, is the fluid density, For time, is the divergence operator, For speed, is the fluid velocity vector, is the gradient operator, For pressure, is the molecular viscosity coefficient, is the subgrid viscosity coefficient, is the acceleration due to gravity, is the subgrid stress tensor, The wind force factor is is the velocity vector of the external wind, is the velocity vector of the wind inside the tunnel, is enthalpy, is the molecular thermal conductivity, is the subgrid eddy thermal conductivity, is the temperature, is the volume heat source term, is the radiation heat flux vector, is the quality score, is the molecular diffusion coefficient, is the subgrid turbulent diffusion coefficient, is the generation rate.

[0060] Furthermore, the Arrhenius equations include:

[0061]

[0062] in, is the generation rate, is the fluid density, is the reaction rate factor, is the activation energy, is the gas constant, is the temperature, is the reactant concentration, is the reaction order.

[0063] Furthermore, the sub-network structure includes:

[0064]

[0065]

[0066]

[0067] in, is the subgrid viscosity coefficient, is the fluid density, is a constant, is the filter width, is the reaction order, is the modulus of shear strain rate, is the subgrid eddy thermal conductivity, is the specific heat capacity at constant pressure, is the turbulent Prandtl number, is the subgrid turbulent diffusion coefficient, is the subgrid viscosity coefficient, is the turbulent Schmidt number.

[0068] S3. Constructing a combustion model of tunnel fire according to the combustion equation.

[0069] S3 specifically includes the following steps:

[0070] S31. Obtain model parameters of the combustion model based on the historical data.

[0071] Specifically, the processor is used to obtain parameters of the neural network according to the nonlinear activation function of the neural network, and the obtained parameters of the neural network are stored in a memory as a data basis for further constructing the combustion model, including:

[0072] Input layer:

[0073]

[0074] Hidden layer 1:

[0075]

[0076] Hidden layer 2:

[0077]

[0078] Hidden layer 3:

[0079]

[0080] Hidden layer 4:

[0081]

[0082] Output layer:

[0083]

[0084] in, For input data, is the output of each hidden layer, is a nonlinear activation function, is the weight, is the bias parameter, is the predicted value.

[0085] S32. Obtain the residuals of the mass, momentum, energy, and mass fraction equations respectively.

[0086] Specifically, the processor is used to obtain the residuals of each equation according to the physical constraint equation, and the obtained residuals are stored in a memory.

[0087] S33: Construct a loss function to obtain the combustion model.

[0088] Specifically, the processor is used to introduce the residual of the physical constraint equation into the loss function as a regularization term with a certain weight, including:

[0089]

[0090] in, is the loss function, is the mean square error between the predicted value and the true value, is the predicted value, is the true value, is the weight coefficient, 、 、 and are the residuals of the mass, momentum, energy and mass fraction equations, respectively, 、 、 and are the weights of each physical residual term.

[0091] S4. Using the standardized data to train the combustion model.

[0092] Specifically, the optimizer is used to set training parameters, and the combustion model is trained based on standardized historical data. The processor is used to minimize the loss function, including:

[0093]

[0094] in, represents minimization, is the weight, is the bias parameter, is the loss function, For input data, is the true value.

[0095] Furthermore, the processor is used to evaluate the prediction performance of the combustion model on the validation set, calculate the mean square error and save it in the memory, including:

[0096]

[0097] in, is the mean square error, is the total number of samples, is the predicted value, is the true value.

[0098] S5. Predicting the temperature field of the tunnel fire based on the trained combustion model, and visualizing the prediction results for dividing the safe area.

[0099] S5 specifically includes the following steps:

[0100] S51. Selecting a suitable combustion model based on monitoring data of detection points in the tunnel.

[0101] Specifically, the processor selects the most appropriate combustion model based on the different input characteristics of the tunnel fire, including:

[0102]

[0103] in, For the combustion model, For input data, is the fluid velocity vector, are tunnel parameters, For pressure.

[0104] Furthermore, a processor is used to construct a hierarchical decision tree for selecting a combustion model, including:

[0105] Determine input features:

[0106]

[0107] in, is the input feature, are different physical quantities in tunnel fire.

[0108] It should be noted that the different physical quantities in the tunnel fire are measured tunnel fire characteristic quantities such as wind speed, temperature and smoke concentration.

[0109] Construct the first-level decision tree to classify the input features of the tunnel fire and determine the general direction of selecting the combustion model, including:

[0110]

[0111] in, is the output category, is the first-level decision tree.

[0112] Based on the first-level decision tree, a second-level decision tree is constructed to further determine the direction of the combustion model, including:

[0113]

[0114] in, is the output category, It is the second-level decision tree.

[0115] Then construct subsequent decision trees at all levels, including:

[0116]

[0117] in, is the output category, For the Layer decision tree.

[0118] Furthermore, a processor is used to connect the combustion model with a multi-layer decision tree in series to quickly select a suitable combustion model, including:

[0119]

[0120] in, For output, For the combustion model, For input, For decision tree.

[0121] S52: Import the monitoring data into the combustion model to perform real-time prediction of the temperature field of the tunnel fire.

[0122] Specifically, the input device is used to input monitoring data of detection points in the tunnel, and the processor uses a combustion model selected by a hierarchical decision tree to make a real-time prediction of the temperature field of the tunnel fire.

[0123] S53: Performing image processing on the prediction result to obtain a three-dimensional visualization image.

[0124] Specifically, a processor is used to construct a continuous temperature field function through an interpolation function, and three-dimensional graphics rendering is performed on the three-dimensional grid temperature data to generate a three-dimensional temperature field visualization image of the tunnel fire and save it in a memory. At the same time, the three-dimensional temperature field visualization image is displayed through an output device, including:

[0125]

[0126] in, For the three-dimensional data visualization process, Indicates three-dimensional image rendering. It is a three-dimensional grid structure. is the constructed continuous temperature field function, is the interpolation function, is a three-dimensional data point set, is the predicted temperature value, are the three-dimensional point coordinates, and is the total number of traversal points.

[0127] See Figure 2 The figure shows a front view comparison diagram of the actual value and predicted value of the tunnel fire temperature field, which shows a front view visualization image of the three-dimensional temperature field of the tunnel fire obtained by the processor, including the actual value temperature field distribution diagram of the tunnel fire and the predicted value temperature field distribution diagram obtained by the present invention.

[0128] See Figure 3 The figure shows a top-down comparison diagram of the actual value and predicted value of the tunnel fire temperature field, which shows a top-down visualization image of the three-dimensional temperature field of the tunnel fire obtained by the processor, including the actual value temperature field distribution diagram of the tunnel fire and the predicted value temperature field distribution diagram obtained by the present invention.

[0129] See Figure 4 The figure shows a comparison diagram of the actual value and predicted value of the tunnel fire temperature field, which shows a three-dimensional tunnel fire temperature field oblique visualization image obtained by the processor, including the tunnel fire actual value temperature field distribution diagram and the predicted value temperature field distribution diagram obtained by the present invention.

[0130] according to Figure 2 、 Figure 3 and Figure 4 The actual temperature field distribution of the tunnel fire is highly consistent with the predicted temperature field distribution obtained by the method of the present invention, which fully reflects the accuracy and feasibility of the method of the present invention and greatly enhances the practical application value of the method of the present invention.

[0131] S54: Divide the safety area according to the three-dimensional visualization image.

[0132] Specifically, a processor is used to divide the safe area in the tunnel fire according to the obtained three-dimensional temperature field visualization image of the tunnel fire, and the safe area is displayed through an output device to provide a scientific reference basis for emergency safety rescue.

[0133] See Figure 5 In an optional embodiment, in order to efficiently execute the tunnel fire temperature field prediction and safety zone division method provided by the present invention, the present invention provides a tunnel fire temperature field prediction and safety zone division system, the system comprising an input device, an output device, a processor, and a memory, wherein the hardware facilities are interconnected, wherein the memory is used to store a computer program, the computer program comprising program instructions, and the processor is configured to call the program instructions to execute the specific steps of the embodiments related to the tunnel fire temperature field prediction and safety zone division method provided by the present invention. The tunnel fire temperature field prediction and safety zone division system provided by the present invention has a complete structure, is objective and stable, and can efficiently execute the tunnel fire temperature field prediction and safety zone division method provided by the present invention, thereby improving the overall applicability and practical application capabilities of the present invention.

[0134] In summary, the method and system for predicting the temperature field and demarcating safe areas for tunnel fires provided by the present invention introduces physical constraints into a neural network to construct a combustion model. This improves the physical consistency of the combustion model, making it applicable to a variety of complex tunnel fire scenarios and significantly improving computational efficiency. By introducing physical constraints and data-driven methods, the constructed combustion model achieves real-time prediction of tunnel fires and further visualizes the prediction results, providing a scientific basis for demarcating safe areas and making timely safety rescue decisions for tunnel fires. The method of the present invention is easy to understand, computationally simple, requires minimal workload, and is readily applicable in practice, providing a theoretical foundation and technical support for timely safety rescue in tunnel fires.

[0135] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and description of the present invention.

Claims

1. A method for predicting temperature field of tunnel fire and dividing safe areas, characterized in that: The steps include: Preprocess the historical data of tunnel fire to obtain standardized data; Establishing a combustion equation for tunnel fire based on the standardized data; Constructing a combustion model of tunnel fire according to the combustion equation; Using the standardized data to train the combustion model; Predicting the temperature field of the tunnel fire based on the trained combustion model, and visualizing the prediction results for dividing the safety area; The preprocessing of the historical data of tunnel fire to obtain standardized data includes: Obtain historical data on tunnel fires; Performing interval mapping on the spatial temperature field of the historical data; Standardizing the historical data; The combustion equation of tunnel fire is established based on the standardized data, including: Obtain the mass conservation expression during combustion; The momentum conservation expression of wind speed effect during combustion is obtained; Construct the energy equation expression for the effect of thermal radiation during combustion; Obtain the mass fraction equation expression considering the burning rate during the combustion process; The method of constructing a combustion model of a tunnel fire according to the combustion equation comprises: Obtaining model parameters of the combustion model based on the historical data; Obtain the residuals of the mass, momentum, energy, and mass fraction equations respectively; Constructing a loss function to obtain the combustion model; The step of constructing a loss function to obtain the combustion model includes: ; in, is the loss function, is the mean square error between the predicted value and the true value, is the predicted value, is the true value, is the weight coefficient, , , and are the residuals of the equations for mass, momentum, energy, and mass fraction, respectively, , , and are the weights of each physical residual term; The step of training the combustion model using the standardized data comprises: Using an optimizer to set training parameters of the combustion model; importing the standardized data into the combustion model; Training the combustion model using the optimizer; The temperature field of the tunnel fire is predicted based on the trained combustion model, and the prediction result is visualized for dividing the safety area, including: Selecting a suitable combustion model based on monitoring data of detection points in the tunnel; Importing the monitoring data into the combustion model to perform real-time prediction of the temperature field of the tunnel fire; Performing image processing on the prediction results to obtain a three-dimensional visualization image; Dividing a safe area according to the three-dimensional visualization image; The step of processing the prediction result into a three-dimensional visualization image includes: ; in, For the three-dimensional data visualization process, Indicates three-dimensional image rendering. It is a three-dimensional grid structure. is the continuous temperature field function constructed, is the interpolation function, is a three-dimensional data point set, is the predicted temperature value, is the three-dimensional point coordinate, and is the total number of traversal points.

2. The method for predicting temperature field and dividing safe areas of tunnel fire according to claim 1 is characterized in that: The combustion equation includes: The mass conservation expression is: ; The expression for conservation of momentum is: ; Energy equation expression: ; The mass fraction equation expression is: ; in, is the fluid density, For time, is the divergence operator, For speed, is the fluid velocity vector, is the gradient operator, For pressure, is the molecular viscosity coefficient, is the subgrid viscosity coefficient, is the acceleration due to gravity, is the subgrid stress tensor, The wind force factor is is the velocity vector of the external wind, is the velocity vector of the wind inside the tunnel, is the enthalpy, is the molecular thermal conductivity, is the sub-grid eddy thermal conductivity, is the temperature, is the volume heat source term, is the radiation heat flux vector, is the quality score, is the molecular diffusion coefficient, is the sub-grid turbulent diffusion coefficient, is the generation rate.

3. Tunnel fire temperature field prediction and safety area division system, characterized by: The system includes an input device, an output device, a processor and a memory, wherein the input device, the output device, the processor and the memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the tunnel fire temperature field prediction and safety area division method according to any one of claims 1-2.

Images