A scene pre-processing method for fire simulation

Abstract

The application discloses a scene preprocessing method for fire simulation, and relates to the technical field of building fire simulation, comprising the following steps: calculating three-dimensional thermal power weight by using a fire spread dynamics model, and generating a thermal map of a fire danger area; extracting a main diffusion direction of the fire and each axial anisotropy parameter by using a tensor field generation algorithm, and generating fire grid parameters; constructing a hexahedral leading grid by using a hybrid grid generator, increasing the grid density of the fire danger area, and obtaining an optimized hybrid grid of the fire; and checking the Reynolds number and matching the boundary layer of the optimized hybrid grid of the fire by using a fire feature self-adaptive algorithm, and iteratively correcting to generate a fire simulation correction grid; by using the hybrid grid generator, the application realizes self-adaptive distribution and smooth connection of the grid density of the fire danger area, so that the grid distribution of the fire danger area is accurately matched with the fire spread characteristics.

Description

Technical Field

[0001] This invention relates to the field of building fire simulation technology, and in particular to a scene preprocessing method for fire simulation. Background Technology

[0002] Numerical fire simulation technology plays a crucial role in building safety assessment and fire protection design. Current technologies primarily rely on BIM models to generate simulation meshes and employ uniform meshes or simple mesh refinement strategies to handle fire scenarios. In recent years, with the improvement in the accuracy of CFD solvers, researchers have begun to introduce dynamic mesh optimization algorithms, such as local refinement methods based on temperature gradients and parametric mesh generation techniques incorporating combustion characteristics, significantly improving the accuracy of fire simulation.

[0003] Existing numerical simulation techniques for fire have failed to effectively quantify the dynamic coupling relationship between the main fire diffusion direction and the anisotropic parameters of the grid. Traditional methods employ uniform grids or static densification strategies, which cannot accurately capture the flow characteristics of the fire along the main diffusion direction. This results in insufficient resolution of the viscous sublayer and turbulent transition zone of the high-temperature smoke flow, which seriously affects the simulation accuracy. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a scene preprocessing method for fire simulation to solve the problem of low simulation accuracy caused by insufficient dynamic coupling between fire directionality and mesh parameters.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a scenario preprocessing method for fire simulation, comprising,

[0008] The system collects and corrects 3D geometric data of buildings, and outputs standardized BIM files and component attribute tables through component semantic annotation and material attribute binding. It calculates 3D thermal weights using a fire propagation dynamics model to generate a heat map of the fire hazard area. A tensor field generation algorithm extracts the main fire diffusion direction and anisotropic parameters of each axis to generate fire mesh parameters. A hybrid mesh generator constructs a hexahedral dominant mesh and increases the mesh density in the fire hazard area to obtain an optimized hybrid mesh. An adaptive fire feature algorithm performs Reynolds number verification and boundary layer matching on the optimized hybrid mesh, and iteratively corrects it to generate a fire simulation correction mesh. The fire simulation correction mesh is encapsulated in FDS format and injected with boundary conditions and material combustion parameters from the standardized BIM file and component attribute table, outputting a high-quality input package that can directly drive fire simulation.

[0009] As a preferred embodiment of the scene preprocessing method for fire simulation described in this invention, the correction includes geometric repair and redundancy cleanup.

[0010] As a preferred embodiment of the scene preprocessing method for fire simulation described in this invention, the step of outputting standardized BIM files and component attribute tables through component semantic annotation and material attribute binding is as follows:

[0011] Based on the corrected 3D geometric data of the building, fire-critical components are labeled with semantic tags for fire resistance rating and ventilation attributes, and a list of fire-critical components with classification tags is output.

[0012] The system automatically matches the combustion characteristic parameters of the list of critical fire-fighting components from the fire parameter library to generate a standardized BIM file containing the fire parameters.

[0013] By parsing the fire-critical component attribute data in the standardized BIM file, the correlation between the fire-critical components and combustion characteristic parameters is extracted, and a component attribute table is generated.

[0014] As a preferred embodiment of the scene preprocessing method for fire simulation described in this invention, the specific steps for calculating three-dimensional thermal weights using a fire propagation dynamics model to generate a heat map of the fire hazard area are as follows:

[0015] A fire propagation dynamics model was trained based on historical standardized BIM files using spatial topology deep learning methods.

[0016] The standardized BIM file is input into the trained fire propagation dynamics model to calculate the heat release rate distribution of the fire source.

[0017] Based on standardized BIM files, component attribute tables, and heat release rate distribution of fire sources, the fire spread path is dynamically simulated, and three-dimensional thermal weights are output.

[0018] The three-dimensional thermal weights are mapped to the three-dimensional Euclidean space of the building's three-dimensional geometry data, and the temperature field and smoke concentration hazard indicators are marked to generate a heat map of the fire hazard area.

[0019] As a preferred embodiment of the scene preprocessing method for fire simulation described in this invention, the steps for extracting the main fire diffusion direction and anisotropic parameters of each axis using a tensor field generation algorithm to generate fire mesh parameters are as follows:

[0020] The tensor field generation algorithm is used to extract the gradient characteristics of the temperature field and smoke concentration hazard index of the thermal map of the fire hazard area, and to analyze the main fire spread direction and output the main fire diffusion direction vector field.

[0021] Based on the main fire spread direction vector field and three-dimensional thermal weights, calculate the ratio of fire spread coefficients in each axis of the three-dimensional Euclidean space of the building's three-dimensional geometric data, and output the anisotropy parameters of each axis.

[0022] By integrating the main fire diffusion direction vector field and anisotropic parameters of each axis, fire grid parameters are generated based on spatial grid numbering.

[0023] As a preferred embodiment of the scene preprocessing method for fire simulation described in this invention, the step of constructing a hexahedral dominant mesh and increasing the mesh density in fire hazard areas using a hybrid mesh generator to obtain a fire-optimized hybrid mesh is as follows:

[0024] Based on the fire grid parameters, an octree adaptive subdivision algorithm is used to analyze the anisotropic parameters of each axis, generate a hybrid grid strategy dominated by hexahedrons, and output the initial grid by combining the building's three-dimensional geometric data to extract the coordinate set of the fire hazard area.

[0025] Based on the initial grid, hierarchical densification is performed within the coordinate set of the fire hazard zone to generate a non-uniform intermediate grid with density gradient.

[0026] The non-uniform intermediate grid is smoothed by Laplacian and the transition zone is connected to output a transition grid. The fire grid parameters are then mapped to the transition grid nodes to generate a fire-optimized hybrid grid.

[0027] As a preferred embodiment of the scene preprocessing method for fire simulation described in this invention, the specific steps for generating the fire simulation correction mesh are as follows:

[0028] The Reynolds number distribution field is calculated based on the main fire diffusion direction vector field and the optimized hybrid grid of the fire.

[0029] Based on the Reynolds number distribution field, the number and thickness of the hybrid grid are dynamically adjusted to optimize the fire situation, and the boundary layer refined grid is output.

[0030] Based on the boundary layer encrypted mesh, a fire simulation correction mesh is generated through multiple rounds of iterative correction using a fire feature adaptive algorithm.

[0031] As a preferred embodiment of the scene preprocessing method for fire simulation described in this invention, the output can directly drive a high-quality input packet for fire simulation. The specific steps are as follows:

[0032] Convert the fire simulation correction mesh to FDS format to generate a fire simulation correction mesh file;

[0033] By injecting the boundary conditions of the standardized BIM file and the correlation between fire-critical components and combustion parameters in the component attribute table into the fire simulation correction mesh file, a high-quality input package that can directly drive fire simulation is obtained.

[0034] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the scenario preprocessing method for fire simulation as described in the first aspect of the present invention.

[0035] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the scenario preprocessing method for fire simulation as described in the first aspect of the present invention.

[0036] The beneficial effects of this invention are as follows: By using a hybrid mesh generator, adaptive allocation of mesh density in fire hazard areas and smooth connection of transition areas are achieved, enabling the mesh distribution in fire hazard areas to accurately match the characteristics of fire spread; by using an adaptive algorithm for fire characteristics, dynamic optimization and iterative correction of boundary layer mesh parameters are completed, ensuring accurate analysis of wall turbulence and smoke diffusion; the two technologies work together to improve the computational efficiency and prediction accuracy of fire simulation, providing highly reliable dynamic fire evolution data for building fire protection design. Attached Figure Description

[0037] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a flowchart of a scenario preprocessing method used for fire simulation.

[0039] Figure 2 A flowchart for generating a heat map of a fire hazard area.

[0040] Figure 3 A flowchart for optimizing the hybrid mesh to output fire intensity.

[0041] Figure 4 A flowchart for outputting a high-quality fire simulation input package. Detailed Implementation

[0042] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0043] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0044] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0045] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a scenario preprocessing method for fire simulation, including the following steps:

[0046] S1. Collect and correct the three-dimensional geometric data of the building.

[0047] Specifically, the three-dimensional geometric data of the building is collected by a laser scanner to collect point cloud data of the building. The point cloud data of the building is then subjected to noise reduction and registration processing to generate a triangular patch model. For example, the laser scanner uses a phase scanner to collect point cloud data with a density of 5mm to obtain the three-dimensional geometric data of the building.

[0048] The correction process involves geometric repair using CAD repair tools to perform topological correction on non-manifold geometry, surface defects, and intersecting faces in the building's 3D geometric data. Redundancy cleanup removes decorative components, duplicate faces, and isolated vertices unrelated to the fire simulation. The geometrically repaired and redundancy-cleaned 3D geometric data then undergoes a BIM format verifier to check for compliance with IFC standards. Data that fails the verification is returned to the geometric repair step for reprocessing, while data that passes is output as a standardized BIM file. For example, non-manifold geometry repair involves stitching broken edges and filling missing surfaces; redundancy cleanup examples include removing sculpted patterns and lighting models from the 3D geometric data to obtain the corrected 3D geometric data.

[0049] Furthermore, the BIM format verifier checks whether it conforms to the IFC standard. The acceptable criteria are: the 3D geometric data of the building, after geometric repair and redundancy cleanup, must completely retain the topological connections of all building components; the geometry of each component must be free of self-intersections or cracks; material properties must be consistent with the type attributes defined in the IFC standard; spatial coordinate accuracy must meet the millimeter-level error range required by the IFC format; and there must be no redundant elements or duplicate entities in the model. The unacceptable criteria are: missing components or topological breaks in the 3D geometric data; self-intersections or non-manifold edges in geometric patches; material properties not classified according to the IFC standard or missing key parameters; spatial coordinate errors exceeding the allowable range; or the presence of uncleaned temporary construction lines, auxiliary surfaces, etc.

[0050] S2. Output standardized BIM files and component attribute tables by binding component semantic annotations and material attributes.

[0051] S2.1 Based on the corrected 3D geometric data of the building, label the fire resistance rating and ventilation attribute semantic tags for fire-critical components, and output a list of fire-critical components with classification tags.

[0052] Specifically, based on the corrected 3D geometric data of the building, wall components, door and window components, and ventilation duct components are selected in the BIM software; fire rating labels are added to each selected fire-critical component in the attribute panel, and the fire rating labels are divided into three levels according to the exemplary specifications: Level A corresponds to an exemplary fire resistance limit of ≥3.0 hours, Level B ≥1.5 hours, and Level C ≥0.5 hours; ventilation attribute labels are added to each fire-critical component in the attribute panel, and the ventilation attribute labels are divided into natural ventilation, mechanical ventilation, and airtightness according to the exemplary conditions.

[0053] By iterating through the fire rating label field and ventilation attribute label field of fire-critical components, the system checks for empty or undefined values. Valid fire rating labels are limited to three enumerated values: A, B, and C. Valid ventilation attribute labels are limited to three enumerated values: natural ventilation, mechanical ventilation, and airtightness. When a field value is empty or exceeds the enumeration range, the corresponding component is marked with a red highlight box in the BIM software's 3D view, triggering a verification failure message requiring manual labeling or correction. The marked fire-critical components are then used to generate a list of fire-critical components with categorized labels, based on component number, component name, fire rating label, and ventilation attribute label. For example, during natural ventilation label verification, the system checks whether door and window components contain valid opening parameters, and during mechanical ventilation label verification, it checks whether ventilation duct components are bound to smoke exhaust equipment numbers.

[0054] S2.3 Automatically match the combustion characteristic parameters of the fire critical component list from the fire parameter library to generate a standardized BIM file containing fire parameters.

[0055] Specifically, the fire-critical component list includes fields for material name, fire rating label, peak heat release rate, smoke generation rate, and CO generation rate. An enumerated value mapping relationship is established between the fire rating label field and the fire rating label field in the fire-critical component list. An example data entry is the material name field "rock wool board," the fire rating label field "Class A," the peak heat release rate field (example value 300 kW / m²), and the smoke generation rate field (example value 0.01 m² / kg). Each fire rating label field in the fire parameter library is bound to at least one combination of the peak heat release rate field, smoke generation rate field, and CO generation rate field.

[0056] The component name field of the fire critical component list is matched exactly with the material name field of the fire parameter library; the fire rating label field of the fire critical component list is matched with the enumerated value field of the fire rating label field of the fire parameter library; if no matching parameter is found for a fire critical component, a parameter missing warning is triggered and logged; after the standardized BIM file with the parameters written is verified by IFC format to check the integrity of the attribute fields, a standardized BIM file containing fire parameters is generated.

[0057] S2.4 By parsing the fire-critical component attribute data in the standardized BIM file, the correlation between the fire-critical components and combustion characteristic parameters is extracted, and a component attribute table is generated.

[0058] Specifically, the process involves parsing standardized BIM files and using the IFC parser to traverse the attribute data of fire-critical components, extracting the component ID field, component name field, fire rating label field, ventilation attribute label field, peak heat release rate field, smoke generation rate field, and CO generation rate field; establishing a mapping relationship between the component ID field and the peak heat release rate field, smoke generation rate field, and CO generation rate field; and establishing numerical association rules between the fire rating label field and the peak heat release rate field.

[0059] The component attribute table is generated based on the verified mapping relationship, according to the component ID field, component name field, fire rating label field, ventilation attribute label field, peak heat release rate field, smoke generation rate field, and CO generation rate field.

[0060] Furthermore, the process of establishing the mapping relationship between the component ID field and the peak heat release rate field, smoke generation rate field, and CO generation rate field involves: extracting the combustion characteristic parameter data block corresponding to each component ID field in the standardized BIM file through the IFC parser; writing the heat release rate value from the combustion characteristic parameter data block into the peak heat release rate field, the smoke generation coefficient into the smoke generation rate field, and the carbon monoxide generation coefficient into the CO generation rate field, thus forming a one-to-one mapping relationship between the component ID and the combustion characteristic parameters; and establishing the numerical association rule between the fire rating label field and the peak heat release rate field involves: based on the material combustion performance limits corresponding to different fire ratings in the building fire protection design code, binding the fire rating label field (such as "Class A") with the maximum allowable heat release rate value, and mapping the label to a specific heat release rate threshold range through a lookup table method.

[0061] S3. Calculate the three-dimensional thermal weights using the fire propagation dynamics model to generate a heat map of the fire hazard area.

[0062] S3.1. Based on historical standardized BIM files, a fire propagation dynamics model is trained using spatial topology deep learning methods.

[0063] Specifically, component attribute tables are extracted from historical standardized BIM files; the connection relationships of historical standardized BIM files are used as edges of the spatial topology map, and node attributes include room volume and peak heat release rate. A spatial topology map is generated, where the nodes are room components and fire-critical components in the building's 3D geometry data, and the edges are the connection relationships between doorway components and ventilation duct components. The node feature vectors of the spatial topology map include the volume field of the room components, the peak heat release rate field of the fire-critical components, the smoke generation rate field, and the CO generation rate field.

[0064] The time series data of fire spread in historical fire cases are aligned with the corresponding spatial topology map. The input of the graph neural network model is the node feature vector and edge feature vector of the spatial topology map, and the output is the predicted temperature field value and smoke concentration value of each room component in the next time step. The mean squared error loss function is used when training the graph neural network model. The loss function calculates the difference between the predicted temperature field and the actual temperature field, and the difference between the predicted smoke concentration and the actual smoke concentration. The training termination condition is set to stop training when the loss of the validation set decreases by less than the exemplary stop training threshold of 0.01 for three consecutive iterations. The trained graph neural network model is saved as a fire propagation dynamics model.

[0065] S3.2 Input the standardized BIM file into the trained fire propagation dynamics model and calculate the heat release rate distribution of the fire source.

[0066] It should be noted that the expression for calculating the heat release rate distribution of a fire source is:

[0067] ;

[0068] in, It is a location The heat release rate distribution at that location It is a fire propagation dynamics model after training. It is a building space topology diagram. It is a material parameter matrix. It is the initial fire source location marker vector. It is the baseline value for the maximum heat release rate in the component property table;

[0069] Specifically, the building space topology map is generated by extracting the room component volume field, door opening area field, peak heat release rate field, smoke generation rate field, and CO generation rate field of fire-critical components from the standardized BIM file; the maximum value of the peak heat release rate field of fire-critical components is read from the component attribute table as the benchmark value of the maximum heat release rate; and the material parameter matrix of fire-critical components is extracted from the standardized BIM file.

[0070] The building space topology map, material parameter matrix, and initial fire source location marker vector are input into the trained fire propagation dynamics model. The fire propagation dynamics model aggregates the peak heat release rate and smoke generation rate data of adjacent nodes through graph convolution operation, and calculates the fire propagation path weight by combining the opening area field of doorway components. The fire propagation dynamics model calculates the influence factors of the cross-sectional area and length fields of ventilation duct components on fire spread through the edge attention mechanism.

[0071] The fire propagation dynamics model outputs the heat release rate weight coefficients of the three-dimensional spatial grid nodes, with an example value of 0-1. The heat release rate weight coefficients are multiplied by the maximum heat release rate benchmark value to obtain the heat release rate distribution values ​​of the three-dimensional spatial grid nodes. The heat release rate distribution values ​​are mapped to the coordinate grid of the building's three-dimensional geometric data to generate the heat release rate distribution of the fire source, which includes the location coordinate field and the heat release rate value field.

[0072] S3.3 Based on standardized BIM files, component attribute tables, and the heat release rate distribution of the fire source, dynamically simulate the fire spread path and output three-dimensional thermal weights.

[0073] Specifically, the process involves analyzing the spatial coordinates and connection relationships of room components from standardized BIM files to understand the three-dimensional geometric data of the building; obtaining the fire resistance rating, peak heat release rate, and ventilation attribute data of key fire-prone components from the component attribute table; initializing the room's temperature field and smoke concentration parameters; iteratively updating the room's temperature field and smoke concentration parameters according to the time step; considering the influence of the door opening area and peak heat release rate of the fire resistance rating of adjacent rooms when updating the temperature field; and determining the diffusion path based on the natural ventilation opening area or mechanical ventilation number of the ventilation attribute.

[0074] The temperature field values ​​of each room component are monitored in real time. When the temperature field value of a room component reaches or exceeds the corresponding fire rating label field, the coordinates of the room component whose temperature field value reaches or exceeds the corresponding fire rating label field are marked as a red polygon area in the building's three-dimensional geometric data, and recorded as an active fire spread area. The heat release rate distribution value of the active fire spread area is adjusted according to the peak heat release rate field of the adjacent room components. The temperature field values ​​and smoke concentration values ​​at each time step are weighted and accumulated according to spatial location. After the time step iteration is completed, a three-dimensional thermal weight is generated.

[0075] S3.4 Map the three-dimensional thermal weights to the three-dimensional Euclidean space of the building's three-dimensional geometric data, and mark the temperature field and smoke concentration hazard indicators to generate a heat map of the fire hazard area.

[0076] Specifically, the three-dimensional thermal weights are matched point by point according to the three-dimensional geometric data of the building; in the three-dimensional Euclidean space of the three-dimensional geometric data of the building, the grid nodes are divided into equally spaced grid nodes according to the grid resolution parameters of the three-dimensional thermal weights within the bounding box of the three-dimensional geometric data of the building, and a mapping table between the grid node coordinates and the row and column numbers of the three-dimensional thermal weight data nodes is established to obtain the spatial grid index.

[0077] Temperature field values ​​are converted to RGB color codes via linear interpolation, with the color gradient transitioning from blue to red corresponding to low to high temperatures. Smoke concentration values ​​are converted to transparency parameters via linear interpolation, with the transparency parameter transitioning from high to low transparency corresponding to low concentration. Triangular mesh surface data of the building's 3D geometry is imported into the 3D rendering engine, and the temperature field color codes and smoke concentration transparency parameters are superimposed onto the mesh surface of the building's 3D geometry. High-temperature areas are identified based on the temperature field color codes, and polygonal boundary outlines are drawn. The polygonal boundary outlines, smoke concentration transparency parameters, and component spatial coordinates of the building's 3D geometry are correlated to generate a heat map of the fire hazard area that includes 3D coordinates, temperature field colors, and smoke concentration transparency.

[0078] S4. Extract the main fire diffusion direction and anisotropic parameters of each axis using the tensor field generation algorithm to generate fire mesh parameters.

[0079] S4.1. Using the tensor field generation algorithm, the gradient characteristics of the temperature field and smoke concentration hazard index of the thermal map of the fire hazard area are extracted, and the main fire spread direction vector field is output by analyzing the main fire spread direction.

[0080] Specifically, the process involves extracting the numerical distribution of temperature field weights and smoke concentration hazard indicators from the heat map of the fire hazard area; performing three-dimensional spatial gradient analysis on the temperature field weights to generate a temperature field gradient vector; performing three-dimensional spatial gradient analysis on the smoke concentration hazard indicator field to generate a smoke concentration gradient vector; superimposing the temperature field gradient vector and the smoke concentration gradient vector according to spatial coordinates to form a joint gradient tensor; performing principal component analysis on the joint gradient tensor to extract the dominant direction vector as a candidate for the main fire diffusion direction; setting a gradient amplitude threshold to filter noise interference, the gradient amplitude threshold is set according to the maximum gradient ratio of the temperature field weights and the smoke concentration weights, for example, a ratio of 70% for the temperature field gradient and 30% for the smoke concentration gradient; filtering effective main fire diffusion direction candidates according to the gradient amplitude threshold; mapping the effective main fire diffusion direction candidates to direction vectors according to the three-dimensional coordinates of the building's three-dimensional geometric data; and generating a main fire diffusion direction vector field according to the spatial coordinates and direction angles of the direction vectors.

[0081] S4.2. Based on the main fire diffusion direction vector field and three-dimensional thermal weight, calculate the ratio of fire diffusion coefficients in each axis of the three-dimensional Euclidean space of the building's three-dimensional geometric data, and output the anisotropy parameters of each axis.

[0082] It should be noted that the expression for the ratio of fire diffusion coefficients along each axis in the three-dimensional Euclidean space of the building's three-dimensional geometric data is as follows:

[0083] ;

[0084] in, , , yes The fire spread coefficient of the axis is higher than that of the axis. , , The main direction vector of fire spread is in The component modulus of the shaft, The exemplary range of the three-dimensional thermal weights of the mesh nodes is [0-1]. It is a very small positive number, for example, 1e-6;

[0085] Specifically, the direction vector field of each three-dimensional coordinate point is extracted from the main fire diffusion direction vector field; the temperature field weight field and smoke concentration weight field of the corresponding coordinate point are extracted from the three-dimensional thermal weight field; and the direction vector field is decomposed into... Axial components, Axial components, Axis components; The axial components are weighted according to the temperature field weight field and the flue gas concentration weight field, calculated by weighting them according to their respective weight ratios (e.g., temperature field weight accounts for 60%, flue gas concentration weight accounts for 40%). Axial fire spread coefficient components; The axial components are weighted according to the same weight ratio as the temperature field weight field and the flue gas concentration weight field. Axial fire spread coefficient components; The axial components are weighted according to the same weight ratio as the temperature field weight field and the flue gas concentration weight field. Axial fire spread coefficient components; for The three axial fire spread coefficient components were normalized to obtain The percentage value of the axial fire spread coefficient ratio; the coordinates of the fire spread points. The axial fire spread coefficient ratio is correlated with the spatial structure of the three-dimensional coordinates and the three-dimensional geometric data of the building; the output includes three-dimensional coordinate fields. Axial fire spread coefficient ratio field Axial fire spread coefficient ratio field The axial fire spread coefficient is a parameter of the axial anisotropy of the field.

[0086] S4.3 Integrate the main fire diffusion direction vector field and the anisotropic parameters of each axis, and generate fire grid parameters based on the spatial grid number.

[0087] Specifically, the direction vector field of the three-dimensional coordinate points is extracted from the main fire spread direction vector field; the corresponding three-dimensional coordinate points are extracted from the anisotropy parameter data table for each axis. Axial fire spread coefficient ratio field Axial fire spread coefficient ratio field Axial fire spread coefficient ratio field; the direction vector field Quantity and Axial fire spread coefficient ratio field combined to generate Axial weighted direction components; the direction vector field Quantity and Axial fire spread coefficient ratio field combined generation Axial weighted direction components; the direction vector field Quantity and Axial fire spread coefficient ratio field combined generation Axial weighted direction components; Axial weighted direction components, Axial weighted direction components, The axial weighted directional components are integrated into a comprehensive fire spread vector based on three-dimensional coordinate points; a fire spread direction vector is generated.

[0088] The temperature field weight field and the smoke concentration weight field in the three-dimensional thermal weighting are combined according to the three-dimensional coordinate points and the fire spread direction vector; the integration of the temperature field weight field and the smoke concentration weight field is used as the comprehensive fire hazard index field; the three-dimensional coordinate points, the unit fire spread direction vector, the fire spread intensity field, and the comprehensive fire hazard index field are sorted according to the spatial grid number to generate fire grid parameters containing the spatial grid number, three-dimensional coordinate field, unit fire spread direction vector field, fire spread intensity field, and comprehensive fire hazard index field.

[0089] S5. Using a hybrid mesh generator, construct a hexahedral dominant mesh and increase the mesh density in fire hazard areas to obtain a fire-optimized hybrid mesh.

[0090] S5.1 Based on the fire grid parameters, an octree adaptive subdivision algorithm is used to analyze the anisotropic parameters of each axis, generate a hybrid grid strategy dominated by hexahedrons, and output the initial grid by combining the three-dimensional geometric data of the building to extract the coordinate set of the fire hazard area.

[0091] Specifically, the process involves extracting the spatial grid number, comprehensive fire hazard index field, and unit fire spread direction vector field from the fire grid parameters; creating an octree root node within the bounding box of the building's 3D geometric data, with the root node size covering the entire spatial range of the building's 3D geometric data; traversing the comprehensive fire hazard index field of the fire grid parameters, and triggering an octree node subdivision operation based on the numerical distribution of the comprehensive fire hazard index field in the fire grid parameters, dividing the node into eight child nodes; repeating the subdivision process until the child node size meets the octree subdivision termination condition.

[0092] Mark the final-level octree nodes as hexahedral meshes; convert the final-level octree nodes of curved surface boundaries in the building 3D geometry data into tetrahedral meshes; merge the hexahedral meshes and tetrahedral meshes to generate a hybrid mesh set; associate each mesh in the hybrid mesh set with the fire hazard index field and the unit fire spread direction vector field in the fire mesh parameters; filter the mesh spatial coordinate fields that meet the fire mesh filtering conditions based on the fire hazard index field; match the filtered mesh spatial coordinate fields with the component spatial coordinates of the building 3D geometry data to generate a fire hazard area coordinate set;

[0093] Furthermore, the octree subdivision termination condition is as follows: subdivision stops when the size of the child node is less than or equal to the preset minimum grid size (e.g., 0.5m³) or the variance of the fire hazard index within the node is less than the octree subdivision termination condition (e.g., 10%).

[0094] Fire grid screening criteria: Grids with a comprehensive fire hazard index exceeding the critical value (e.g., ≥0.7) are judged to meet the fire grid screening criteria and are used to match component coordinates to generate hazardous areas.

[0095] S5.2. Based on the initial grid, perform hierarchical densification within the coordinate set of the fire hazard zone to generate a non-uniform intermediate grid with density gradient.

[0096] Specifically, the three-dimensional coordinate field and the comprehensive fire hazard index field of the fire hazard area coordinate set are extracted from the initial grid; the hierarchical encryption rule is set so that the value of the comprehensive fire hazard index field is positively correlated with the encryption level; the coordinates of each grid in the fire hazard area coordinate set are traversed, the range of the comprehensive fire hazard index is determined and the corresponding subdivision level is assigned; and an octree subdivision operation is performed on the grid with the assigned level.

[0097] The subdivided subgrids inherit the fire hazard index field from the parent grid; the density gradient parameter is set as the subdivision level exponential function, with example density gradient parameters of density base 1 for level 0, density base 4 for level 1, density base 16 for level 2, and density base 64 for level 3; the original grid and the subdivided subgrids are merged according to spatial coordinates to generate a non-uniform intermediate grid set; each grid in the non-uniform intermediate grid set is associated with the fire hazard index field and the density gradient field; the non-uniform intermediate grid data file is output, with file fields including grid number, 3D coordinates, density gradient, and fire hazard index, in an example CSV table format, with the field order as number, ... coordinate, coordinate, Coordinates, density gradient, and comprehensive fire hazard index are used to generate a non-uniform intermediate grid with density gradient.

[0098] S5.3. Perform Laplacian smoothing and transition zone connection processing on the non-uniform intermediate grid, output the transition grid, and map the fire grid parameters to the transition grid nodes to generate the fire optimized hybrid grid.

[0099] Specifically, the process involves traversing each non-uniform intermediate grid node, obtaining the coordinates of adjacent nodes, calculating the average of the adjacent node coordinates, and updating the current node coordinates to the average value. This process is repeated until the node displacement converges. Regions with significant differences in density gradients between adjacent grids in the non-uniform intermediate grid set are detected and marked as transition zone boundaries. Intermediate layer grid nodes are inserted at these transition zone boundaries, with their coordinates generated through linear interpolation and the density gradient transitioning uniformly according to the difference between adjacent grids. The smoothed grid nodes are merged with the inserted intermediate layer grid nodes to generate a transition grid set. The comprehensive fire hazard index field and the unit fire spread direction vector field are extracted from the fire grid parameters and mapped using bilinear interpolation based on the 3D coordinates of the transition grid nodes. The consistency between the mapped comprehensive fire hazard index field and the unit fire spread direction vector data is verified. Abnormal nodes are re-interpolated and updated. The verified transition grid nodes are integrated with the mapped comprehensive fire hazard index field and unit fire spread direction vector field to generate a fire-optimized hybrid grid.

[0100] S6. Using the fire characteristic adaptive algorithm, the fire optimization hybrid mesh is verified for Reynolds number and matched with the boundary layer, and the fire simulation correction mesh is generated iteratively.

[0101] S6.1 Calculate the Reynolds number distribution field based on the main fire diffusion direction vector field and the optimized hybrid grid of the fire.

[0102] It should be noted that the expression for calculating the Reynolds number distribution field is:

[0103] ;

[0104] in, It is a fire-optimized hybrid grid Reynolds number distribution field at that location, It is the density of the smoke. It is the dynamic viscosity of flue gas. It is a fire-optimized hybrid grid The fire spread velocity vector at the location, It is a fire-optimized hybrid grid Scale parameters along the main direction of the fire;

[0105] Specifically, the process involves extracting the unit fire spread direction vector field and fire spread intensity field from the fire spread direction vector field of the optimized fire hybrid grid nodes; extracting the three-dimensional coordinate field and comprehensive fire hazard index field from the optimized fire hybrid grid nodes; defining the magnitude of the fire spread velocity vector as the product of the magnitude of the unit fire spread direction vector field and the fire spread intensity field; calculating the average spatial interval of the nodes in the optimized fire hybrid grid to obtain the scale parameter, which is obtained by the arithmetic mean of the distance between the three-dimensional coordinates of the node and the distance between adjacent nodes; defining the smoke density parameter as the corresponding value of the comprehensive fire hazard index field mapped to the preset smoke density relationship table; defining the smoke dynamic viscosity parameter as a constant experimentally measured value; traversing each node of the optimized fire hybrid grid, and substituting the magnitude of the fire spread velocity vector, scale parameter, smoke density parameter, and smoke dynamic viscosity parameter into the Reynolds number calculation formula to generate the Reynolds number value of the current node, thus completing the calculation of the Reynolds number distribution field;

[0106] Furthermore, the specific construction process of the preset smoke density relationship table is as follows: Based on historical fire case data, the measured smoke density values ​​corresponding to different fire comprehensive hazard index intervals are extracted; the fire comprehensive hazard index is divided into several continuous intervals, each interval corresponding to a median smoke density value; a mapping relationship table between the fire comprehensive hazard index and smoke density is obtained through statistical analysis methods; a linear interpolation method is used to process the transition values ​​at the interval boundaries to ensure the continuity of the mapping relationship; finally, a preset smoke density relationship table containing the fire comprehensive hazard index field and the smoke density field is generated.

[0107] S6.2 Based on the Reynolds number distribution field, dynamically adjust the number and thickness of the fire optimization mixed mesh layer, and output the boundary layer refined mesh.

[0108] Specifically, the Reynolds number and 3D coordinate fields of the fire-optimized hybrid grid nodes are extracted from the Reynolds number distribution field; the near-wall densification rule is defined as a dynamic association between the Reynolds number and the number of grid layers and thickness parameters, with regions having higher Reynolds numbers corresponding to larger number of layers and smaller thickness parameters; each node of the fire-optimized hybrid grid is traversed, and the number of layers and thickness parameters are matched according to the Reynolds number; an octree subdivision operation is performed on the nodes with matched parameters, with the number of subdivisions equal to the number of layers, and each subdivision generates a sub-grid and updates the 3D coordinate field; the subdivided sub-grids inherit the Reynolds number field of the parent grid; the thickness parameter is adjusted so that the spacing between each grid layer is gradually reduced according to the thickness parameter; transition grids are inserted in regions with significant differences in the spacing between adjacent grid layers, and the spacing between transition grids is generated by linear interpolation; the subdivided densified grids and transition grids are merged to generate a boundary layer densified grid set; each grid in the boundary layer densified grid set is associated with the Reynolds number field, the number of layers, and the thickness field; differences in the number of layers between adjacent grids are detected and smoothed, with the smoothing method being to redistribute the number of grid layers in the difference region according to the average value of the nearest neighbor; the smoothed boundary layer densified grid set is output as a structured data file to generate the boundary layer densified grid.

[0109] S6.3. Based on the boundary layer encrypted mesh, the fire simulation correction mesh is generated through multiple rounds of iterative correction using the fire feature adaptive algorithm.

[0110] Specifically, the process involves extracting grid number, 3D coordinates, Reynolds number, number of layers, and thickness data from the boundary layer densification grid; loading time series data of temperature fields and smoke concentrations from historical fire cases as validation benchmarks; inputting the 3D coordinates, number of layers, and thickness parameters of the boundary layer densification grid; calculating the absolute percentage errors between the predicted temperature field and smoke concentration values ​​and the benchmark values; traversing the grid nodes of the boundary layer densification grid and generating error weighting coefficients based on the percentage errors of temperature field and smoke concentration predictions, where the error weighting coefficient is the average of the two percentages; dynamically adjusting the layer number field parameters according to the error weighting coefficient, for example: an increase of 1% in the error weighting coefficient corresponds to an increase of 0.1 layers in the number of layers, and a decrease of 1% in the error weighting coefficient corresponds to a decrease of 0.1 layers in the number of layers; and dynamically adjusting the thickness field according to the error weighting coefficient, for example: an increase of 1% in the error weighting coefficient corresponds to a decrease of 0.1% in the thickness, and a decrease of 1% in the error weighting coefficient corresponds to an increase of 0.1% in the thickness. An octree subdivision operation is performed on the adjusted layer number segment and thickness field parameters to generate new sub-mesh and update the 3D coordinate field; the new mesh data is input into the fire simulation model to execute the next iteration; the process of error weight allocation, dynamic parameter adjustment, mesh subdivision, and simulation verification is repeated until the rate of change of the error weight coefficient is lower than the preset convergence condition or the preset iteration round is reached; the 3D coordinate field, layer number segment, and thickness field of the final mesh node are extracted; a fire simulation corrected mesh data file is generated in CSV format, with the field order being mesh number, layer number, thickness field, and thickness field. Coordinates, number of layers, and thickness; output fire simulation correction mesh;

[0111] It should be noted that the process for setting the preset convergence condition is as follows: Based on the error weight coefficient change rate data of historical fire simulation cases, the average change rate of the error weight coefficient in multiple iterations is obtained; the convergence condition is defined as the absolute value of the error weight coefficient change rate in three consecutive iterations being less than 1% of the historical average change rate, which is considered as the error change tending to stabilize; the process for setting the preset iteration rounds is as follows: The average number of iterations required to reach the convergence condition in historical fire simulation cases is statistically analyzed, with an example result of 8 rounds; the preset iteration rounds are set to 1.5 times the average number of iterations, with an example value of 12 rounds; through experiments, it is verified that when the iteration rounds reach 12, the proportion of cases that have not converged is less than 5%, confirming the effectiveness of this upper limit of rounds.

[0112] S7. Encapsulate the fire simulation correction mesh into FDS format, and inject the boundary conditions and material combustion parameters of the standardized BIM file and component attribute table, outputting a high-quality input package that can directly drive fire simulation.

[0113] S7.1 Convert the fire simulation correction mesh to FDS format to generate a fire simulation correction mesh file.

[0114] Specifically, this involves extracting the grid number field, 3D coordinate field, layer number segment, thickness field, predicted temperature field, and predicted smoke concentration field from the fire simulation correction grid data file; defining grid structure parameters according to the FDS format specification, including grid start coordinate field, grid end coordinate field, and grid cell size field; converting the 3D coordinate field of the fire simulation correction grid into FDS format grid start coordinate field and grid end coordinate field, with the conversion rule being that the start coordinate of each grid is the 3D coordinate field minus 50% of the thickness field, and the end coordinate is the 3D coordinate field plus 50% of the thickness field; converting the layer number segment into FDS format grid cell size field, and the grid cell size field... The element size field equals the thickness field divided by the layer number field; the temperature field prediction field is mapped to the FDS format temperature field data field, and the flue gas concentration prediction field is mapped to the FDS format flue gas concentration data field; the mesh structure parameters, temperature field data field, and flue gas concentration data field are written to a text file according to the FDS input file syntax rules; the mesh cell size field of the text file is verified to meet the mesh continuity requirements of the FDS format, which requires that the size difference between adjacent mesh cells does not exceed 10% of the size field; the verified text file is saved as an FDS format fire simulation correction mesh file with the file extension .fds; the fire simulation correction mesh file is output to the preset storage path.

[0115] S7.2. Inject the boundary conditions of the standardized BIM file and the correlation between fire-critical components and combustion parameters in the component attribute table into the fire simulation correction mesh file to obtain a high-quality input package that can directly drive fire simulation.

[0116] It should be explained that the following steps are performed: extract the boundary condition fields from the standardized BIM file and the fire-critical component number field, fire rating label field, peak heat release rate field, and ventilation attribute label field from the component attribute table; extract the grid number field, 3D coordinate field, temperature field prediction field, and smoke concentration prediction field from the fire simulation correction mesh file; establish a spatial association between the fire-critical component number field of the standardized BIM file and the grid number field of the fire simulation correction mesh file through 3D coordinate field matching; map the fire rating label field to the material thermal conductivity coefficient field of the fire simulation correction mesh file, map the peak heat release rate field to the heat release rate distribution field of the fire simulation correction mesh file, and map the ventilation attribute label field to the ventilation opening area field of the fire simulation correction mesh file; and map the material thermal conductivity coefficient field according to the FDS format specification. The heat release rate distribution field and ventilation opening area field are converted into FDS input parameter fields. The converted FDS input parameter fields are verified to meet the input constraints of the fire simulation engine, including the non-negativity of the heat release rate distribution field, the validity of the numerical range of the ventilation opening area field, and the spatial topological consistency between the mesh cells and boundary conditions. The verified material thermal conductivity coefficient field, heat release rate distribution field, ventilation opening area field, temperature field prediction field, and smoke concentration prediction field are merged into a structured data table according to the FDS input file format. The structured data table is integrated with the mesh number field and 3D coordinate field of the fire simulation correction mesh file and written into the FDS input file. The FDS input file, along with the component attribute table and boundary condition table of the standardized BIM file, are packaged into a ZIP compressed file to generate a high-quality input package.

[0117] This embodiment also provides a computer device applicable to a scenario preprocessing method for fire simulation, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the scenario preprocessing method for fire simulation as proposed in the above embodiment.

[0118] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0119] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the scenario preprocessing method for fire simulation as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0120] In summary, this invention achieves adaptive allocation of mesh density in fire hazard areas and smooth transition between areas through a hybrid mesh generator, ensuring that the mesh distribution in fire hazard areas accurately matches the characteristics of fire spread. Furthermore, through an adaptive fire feature algorithm, it completes dynamic optimization and iterative correction of boundary layer mesh parameters, ensuring accurate analysis of wall turbulence and smoke diffusion. These two technologies work synergistically to improve the computational efficiency and prediction accuracy of fire simulation, providing highly reliable dynamic fire evolution data for building fire protection design.

[0121] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for scene pre-processing for fire simulation, characterized in that: comprising, collecting building three-dimensional geometric data, and performing correction, outputting standardized BIM files and component attribute tables through component semantic labeling and material attribute binding; calculating three-dimensional thermal weight using fire spread dynamics model to generate fire hazard area thermal map; extracting fire main diffusion direction and each axial anisotropy parameter through tensor field generation algorithm to generate fire grid parameter; constructing hexahedral dominant grid and increasing fire hazard area grid density through hybrid grid generator to obtain fire optimized hybrid grid; performing Reynolds number verification and boundary layer matching on fire optimized hybrid grid through fire characteristic self-adaptive algorithm, and iteratively correcting to generate fire simulation correction grid; packaging fire simulation correction grid into FDS format, and injecting boundary conditions and material combustion parameters of standardized BIM files and component attribute tables to output high-quality input package that can directly drive fire simulation.

2. The scene pre-processing method for fire simulation of claim 1, wherein: The correction includes geometric repair and redundancy cleaning.

3. The scene pre-processing method for fire simulation of claim 2, wherein: The outputting standardized BIM files and component attribute tables through component semantic labeling and material attribute binding specifically includes the following steps, based on the corrected building three-dimensional geometric data, labeling fireproof grade and ventilation attribute semantic tags for fire key components to output fire key component list with classification tags; automatically matching combustion characteristic parameters of fire key component list from fire parameter library to generate standardized BIM files containing fire parameters; extracting the association between fire key components and combustion characteristic parameters by analyzing fire key component attribute data in standardized BIM files to generate component attribute table.

4. The scene pre-processing method for fire simulation of claim 3, wherein: The steps of calculating three-dimensional thermal weight using fire spread dynamics model to generate fire hazard area thermal map include the following steps, training fire spread dynamics model based on historical standardized BIM files through spatial topology deep learning method; inputting standardized BIM files into trained fire spread dynamics model to calculate heat release rate distribution of fire source; based on standardized BIM files, component attribute table and heat release rate distribution of fire source, dynamically simulating fire spread path to output three-dimensional thermal weight; mapping three-dimensional thermal weight to three-dimensional Euclidean space of building three-dimensional geometric data and marking temperature field and smoke concentration hazard indicators to generate fire hazard area thermal map.

5. The scene pre-processing method for fire simulation of claim 4, wherein: The steps of extracting fire main diffusion direction and each axial anisotropy parameter through tensor field generation algorithm to generate fire grid parameter include the following steps, using tensor field generation algorithm to extract gradient features of temperature field and smoke concentration hazard indicators of fire hazard area thermal map, analyzing fire main spread direction to output fire main diffusion direction vector field; calculating each axial fire spread coefficient ratio of three-dimensional Euclidean space of building three-dimensional geometric data according to fire main diffusion direction vector field and three-dimensional thermal weight to output each axial anisotropy parameter; integrating fire main diffusion direction vector field and each axial anisotropy parameter to generate fire grid parameter based on spatial grid number.

6. The scene pre-processing method for fire simulation of claim 5, wherein: The steps of constructing hexahedral dominant grid and increasing fire hazard area grid density through hybrid grid generator to obtain fire optimized hybrid grid include the following steps, Based on the fire grid parameters, the octree adaptive subdivision algorithm is adopted to analyze the axial anisotropy parameters, generate the hexahedral dominant hybrid grid strategy, and output the initial grid combined with the three-dimensional geometric data of the building, and extract the fire danger area coordinate set; Based on the initial grid, hierarchical encryption is performed within the scope of the fire danger area coordinate set to generate a non-uniform intermediate grid with a density gradient; The non-uniform intermediate grid is subjected to Laplacian smoothing and transition zone connection processing to output a transition grid, and the fire grid parameters are mapped to the transition grid nodes to generate a fire optimization hybrid grid.

7. The scene pre-processing method for fire simulation of claim 6, wherein: The specific steps of generating the fire simulation correction grid are as follows, Based on the fire main diffusion direction vector field and the fire optimization hybrid grid, the Reynolds number distribution field is calculated; According to the Reynolds number distribution field, the number of layers and the thickness of the fire optimization hybrid grid are dynamically adjusted, and a boundary layer encryption grid is output; Based on the boundary layer encryption grid, the fire simulation correction grid is generated through multiple rounds of iteration correction by the fire feature adaptive algorithm.

8. The scene pre-processing method for fire simulation of claim 7, wherein: The specific steps of outputting the high-quality input package that can directly drive the fire simulation are as follows, The fire simulation correction grid is converted into FDS format to generate a fire simulation correction grid file; The boundary conditions and component attribute table of the standardized BIM file are associated with the fire key components and the burning parameters, and the association relationship is injected into the fire simulation correction grid file to obtain a high-quality input package that can directly drive the fire simulation. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The processor executes the computer program to realize the steps of the scene preprocessing method for fire simulation according to any one of claims 1-8.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to realize the steps of the scene preprocessing method for fire simulation according to any one of claims 1-8.

Images