Bridge cable vehicle fire temperature-height curve generation method and system

Abstract

This invention discloses a method and system for generating temperature-height curves in bridge cable vehicle fires, relating to the field of bridge fire prevention technology. The method includes the following steps: constructing a bridge vehicle fire simulation model; inputting combined operating parameters into the simulation model to calculate output data; constructing a dataset from the simulation model output data, experimental data, and operating parameters; preprocessing the dataset; constructing a bridge cable vehicle fire temperature prediction model based on a conditional generative adversarial network (GAN); training the GAN model using a weighted combined loss function that includes adversarial loss and physical constraints; obtaining operating parameters of a bridge vehicle fire scenario using a semantic recognition model; inputting the recognized operating parameters into the fully trained bridge vehicle fire temperature prediction model, calculating and outputting the predicted temperature value corresponding to a set standard height point within a preset height range; and generating a bridge cable vehicle fire temperature-height curve based on the predicted temperature value.

Description

Technical Field

[0001] This invention relates to the field of bridge fire prevention technology, and in particular to a method and system for generating temperature-height curves in bridge cable-stayed vehicles fires. Background Technology

[0002] In recent years, with the widespread application of long-span cable-stayed bridges in important projects such as those spanning seas and rivers, fire safety issues have become increasingly prominent. Bridge-vehicle fires have occurred frequently in recent years, and the high-temperature flames and hot smoke generated by burning vehicles pose a serious threat to the bridge's critical load-bearing components (such as main cables, suspenders, stay cables, and bridge towers). The mechanical properties of steel are extremely sensitive to temperature. When the temperature of the cable exceeds its material critical value, its strength and stiffness will significantly degrade, potentially leading to overall structural failure and catastrophic consequences.

[0003] Therefore, in bridge fire protection design, scientifically determining the temperature effect of high temperatures during a fire on the bridge cables—that is, developing a temperature design curve—is the prerequisite and key basis for fire-resistant design and the implementation of targeted fire prevention measures. Developing a temperature design curve requires a large number of temperature-height curves under various operating conditions. Currently, this field mainly relies on fire dynamics simulation software (FDS) based on computational fluid dynamics (CFD). This method establishes a detailed physical model, sets boundary conditions such as fire source power, wind speed, and distance, and numerically solves the Navier-Stokes equations to obtain the spatial temperature field distribution under fire scenarios.

[0004] However, this traditional physics-based simulation method has many limitations: 1. Extremely low computational efficiency: A typical FDS simulation of a bridge cable vehicle fire scenario, due to its large computational domain, numerous meshes, and the need to calculate to a stable combustion state, typically requires several hours or even days to complete a single calculation. Furthermore, obtaining the envelope-type fortification curve for design requires extensive parametric analysis of various operating conditions, leading to an exponential increase in computational costs. This severely restricts design efficiency and fails to meet the demands of rapid design in modern engineering.

[0005] 2. High technical threshold and limited adoption: FDS software is complex to operate, requiring users to have a strong background in fluid mechanics and thermodynamics to correctly set parameters, generate meshes, and interpret results. This limits the widespread adoption and application of this technology in engineering design.

[0006] In recent years, artificial intelligence technology, especially deep learning, has demonstrated its enormous potential to replace traditional numerical simulations in many fields of scientific computing. Generative adversarial networks and their variants have achieved remarkable results in generating complex high-dimensional data distributions. However, applying this technology to the specific engineering scenario of a bridge cable vehicle fire and directly generating temperature-height curves remains a gap.

[0007] Therefore, developing a new method that can overcome the above-mentioned defects has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0008] This solution addresses the problems and needs raised above by proposing a method and system for generating temperature-height curves in bridge cable-stayed vehicle fires. The above-mentioned technical objectives are achieved by adopting the following technical features, and several other technical benefits are also brought about.

[0009] One object of the present invention is to provide a method for generating a temperature-height curve in a bridge cable vehicle fire, comprising the following steps: S10: Construct a bridge vehicle fire simulation model, and output temperature-height data under different working conditions based on the simulation model. S20: Construct datasets from the temperature-height data and experimental data output by the simulation model and their corresponding operating parameters; preprocess the datasets and divide them into training, validation and test sets; S30: Construct a temperature prediction model for bridge cable vehicle fires based on conditional generative adversarial networks. S40: Train, validate, and test the fire temperature prediction model using the loss function using the training set, validation set, and test set; S50: Use a semantic recognition model to obtain the operating parameters of a bridge vehicle fire scenario; S60: Input the identified working condition parameters into the well-trained bridge vehicle fire temperature prediction model, calculate and output the temperature prediction value corresponding to the set standard height point within the preset height range. S70: Based on the predicted temperature value, generate a temperature-height curve for a bridge cable vehicle fire.

[0010] In addition, the method for generating the temperature-height curve of a bridge cable vehicle fire according to the present invention may also have the following technical features: In one example of the present invention, step S10, constructing the bridge vehicle fire simulation model specifically includes: The two-dimensional cable plane is simplified into a one-dimensional vertical bar, and the arbitrary position of the cable is represented by the height position of the vertical bar, thus simplifying the two-dimensional spatial coordinates into a one-dimensional position. The vehicle is simplified into a cube with a fire source on its surface. Recommended maximum grid sizes are given for different vehicle types: 20cm for a maximum heat release rate of 0-5MW; 25cm for 5-10MW; 30cm for 10-15MW; 35cm for 15-20MW; and 50cm for 20-200MW. The maximum heat release efficiency of the flame is simulated for 8-12 minutes to equivalently replace the highest temperature of the actual flame simulated for 1.5 hours. Temperature measuring points are set at equal intervals along the height direction on the surface of the vertical rod according to the required accuracy.

[0011] In one example of the present invention, in step S10, the operating parameters include: fire source intensity, bridge deck wind speed, and fire source distance; wherein, the fire source intensity is densely sampled in a uniform distribution in the range of 0-50MW, and the fire source intensity is sparsely sampled in a uniform distribution in the range of 50-200MW; the bridge deck wind speed is sampled in a uniform distribution in the range of 0-15m / s; and the fire source distance is the distance between the emergency lane and the adjacent lane from the edge of the main cable.

[0012] In one example of the present invention, in step S20, the preprocessing of the dataset includes: mapping the non-uniform temperature-height data output by the simulation model to a standard height sampled at equal intervals within 0-35m using a linear interpolation method; and normalizing the working condition parameter matrix and the temperature value matrix respectively.

[0013] In one example of the present invention, the fire temperature prediction model includes a generator and a discriminator, wherein the generator is configured to generate a predicted temperature curve based on a noise vector and normalized working condition parameters; and the discriminator is configured to distinguish between the generated temperature curve and the actual temperature curve and output a discrimination probability value.

[0014] In one example of the present invention, the generator includes an input layer, three first fully connected hidden layers, and a first output layer; the input layer is a concatenated vector of a noise vector and normalized working condition parameters, used for data input and feature initialization; the three first fully connected hidden layers are connected sequentially, each followed by a batch normalization layer and a LeakReLU activation function, used for feature extraction and nonlinear transformation at different levels, respectively; the first output layer includes a first fully connected layer and a Tanh activation function, used to output temperature prediction values ​​for multiple standard altitude points; The discriminator is a two-stream network structure, including a conditional stream, a data stream, a concatenation layer, two second fully connected hidden layers, and a second output layer. The conditional stream includes a second fully connected layer and a LeakReLU activation function, used to process the operating condition vector and extract conditional features. The data stream includes a third fully connected layer and a LeakReLU activation function, used to process the temperature curve vector and extract data features. The concatenation layer concatenates the feature vectors output from the conditional stream and the data stream into a fused feature vector. The two second fully connected hidden layers are connected sequentially, with each second fully connected hidden layer followed by a LeakReLU activation function, used for feature fusion and discriminant analysis, respectively. The second output layer includes a fourth fully connected layer and a Sigmoid activation function, used to output the discriminant probability value.

[0015] In one example of the present invention, in step S40, the weighted combined loss function L used by the generator in the fire temperature prediction model... total From the counter-loss L adv and physical constraint loss L smooth Weighted according to the following formula: Among them, the resistance loss L adv The calculation formula is: Among them, physical constraint loss L smooth The calculation formula is: In the formula, λ is the hyperparameter; E represents the mathematical expectation; P Z (Z) represents the probability distribution of random noise; P data (C) represents the data distribution under actual operating conditions; D(.) represents the output probability of the discriminator; G(.) represents the output of the generator; N represents the length of the temperature sequence; T i This represents the predicted temperature value at the i-th elevation point; The loss function used by the discriminator The expression is: In the formula, This represents a discriminator that outputs the probability of "true" under given operating conditions.

[0016] In one example of the present invention, step S40 specifically includes the following: An optimization algorithm based on gradient descent and its hyperparameter configuration strategy are adopted. The network is trained on the training set using an alternating discriminator and generator update strategy, and then trained on the test set using the coefficient of determination R. 2 The model was evaluated using mean absolute error (MAE); training parameters were adjusted during training by monitoring the training effect on the validation set, with a hyperparameter learning efficiency of 2 × 10⁻⁶. -4 The minimum number of training epochs is 500 to ensure prediction accuracy, and the maximum is 1000 to prevent overfitting. The coefficient of determination R is used on the test set. 2 The model is evaluated using the mean absolute error (MAE) and the coefficient of determination (R²). 2 >0.9, mean absolute error (MAE) <50℃.

[0017] In one example of the present invention, in step S50, the semantic recognition model includes a predefined semantic template and a rule parser. The predefined semantic template is configured to identify the working condition parameters and their units in the user's text. The rule parser is configured to parse the natural language text input by the user, apply template matching rules and syntax analysis algorithms, extract the working condition parameter values, and convert them into a standardized data format that the model can process.

[0018] Another object of the present invention is to provide a temperature-height curve generation system for bridge cable vehicle fires, comprising: The simulation model module is configured to build a bridge vehicle fire simulation model, and the simulation model calculates and outputs temperature-height data under different working conditions. The data processing and partitioning module is configured to construct datasets from the temperature-height data and experimental data output by the simulation model and their corresponding operating parameters; preprocess the datasets and partition them into training, validation and test sets. The fire temperature prediction module is configured to build a bridge cable vehicle fire temperature prediction model based on a conditional generative adversarial network. The model training module is configured to train, validate, and test a fire temperature prediction model using a loss function using a training set, a validation set, and a test set. The working condition parameter acquisition module is configured to acquire working condition parameters for bridge vehicle fire scenarios using a semantic recognition model. The predicted temperature calculation module is configured to input the identified working condition parameters into the well-trained bridge vehicle fire temperature prediction model, calculate and output the predicted temperature value corresponding to the set standard height point within the preset height range. The temperature-height curve module is configured to generate a temperature-height curve for bridge cable vehicle fires based on the predicted temperature values.

[0019] Compared with the prior art, the present invention has the following beneficial effects: 1. Improved computational efficiency: Traditional FDS-based numerical simulation methods often require hours or even days to calculate a temperature-height curve, severely limiting engineering design efficiency. This invention uses a pre-trained conditional generative adversarial network model for forward prediction, enabling the generation of a high-precision temperature-height curve within milliseconds. This improves computational efficiency by several orders of magnitude, achieving a leap from "hours" to "milliseconds," thus meeting the needs of engineering design.

[0020] 2. Strong generalization ability and wide engineering applicability: This invention trains the model by generating a massive number of random working conditions, enabling it to cover unknown working conditions far exceeding the training sample range. The temperature protection curve generated by this invention is based on the optimization results of the envelope of a large number of random working conditions, rather than the calculation results of a single working condition. Therefore, its protection recommendations are more universal and safer, and can provide scientific and customized fire protection design basis for bridges of different sizes and fire risk levels.

[0021] 3. High degree of automation and intelligence: This invention achieves full automation from input of operating conditions to curve generation. Once the model training is complete, users only need to provide key operating condition parameters (or extract them automatically through semantic recognition) to generate temperature-height curves with one click. This eliminates the need for complex simulation software operations and in-depth knowledge of fluid dynamics, lowering the technical threshold and improving the automation level of the design work.

[0022] 4. Innovative application of AI and large-scale model technology to the traditional field of engineering safety: This invention successfully applies the cutting-edge conditional generative adversarial network (GAN) technology—a "large-scale model"—to the specific engineering scenario of fire protection for bridges, cables, and vehicles. It not only solves the efficiency bottleneck of traditional methods but, more importantly, provides a new paradigm of "physical mechanism + data-driven approach."

[0023] The preferred embodiments of the invention will be described in more detail below with reference to the accompanying drawings, so as to facilitate an understanding of the features and advantages of the invention. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. The drawings are merely illustrative of some embodiments of the present invention and are not intended to limit the scope of the present invention to all embodiments.

[0025] Figure 1 This is a flowchart of a method for generating a temperature-height curve in a bridge cable vehicle fire according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the generator according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of the discriminator according to an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the influence of changes in fire source intensity, wind speed, and distance on temperature distribution according to an embodiment of the present invention. Figure 5 This is a schematic diagram comparing the temperature-height curves under different operating conditions with the model prediction results according to an embodiment of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. The same reference numerals in the drawings represent the same components. It should be noted that the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0027] Unless otherwise defined, the technical or scientific terms used herein shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms “first,” “second,” and similar terms used in this patent application specification and claims do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, “an” or “a” and similar terms do not necessarily indicate a quantity limitation. Terms such as “comprising” or “including” mean that the element or object preceding the word encompasses the element or object listed following the word and its equivalents, without excluding other elements or objects. Terms such as “connected” or “linked” are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as “upper,” “lower,” “left,” and “right” are used only to indicate relative positional relationships; these relative positional relationships may change accordingly when the absolute position of the described object changes.

[0028] A method for generating a temperature-height curve for a bridge cable vehicle fire according to a first aspect of the present invention, such as... Figure 1 As shown, it includes the following steps: S10: Construct a bridge vehicle fire simulation model, and output temperature-height data under different working conditions based on the simulation model. S20: Construct datasets from the temperature-height data and experimental data output by the simulation model and their corresponding operating parameters; preprocess the datasets and divide them into training, validation and test sets; S30: Construct a temperature prediction model for bridge cable vehicle fires based on conditional generative adversarial networks. This model is a generator-discriminator dual-path system. S40: Train, validate, and test the fire temperature prediction model using the loss function using the training set, validation set, and test set; S50: Use a semantic recognition model to obtain the operating parameters of a bridge vehicle fire scenario; S60: Input the identified working condition parameters into the well-trained bridge vehicle fire temperature prediction model, calculate and output the temperature prediction value corresponding to the set standard height point within the preset height range. S70: Based on the predicted temperature value, generate a temperature-height curve for a bridge cable vehicle fire.

[0029] The improved computational efficiency of this curve generation method: Traditional FDS-based numerical simulation methods often require hours or even days to calculate a temperature-elevation curve, severely limiting engineering design efficiency. This curve generation method, through forward prediction using a trained conditional generative adversarial network model, can generate a high-precision temperature-elevation curve in milliseconds, improving computational efficiency by several orders of magnitude. This leap from "hours" to "milliseconds" meets the needs of engineering design.

[0030] This curve generation method boasts strong generalization ability and wide engineering applicability: it trains the model by generating a massive number of random operating conditions, enabling it to cover unknown operating conditions far exceeding the training sample range. The temperature protection curves generated by this method are based on the optimization results of envelopes from a large number of random operating conditions, rather than calculations for a single condition. Therefore, its protection recommendations are more universal and secure, providing scientific and customized fire protection design basis for bridges of different sizes and fire risk levels.

[0031] This curve generation method boasts a high degree of automation and intelligence: it automates the entire process from inputting operating conditions to generating curves. Once the model training is complete, users only need to provide key operating parameters (or extract them automatically through semantic recognition) to generate temperature-height curves with a single click. This eliminates the need for complex simulation software operations and extensive fluid dynamics expertise, lowering the technical barrier to entry and enhancing the automation level of the design process.

[0032] This curve generation method innovatively applies AI and large-model technology to the traditional field of engineering safety: it successfully applies the cutting-edge conditional generative adversarial network (GAN) technology—a "large-model" approach—to the specific engineering scenario of fire protection for bridges, cables, and vehicles. It not only solves the efficiency bottleneck of traditional methods but, more importantly, provides a new paradigm of "physical mechanism + data-driven approach."

[0033] In one example of the present invention, step S10, constructing the bridge vehicle fire simulation model specifically includes: The two-dimensional cable plane is simplified into a one-dimensional vertical bar, and the arbitrary position of the cable is represented by the height position of the vertical bar, thus simplifying the two-dimensional spatial coordinates into a one-dimensional position. The vehicle is simplified into a cube with a fire source on its surface. Recommended maximum grid sizes are given for different vehicle types: 20cm for a maximum heat release rate of 0-5MW; 25cm for 5-10MW; 30cm for 10-15MW; 35cm for 15-20MW; and 50cm for 20-200MW. The maximum heat release efficiency of the flame is simulated for 8-12 minutes to equivalently replace the highest temperature of the actual flame simulated for 1.5 hours. Temperature measuring points are set at equal intervals along the height direction on the surface of the vertical rod according to the required accuracy.

[0034] In one example of the present invention, in step S10, the operating parameters include: fire source intensity, bridge deck wind speed, and fire source distance; wherein, the fire source intensity is densely sampled in a uniform distribution from 0-50MW, and sparsely sampled in a uniform distribution from 50-200MW, so as to match the actual distribution of the number of vehicles; the bridge deck wind speed is sampled in a uniform distribution in the range of 0-15m / s; the fire source distance is the distance between the emergency lane and the adjacent lane from the edge of the main cable; these three operating parameters are arranged and combined to achieve full coverage of the sample operating conditions.

[0035] In one example of this invention, in step S20, experimental data is collected using the Chinese invention patent application number 2025111956407, "Real-time Hybrid Test Platform and Implementation Method for Vehicle-Fire-Wind Force Mixed Test of Substructure of Long-Span Bridge," to conduct a full-scale or large-scale model fire resistance test of the cable under controllable parameter wind, fire, and force coupling conditions, directly collecting temperature field data at height points. The repeatable, measurable, and high-fidelity fire environment provided by this platform greatly ensures the diversity and reliability of the original data, laying the foundation for a sample set for training the temperature field generation model.

[0036] In one example of the present invention, in step S20, the preprocessing of the dataset includes: mapping the non-uniform temperature-height data output by the simulation model to 36 standard heights sampled at equal intervals of 1m within the range of 0-35m using linear interpolation; and normalizing the working condition parameter matrix and the temperature value matrix respectively.

[0037] In one example of the present invention, the fire temperature prediction model includes a generator and a discriminator, wherein the generator is configured to generate a predicted temperature curve based on a noise vector and normalized working condition parameters; and the discriminator is configured to distinguish between the generated temperature curve and the actual temperature curve and output a discrimination probability value.

[0038] In one example of the present invention, such as Figure 2As shown, the generator includes an input layer, three first fully connected hidden layers, and a first output layer. The input layer is a concatenated vector of a received noise vector and normalized working condition parameters, used for data input and feature initialization. The three first fully connected hidden layers are connected sequentially, with each first fully connected hidden layer followed by a batch normalization layer and a LeakReLU activation function, used for feature extraction and nonlinear transformation at different levels, respectively. The first output layer includes a first fully connected layer and a Tanh activation function, used to output temperature prediction values ​​for multiple standard altitude points. like Figure 3 As shown, the discriminator is a two-stream network structure, including a conditional stream, a data stream, a concatenation layer, two second fully connected hidden layers, and a second output layer. The conditional stream includes a second fully connected layer and a LeakReLU activation function, used to process the operating condition vector and extract conditional features. The data stream includes a third fully connected layer and a LeakReLU activation function, used to process the temperature curve vector and extract data features. The concatenation layer concatenates the feature vectors output from the conditional stream and the data stream into a fused feature vector for feature fusion. The two second fully connected hidden layers are connected sequentially, with each second fully connected hidden layer followed by a LeakReLU activation function, used for feature fusion and discriminant analysis, respectively. The second output layer includes a fourth fully connected layer and a Sigmoid activation function, used to output the discriminant probability value.

[0039] Specifically, the generator network is structured as follows: the input layer receives a concatenated vector of a 100-dimensional noise vector and a 3-dimensional normalized condition parameter vector; it passes through a first fully connected hidden layer of 256 nodes, 512 nodes, and 256 nodes in sequence, with each first fully hidden layer followed by a batch normalization layer and a LeakyReLU activation function; the output layer is a 36-node fully connected layer using the Tanh activation function.

[0040] The discriminator network has a dual-branch structure: a second fully connected layer with 64 nodes for processing 3D operating condition vectors; a third fully connected layer with 64 nodes for processing 36D temperature curves; the 64-dimensional feature vectors output from the two branches are concatenated into a 128-dimensional vector, which is then passed through a fourth fully connected layer with 256 nodes and a fourth fully connected layer with 128 nodes in sequence. Each fourth fully connected layer is followed by a batch normalization layer and a LeakyReLU activation function, and finally a discriminant probability value is output through a Sigmoid activation function.

[0041] In one example of the present invention, in step S40, the weighted combined loss function L used by the generator in the fire temperature prediction model... total From the counter-loss L adv and physical constraint loss L smooth Weighted according to the following formula: Among them, the resistance loss L adv The calculation formula is: Among them, physical constraint loss L smooth The calculation formula is: In the formula, λ is the hyperparameter; E represents the mathematical expectation; P Z (Z) represents the probability distribution of random noise; P data (C) represents the data distribution under actual operating conditions; D(.) represents the output probability of the discriminator; G(.) represents the output of the generator; N represents the length of the temperature sequence; T i This represents the predicted temperature value at the i-th elevation point; The loss function used by the discriminator The expression is: In the formula, This represents a discriminator that outputs the probability of "true" under given operating conditions.

[0042] In one example of the present invention, step S40 specifically includes the following: An optimization algorithm based on gradient descent and its hyperparameter configuration strategy are adopted. The network is trained on the training set using an alternating discriminator and generator update strategy, and then trained on the test set using the coefficient of determination R. 2 The model was evaluated using mean absolute error (MAE); training parameters were adjusted during training by monitoring the training effect on the validation set, with a hyperparameter learning efficiency of 2 × 10⁻⁶. -4 The minimum number of training epochs is 500 to ensure prediction accuracy, and the maximum is 1000 to prevent overfitting. The coefficient of determination R is used on the test set. 2 The model is evaluated using the mean absolute error (MAE) and the coefficient of determination (R²). 2 >0.9, mean absolute error (MAE) <50℃.

[0043] In one example of the present invention, in step S50, the semantic recognition model includes a predefined semantic template and a rule parser. The predefined semantic template is configured to identify the working condition parameters and their units in the user's text. The rule parser is configured to parse the natural language text input by the user, apply template matching rules and syntax analysis algorithms, extract the working condition parameter values, and convert them into a standardized data format that the model can process.

[0044] The role and function of predefined semantic templates: Predefined semantic templates are a core component of the semantic recognition model. Their functions include: standardized input guidance and keyword pattern matching. Standardized input guidance provides structured input examples (such as "fire source intensity: X MW", "wind speed: Y m / s", "distance: Z m") to guide users to input text in a standardized format, reducing ambiguity. Keyword pattern matching utilizes a built-in keyword library related to operating parameters (such as "fire source", "power", "wind speed", "distance") and unit identifiers (such as "MW", "m / s", "m") to quickly locate parameters through pattern matching. Through template-based design, this component improves the accuracy and efficiency of parameter recognition.

[0045] The role and functions of the rule parser: The rule parser is the execution component of the semantic recognition model. Its functions include: text parsing and feature extraction, rule application and validation, and data standardization output. Text parsing and feature extraction refer to using a rule engine (such as regular expressions) or lightweight natural language processing (NLP) techniques to segment user input, identify entities, and extract relationships, recognizing parameter values ​​and units. Rule application and validation refer to using rules based on predefined semantic templates (such as numerical format validation and unit conversion rules) to logically validate the extracted parameters, ensuring data validity. Data standardization output refers to converting the extracted parameters into the normalized format required by the model (such as converting "10 megawatts" to the floating-point number 10.0) and outputting it to the fire temperature prediction model. This component realizes the automated conversion from unstructured text to structured data, reducing the user's operational threshold.

[0046] A bridge cable vehicle fire temperature-height curve generation system according to a second aspect of the present invention includes: The simulation model module is configured to build a bridge vehicle fire simulation model, and the simulation model calculates and outputs temperature-height data under different working conditions. The data processing and partitioning module is configured to construct datasets from the temperature-height data and experimental data output by the simulation model and their corresponding operating parameters; preprocess the datasets and partition them into training, validation and test sets. The fire temperature prediction module is configured to build a bridge cable vehicle fire temperature prediction model based on a conditional generative adversarial network. This model is a generator-discriminator dual-path system. The model training module is configured to train, validate, and test a fire temperature prediction model using a loss function using a training set, a validation set, and a test set. The working condition parameter acquisition module is configured to acquire working condition parameters for bridge vehicle fire scenarios using a semantic recognition model. The predicted temperature calculation module is configured to input the identified working condition parameters into the well-trained bridge vehicle fire temperature prediction model, calculate and output the predicted temperature value corresponding to the set standard height point within the preset height range. The temperature-height curve module is configured to generate a temperature-height curve for bridge cable vehicle fires based on the predicted temperature values.

[0047] The improved computational efficiency of this curve generation system: Traditional FDS-based numerical simulation methods often require hours or even days to calculate a temperature-elevation curve, severely limiting engineering design efficiency. This curve generation system, through forward prediction using a trained conditional generative adversarial network model, can generate a high-precision temperature-elevation curve in milliseconds, improving computational efficiency by several orders of magnitude. This represents a leap from "hours" to "milliseconds," meeting the needs of engineering design.

[0048] This curve generation system boasts strong generalization capabilities and wide engineering applicability. By generating a massive number of random operating conditions for training, the model can cover unknown operating conditions far exceeding the training sample range. The temperature protection curves generated by this system are based on the optimization results of the envelopes of a large number of random operating conditions, rather than the calculation results of a single condition. Therefore, its protection recommendations are more universal and safer, providing scientific and customized fire protection design basis for bridges of different sizes and fire risk levels.

[0049] This curve generation system boasts a high degree of automation and intelligence: it automates the entire process from inputting operating conditions to generating curves. Once the model training is complete, users only need to provide key operating parameters (or extract them automatically through semantic recognition) to generate temperature-height curves with a single click. This eliminates the need for complex simulation software operations and extensive fluid dynamics expertise, lowering the technical barrier to entry and enhancing the automation level of the design process.

[0050] This curve generation system innovatively applies AI and large-scale model technology to the traditional field of engineering safety: it successfully applies the cutting-edge conditional generative adversarial network (GAN) technology—a "large-scale model"—to the specific engineering scenario of fire protection for bridges, cables, and vehicles. It not only solves the efficiency bottleneck of traditional methods but, more importantly, provides a new paradigm of "physical mechanism + data-driven approach."

[0051] It should be noted that the bridge cable vehicle fire temperature-height curve generation system of the present invention can also perform any of the processing described in the previously described bridge cable vehicle fire temperature-height curve generation method, the specific details of which will not be repeated here.

[0052] Specific examples: This embodiment takes the main cable of a typical long-span cable-stayed bridge as the research object, considering the temperature distribution of the cable along the height direction caused by a vehicle fire on the bridge deck. The goal is to quickly generate the temperature-height curve of the main cable within the range of 0-35 m under given operating parameters (fire source intensity, bridge deck wind speed, and fire source distance). The curve generation process includes the following steps in sequence: fire simulation modeling and data generation, dataset construction and preprocessing, conditional generative adversarial network model construction and training, semantic recognition to obtain operating conditions, model inference to output temperature values ​​at standard height points, and generation of temperature-height curves. The specific steps are as follows: S10: Bridge Vehicle Fire Simulation Model Construction and Data Generation: In the numerical simulation modeling of fire, the two-dimensional cable plane is simplified into a one-dimensional vertical bar, and the height of the vertical bar is used to represent any position of the cable; the vehicle is simplified into a cube and a fire source is set on its surface; temperature measuring points are arranged along the height direction on the surface of the vertical bar according to the accuracy requirements; at the same time, the recommended maximum grid size is selected according to the fire source intensity, and the simulation method of using the maximum heat release efficiency flame for 10 minutes to replace the actual flame's highest temperature for 1.5 hours is adopted to obtain temperature data in the stable phase. This embodiment selects a set of working conditions: fire source intensity: 30 MW; bridge deck wind speed: 5 m / s; fire source distance: 2.5 m (example of the distance between the emergency lane and the adjacent lane and the edge of the main cable). The above three types of working condition parameters belong to the input parameter type and range system defined by this method, and the fire source intensity, bridge deck wind speed, and fire source distance can be combined and covered according to preset intervals during the sample generation stage.

[0053] S20: Dataset Construction, Preprocessing, and Partitioning: Samples are constructed by combining the "height-temperature" data output from numerical simulations with corresponding operating parameters. Experimental data is also introduced as a supplementary sample source; this data can be directly collected from a full-scale or large-scale model fire-resistant test platform under controlled parameter wind, fire, and force coupling conditions. The dataset is preprocessed by mapping the non-uniform height-temperature data output from simulations to 36 standard height points within the range of 0–35 m at 1-meter intervals using linear interpolation. The operating parameter matrix and temperature value matrix are then normalized. Subsequently, the training set, validation set, and test set are partitioned in an 8:1:1 ratio.

[0054] S30: Fire Temperature Prediction Model Construction: Construct a temperature prediction model based on conditional generative adversarial networks. The model is a generator-discriminator dual-path system: the generator generates temperature curve prediction values ​​based on noise vectors and normalized chemical condition parameters, and the discriminator is used to distinguish between the generated temperature curve and the real temperature curve and output the discrimination probability value.

[0055] The generator structure is as follows: The input is a concatenated vector of a 100-dimensional noise vector and a 3-dimensional normalized working condition parameter vector; it passes through a first fully connected hidden layer with 256, 512, and 256 nodes in sequence, followed by batch normalization and LeakyReLU after each first fully connected hidden layer; the output layer is a 36-node fully connected layer with Tanh activation to output the temperature prediction values ​​of 36 standard altitude points.

[0056] Discriminator structure: A dual-branch structure is adopted. The conditional flow uses a 64-node second fully connected layer to process the 3D operating condition vector, and the data flow uses a 64-node third fully connected layer to process the 36-dimensional temperature curve. The outputs of the two branches are concatenated into a 128-dimensional vector, and then passed through a 256-node and 128-node fourth fully connected layer (each layer is followed by batch normalization and LeakyReLU activation function). Finally, the discriminant probability value is output through Sigmoid.

[0057] S40: Model Training and Evaluation: During the training phase, a weighted combined loss function, incorporating adversarial loss and physical constraints, is used to train the model. The physical constraints ensure the temperature curve exhibits smoothness with altitude variation. Training employs an alternating iterative update strategy of "discriminator first, generator second," and the performance is monitored on the validation set to adjust training parameters. Example hyperparameter: learning rate. Training rounds: 700; Prediction accuracy: MAE=18.40℃, RMSE=24.98℃, R2=0.9908.

[0058] S50: Semantic Recognition for Obtaining Operating Condition Parameters: In engineering applications, users can input fire scenarios using natural language, such as: "Fire source intensity 30MW, bridge surface wind speed 5m / s, fire source distance 2.5m". The semantic recognition model identifies parameters and units through predefined semantic templates, and the rule parser extracts the values, completes unit processing and standardization, and outputs structural operating condition parameters that the model can handle.

[0059] S60~S70: The model inference outputs temperature values ​​and generates temperature-height curves. The working condition parameters obtained through semantic recognition and normalized are input into the fully trained temperature prediction model. The generator outputs normalized temperature prediction values ​​corresponding to 36 standard height points (0~35 m, step size 1 m), and then performs inverse normalization to restore the actual temperature values, and draws temperature-height curves accordingly.

[0060] Comparative Validation: To verify the effectiveness of the model, several representative test conditions were selected for comparison. Numerical Comparison: For the selected test conditions, numerical simulations such as FDS were used to obtain the actual temperature-height curves, which were then compared with the model-predicted curves. The results show that the two curves are generally consistent within the range of 0–35 m, accurately reflecting the influence trends of changes in fire source intensity, wind speed, and distance on temperature distribution. The corresponding comparison results are as follows: Figure 4 As shown.

[0061] Experimental Comparison: Temperature-height curves collected from the experimental platform under different operating conditions were compared with the model prediction results. The results show that the model prediction curves and the experimental curves have consistent overall shape and attenuation patterns. The corresponding comparison results are as follows: Figure 5 As shown.

[0062] Conclusion: This embodiment demonstrates that the proposed method can directly output predicted temperature values ​​at 36 standard altitude points within the range of 0–35 m and generate temperature-altitude curves after inputting operating parameters, exhibiting good consistency in both numerical simulation and experimental data comparisons. Furthermore, the model training employs a weighted combination of adversarial loss and physical constraints, which helps ensure the smoothness and engineering usability of the generated curves. (The last sentence appears to be incomplete and possibly refers to a separate point about R.) 2 Under the premise of an accuracy evaluation index of >0.9 and MAE<50 ℃, the cost of traditional simulation calculations per working condition can be significantly reduced, enabling the rapid generation of temperature-height curves for bridge cable fire protection design and defense curve formulation.

[0063] The foregoing description, with reference to preferred embodiments, details an exemplary implementation of the method and system for generating temperature-height curves in bridge cable vehicle fires proposed by the present invention. However, those skilled in the art will understand that various modifications and alterations can be made to the above specific embodiments without departing from the concept of the present invention, and various combinations can be made to the various technical features and structures proposed by the present invention without exceeding the protection scope of the present invention, which is determined by the appended claims.

Claims

1. A method for generating a temperature-height curve in a bridge cable vehicle fire, characterized in that, Includes the following steps: S10: Construct a bridge vehicle fire simulation model, and output temperature-height data under different working conditions based on the simulation model. S20: Construct datasets from the temperature-height data and experimental data output by the simulation model and their corresponding operating parameters; preprocess the datasets and divide them into training, validation and test sets; S30: Construct a temperature prediction model for bridge cable vehicle fires based on conditional generative adversarial networks. S40: Train, validate, and test the fire temperature prediction model using the loss function using the training set, validation set, and test set; S50: Use a semantic recognition model to obtain the operating parameters of a bridge vehicle fire scenario; S60: Input the identified working condition parameters into the well-trained bridge vehicle fire temperature prediction model, calculate and output the temperature prediction value corresponding to the set standard height point within the preset height range. S70: Based on the predicted temperature value, generate a temperature-height curve for a bridge cable vehicle fire.

2. The method for generating a temperature-height curve for a bridge cable vehicle fire according to claim 1, characterized in that, In step S10, constructing the bridge vehicle fire simulation model specifically includes: The two-dimensional cable plane is simplified into a one-dimensional vertical bar, and the arbitrary position of the cable is represented by the height position of the vertical bar, thus simplifying the two-dimensional spatial coordinates into a one-dimensional position. The vehicle is simplified into a cube with a fire source on its surface. Recommended maximum grid sizes are given for different vehicle types: 20cm for a maximum heat release rate of 0-5MW; 25cm for 5-10MW; 30cm for 10-15MW; 35cm for 15-20MW; and 50cm for 20-200MW. The maximum heat release efficiency of the flame is simulated for 8-12 minutes to equivalently replace the highest temperature of the actual flame simulated for 1.5 hours. Temperature measuring points are set at equal intervals along the height direction on the surface of the vertical rod according to the required accuracy.

3. The method for generating a temperature-height curve for a bridge cable vehicle fire according to claim 1, characterized in that, In step S10, the operating parameters include: fire source intensity, bridge deck wind speed, and fire source distance; wherein, the fire source intensity is densely sampled in a uniform distribution in the range of 0-50MW, and sparsely sampled in a uniform distribution in the range of 50-200MW; the bridge deck wind speed is sampled in a uniform distribution in the range of 0-15m / s; and the fire source distance is the distance between the emergency lane and the adjacent lane from the edge of the main cable.

4. The method for generating a temperature-height curve for a bridge cable vehicle fire according to claim 1, characterized in that, In step S20, the preprocessing of the dataset includes: mapping the non-uniform temperature-height data output by the simulation model to a standard height sampled at equal intervals within 0-35m using linear interpolation; and normalizing the working condition parameter matrix and the temperature value matrix respectively.

5. The method for generating a temperature-height curve for a bridge cable vehicle fire according to claim 1, characterized in that, The fire temperature prediction model includes a generator and a discriminator. The generator is configured to generate predicted temperature curves based on a noise vector and normalized working condition parameters. The discriminator is configured to distinguish between the generated temperature curves and the actual temperature curves and output a discrimination probability value.

6. The method for generating a temperature-height curve for a bridge cable vehicle fire according to claim 5, characterized in that, The generator comprises an input layer, three first fully connected hidden layers, and a first output layer. The input layer is a concatenated vector of a noise vector and normalized working condition parameters, used for data input and feature initialization. The three first fully connected hidden layers are connected sequentially, with each first fully connected hidden layer followed by a batch normalization layer and a LeakyReLU activation function, used for feature extraction and nonlinear transformation at different levels, respectively. The first output layer comprises a first fully connected layer and a Tanh activation function, used to output predicted temperature values ​​at multiple standard altitude points. The discriminator is a two-stream network structure, including a conditional stream, a data stream, a splicing layer, two second fully connected hidden layers, and a second output layer. The conditional stream includes a second fully connected layer and a LeakyReLU activation function, which are used to process the operating condition vector and extract conditional features. The data stream includes a third fully connected layer and a LeakyReLU activation function, which are used to process the temperature curve vector and extract data features. The concatenation layer concatenates the feature vectors output from the conditional stream and the data stream into a fused feature vector; the two second fully connected hidden layers are connected sequentially, with each second fully connected hidden layer followed by a LeakyReLU activation function, which are used for feature fusion and discriminant analysis, respectively; the second output layer includes a fourth fully connected layer and a Sigmoid activation function, which are used to output the discriminant probability value.

7. The method for generating a temperature-height curve for a bridge cable vehicle fire according to claim 5, characterized in that, In step S40, the weighted combined loss function L used by the generator in the fire temperature prediction model... total From the counter-loss L adv and physical constraint loss L smooth Weighted according to the following formula: Among them, the resistance loss L adv The calculation formula is: Among them, physical constraint loss L smooth The calculation formula is: In the formula, λ is the hyperparameter; E represents the mathematical expectation; P Z (Z) represents the probability distribution of random noise; P data (C) represents the data distribution under actual operating conditions; D(.) represents the output probability of the discriminator; G(.) represents the output of the generator; N represents the length of the temperature sequence; T i This represents the predicted temperature value at the i-th elevation point; The loss function used by the discriminator The expression is: In the formula, This represents a discriminator that outputs the probability of "true" under given operating conditions.

8. The method for generating a temperature-height curve for a bridge cable vehicle fire according to claim 1, characterized in that, Step S40 specifically includes the following: An optimization algorithm based on gradient descent and its hyperparameter configuration strategy are adopted. The network is trained on the training set using an alternating discriminator and generator update strategy, and then trained on the test set using the coefficient of determination R. 2 The model was evaluated using mean absolute error (MAE); training parameters were adjusted during training by monitoring the training effect on the validation set, with a hyperparameter learning efficiency of 2 × 10⁻⁶. -4 The minimum number of training epochs is 500 to ensure prediction accuracy, and the maximum is 1000 to prevent overfitting. The coefficient of determination R is used on the test set. 2 The model is evaluated using the mean absolute error (MAE) and the coefficient of determination (R²). 2 >0.9, mean absolute error (MAE) <50℃.

9. The method for generating a temperature-height curve for a bridge cable vehicle fire according to claim 1, characterized in that, In step S50, the semantic recognition model includes a predefined semantic template and a rule parser. The predefined semantic template is configured to identify the working condition parameters and their units in the user's text. The rule parser is configured to parse the natural language text input by the user, apply template matching rules and syntax analysis algorithms, extract the working condition parameter values, and convert them into a standardized data format that the model can process.

10. A system for generating temperature-height curves in bridge cable-stayed vehicles fires, characterized in that, include: The simulation model module is configured to build a bridge vehicle fire simulation model, and the simulation model calculates and outputs temperature-height data under different working conditions. The data processing and partitioning module is configured to construct datasets from the temperature-height data and experimental data output by the simulation model and their corresponding operating parameters; preprocess the datasets and partition them into training, validation and test sets. The fire temperature prediction module is configured to build a bridge cable vehicle fire temperature prediction model based on a conditional generative adversarial network. The model training module is configured to train, validate, and test a fire temperature prediction model using a loss function using a training set, a validation set, and a test set. The working condition parameter acquisition module is configured to acquire working condition parameters for bridge vehicle fire scenarios using a semantic recognition model. The predicted temperature calculation module is configured to input the identified working condition parameters into the well-trained bridge vehicle fire temperature prediction model, calculate and output the predicted temperature value corresponding to the set standard height point within the preset height range. The temperature-height curve module is configured to generate a temperature-height curve for bridge cable vehicle fires based on the predicted temperature values.

Images