Tank fire digital twin deduction method and device

Abstract

This invention relates to the field of computer simulation technology, and provides a method and apparatus for digital twin simulation of storage tank fires. The method includes: acquiring a dataset of standard multi-source parameters of the target storage tank at a fire scene, including its structure, environment, and fire intensity; constructing mapping relationships based on multi-layer coding rules to generate multi-dimensional and joint codes, and establishing a coding index library; extracting parameter correlation relationships to construct a multi-parameter coupled model; performing fire simulation based on the model; constructing a three-dimensional twin model and mapping the codes and simulation results in real time; receiving fire extinguishing tactical parameters for dynamic pre-simulation and evaluation; and generating a standardized emergency decision report. This achieves real-time simulation of strong coupling of multiple parameters related to storage tank structure, environment, and fire intensity, as well as dynamic quantitative evaluation and optimization of fire extinguishing tactics.

Description

Technical Field

[0001] This invention relates to the field of computer simulation technology, and in particular to a method and apparatus for digital twin simulation of storage tank fires. Background Technology

[0002] In the current field of emergency response and tactical development for tank fires, mainstream technologies mostly adopt single-dimensional monitoring or modeling methods, combined with historical case experience to develop fire extinguishing plans. These technologies have been applied to the safety management of tanks in industries such as oil and gas and chemicals.

[0003] In existing technologies, some solutions focus only on monitoring fire parameters or the structural safety of storage tanks, lacking the collaborative collection and integration of multiple parameters related to structure, environment, and fire. This makes it impossible to quantify the impact of multi-factor interactions on fire evolution and structural safety. Some coupled modeling solutions use loose coupling, resulting in delayed parameter linkage and a lack of clear constraint rules and conflict resolution mechanisms. This makes it difficult to balance computational accuracy and real-time performance, and it cannot adapt to the dynamic scenarios of rapidly developing fires. At the same time, existing technologies lack a unified digital representation method for multi-source parameters. The large differences in parameter dimensions and the difficulty in fusion lead to low modeling efficiency and error-prone parameter matching. Furthermore, traditional tactical formulation relies heavily on human experience and lacks digital twin visualization and quantitative evaluation. Tactical optimization lacks data support, which can easily lead to untimely response and poor tactical adaptability. This makes it difficult to meet the needs of efficient and accurate emergency response to large storage tank fires and cannot effectively avoid secondary accident risks such as tank wall rupture and fire spread. Summary of the Invention

[0004] This invention provides a digital twin simulation method and device for tank fires, which addresses the shortcomings of existing technologies in emergency response to tank fires, such as reliance on single-dimensional monitoring, difficulty in integrating multi-source heterogeneous parameters, lack of unified digital characterization methods, reliance on human experience in tactical formulation, and inability to quantify and simulate fires. It enables real-time simulation of tank structure, environment, and fire intensity with strong coupling of multiple parameters, and dynamic quantitative evaluation and optimization of fire extinguishing tactics, providing scientific decision support for emergency rescue.

[0005] This invention provides a digital twin simulation method for tank fires, comprising: Obtain a standard multi-source parameter dataset of the fire scene, including the tank structure, fire environment, and fire intensity characteristics of the target storage tank; Based on the standard multi-source parameter dataset and the preset multi-layer coding rules, a mapping relationship between parameters and coding rules is constructed, and multi-dimensional codes of each parameter and joint codes of associated parameters are generated based on the mapping relationship. A coding index library is also established based on the multi-dimensional codes and the joint codes. Based on the encoding index library, the correlation between each parameter is extracted, and a multi-parameter coupling model is constructed based on the correlation. Fire simulations were performed based on the multi-parameter coupled model, and the simulation results were obtained. A three-dimensional twin model of the target storage tank and its surrounding environment is constructed, and the multidimensional encoding of each parameter and the inference calculation results are mapped to the three-dimensional twin model in real time. The received fire extinguishing tactical parameters are dynamically simulated in the mapped 3D twin model to obtain fire extinguishing tactical simulation results and evaluation results; Based on the results of the firefighting tactical drills and assessments, a standardized emergency decision-making report is generated, which includes detailed tactical parameters, execution procedures, and risk warnings.

[0006] In one possible implementation, the method further includes: The relationships between the parameters include the correlation threshold level, the mutual influence coefficient level, the constraint priority, and the coupling topology. Parse the constraint rule layer encoding of each parameter in the encoding index library, and extract the correlation threshold level, mutual influence coefficient level and constraint priority between each parameter; The correlation threshold level, mutual influence coefficient level and constraint priority of each parameter are converted into physical constraint threshold and coupling weight coefficient to obtain the parameter coupling topology. Based on the parameter coupling topology, a structural mechanical model, a fire dynamics model, and an environmental impact model of the storage tank are constructed respectively, and a bidirectional data exchange interface is established between the structural mechanical model, the fire dynamics model, and the environmental impact model of the storage tank. Based on the dynamic feature layer encoding in the encoding index library, the coupling trigger condition is set. When the parameter change rate or threshold range of the real-time monitoring data reaches the coupling trigger condition, cross-model coupling calculation is started to obtain the multi-parameter coupling model.

[0007] In one possible implementation, the method further includes: Extreme fire scenario simulation data is obtained based on the multi-parameter coupling model. The extreme fire scenario simulation data is the simulation results of full tank combustion and / or strong wind extreme conditions based on the simulation software in the multi-parameter coupling model. Based on the sliding window algorithm, the extreme fire scenario simulation data and the standard multi-source parameter dataset are compared in real time at a preset period. When the deviation exceeds a preset threshold, the basic parameters of the multi-parameter coupled model are corrected based on the standard multi-source parameter dataset using the least squares method, and the extreme condition adaptation parameters are corrected based on the extreme fire scenario simulation data using Bayesian estimation. The calibrated basic parameters and extreme condition adaptation parameters are obtained and fed back to update the multi-parameter coupled model.

[0008] In one possible implementation, the method further includes: Based on the coupling weight coefficients defined in the constraint rule layer of the coding index library, a two-way coupled deduction of the tank structure mechanical model and the fire dynamics model is performed to obtain the deduction calculation results of the tank wall rupture risk level. The fire environment parameters are mapped to the fire dynamics model calculation grid using the Kriging interpolation method to correct the flame spread parameters, and the fire characteristic parameters are fed back to the environmental impact model to update the local environmental parameters. The two-way coupled deduction of the environmental impact model and the fire dynamics model is performed to obtain the deduction calculation results of the fire spread prediction trajectory. Environmental loads are applied to the structural mechanics model of the storage tank to correct the tank wall stress, and the thermal deformation parameters of the tank wall are fed back to the environmental impact model to optimize the airflow field calculation. The two-way coupled deduction of the structural mechanics model and the environmental impact model of the storage tank is performed to obtain the deduction calculation results of thermal radiation distribution and structural stability index.

[0009] In one possible implementation, the method further includes: The static structural data of the target storage tank is imported using a modeling tool. Based on the fusion of neural radiation field and real-time fire scene image data, a three-dimensional twin model of the target storage tank and its surrounding environment is constructed, and a real-time synchronization mechanism between the three-dimensional twin model and the physical scene is established. The basic attribute layer and dynamic feature layer in the multidimensional encoding are analyzed, the spatial location identifier and type attribute of each parameter are extracted, the spatial location identifier is associated with the spatial coordinates of the three-dimensional twin model, and the mapping position of each parameter in three-dimensional space is determined. The inference and calculation results are mapped to the corresponding spatial coordinates of the three-dimensional twin model in real time.

[0010] In one possible implementation, the method further includes: The system receives multiple sets of fire extinguishing tactical parameters input from the user terminal, including the type of extinguishing agent, spray flow rate, spray angle, and rescue deployment location. Each set of fire extinguishing tactical parameters is input into the multi-parameter coupled model, which drives the model to calculate the fire evolution and structural response after the tactical execution. The fire extinguishing tactical execution process is dynamically demonstrated in the mapped three-dimensional twin model to obtain the corresponding fire extinguishing tactical pre-simulation results. The fire extinguishing tactical execution process includes flame suppression effect, smoke diffusion change and tank wall temperature drop process. Based on a pre-set evaluation index system, the results of each group of fire-fighting tactical drills are quantitatively evaluated. The weight of each evaluation index is determined by the analytic hierarchy process and the comprehensive score is calculated to obtain the evaluation results of each group of fire-fighting tactical parameters, which include multi-dimensional scores. The evaluation indexes include fire control efficiency, structural protection level, rescue safety and fire-fighting timeliness.

[0011] In one possible implementation, the method further includes: Analyze the physical relationships between parameters in the standard multi-source parameter dataset, identify parameter pairs with coupling relationships, and determine the association threshold, mutual influence coefficient, and constraint priority of each parameter pair; Based on preset multi-layer coding rules, hierarchical coding is performed on each parameter in the standard multi-source parameter dataset. The hierarchical coding includes basic attribute layer coding, dynamic feature layer coding, and constraint rule layer coding, and the hierarchical coding results are combined to generate multi-dimensional coding. The parameter pairs with coupling relationships are jointly encoded to establish a correspondence between parameter names and multidimensional encoding fields, and to construct a mapping relationship between parameters and encoding rules. An encoding index library is established based on the multidimensional encoding, the joint encoding, and the mapping relationship. The encoding index library stores the bidirectional mapping relationship between the encoding and the original parameters, parameter attributes, and the correlation between parameters.

[0012] In one possible implementation, the method further includes: A hybrid acquisition mode combining fixed monitoring and mobile supplementation is adopted to collect in real time the structural parameters of the target storage tank, fire environmental parameters, and fire characteristic parameters of the fire scene. The collected tank structural parameters, fire environment parameters, and fire characteristic parameters are preprocessed, and the parameters of different dimensions are uniformly mapped to a preset range using a standardization method to obtain the standard multi-source parameter dataset.

[0013] The present invention also provides a digital twin simulation device for tank fires, comprising the following modules: The acquisition module is used to acquire a standard multi-source parameter dataset of the fire scene, including the tank structure, fire environment, and fire intensity characteristics of the target storage tank. The construction module is used to construct a mapping relationship between parameters and encoding rules based on the standard multi-source parameter dataset and preset multi-layer encoding rules, generate multi-dimensional encodings of each parameter and joint encodings of associated parameters based on the mapping relationship, and establish an encoding index library based on the multi-dimensional encodings and the joint encodings. A construction module is used to extract the correlation between parameters based on the encoding index library, and to construct a multi-parameter coupling model based on the correlation. The simulation module is used to perform fire simulation based on the multi-parameter coupled model and obtain the simulation calculation results. The mapping module is used to construct a three-dimensional twin model of the target storage tank and its surrounding environment, and to map the multidimensional encoding of each parameter and the inference calculation results to the three-dimensional twin model in real time. The pre-simulation module is used to receive fire extinguishing tactical parameters and perform dynamic pre-simulation in the mapped 3D twin model to obtain fire extinguishing tactical pre-simulation results and evaluation results; The generation module is used to generate a standardized emergency decision report based on the fire extinguishing tactical simulation results and assessment results, which includes details of tactical parameters, execution procedures and risk warnings.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the digital twin simulation method for tank fires as described above.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the digital twin simulation method for tank fires as described above.

[0016] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the digital twin simulation method for tank fires as described above.

[0017] The present invention provides a digital twin simulation method and apparatus for tank fires. This method acquires a standard multi-source parameter dataset containing the tank structure, fire environment, and fire characteristics of the target tank at a fire scene. Based on the standard multi-source parameter dataset and preset multi-layer coding rules, a mapping relationship between parameters and coding rules is constructed. Based on the mapping relationship, multi-dimensional codes for each parameter and joint codes for associated parameters are generated, and a coding index library is established based on the multi-dimensional codes and joint codes. Based on the coding index library, the correlation relationships between parameters are extracted, and a multi-parameter coupling model is constructed based on these relationships. Fire simulation is performed based on the multi-parameter coupling model to obtain simulation calculation results. A three-dimensional twin model of the target tank and its surrounding environment is constructed, and the multi-dimensional codes of each parameter and the simulation calculation results are mapped to the three-dimensional twin model in real time. Firefighting tactical parameters are received and dynamically simulated in the mapped three-dimensional twin model to obtain firefighting tactical simulation results and evaluation results. Based on the firefighting tactical simulation results and evaluation results, a standardized emergency decision-making report containing detailed tactical parameters, execution procedures, and risk warnings is generated. Compared to existing technologies that rely on single-dimensional monitoring for emergency response to tank fires, have difficulties in integrating multi-source heterogeneous parameters, lack unified digital characterization methods, and rely on human experience for tactical formulation without quantitative simulation, this solution enables real-time simulation of strong coupling of multiple parameters of tank structure, environment, and fire, as well as dynamic quantitative assessment and optimization of fire extinguishing tactics, providing scientific decision support for emergency rescue. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the digital twin simulation method for tank fires provided by the present invention.

[0020] Figure 2 This is a schematic diagram of the architecture of the digital twin simulation system for tank fires provided by the present invention.

[0021] Figure 3 This is a schematic diagram of the structure of the digital twin simulation device for tank fires provided by the present invention.

[0022] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0024] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of the present invention.

[0025] Figure 1 This is a flowchart illustrating the digital twin simulation method for tank fires provided by the present invention, as shown below. Figure 1 As shown, the method includes the following: S11. Obtain a standard multi-source parameter dataset of the fire scene, including the tank structure, fire environment, and fire characteristics of the target tank.

[0026] This step aims to establish a data bridge between the physical fire scene and the digital system, resolving the issues of multi-source heterogeneous data acquisition and standardization. In specific implementation, for target storage tanks experiencing a fire, a hybrid acquisition mode combining fixed monitoring and mobile supplementation can be adopted to collect all three types of key parameters: Regarding the acquisition of tank structural parameters, strain sensors and temperature sensors deployed at key parts of the tank wall (tank top, tank body, and tank bottom welds) are used to acquire dynamic structural parameters such as tank wall thermal deformation, temperature distribution, and stress changes in real time at a variable sampling frequency of 1-10Hz. At the same time, tank design drawings can be imported through Building Information Modeling (BIM) modeling tools to extract static structural parameters such as tank wall thickness, material, tank volume, and supporting structure dimensions, forming a structural parameter library.

[0027] Regarding the collection of fire environmental parameters, meteorological sensors and environmental monitoring terminals deployed in and around the storage tank area continuously collect meteorological parameters such as wind speed, wind direction, ambient temperature, humidity, and atmospheric pressure at a sampling frequency of 0.5-5Hz. Combined with UAV aerial photography data and Geographic Information System (GIS), spatial environmental parameters such as surrounding terrain undulations, building distribution, location and operating conditions of fire-fighting facilities (fire hydrants, foam stations), and evacuation route layout are collected.

[0028] Regarding the acquisition of fire characteristic parameters, dynamic fire parameters such as flame temperature, thermal radiation intensity, smoke concentration and composition, flame height, and fire spread rate are collected in real time using equipment such as infrared thermal imagers, flame detectors, smoke sensors, and thermal radiation sensors. For parameters that are difficult to acquire directly, such as heat release rate and flame pulse frequency, simulations can be performed based on the real-time calculation results of the Fire Dynamics Simulator (FDS) tool, and the acquired parameters can be initially calibrated by combining them with historical exercise data.

[0029] After obtaining the above multi-source parameters, data preprocessing is performed: first, noise reduction and outlier removal are carried out to eliminate sensor errors and electromagnetic interference; for data loss caused by possible equipment disconnection, missing values ​​are filled in using a cache unit; finally, the Z-score standardization method is used to uniformly map parameters of different dimensions (such as temperature ℃, pressure Pa, velocity m / s) to the [0,1] interval to eliminate modeling bias caused by differences in dimensions, and finally output a standardized multi-source parameter dataset, providing a data foundation with a unified format for subsequent encoding.

[0030] S12. Based on the standard multi-source parameter dataset and the preset multi-layer coding rules, construct the mapping relationship between parameters and coding rules, generate multi-dimensional codes for each parameter and joint codes for associated parameters based on the mapping relationship, and establish a coding index library based on the multi-dimensional codes and the joint codes.

[0031] The core of this step lies in solving the problem of "inconsistent semantics and unknown correlations" among multi-source parameters, and achieving standardized management of parameters by constructing a digital representation system. First, the physical correlations between parameters in the standard multi-source parameter dataset are analyzed to identify parameter pairs with coupling relationships (such as wind speed and flame spread rate, tank wall temperature and structural strength), and the correlation thresholds, mutual influence coefficients, and constraint priorities of each parameter pair are determined to provide a physical basis for the coding of constraint rules.

[0032] Based on this, and using a pre-defined three-layer binary encoding architecture of "basic attribute layer - dynamic feature layer - constraint rule layer", layered encoding is performed on each parameter: The basic attribute layer uses 8-bit binary encoding, where the first 2 bits identify the parameter type (00=structural parameter, 01=environmental parameter, 10=fire parameter), and the last 6 bits consist of a 3-bit parameter name encoding and a 3-bit precision level encoding, which are used to characterize the core basic information of the parameter. The dynamic feature layer uses 16-bit binary encoding. The first 4 bits represent the parameter change rate, the middle 4 bits represent the fluctuation range, the last 4 bits represent the threshold range (safe / warning / danger), and the last 4 bits correspond to the data acquisition frequency, which is used to capture the dynamic evolution characteristics of the parameters. The constraint rule layer uses 8-bit binary encoding. The first 2 bits indicate the constraint priority, the middle 3 bits indicate the mutual influence coefficient level, and the last 3 bits indicate the association threshold level, which is used to characterize the physical constraint relationship between parameters.

[0033] Subsequently, the parameter pairs with coupling relationships are jointly encoded to establish a one-to-one correspondence between parameter names and multidimensional encoding fields, and a structured parameter-encoding rule mapping table is constructed to achieve automatic matching of parameter attributes and encoding fields.

[0034] Finally, an index library is established based on the 32-bit multidimensional encoding (three-layer encoding combination) of each parameter and the joint encoding between parameters. This index library not only stores the bidirectional mapping relationship between encoding and original parameters and parameter attributes, supporting fast querying, parsing and dynamic updating of encoding, but more importantly, it provides a calling interface for parameter association relationships for subsequent construction of multi-parameter coupling models, enabling coupling calculations to directly retrieve the association thresholds and influence coefficients between parameters as the basis for calculation.

[0035] S13. Based on the encoding index library, extract the correlation between each parameter, and construct a multi-parameter coupling model based on the correlation.

[0036] This step aims to build a core computing engine capable of reflecting the complex physical processes of tank fires. First, the coding index library established by S12 is called to parse the constraint rule layer in the three-dimensional coding of each parameter, extract the correlation threshold level, mutual influence coefficient level and constraint priority between parameters, convert them into specific physical constraint thresholds and coupling weight coefficients, obtain the parameter coupling topology, and clarify which parameters need to be calculated together and their influence intensity.

[0037] Based on this coupled topological relationship, three fundamental mechanism models are constructed: a structural mechanical model of the storage tank based on elasticity and fracture mechanics to calculate the thermal stress, strain, and structural stability of the tank wall; a fire dynamics model based on the FDS core algorithm to deduce the flame spread, heat radiation distribution, and smoke diffusion patterns; and an environmental impact model based on the analytic hierarchy process (AHP) to determine the weights of various environmental parameters (such as wind speed 0.5 and humidity 0.2) to quantify the effect of environmental factors on the fire.

[0038] The key lies in establishing bidirectional data exchange interfaces between the three types of models, forming a strongly coupled closed loop: a structure-fire bidirectional coupling interface, an environment-fire bidirectional coupling interface, and a structure-environment bidirectional coupling interface. Simultaneously, to achieve a balance between computational accuracy and real-time performance, a hybrid architecture of "mechanistic model + AI proxy model" is adopted: critical areas (such as the high-temperature zone of the tank wall and the flame core zone) are precisely calculated using the three mechanistic models; non-critical areas are rapidly extrapolated using an AI proxy model built based on Transformer and Random Forest algorithms, ensuring that the computational response time for non-critical areas is controlled within 0.1 seconds.

[0039] In addition, based on the coupling triggering conditions set by the dynamic feature layer encoding in the encoding index library (such as tank wall temperature change ≥5℃, wind speed change ≥1m / s), when the real-time monitoring data shows that the parameter change rate or threshold range reaches the preset trigger threshold, cross-model coupling calculation is automatically started to realize the two-way linkage and real-time iteration of structure-environment-fire, and finally obtain a multi-parameter coupling model that can dynamically reflect the fire evolution and structural response.

[0040] S14. Fire simulation is performed based on the multi-parameter coupling model to obtain the simulation calculation results.

[0041] This step performs actual fire evolution prediction calculations. First, the dynamic feature layer codes of each parameter are parsed using the coding index library. When the parameter change rate or threshold range is detected to reach the preset trigger condition, the strongly coupled closed-loop calculation of the multi-parameter coupled model is initiated.

[0042] During the calculation process, based on the coupling weight coefficients defined in the constraint rule layer of the coded index library, three types of bidirectional coupling logic are executed: In the structure-fire coupling, the thermal radiation intensity and temperature distribution output from the fire dynamics model are input as thermal loads into the tank structural mechanics model, driving the recalculation of tank wall stress and structural stability; simultaneously, the tank wall rupture risk level output from the structural mechanics model is fed back to the fire dynamics model, triggering fire leakage and diffusion calculations when the risk reaches the warning level, correcting the flame spread trajectory and heat release rate. In the environment-fire coupling, Kriging interpolation is used to map environmental parameters to the fire dynamics model calculation grid, correcting the flame height and smoke diffusion path; simultaneously, fire thermal gas flow parameters are fed back to the environmental impact model to update local wind speed and direction. In the structure-environment coupling, environmental loads such as atmospheric pressure and humidity are loaded into the tank structural mechanics model to correct tank wall stress; simultaneously, tank wall thermal deformation parameters are fed back to the environmental impact model to optimize local airflow calculations.

[0043] When the deviation of the same parameter output by multiple models exceeds the association threshold defined in the coding index library, a weighted voting method (e.g., 0.4 for the structural mechanics model, 0.4 for the fire dynamics model, and 0.2 for the environmental impact model) is used to reconcile the conflict, and the weighted average value is calculated as the final parameter value. Through the above strongly coupled closed-loop calculation, the inferred calculation results including the tank wall rupture risk level, fire spread prediction trajectory, thermal radiation distribution, structural stability index, and smoke diffusion range are obtained, providing data support for subsequent visualization mapping.

[0044] S15. Construct a three-dimensional twin model of the target storage tank and its surrounding environment, and map the multidimensional encoding of each parameter and the calculation results to the three-dimensional twin model in real time.

[0045] This step enables physical spatial visualization and real-time synchronization of the calculation results. First, static structural data (tank wall, welds, interface geometry) of the target tank is imported based on BIM technology. Then, real-time image data of the fire scene is fused using Neural Radiance Fields (NeRF) technology to construct a high-precision 3D twin model of the target tank and its surrounding environment (fire protection facilities, terrain, buildings). A real-time synchronization mechanism with the physical scene is established to ensure that the digital image is consistent with the actual fire scene.

[0046] Subsequently, the basic attribute layer and dynamic feature layer in the multidimensional code generated by S12 are analyzed, the spatial location identifier and type attribute of each parameter are extracted, the spatial location identifier is associated with the spatial coordinates of the three-dimensional twin model, and the mapping position of the parameter in three-dimensional space is determined (such as the 3rd monitoring point on the north side of the tank wall, the southeast edge of the flame core area).

[0047] Finally, the state values ​​from the simulation calculations obtained in S14 are mapped to corresponding spatial coordinates in real time: the tank wall stress value and rupture risk level are mapped to the color gradient of the tank wall surface (green - safe, yellow - warning, red - danger); the thermal radiation intensity is mapped to a thermal map overlay; the fire spread range is mapped to a dynamic flame particle effect; and the smoke diffusion path is mapped to a smoke flow animation. Through the above mapping, the real-time visualization of the safety status of the storage tank structure and the evolution of the fire is achieved, resulting in a mapped three-dimensional twin model, providing a visualization carrier for tactical rehearsals.

[0048] S16. Receive fire extinguishing tactical parameters and perform dynamic pre-simulation in the mapped 3D twin model to obtain fire extinguishing tactical pre-simulation results and evaluation results.

[0049] This step enables the visualization, simulation, and quantitative evaluation of firefighting plans. First, it receives multiple sets of firefighting tactical parameters, including extinguishing agent type (such as protein foam or fluoroprotein foam), spray flow rate, spray angle, spray location, and rescue deployment plan.

[0050] These tactical parameters are input into a multi-parameter coupled model, driving the model to calculate the fire evolution and structural response under different tactical interventions: the model calculates the flame temperature drop curve based on the extinguishing agent flow rate, the interaction between the extinguishing agent coverage area and the flame based on the spray angle, and the degree of structural stress relief based on the tank wall cooling effect. Simultaneously, the tactical execution process is dynamically demonstrated in the mapped 3D twin model: displaying the extinguishing agent spray trajectory, flame suppression effect, smoke concentration changes, tank wall temperature drop process, and fire spread obstruction, providing intuitive results of the fire extinguishing tactical simulation.

[0051] The results of the pre-drill were quantitatively evaluated based on a pre-set evaluation index system. This system covers four dimensions: fire control efficiency (flame temperature drop rate, spread rate suppression rate), structural protection level (reduction in tank wall stress, change in rupture risk level), rescue safety (firefighter heat exposure dose, availability of evacuation routes), and fire extinguishing timeliness (extinguishing time, extinguishing agent consumption). The analytic hierarchy process (AHP) was used to determine the weight of each index, automatically calculating the comprehensive score of each tactical plan to obtain an evaluation result containing multi-dimensional scores, thus clarifying the advantages and disadvantages of each plan.

[0052] S17. Based on the results of the fire extinguishing tactical rehearsal and assessment, generate a standardized emergency decision-making report that includes detailed tactical parameters, execution procedures, and risk warnings.

[0053] This step completes the transformation from simulation and evaluation to actual decision-making. Based on the fire-fighting tactical simulation and evaluation results obtained in S16, multiple schemes are automatically compared and screened: if a certain scheme has the highest evaluation score and meets the structural safety constraints, it is determined as the optimal tactical scheme; if the scores of multiple schemes are close, the index weights are adjusted according to user needs (such as prioritizing structural safety or prioritizing improving fire-fighting efficiency) for secondary screening.

[0054] The final tactical plan was optimized and fine-tuned to ensure that parameters such as spray angle and flow rate were within the adjustable range of the actual equipment. Subsequently, the optimal tactical plan, pre-simulation data, and coupled calculation results were integrated to generate a standardized emergency decision-making report. The report includes detailed tactical parameters (fire extinguishing agent type, spray flow rate, angle, and location), execution procedures (phased operation steps and time nodes), risk warnings (risk points of tank wall rupture, personnel heat radiation exposure areas, and early warning of sudden fire spread), and disposal suggestions (alternative evacuation routes and suggestions for reinforcement deployment).

[0055] Finally, through standardized data and control interfaces, the decision report is pushed to the emergency command platform for display, while the optimized tactical parameters (such as spray angle and flow rate setpoints) are directly sent to the on-site fire control system (such as remote-controlled fire monitors and drone bombing systems) to achieve automated execution of tactical parameters and complete the transition from digital simulation to physical implementation.

[0056] Optionally, after generating a standardized emergency decision report and issuing tactical parameters, the system remains in continuous operation, receiving updated data from the fire scene in real time (including dynamic parameters such as newly collected tank wall temperature changes, new locations of fire spread, and sudden changes in wind speed).

[0057] Based on this updated data, the system performs a three-level synchronous update: First, it updates the dynamic feature layer encoding of each parameter in the encoding index library (e.g., adjusting the "rate of change" of the temperature parameter from 0000 to 1111, and adjusting the "threshold range" from safe to warning), refreshing the joint encoding relationship between parameters; Second, it dynamically calibrates the multi-parameter coupled model based on the new data (e.g., using a sliding window algorithm to compare the calculation results with the measured data, and starting the least squares method or Bayesian estimation to correct the model parameters when the deviation exceeds the threshold), ensuring that the model calculation accuracy is adapted to the dynamic fire scenario; Third, it synchronously updates the three-dimensional twin mirror, mapping the latest physical scene state (e.g., newly appearing deformations of the tank wall, new areas of flame spread) to the virtual model in real time, maintaining the consistency between the digital twin and the physical scene.

[0058] After the above updates are completed, the system automatically repeats steps S13 to S17 (i.e., reconstructing the multi-parameter coupled model, performing fire simulation calculations, updating the 3D twin mapping, conducting tactical pre-simulation assessments, and generating new decision reports), forming a cycle of "data acquisition - code update - model calibration - simulation pre-simulation - decision output." This dynamic iterative update mechanism ensures that emergency decision-making plans can be adjusted and optimized in real time as the fire situation evolves, continuously providing fire and rescue services with the latest tactical guidance matching the current fire scene until the fire is completely extinguished.

[0059] The digital twin simulation method for tank fires provided by this invention acquires a standard multi-source parameter dataset containing the tank structure, fire environment, and fire characteristics of the target tank at the fire scene; based on the standard multi-source parameter dataset and preset multi-layer coding rules, a mapping relationship between parameters and coding rules is constructed, and multi-dimensional codes for each parameter and joint codes for associated parameters are generated based on the mapping relationship; a coding index library is established based on the multi-dimensional codes and the joint codes; based on the coding index library, the correlation relationships between each parameter are extracted, and a multi-parameter coupling model is constructed based on the correlation relationships; fire simulation is performed based on the multi-parameter coupling model to obtain simulation calculation results; a three-dimensional twin model of the target tank and its surrounding environment is constructed, and the multi-dimensional codes of each parameter and the simulation calculation results are mapped to the three-dimensional twin model in real time; fire extinguishing tactical parameters are received and dynamically simulated in the mapped three-dimensional twin model to obtain fire extinguishing tactical simulation results and evaluation results; based on the fire extinguishing tactical simulation results and evaluation results, a standardized emergency decision report containing tactical parameter details, execution procedures, and risk warnings is generated. Compared to existing technologies that rely on single-dimensional monitoring for emergency response to tank fires, have difficulties in integrating multi-source heterogeneous parameters, lack unified digital characterization methods, and rely on human experience for tactical formulation without quantitative simulation, this method enables real-time simulation of strong coupling of multiple parameters of tank structure, environment, and fire, as well as dynamic quantitative assessment and optimization of fire extinguishing tactics, providing scientific decision support for emergency rescue.

[0060] Figure 2 This is a schematic diagram of the architecture of the digital twin simulation system for tank fires provided by the present invention. The system adopts a layered architecture design, which is divided into a data acquisition layer, a three-dimensional encoding layer, a multi-parameter coupling layer, a digital twin simulation layer, and a tactical decision-making layer from bottom to top. Each layer works together to achieve a closed loop of the entire process from parameter acquisition to tactical output.

[0061] Data Acquisition Layer The data acquisition layer is used to acquire full data on the tank structure, fire environment, and fire characteristics, providing data support for subsequent modeling and simulation. It adopts a hybrid acquisition mode of "fixed monitoring + mobile supplementation" to balance the comprehensiveness and real-time nature of data acquisition. Specifically, it includes three major acquisition modules and a data preprocessing unit.

[0062] (1) Structural parameter acquisition module: Import tank design drawings through BIM modeling tools to extract static structural parameters such as tank wall thickness, material, tank volume, supporting structure dimensions, tank bottom slope, and interface location to form a structured parameter library; deploy strain sensors and temperature sensors at key parts of the tank wall (tank top, tank body, and tank bottom weld seams) to collect dynamic structural parameters such as tank wall thermal deformation, temperature distribution, and stress changes in real time. The sampling frequency is set to 1-10Hz, which can be dynamically adjusted according to the fire development stage (for example, 1-3Hz is used to reduce data redundancy when the fire is slow, and 5-10Hz is increased when the fire spreads rapidly) to adapt to the needs of rapid parameter changes in fire scenarios and ensure dynamic updates of structural parameters.

[0063] (2) Environmental parameter acquisition module: Meteorological sensors and environmental monitoring terminals are deployed in and around the storage tank area to continuously collect meteorological and environmental parameters such as wind speed, wind direction, ambient temperature, humidity, atmospheric pressure, and atmospheric particulate matter concentration; combined with UAV aerial photography data and GIS geographic information system, spatial environmental parameters such as terrain undulation, distribution of surrounding buildings, location and operating conditions of fire-fighting facilities (fire hydrants, foam stations, water pump stations), and layout of evacuation routes are supplemented. The sampling frequency of meteorological parameters is set to 0.5-5Hz, and the spatial environmental parameters are updated regularly (the update cycle does not exceed 30s) to ensure that the environmental parameters match the actual scene.

[0064] (3) Fire feature acquisition module: Through devices such as infrared thermal imager, flame detector, smoke sensor, and thermal radiation sensor, dynamic fire parameters such as flame temperature, thermal radiation intensity, smoke concentration and composition, flame height, flame shape, and fire spread speed are collected in real time; for parameters that are difficult to collect directly, such as heat release rate and flame pulse frequency, the real-time calculation results of the FDS fire simulation tool are used for extrapolation and supplementation. At the same time, the collected and extrapolated fire parameters are initially calibrated by combining historical tank fire drill data and similar fire case data, optimizing the collection range and accuracy threshold to ensure the accuracy of fire parameters.

[0065] (4) Data preprocessing unit: It has built-in denoising, outlier removal, missing value completion and standardization processing logic. It uses the Z-score standardization method to uniformly map parameters of different dimensions to the [0,1] interval, eliminating modeling bias caused by parameter dimension differences. In response to the potential problems of equipment disconnection and data loss in fire scenarios, a data caching unit is set up with a caching time of not less than 5 minutes. The cached data can be automatically synchronized after the equipment recovers the connection to ensure data continuity. At the same time, the integrity of the preprocessed data is checked, invalid data is removed, and a standardized multi-source parameter dataset is output for subsequent three-dimensional coding layer to call.

[0066] 3D coding layer The three-dimensional coding layer is used to realize the unified digital representation of three types of parameters: tank structure, environment, and fire. It solves the technical pain points of large differences in the dimensions of multi-source parameters, difficulty in integration, and weak linkage. It adopts a hierarchical coding architecture to realize the standardization and normalization of parameter management and provide a unified data interaction foundation for multi-parameter coupled modeling. Specifically, it includes coding rule definition unit, parameter mapping unit, and coding generation unit.

[0067] (1) Encoding rule definition unit: A three-layer encoding standard of "basic attribute layer - dynamic feature layer - constraint rule layer" is formulated, taking into account the representation of basic parameter information, dynamic change characteristics and physical constraint relationships. A binary encoding format is adopted to ensure the simplicity and parsability of the encoding. The encoding rules of each layer are clear and can be implemented. Basic Attribute Layer: This layer represents the core basic information of the parameters, covering parameter type (structure / environment / fire), parameter name, dimensions, data acquisition / inference source, and accuracy level. It uses 8-bit binary encoding, where the first two bits identify the parameter type (00 = structural parameter, 01 = environmental parameter, 10 = fire parameter), and the last six bits consist of a 3-bit parameter name code and a 3-bit accuracy level code (e.g., 00000001 corresponds to tank wall thickness and accuracy level 1 in structural parameters; 10003002 corresponds to thermal radiation intensity and accuracy level 2 in fire parameters). This allows for rapid parameter classification and traceability, avoiding confusion between parameters from multiple sources. Table 1 below provides examples of accuracy level definitions for reference; it is recommended to set these based on the specific scenario of a tank fire. Table 1 Accuracy level error Recommended application scenarios Level 1 ≤±1% Applicable to key parameters (such as tank wall stress, flame core temperature), corresponding to high-precision sensor acquisition or professional simulation data. Level 2 ≤±5% Applicable to common parameters (such as ambient temperature, fire spread rate). Level 3 ≤±10% Applicable to auxiliary parameters (such as atmospheric particulate matter concentration). Dynamic Feature Layer: This layer characterizes the dynamic changes of parameters, including the rate of change, fluctuation range, threshold intervals (safe threshold, dangerous threshold), and update cycle. It employs 16-bit binary encoding to meet the real-time update requirements of dynamic parameters. This allows for accurate capture of dynamic fluctuations in parameters such as fire spread rate, tank wall thermal deformation, and wind speed, providing data support for real-time triggering of subsequent coupled modeling and ensuring that coupled calculations closely reflect dynamic parameter changes. The recommended 16-bit encoding scheme is shown in Table 2 below. Table 2 Encoding splitting name Detailed description 4 people Parameter change rate Characterizing the rate of change, such as 0000 = slow, 1111 = fast, corresponding to specific rate ranges. 4 people Parameter fluctuation range Characterizes the fluctuation range, such as 0000 = small fluctuation, 1111 = large fluctuation, adapting to parameters such as wind speed and tank wall deformation. 4 people Threshold range Integrate safety / hazard threshold indicators, such as 0000 = safe range, 1000 = warning range, 1111 = danger range. 4 people Update cycle The corresponding data acquisition frequencies, such as 0000=0.5Hz, 0010=1Hz…1111=10Hz+, cover the frequency range of the acquisition layer. Constraint Rule Layer: This layer characterizes the physical constraint relationships between parameters, including correlation thresholds, mutual influence coefficients, and constraint priorities. It uses 8-bit binary encoding and is specifically designed for constraint matching in subsequent multi-parameter coupled modeling. Examples include the correlation coefficient range between wind speed and flame spread rate, the constraint threshold between tank wall temperature and structural strength, and the correlation between extinguishing agent dosage and fire extinguishing efficiency. This provides clear rule-based support for strongly coupled calculations. The suggested 8-bit encoding scheme is shown in Table 3 below. Table 3 (2) Parameter mapping unit: Establish a one-to-one correspondence between the standardized parameters output by the data acquisition layer and the three-dimensional coding rules, construct a structured parameter-code mapping table, realize the automatic matching of parameter attributes and coding fields without manual intervention; for parameters that cannot be directly collected and need to be obtained through simulation (such as heat release rate and tank wall ultimate stress), establish a separate mapping interface between simulation parameters and coding rules, supplement and improve the mapping table, and ensure that the coding covers all structural, environmental and fire parameters without omission or redundancy.

[0068] (3) Encoding generation unit: Based on the parameter-encoding mapping table, a 32-bit three-dimensional code for each parameter is automatically generated (8 bits for basic attribute layer + 16 bits for dynamic feature layer + 8 bits for constraint rule layer). The uniqueness of the code can be ensured by the combination verification of parameter type and parameter name. At the same time, an encoding index library is established to store the association information between the code and the original parameter and parameter attribute, realize the bidirectional mapping between the code and the original parameter, support the fast query, parsing, updating and deletion of the code, and adapt to the scenario of dynamic parameter update.

[0069] Multi-parameter coupling layer The multi-parameter coupling layer is the core of this invention. Its innovation lies in constructing a multi-parameter physical constraint coupling method for structure, environment, and fire, integrating existing mature algorithms and optimizing them for tank fire scenarios. It combines key formulas to achieve accurate calculations of multi-factor interactions, employing a hybrid architecture of "mechanism model + AI agent model" to balance accuracy and real-time performance, adapting to the rapid decision-making needs of emergency rescue. Specifically, it includes a model building unit, a parameter coupling unit, and a model calibration unit. The core content of each unit is as follows: (1) Model building unit: Based on the parameter encoding and constraint rules output by the 3D encoding layer, three basic models are built by integrating existing mature mechanism algorithms. These models are then combined with AI proxy models to accelerate computation, forming a complete multi-parameter coupled model framework that balances versatility and scenario adaptability. Tank Structural Mechanics Model: Integrating mature algorithms from elasticity and fracture mechanics, and combining them with the core formula for thermo-mechanical coupling of the tank wall, this model enables collaborative analysis of tank wall temperature and stress under fire scenarios. It calculates tank wall stress, strain, and structural stability, predicts tank wall rupture risks, and adapts to dynamic changes in structural parameters. The core formula uses the tank wall thermal stress calculation formula: σ=E(T)·ε, where σ is the tank wall stress (Pa), and E(T) is the elastic modulus of the tank material at temperature T (Pa). Its value decreases non-linearly with increasing temperature; the specific decay law is determined through experimental data fitting based on material properties (e.g., the elastic modulus of Q235 steel decreases by about 15% at 200℃ compared to room temperature). Temperature affects interatomic bonding forces by altering the material's lattice vibration energy, causing the elastic modulus to decrease with increasing temperature. In fire scenarios, piecewise functions or polynomial fitting are required (e.g., E(T)=E0·(1-β·T-γ·T)). 2(where E0 is the elastic modulus at room temperature, and β and γ are temperature coefficients) Dynamic calculations are performed to accurately quantify the impact of high temperatures on the tank wall structure. ε is the total strain of the tank wall, including the thermal strain α·ΔT caused by temperature changes and the mechanical strain caused by mechanical loads, i.e., ε = α·ΔT + ε m Where α is the coefficient of thermal expansion (1 / ℃), ΔT is the change in tank wall temperature (℃), and ε m For mechanical strain (dimensionless).

[0070] Fire dynamics model: Based on the core algorithm of FDS fire simulation, it integrates mature technologies related to turbulence simulation and heat radiation transfer, and combines the fluid motion continuity equation. ( Where is the fluid density, and t is time. For fluid velocity vector, is the partial derivative of density with respect to time, representing the rate of change of fluid density over time, and ▽ is the Hamiltonian operator), enabling the deduction of flame spread, heat radiation distribution, and smoke diffusion patterns, accurately capturing the characteristics of fire evolution, and adapting to complex scenarios of storage tank fires.

[0071] Environmental impact model: The analytic hierarchy process (AHP) is used to determine the influence weights of each environmental parameter (wind speed 0.5, humidity 0.2, atmospheric pressure 0.2, other environmental parameters 0.1). A parameter coupling correction algorithm is integrated, combined with the flame spread rate correction formula. ( The corrected flame spread rate is given by v, which is the initial calculated value. For environmental parameter weights, The correction coefficient is i, which is the index of the environmental parameter (the i-th environmental parameter), and n is the total number of environmental parameters. This quantifies the impact of environmental factors such as wind speed and humidity on the fire evolution and tank structure, thereby improving the model's adaptability to different scenarios.

[0072] AI proxy model: Integrating mature machine learning algorithms such as Transformer and Random Forest, it is trained based on the calculation results of three major mechanism models in multiple scenarios (conventional fires and extreme fires), constructing a hybrid computing mode of "accurate calculation of mechanisms in key areas + rapid AI inference in non-key areas". The calculation response time for non-key areas is ≤0.1s, significantly improving the overall computing efficiency. At the same time, through regular iterative training, model drift is avoided, ensuring calculation accuracy, with the coefficient of determination (R²) as the benchmark. 2 ≥0.95) and root mean square error (RMSE≤5%) are the accuracy control standards.

[0073] (2) Parameter coupling unit: An innovative closed-loop method for strong coupling between structure, environment and fire is constructed. Based on three-dimensional coding constraint rules, cross-model parameter linkage and real-time iterative coupling are realized. The core logic is clear and feasible, as follows: Coupling Trigger: Parameter changes are monitored in real time according to the data acquisition cycle (0.5-10Hz). Preset trigger thresholds for key parameters (tank wall temperature change ≥5℃, wind speed change ≥1m / s, fire spread speed change ≥0.05m / s). When any model output parameter reaches the trigger threshold, cross-model coupling calculation is immediately started to ensure the real-time nature of coupling interaction and avoid calculation deviations caused by parameter lag.

[0074] Two-way coupling: Three types of two-way coupling logic are constructed to form a complete closed loop. Structure-fire coupling: The thermal radiation intensity and temperature distribution parameters output from the fire dynamics model are converted into a three-dimensional encoded format and then input into the tank structural mechanics model as thermal load-driven stress, strain, and structural stability recalculation. Simultaneously, parameters such as the tank wall rupture risk level and strain limit output from the structural mechanics model are fed back to the fire dynamics model. If a warning level or higher is reached, fire leakage and diffusion calculations are triggered, correcting the flame spread trajectory, heat release rate, and thermal radiation distribution results. Environment-fire coupling: Kriging interpolation is used to map environmental parameters to the fire dynamics model's calculation grid, ensuring the continuous distribution of environmental parameters within the calculation area. The core interpolation formula is... ( For interpolation point parameter values, For sampling point weights, The values ​​are measured at the sampling points, where i is the index of the sampling point (the i-th known sampling point), and n is the total number of known sampling points. The interpolated environmental parameters are substituted into the fire dynamics model to correct results such as flame height and smoke diffusion path. Simultaneously, the fire heat flow parameters are fed back to the environmental impact model to update local environmental parameters. Structure-environment coupling: Environmental parameters such as atmospheric pressure and humidity are used as loads and applied to the tank structural mechanics model to correct the tank wall stress calculation results (e.g., by superimposing the pressure difference between the inside and outside of the tank wall). Simultaneously, the tank wall thermal deformation parameters output by the structural mechanics model are fed back to the environmental impact model to optimize local airflow calculations and adapt to airflow disturbances caused by tank wall deformation.

[0075] Conflict resolution: When the deviation of the same parameter output by multiple models exceeds 10%, a weighted voting method is used to coordinate the conflict. The weights are assigned based on the calculation accuracy of each model (0.4 for structural mechanics model, 0.4 for fire dynamics model, and 0.2 for environmental impact model), and the weighted average value is calculated as the final parameter value. At the same time, the conflict information is recorded to provide a basis for subsequent model calibration.

[0076] (3) Model calibration unit: Construct a calibration method of "dual data source joint calibration + dynamic iterative optimization", integrate mature calibration algorithms such as least squares method and Bayesian estimation, continuously improve the model calculation accuracy, and adapt to the dynamic changes of fire scenarios, as detailed below: Dual Data Source Calibration: Comprehensive calibration is performed using dual data sources to ensure the model is adaptable to various scenarios. Experimental Data Calibration: Measured data (tank wall temperature, flame spread rate, thermal radiation intensity, tank wall strain, etc.) from large-scale tank fire drills at Zhongyuan Oilfield and Shandong Port are collected to construct a measured dataset; the least squares method (core formula) is employed. , Measured values (The calculated value is m, where m is the total number of measured data points) The model parameters are corrected to ensure that the deviation between the model calculation results and the measured data is ≤3%. Simulation data calibration: Using professional simulation software such as ANSYS and FDS, extreme fire scenarios (full tank combustion, strong winds and rainstorms, multi-tank linkage fires) are simulated to obtain simulation datasets; Bayesian estimation (core formula) is used. , For the posterior probability of the parameter, Let be the likelihood function. The initial distribution of parameters obtained based on conventional fire drill data is used to perform secondary calibration of p(D) as evidence (marginal likelihood) to improve the model's adaptability to extreme working conditions.

[0077] Dynamic calibration: Establish a dynamic calibration process of "real-time monitoring - deviation judgment - iterative update", with a cycle of 5-10 seconds (shortened to within 5 seconds when the fire develops rapidly), based on the sliding window algorithm (window size of 10 sets of data), to compare the model calculation results with the updated data of the data acquisition layer in real time; if the deviation exceeds 5%, parameter iterative correction is initiated, calibration logs are kept to record parameter changes and deviations for model optimization traceability.

[0078] Accuracy verification: Establish a multi-dimensional verification indicator system (deviation rate, R-value, etc.) 2 RMSE), in typical scenarios the deviation rate is required to be ≤3%, R 2 ≥0.95, RMSE≤5%, extreme scenarios require deviation rate≤5%, R 2 ≥0.9, RMSE≤8%; if the standards are not met, restart the dual data source calibration process to ensure that the model calculation deviation meets the emergency decision-making requirements.

[0079] Digital Twin Pre-rehearsal Layer The digital twin pre-simulation layer is used to construct a digital twin image of tank fire extinguishing. Based on the calculation results of the multi-parameter coupling model, it completes the dynamic pre-simulation and quantitative evaluation of fire extinguishing tactics, providing visualization and quantitative support for tactical optimization. Specifically, it includes a twin modeling unit, a pre-simulation control unit, and an effect evaluation unit.

[0080] (1) Twin Modeling Unit: Based on BIM+NeRF 3D modeling technology, a high-precision 3D twin model of the storage tank and its surrounding environment is constructed to restore the structural details of the storage tank (tank wall, tank top, welds, interfaces), the layout and working conditions of fire protection facilities, terrain undulations, surrounding buildings and obstacles, evacuation routes and other scene elements, so as to realize the transparent display of the building and support the synchronous visualization of the internal structure of the storage tank and the external environment; the parameter encoding of the 3D encoding layer and the calculation results of the multi-parameter coupling model are accurately associated with the spatial position of the twin model to realize the visualization mapping of parameters (such as using different color gradients to mark the temperature distribution of the tank wall, the flame spread area and intensity); a real-time synchronization mechanism with the physical scene is established, and the synchronization cycle is consistent with the data acquisition cycle to ensure the dynamic consistency between the twin mirror and the actual scene.

[0081] (2) Pre-simulation control unit: Provides a visual tactical parameter input interface, supports batch input and iterative adjustment of multiple sets of fire extinguishing tactical parameters (fire extinguishing agent type, spray flow rate, spray angle, rescue deployment, etc.); drives the coupled model calculation and dynamically demonstrates the tactical execution process in the twin model, supports multi-angle viewing, timeline backtracking and speed-up demonstration, and adapts to rapid comparative analysis of different tactical schemes.

[0082] (3) Effectiveness evaluation unit: Establish a quantitative evaluation index system covering fire control, structural protection, rescue safety and fire extinguishing efficiency, use the analytic hierarchy process to determine the index weights, automatically calculate the comprehensive score of the tactical plan, generate an evaluation report, clarify the advantages and disadvantages of the plan, and provide a quantitative basis for tactical optimization.

[0083] Tactical decision-making level The tactical decision-making layer is used to optimize, generate, and output firefighting tactics, providing scientific support for on-site emergency rescue. Specifically, it includes a tactical optimization unit, a decision generation unit, and an interface output unit.

[0084] (1) Tactical optimization unit: Based on the comprehensive evaluation results of the digital twin pre-drill layer, combined with the three-dimensional coding constraint rules and the calculation logic of the multi-parameter coupling model, the fire extinguishing tactical parameters are automatically optimized and adjusted; it supports the comparison and screening of multiple schemes, and can adjust the weight of evaluation indicators according to user needs (prioritizing structural safety and prioritizing the improvement of fire extinguishing efficiency) to complete the iterative optimization of tactical schemes and ensure the feasibility of the schemes.

[0085] (2) Decision generation unit: integrates the best tactical plan, pre-rehearsal data, and coupled calculation results to generate a standardized emergency decision report, covering details of tactical parameters, execution process, risk warnings, and disposal suggestions. It supports export and printing and is adapted to the needs of on-site command.

[0086] (3) Interface output unit: Set up standardized data interface and control interface, which can realize data communication and linkage control with the on-site fire control system, UAV rescue system and emergency command platform; send optimized tactical parameters to the on-site fire-fighting facilities and push decision reports to the emergency command platform to improve the coordination and efficiency of emergency response.

[0087] Compared with existing technologies, this system has significant advantages and practical effects, as detailed below: First, the multi-source parameter management is comprehensive and efficient. It adopts a "fixed monitoring + mobile supplement" acquisition mode, covering all parameters of tank structure, environment and fire. Combined with Z-score standardized preprocessing and three-dimensional coding fusion technology, it completely solves the problems of fragmentation and fusion difficulties of multi-source heterogeneous parameters, realizes standardized management and rapid traceability of parameters, and lays a solid foundation for subsequent coupled modeling.

[0088] Second, the coupled modeling is accurate and efficient. The innovative strongly coupled closed-loop method realizes two-way real-time linkage between structure, environment and fire. Combined with the hybrid architecture of "mechanism model + AI agent model", it not only ensures the calculation accuracy of key areas, but also compresses the simulation response time of non-critical areas to within 0.1s, meeting the real-time requirements of emergency decision-making. The dual data source dynamic calibration scheme further improves the model's adaptability to conventional and extreme fire scenarios, ensuring that the calculation deviation meets the preset standards.

[0089] Third, the tactics are scientifically formulated and feasible. Through digital twin mirroring, parameters are visualized and mapped and tactics are dynamically rehearsed. Combined with a multi-dimensional quantitative evaluation system, multiple tactical plans can be quickly compared and optimized, eliminating reliance on human experience. The generated tactical plans are more in line with actual fire scenarios, improving the accuracy of emergency response.

[0090] Fourth, it has strong versatility and scalability, adapting to large storage tanks of different specifications and storage media. It can be linked with the on-site fire control system and emergency command platform to achieve efficient coordination in the distribution of tactical parameters and the push of decision reports, effectively reducing the risk of tank fires spreading and minimizing casualties and property losses. At the same time, the coding rules and modeling methods can be adapted to subsequent new parameters and scenarios, and have good scalability.

[0091] Fifth, the entire process is closed-loop and adaptable to dynamic scenarios. From parameter acquisition, code updates, and model calibration to tactical optimization and iterative updates, a complete closed loop is formed. It can synchronize physical scene changes in real time, continuously optimize decision-making schemes, and ensure the dynamic adaptability and reliability of emergency response.

[0092] The digital twin simulation device for tank fires provided by the present invention will be described below. The digital twin simulation device for tank fires described below can be referred to in correspondence with the digital twin simulation method for tank fires described above.

[0093] Figure 3 This is a schematic diagram of the structure of the digital twin simulation device for storage tank fires provided by the present invention, specifically including: The acquisition module 301 is used to acquire a standard multi-source parameter dataset of the fire scene, including the tank structure, fire environment, and fire intensity characteristics of the target storage tank. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0094] The construction module 302 is used to construct a mapping relationship between parameters and encoding rules based on the standard multi-source parameter dataset and preset multi-layer encoding rules, generate multi-dimensional codes for each parameter and joint codes for associated parameters based on the mapping relationship, and establish an encoding index library based on the multi-dimensional codes and the joint codes. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0095] The construction module 302 is used to extract the correlation relationships between various parameters based on the encoding index library, and to construct a multi-parameter coupled model based on the correlation relationships. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0096] The simulation module 303 is used to perform fire simulation based on the multi-parameter coupled model and obtain the simulation calculation results. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0097] The mapping module 304 is used to construct a three-dimensional twin model of the target storage tank and its surrounding environment, and to map the multidimensional encoding of each parameter and the inference calculation results to the three-dimensional twin model in real time. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0098] The pre-simulation module 305 is used to receive fire extinguishing tactical parameters and perform dynamic pre-simulation in the mapped 3D twin model to obtain fire extinguishing tactical pre-simulation results and evaluation results. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0099] The generation module 306 is used to generate a standardized emergency decision-making report containing detailed tactical parameters, execution procedures, and risk warnings based on the fire-fighting tactical simulation results and assessment results. For detailed explanations, please refer to the relevant descriptions in the above method embodiments; they will not be repeated here.

[0100] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4As shown, the electronic device may include: a processor 410, a communications interface 420, a memory 430, and a communications bus 440, wherein the processor 410, the communications interface 420, and the memory 430 communicate with each other through the communications bus 440. The processor 410 can call logic instructions in the memory 430 to execute a digital twin simulation method for tank fires. This method includes: acquiring a standard multi-source parameter dataset containing the tank structure, fire environment, and fire characteristics of the target tank at the fire scene; constructing a mapping relationship between parameters and coding rules based on the standard multi-source parameter dataset and preset multi-layer coding rules, generating multi-dimensional codes for each parameter and joint codes for associated parameters based on the mapping relationship, and establishing a coding index library based on the multi-dimensional codes and the joint codes; extracting the correlation relationships between parameters based on the coding index library, and constructing a multi-parameter coupling model based on the correlation relationships; performing fire simulation based on the multi-parameter coupling model to obtain simulation calculation results; constructing a three-dimensional twin model of the target tank and its surrounding environment, and mapping the multi-dimensional codes of each parameter and the simulation calculation results to the three-dimensional twin model in real time; receiving fire extinguishing tactical parameters for dynamic pre-simulation in the mapped three-dimensional twin model to obtain fire extinguishing tactical pre-simulation results and evaluation results; and generating a standardized emergency decision report containing tactical parameter details, execution procedures, and risk warnings based on the fire extinguishing tactical pre-simulation results and evaluation results.

[0101] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0102] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the digital twin simulation method for tank fires provided by the above methods. This method includes: acquiring a standard multi-source parameter dataset of the tank structure, fire environment, and fire characteristics of the target tank at the fire scene; constructing a mapping relationship between parameters and coding rules based on the standard multi-source parameter dataset and preset multi-layer coding rules, and generating multi-dimensional codes for each parameter and joint codes for associated parameters based on the mapping relationship; and generating multi-dimensional codes for each parameter and joint codes for associated parameters based on the multi-dimensional codes. A coding index library is established with the joint coding; based on the coding index library, the correlation between each parameter is extracted, and a multi-parameter coupling model is constructed based on the correlation; fire simulation is performed based on the multi-parameter coupling model to obtain simulation calculation results; a three-dimensional twin model of the target storage tank and its surrounding environment is constructed, and the multi-dimensional coding of each parameter and the simulation calculation results are mapped to the three-dimensional twin model in real time; fire extinguishing tactical parameters are received and dynamically simulated in the mapped three-dimensional twin model to obtain fire extinguishing tactical simulation results and evaluation results; a standardized emergency decision report containing tactical parameter details, execution procedures, and risk warnings is generated based on the fire extinguishing tactical simulation results and evaluation results.

[0103] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the digital twin simulation method for tank fires provided by the above methods. The method includes: acquiring a standard multi-source parameter dataset of the tank structure, fire environment, and fire characteristics of the target tank at the fire scene; constructing a mapping relationship between parameters and coding rules based on the standard multi-source parameter dataset and preset multi-layer coding rules, generating multi-dimensional codes for each parameter and joint codes for associated parameters based on the mapping relationship, and establishing a coding index library based on the multi-dimensional codes and the joint codes. Based on the coded index library, the correlation between each parameter is extracted, and a multi-parameter coupled model is constructed based on the correlation. Fire simulation is performed based on the multi-parameter coupled model to obtain simulation calculation results. A three-dimensional twin model of the target storage tank and its surrounding environment is constructed, and the multi-dimensional codes of each parameter and the simulation calculation results are mapped to the three-dimensional twin model in real time. Fire extinguishing tactical parameters are received and dynamically simulated in the mapped three-dimensional twin model to obtain fire extinguishing tactical simulation results and evaluation results. Based on the fire extinguishing tactical simulation results and evaluation results, a standardized emergency decision report containing tactical parameter details, execution procedures, and risk warnings is generated.

[0104] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0105] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0106] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A digital twin simulation method for storage tank fires, characterized in that, include: Obtain a standard multi-source parameter dataset of the fire scene, including the tank structure, fire environment, and fire intensity characteristics of the target storage tank; Based on the standard multi-source parameter dataset and the preset multi-layer coding rules, a mapping relationship between parameters and coding rules is constructed, and multi-dimensional codes of each parameter and joint codes of associated parameters are generated based on the mapping relationship. A coding index library is also established based on the multi-dimensional codes and the joint codes. Based on the encoding index library, the correlation between each parameter is extracted, and a multi-parameter coupling model is constructed based on the correlation. Fire simulations were performed based on the multi-parameter coupled model, and the simulation results were obtained. A three-dimensional twin model of the target storage tank and its surrounding environment is constructed, and the multidimensional encoding of each parameter and the inference calculation results are mapped to the three-dimensional twin model in real time. The received fire extinguishing tactical parameters are dynamically simulated in the mapped 3D twin model to obtain fire extinguishing tactical simulation results and evaluation results; Based on the results of the firefighting tactical drills and assessments, a standardized emergency decision-making report is generated, which includes detailed tactical parameters, execution procedures, and risk warnings.

2. The method according to claim 1, characterized in that, The relationships between the parameters include the correlation threshold level, the mutual influence coefficient level, the constraint priority, and the coupling topology. The step of extracting the correlation relationships between parameters based on the encoded index library and constructing a multi-parameter coupling model based on the correlation relationships includes: Parse the constraint rule layer encoding of each parameter in the encoding index library, and extract the correlation threshold level, mutual influence coefficient level and constraint priority between each parameter; The correlation threshold level, mutual influence coefficient level and constraint priority of each parameter are converted into physical constraint threshold and coupling weight coefficient to obtain the parameter coupling topology. Based on the parameter coupling topology, a structural mechanical model, a fire dynamics model, and an environmental impact model of the storage tank are constructed respectively, and a bidirectional data exchange interface is established between the structural mechanical model, the fire dynamics model, and the environmental impact model of the storage tank. Based on the dynamic feature layer encoding in the encoding index library, the coupling trigger condition is set. When the parameter change rate or threshold range of the real-time monitoring data reaches the coupling trigger condition, cross-model coupling calculation is started to obtain the multi-parameter coupling model.

3. The method according to claim 2, characterized in that, The method further includes: Extreme fire scenario simulation data is obtained based on the multi-parameter coupling model. The extreme fire scenario simulation data is the simulation results of full tank combustion and / or strong wind extreme conditions based on the simulation software in the multi-parameter coupling model. Based on the sliding window algorithm, the extreme fire scenario simulation data and the standard multi-source parameter dataset are compared in real time at a preset period. When the deviation exceeds a preset threshold, the basic parameters of the multi-parameter coupled model are corrected based on the standard multi-source parameter dataset using the least squares method, and the extreme condition adaptation parameters are corrected based on the extreme fire scenario simulation data using Bayesian estimation. The calibrated basic parameters and extreme condition adaptation parameters are obtained and fed back to update the multi-parameter coupled model.

4. The method according to claim 2, characterized in that, The fire simulation based on the multi-parameter coupled model yields the following simulation results: Based on the coupling weight coefficients defined in the constraint rule layer of the coding index library, a two-way coupled deduction of the tank structure mechanical model and the fire dynamics model is performed to obtain the deduction calculation results of the tank wall rupture risk level. The fire environment parameters are mapped to the fire dynamics model calculation grid using the Kriging interpolation method to correct the flame spread parameters, and the fire characteristic parameters are fed back to the environmental impact model to update the local environmental parameters. The two-way coupled deduction of the environmental impact model and the fire dynamics model is performed to obtain the deduction calculation results of the fire spread prediction trajectory. Environmental loads are applied to the structural mechanics model of the storage tank to correct the tank wall stress, and the thermal deformation parameters of the tank wall are fed back to the environmental impact model to optimize the airflow field calculation. The two-way coupled deduction of the structural mechanics model and the environmental impact model of the storage tank is performed to obtain the deduction calculation results of thermal radiation distribution and structural stability index.

5. The method according to claim 4, characterized in that, The construction of a three-dimensional twin model of the target storage tank and its surrounding environment, and the real-time mapping of the multidimensional encoding of each parameter and the inference calculation results to the three-dimensional twin model, includes: The static structural data of the target storage tank is imported using a modeling tool. Based on the fusion of neural radiation field and real-time fire scene image data, a three-dimensional twin model of the target storage tank and its surrounding environment is constructed, and a real-time synchronization mechanism between the three-dimensional twin model and the physical scene is established. The basic attribute layer and dynamic feature layer in the multidimensional encoding are analyzed, the spatial location identifier and type attribute of each parameter are extracted, the spatial location identifier is associated with the spatial coordinates of the three-dimensional twin model, and the mapping position of each parameter in three-dimensional space is determined. The inference and calculation results are mapped to the corresponding spatial coordinates of the three-dimensional twin model in real time.

6. The method according to claim 5, characterized in that, The received firefighting tactical parameters are dynamically simulated in the mapped 3D twin model to obtain firefighting tactical simulation results and evaluation results, including: The system receives multiple sets of fire extinguishing tactical parameters input from the user terminal, including the type of extinguishing agent, spray flow rate, spray angle, and rescue deployment location. Each set of fire extinguishing tactical parameters is input into the multi-parameter coupled model, which drives the model to calculate the fire evolution and structural response after the tactical execution. The fire extinguishing tactical execution process is dynamically demonstrated in the mapped three-dimensional twin model to obtain the corresponding fire extinguishing tactical pre-simulation results. The fire extinguishing tactical execution process includes flame suppression effect, smoke diffusion change and tank wall temperature drop process. Based on a pre-set evaluation index system, the results of each group of fire-fighting tactical drills are quantitatively evaluated. The weight of each evaluation index is determined by the analytic hierarchy process and the comprehensive score is calculated to obtain the evaluation results of each group of fire-fighting tactical parameters, which include multi-dimensional scores. The evaluation indexes include fire control efficiency, structural protection level, rescue safety and fire-fighting timeliness.

7. The method according to claim 1 or 2, characterized in that, The process of constructing a mapping relationship between parameters and encoding rules based on the standard multi-source parameter dataset and preset multi-layer encoding rules, generating multi-dimensional codes for each parameter and joint codes for associated parameters based on the mapping relationship, and establishing an encoding index library based on the multi-dimensional codes and the joint codes includes: Analyze the physical relationships between parameters in the standard multi-source parameter dataset, identify parameter pairs with coupling relationships, and determine the association threshold, mutual influence coefficient, and constraint priority of each parameter pair; Based on preset multi-layer coding rules, hierarchical coding is performed on each parameter in the standard multi-source parameter dataset. The hierarchical coding includes basic attribute layer coding, dynamic feature layer coding, and constraint rule layer coding, and the hierarchical coding results are combined to generate multi-dimensional coding. The parameter pairs with coupling relationships are jointly encoded to establish a correspondence between parameter names and multidimensional encoding fields, and to construct a mapping relationship between parameters and encoding rules. An encoding index library is established based on the multidimensional encoding, the joint encoding, and the mapping relationship. The encoding index library stores the bidirectional mapping relationship between the encoding and the original parameters, parameter attributes, and the correlation between parameters.

8. The method according to claim 1 or 2, characterized in that, The standard multi-source parameter dataset obtained at the fire scene includes the tank structure, fire environment, and fire intensity characteristics of the target storage tank, including: A hybrid acquisition mode combining fixed monitoring and mobile supplementation is adopted to collect in real time the structural parameters of the target storage tank, fire environmental parameters, and fire characteristic parameters of the fire scene. The collected tank structural parameters, fire environment parameters, and fire characteristic parameters are preprocessed, and the parameters of different dimensions are uniformly mapped to a preset range using a standardization method to obtain the standard multi-source parameter dataset.

9. A digital twin simulation device for tank fires, characterized in that, include: The acquisition module is used to acquire a standard multi-source parameter dataset of the fire scene, including the tank structure, fire environment, and fire intensity characteristics of the target storage tank. The construction module is used to construct a mapping relationship between parameters and encoding rules based on the standard multi-source parameter dataset and preset multi-layer encoding rules, generate multi-dimensional encodings of each parameter and joint encodings of associated parameters based on the mapping relationship, and establish an encoding index library based on the multi-dimensional encodings and the joint encodings. A construction module is used to extract the correlation between parameters based on the encoding index library, and to construct a multi-parameter coupling model based on the correlation. The simulation module is used to perform fire simulation based on the multi-parameter coupled model and obtain the simulation calculation results. The mapping module is used to construct a three-dimensional twin model of the target storage tank and its surrounding environment, and to map the multidimensional encoding of each parameter and the inference calculation results to the three-dimensional twin model in real time. The pre-simulation module is used to receive fire extinguishing tactical parameters and perform dynamic pre-simulation in the mapped 3D twin model to obtain fire extinguishing tactical pre-simulation results and evaluation results; The generation module is used to generate a standardized emergency decision report based on the fire extinguishing tactical simulation results and assessment results, which includes details of tactical parameters, execution procedures and risk warnings.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the digital twin simulation method for tank fires as described in any one of claims 1 to 8.

Images