Ruin simulation method for earthquake rescue

Abstract

The invention relates to the technical field of geological earthquake simulation ruins, and discloses a ruin simulation method for earthquake rescue, and the method comprises the steps: firstly building a main structure and a scene simulation engine, then building an environment simulation device, transmitting the data collected in the environment simulation device to an intelligent sensing network, and transmitting the data to the intelligent sensing network; the intelligent sensing network transmits the real-time data of the personnel to the scene simulation engine, and finally transmits all the data to the data management platform. According to the ruin simulation method for earthquake rescue, simulation parameters are adjusted in real time through behaviors of rescue workers, risks and challenges facing by participants in training are close to a real disaster site, psychological adaptability and emergency response ability are improved, an AI system adjusts scene complexity or triggers events according to actions, decisions and speeds of each rescue worker, and the simulation accuracy is improved. The difficulty can be set individually for persons with different skill levels, and the training efficiency and effect are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of geological earthquake simulation ruins, in particular to a ruin simulation method for earthquake rescue. BACKGROUND

[0002] The geological earthquake simulation ruins technology is a technology that reproduces the geological structure and building destruction process under the action of an earthquake through numerical simulation, physical modeling and experimental means. It can analyze soil deformation, fault activity, building collapse and ruin accumulation characteristics caused by an earthquake, and provide scientific basis and visual reference for urban earthquake resistance planning, civil engineering design, disaster emergency drill and earthquake disaster research.

[0003] The existing geological earthquake simulation ruins technology usually uses steel structures or high-strength concrete to build the main frame, simulates different building collapse forms through modular splicing, quickly adjusts the layout of the ruins, adapts to various rescue scene training, and uses flexible materials or hydraulic drive devices for some walls to simulate the collapse and extrusion process of the walls in an earthquake, so that rescue personnel can experience the dynamic changes of the dangerous environment.

[0004] However, the modular structure of the existing geological earthquake simulation ruins technology is flexible, but it is difficult to completely reproduce the randomness and complexity of building collapse in an earthquake, such as irregular steel twisting and concrete breaking patterns, resulting in differences between the training scene and the actual disaster. In view of this, we propose a ruin simulation method for earthquake rescue. SUMMARY

[0005] The present application aims to provide a ruin simulation method for earthquake rescue to solve the problems raised in the background art.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solution: a ruin simulation method for earthquake rescue, comprising: S1.1, main structure building; S1.2, scene simulation engine, development software system; S2, environment simulation equipment; S3, intelligent sensing network, obtaining dynamic data; S4, data management platform, generating training report.

[0007] Preferably, the main structure includes a modular frame and a variable wall system. The modular frame uses high-strength Q345B steel as the main frame, with standard-sized modules of 3m×3m×3m, which are quickly assembled using bolts. Key load-bearing parts, such as columns and beams, have built-in stress sensors with an accuracy of ±0.1MPa. These stress sensors can simulate the collapse modes of different building types, such as residential buildings, shopping malls, and factories. The high-strength steel as the main frame can withstand large loads, ensuring structural stability during the simulation. It can withstand extreme stresses from shaking tables or physical experiments within a safe range, preventing premature model failure and ensuring the accuracy of experimental data. The built-in stress sensors can monitor stress changes in the structure in real time under simulated earthquakes or external forces.

[0008] Preferably, the variable wall system includes a portion of the wall made of flexible composite material and detachable precast concrete slabs. The flexible composite material is made of aramid fiber reinforced board. The flexible composite material achieves 0-90 degree tilt deformation through hydraulic push rods to simulate the wall collapse process. It can accurately simulate the wall collapse process under earthquakes or other disasters. By controlling the tilt angle and speed, different failure modes can be reproduced, such as overall collapse or partial damage. The detachable precast concrete slabs have built-in steel mesh and are equipped with a buffer device to prevent equipment damage, simulating the floor slab falling scenario through an electric hook. This enhances the experiment's ability to reproduce real ruin conditions. The electric hook can control the falling process, increasing safety and repeatability.

[0009] Preferably, the scene simulation engine includes a physics engine module and an AI decision-making system. The physics engine module is a dynamic simulation system developed based on Unity3D, with a built-in building collapse algorithm library. It supports rigid body dynamics simulation of components such as prefabricated slabs and walls, realizing the effects of debris splashing and chain collapse of structures. It integrates fluid simulation algorithms to dynamically render the diffusion paths of smoke and water. The integrated fluid simulation of smoke and water diffusion makes the disaster scene more realistic and helps to assess the difficulty of rescue and safety risks. Dynamic rendering not only provides visualization effects but also supports intuitive verification of experimental data and virtual scenes. The AI ​​decision-making system trains a reinforcement learning-based agent and builds a secondary disaster association model. The agent adjusts the simulation parameters in real time according to the behavior of rescuers. For example, when rescuers pry open a dangerous wall, the AI ​​triggers a partial collapse of the wall to simulate actual danger and train personnel's emergency response. If the rescue timeout is detected, the "aftershock intensification" event is automatically activated to enhance the sense of urgency and improve the realism of the exercise. The secondary disaster association model includes the dynamic relationship between fire spread speed and ventilation conditions and combustible material distribution, which can predict the risk of chain disasters and provide a basis for optimizing emergency plans.

[0010] Preferably, the environmental simulation equipment includes a secondary disaster simulation device and an audio-visual environment system. The secondary disaster simulation device includes a vibration system, a smoke and dust system, and a gas simulation module. The vibration system, through the installation of a 6-DOF electric vibration platform with a platform size of 6m×6m, can support large-scale building or model experiments, simulating earthquake waveforms of magnitude 3-8 on the Richter scale, covering small to strong earthquakes, and realistically reproducing the impact of earthquakes on structures and personnel. The audio-visual environment system includes the installation of adjustable color temperature LED light strips and the deployment of an infrared thermal imager, which can simulate day and night changes, firelight or flashing effects, creating a realistic disaster scene lighting environment. It is equipped with high-power speakers to play environmental sound effects such as earthquake roars and cries for help, improving the immersion of drills or experiments, allowing participants to visually and aurally approach real disaster scenarios, enhancing training effectiveness and psychological adaptability. The infrared thermal imager is linked with a visible light camera to monitor and collect data in low-light / dark environments, ensuring data integrity and blind-spot-free safety monitoring during experiments and drills, and can also analyze rescue behaviors under low-light conditions.

[0011] Preferably, the environmental simulation equipment also includes a human-computer interaction system. The human-computer interaction system includes an AR auxiliary terminal, which uses customized AR glasses to overlay virtual information in real time, such as virtual markers of trapped personnel, structural stress cloud maps, and rescue operation guidelines. Rescuers can see key information in the real environment without the need for additional equipment operation. It also supports gesture recognition and voice command control to achieve non-contact interaction. The human-computer interaction system also includes a VR pre-rehearsal platform, which develops VR training sandbox scenarios containing 10 typical earthquake rescue tasks, such as shaft rescue and tunnel collapse. It supports multi-person collaborative training, allowing personnel to experience the spatial pressure, obstacle complexity, and urgency of actual rescue. It is equipped with a force feedback vest and vibration foot pedals to simulate physical feedback such as aftershocks and object collisions. The force feedback vest is equipped with 16 tactile feedback units, making the training experience closer to real disaster scenarios and improving psychological and physical adaptability.

[0012] Preferably, the intelligent sensing network includes positioning and motion capture as well as environmental monitoring. The positioning and motion capture employs the deployment of UWB positioning base stations and rescue personnel wearing wristbands with integrated IMUs. The IMU wristbands supplement angular velocity and acceleration information to achieve motion and posture capture. The positioning accuracy of the UWB positioning base stations is ±10cm, enabling real-time tracking of the rescue personnel's position in complex environments. The positioning and motion capture also includes the installation of optical motion capture cameras in key rescue areas to capture millimeter-level motion details. The key rescue areas include narrow passages. The optical motion capture cameras have a resolution of 1920×1080 and a frame rate of 120fps, accurately analyzing the rescue personnel's behavior and motion efficiency, and assessing the safety of dangerous operations, such as the risks of prying open dangerous walls or traversing narrow spaces.

[0013] Preferably, the environmental monitoring employs temperature and humidity sensors and dust concentration sensors deployed every 10 square meters. The temperature and humidity sensors have an accuracy of ±0.5℃ and ±2%RH, and the dust concentration sensors have a measurement range of 0-1000 μg / m³. 3 The data from the temperature and humidity sensors and dust concentration sensors are wirelessly transmitted to the central control room via LoRa to monitor the internal environment of the simulated ruins in real time, ensuring personnel safety. It can analyze the impact of environmental changes on rescue efficiency and human health, and supports the collection and analysis of experimental data in multiple scenarios. The environmental monitoring also includes setting up vital sign monitoring gates at the entrances and exits of the simulated ruins to detect personnel's heart rate and body temperature in a non-contact manner, instantly detecting physiological abnormalities of rescue personnel, such as overheating or abnormal heart rate, thereby improving safety. The data can be used for training stress assessment and fatigue management to improve rescue efficiency and safety.

[0014] Preferably, the data management platform includes a real-time database and an intelligent analysis module. The real-time database uses the time-series database InfluxDB to store sensor data. The real-time database supports millisecond-level data acquisition and querying, and can process massive amounts of sensor data, including location, action, environmental parameters, and vital signs. It supports millisecond-level data querying and anomaly warning, improving the safety of training and experiments and reducing unexpected risks. The real-time database establishes a personnel behavior database, recording information such as operation steps, decision-making time, and interactive dialogues, monitoring the status of rescue training or experiments in real time, promptly detecting abnormal events, and providing complete and continuous data records for subsequent analysis and optimization.

[0015] Preferably, the intelligent analysis module utilizes a decision tree algorithm to evaluate the rationality of rescue priorities, analyzes team collaboration efficiency through Bayesian networks, provides a scientific basis for optimizing rescue strategies and personnel division of labor, supports targeted feedback on decision-making and collaboration during training, and then generates a visual training report, including a heat map and a risk assessment radar chart. The heat map represents personnel activity areas, showing personnel activity areas and high-frequency operation locations, making it easy to identify bottlenecks or high-risk areas. The risk assessment radar chart includes data on decision-making, operation, communication, and other dimensions, quantifying the performance of decision-making, operation, communication, and other dimensions, providing an intuitive basis for training evaluation, and intuitively presenting complex data, making it easy for instructors or researchers to analyze training effects.

[0016] Compared with existing technologies, the present invention provides a method for simulating rubble in earthquake rescue, which has the following beneficial effects: 1. This rubble simulation method for earthquake rescue uses an AI decision-making system to adjust simulation parameters in real time based on the behavior of rescuers. For example, it can trigger local collapse when prying open a dangerous wall, or trigger intensified aftershocks when the rescue exceeds the time limit. The risks and challenges faced by participants in the training are close to those of a real disaster site, which improves their psychological adaptability and emergency response capabilities. The AI ​​system adjusts the complexity of the scene or the triggering events based on the actions, decisions and speed of each rescuer. It can also personalize the difficulty for people with different skill levels, thereby improving training efficiency and effectiveness.

[0017] 2. This rubble simulation method for earthquake rescue improves the safety of training and experimentation and reduces unexpected risks through a data management platform. A real-time database is established to create a personnel behavior database, recording information such as operation steps, decision-making time, and interactive dialogues. The system monitors the status of rescue training or experimentation in real time, promptly detects abnormal events, and provides complete and continuous data records for subsequent analysis and optimization.

[0018] 3. This rubble simulation method for earthquake rescue combines VR and AR devices to achieve a virtual-real integrated training experience, enhancing the realism and practicality of the training. The integration and interaction of virtual reality and augmented reality minimizes the risk of personal injury during training, while enhancing immersion and improving the stress adaptability and reaction speed of rescue personnel. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0020] like Figure 1 As shown, the present invention provides a technical solution: a method for simulating rubble for earthquake rescue, comprising: S1.1 Main structure construction; S1.2, Scene simulation engine, for developing software systems; S2, Environmental simulation equipment; S3, intelligent sensing network, to acquire dynamic data; S4, the data management platform, generates training reports.

[0021] The main structure includes a modular frame and a variable wall system. The modular frame uses high-strength Q345B steel as the main frame, with standard-sized modules of 3m×3m×3m that can be quickly assembled using bolts. Key load-bearing parts, such as columns and beams, have built-in stress sensors with an accuracy of ±0.1MPa. These sensors can simulate the collapse patterns of different building types, such as residential buildings, shopping malls, and factories. The high-strength steel as the main frame can withstand large loads, ensuring structural stability during the simulation. It can withstand extreme stresses from shaking tables or physical experiments within a safe range, preventing premature model failure and ensuring the accuracy of experimental data. The built-in stress sensors can monitor stress changes in the structure in real time under simulated earthquakes or external forces.

[0022] The variable wall system includes sections of walls made of flexible composite materials and detachable precast concrete slabs. The flexible composite material is made of aramid fiber reinforced panels. The flexible composite material achieves 0-90 degree tilt deformation through hydraulic push rods to simulate the wall collapse process. It can accurately simulate the collapse process of walls under earthquakes or other disasters. By controlling the tilt angle and speed, different failure modes can be reproduced, such as overall collapse or partial damage. The detachable precast concrete slabs have built-in steel mesh and are equipped with a buffer device to prevent equipment damage, which enhances the ability of the experiment to reproduce the real ruin situation. The electric hook can control the fall process, increasing safety and repeatability.

[0023] The scenario simulation engine includes a physics engine module and an AI decision-making system. The physics engine module is a dynamic simulation system developed based on Unity3D, with a built-in building collapse algorithm library. It supports rigid body dynamics simulation of components such as prefabricated slabs and walls, realizing the effects of debris splashing and chain collapse of structures. It integrates fluid simulation algorithms to dynamically render the diffusion paths of smoke and water, making the disaster scene more realistic and helping to assess the difficulty of rescue and safety risks. Dynamic rendering not only provides visualization effects but also supports intuitive verification of experimental data and virtual scenes. The AI ​​decision-making system trains a reinforcement learning-based agent and builds a secondary disaster association model. The agent adjusts the simulation parameters in real time according to the behavior of rescuers. For example, when rescuers pry open a dangerous wall, the AI ​​triggers a partial collapse of the wall to simulate actual danger and train personnel's emergency response. If the rescue timeout is detected, the "aftershock intensification" event is automatically activated to enhance the sense of urgency and improve the realism of the exercise. The secondary disaster association model includes the dynamic relationship between fire spread speed and ventilation conditions and combustible material distribution, which can predict the risk of chain disasters and provide a basis for optimizing emergency plans.

[0024] The environmental simulation equipment includes a secondary disaster simulation device and an audio-visual environment system. The secondary disaster simulation device includes a vibration system, a smoke and dust system, and a gas simulation module. The vibration system uses a 6-DOF electric vibration platform with a platform size of 6m×6m to support large-scale building or model experiments, simulating earthquake waveforms of magnitude 3-8 on the Richter scale, covering small to strong earthquakes, and realistically reproducing the impact of earthquakes on structures and personnel. The audio-visual environment system includes adjustable color temperature LED light strips and an infrared thermal imager, which can simulate day and night changes, firelight or flashing effects, creating a realistic disaster scene lighting environment. It is equipped with high-power speakers to play environmental sound effects such as earthquake roars and cries for help, improving the immersion of drills or experiments, allowing participants to visually and aurally approach real disaster scenarios, enhancing training effectiveness and psychological adaptability. The infrared thermal imager is linked with a visible light camera to monitor and collect data in low-light / dark environments, ensuring data integrity and blind-spot-free safety monitoring during experiments and drills, and can also analyze rescue behaviors under low-light conditions.

[0025] The environmental simulation equipment also includes a human-computer interaction system, which includes an AR auxiliary terminal. Through customized AR glasses, virtual information is overlaid in real time, such as virtual markers of trapped personnel, structural stress cloud maps, and rescue operation guidelines. Rescuers can see key information in the real environment without the need for additional equipment operation. It also supports gesture recognition and voice command control to achieve contactless interaction. The human-computer interaction system also includes a VR pre-rehearsal platform. By developing VR training sandbox scenarios, it includes 10 typical earthquake rescue tasks, such as shaft rescue and tunnel collapse. It supports multi-person collaborative training. Through multi-person collaborative training, personnel can feel the spatial pressure, obstacle complexity, and urgency of actual rescue. Equipped with force feedback vests and vibration foot pedals, it simulates physical feedback such as aftershocks and object collisions. The force feedback vest is equipped with 16 tactile feedback units, making the training experience closer to real disaster scenarios and improving psychological and physical adaptability.

[0026] The intelligent sensing network includes positioning and motion capture as well as environmental monitoring. Positioning and motion capture utilizes the deployment of UWB positioning base stations and rescue personnel wearing wristbands with integrated IMUs. The IMU wristbands supplement angular velocity and acceleration information to achieve motion and posture capture. The positioning accuracy of the UWB positioning base stations is ±10cm, enabling real-time tracking of the rescue personnel's position in complex environments. Positioning and motion capture also includes setting up optical motion capture cameras in key rescue areas to capture millimeter-level motion details. Key rescue areas include narrow passages. The optical motion capture cameras have a resolution of 1920×1080 and a frame rate of 120fps, accurately analyzing the rescue personnel's behavior and action efficiency, and assessing the safety of dangerous operations, such as the risks of prying open dangerous walls or traversing narrow spaces.

[0027] Environmental monitoring employs temperature and humidity sensors and dust concentration sensors deployed every 10 square meters. The temperature and humidity sensors have an accuracy of ±0.5℃ and ±2%RH, while the dust concentration sensors have a range of 0-1000 μg / m³. 3 Data from temperature and humidity sensors and dust concentration sensors are wirelessly transmitted to the central control room via LoRa to monitor the internal environment of the simulated ruins in real time, ensuring personnel safety. It can analyze the impact of environmental changes on rescue efficiency and human health, and supports data collection and analysis for multi-scenario experiments. Environmental monitoring also includes setting up vital sign monitoring gates at the entrances and exits of the simulated ruins to detect personnel's heart rate and body temperature in a non-contact manner, instantly detecting physiological abnormalities of rescue personnel, such as overheating or abnormal heart rate, thereby improving safety. The data can be used for training stress assessment and fatigue management to improve rescue efficiency and safety.

[0028] The data management platform includes a real-time database and an intelligent analysis module. The real-time database uses the time-series database InfluxDB to store sensor data. It supports millisecond-level data acquisition and querying, and can process massive amounts of sensor data, including location, motion, environmental parameters, and vital signs. It supports millisecond-level data querying and anomaly warning, improving the safety of training and experiments and reducing accidental risks. The real-time database establishes a personnel behavior database, recording information such as operation steps, decision-making time, and interactive dialogues, and monitors the status of rescue training or experiments in real time, promptly detecting abnormal events and providing complete and continuous data records for subsequent analysis and optimization.

[0029] The intelligent analysis module utilizes a decision tree algorithm to assess the rationality of rescue priorities and analyzes team collaboration efficiency through Bayesian networks. This provides a scientific basis for optimizing rescue strategies and personnel assignments, supports targeted feedback on decision-making and collaboration during training, and generates a visual training report. The report includes a heatmap and a risk assessment radar chart. The heatmap shows personnel activity areas and high-frequency operation locations, facilitating the identification of bottlenecks or high-risk areas. The risk assessment radar chart includes data on decision-making, operation, and communication, quantifying performance in these dimensions and providing an intuitive basis for training evaluation. It presents complex data visually, making it easier for instructors or researchers to analyze training effectiveness.

[0030] In one embodiment of the present invention, firstly, a main structure is constructed, consisting of a high-strength steel frame and a variable wall system. Combined with stress sensors, hydraulically tiltable flexible walls, and detachable prefabricated panels, it achieves highly realistic physical simulations of collapses and floor falls in different building types. A scene simulation engine is then built, dynamically presenting building damage, debris splashing, and the spread of smoke and water through Unity3D physics dynamics and an AI decision-making system. It can also trigger local collapses or aftershock simulations based on rescue actions, while simultaneously analyzing secondary disaster risks. The main structure constitutes the hardware environment of the environmental simulation equipment, and the scene simulation engine constitutes the software environment. The environmental simulation equipment utilizes a 6-DOF vibration platform, a smoke and dust gas system, and an acoustic and visual environment, combined with LED lighting and infrared thermal imaging, to provide an immersive experimental environment. To ensure monitoring security, the intelligent sensing network uses UWB positioning, IMU wristbands, and optical motion capture environmental simulation equipment to track trainees' rescue movements at the millimeter level. Simultaneously, temperature, humidity, and dust sensors, along with vital sign monitoring gates, are deployed to collect environmental and personnel health data in real time. The data collected by the intelligent sensing network is transmitted to a data management platform. This platform utilizes the InfluxDB time-series database and intelligent analysis modules, supporting millisecond-level data queries and anomaly warnings. It also evaluates rescue strategies and collaboration efficiency through decision trees and Bayesian networks. The platform generates heatmaps and risk radar charts for visualization, providing precise and quantifiable data support for training effectiveness evaluation, rescue strategy optimization, and scientific research. This achieves a comprehensive closed-loop training and experimentation process, from physical simulation to behavioral analysis.

[0031] The intelligent sensing network transmits the behavioral data of rescuers into the scene simulation engine. The AI ​​decision-making system adjusts the simulation parameters in real time based on the rescuers' behavior. For example, it can trigger a partial collapse when prying open a dangerous wall, or trigger an intensified aftershock when the rescue exceeds the time limit. The risks and challenges faced by participants in the training are close to those of a real disaster site, which improves their psychological adaptability and emergency response capabilities. The AI ​​system adjusts the complexity of the scene or the events triggered based on each rescuer's actions, decisions, and speed. It can also personalize the difficulty for people with different skill levels, thereby improving training efficiency and effectiveness.

[0032] In addition, VR equipment is provided to rescue personnel to conduct pre-training exercises in virtual earthquake ruins. In the VR environment, various complex rescue tasks can be simulated, such as rescue in confined spaces and climbing high-rise ruins, avoiding the safety risks in real-world scenarios. At the same time, peripherals such as force feedback vests and vibration foot pedals provide realistic tactile feedback, enhancing immersion. In the real simulated ruins, AR glasses provide rescue personnel with auxiliary information, such as location markers for trapped personnel, rescue route planning, and equipment operation guides. AR technology can also overlay virtual secondary disaster effects onto the real scene, achieving a virtual-real integrated training experience, improving the realism and practicality of the training. The integration and interaction of virtual reality and augmented reality minimizes the risk of personal injury during training, while enhancing immersion and improving the rescue personnel's stress adaptability and actual operational reaction speed.

[0033] In this invention, the main structure and the scene simulation engine are first built separately, and then the environment simulation device is built. The data collected in the environment simulation device is transmitted to the intelligent sensing network. The intelligent sensing network transmits the real-time data of the personnel to the scene simulation engine. The scene simulation engine then adjusts the simulation parameters in real time according to the personnel behavior. Finally, all the data is transmitted to the data management platform to evaluate the rescue strategy and collaboration efficiency.

[0034] The present invention has been described in detail above. However, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, any modifications or improvements that do not depart from the spirit of the present invention are within the scope of protection of the present invention.

Claims

1. A method for simulating rubble in earthquake rescue, characterized in that: include: S1.1 Main structure construction; S1.2, Scene simulation engine, for developing software systems; S2, Environmental simulation equipment; S3, intelligent sensing network, to acquire dynamic data; S4, the data management platform, generates training reports.

2. The rubble simulation method for earthquake rescue according to claim 1, characterized in that: The main structure includes a modular frame and a variable wall system. The modular frame uses high-strength Q345B steel as the main frame and is designed with standard-sized modules, each 3m×3m×3m in size. These modules are quickly assembled using bolts. Key load-bearing parts, such as columns and beams, have built-in stress sensors with an accuracy of ±0.1MPa.

3. The rubble simulation method for earthquake rescue according to claim 2, characterized in that: The variable wall system includes a portion of the wall made of flexible composite material and a detachable precast concrete slab. The flexible composite material is made of aramid fiber reinforced board. The flexible composite material can achieve 0-90 degree tilt deformation through hydraulic push rods to simulate the wall collapse process. The detachable precast concrete slab has a built-in steel mesh and is equipped with a buffer device to prevent equipment damage.

4. The rubble simulation method for earthquake rescue according to claim 1, characterized in that: The scene simulation engine includes a physics engine module and an AI decision-making system. The physics engine module is a dynamic simulation system developed based on Unity3D, with a built-in building collapse algorithm library. It supports rigid body dynamics simulation of components such as prefabricated slabs and walls, integrates fluid simulation algorithms, and dynamically renders the diffusion paths of smoke and water. The AI ​​decision-making system trains an agent based on reinforcement learning and constructs a secondary disaster association model. The agent adjusts simulation parameters in real time according to the behavior of rescuers. For example, when rescuers pry open a dangerous wall, the AI ​​triggers a partial collapse of the wall. If the rescue timeout is detected, the AI ​​automatically initiates an "aftershock intensification" event. The secondary disaster association model includes the dynamic relationship between fire spread speed and ventilation conditions and the distribution of combustibles.

5. The rubble simulation method for earthquake rescue according to claim 1, characterized in that: The environmental simulation equipment includes a secondary disaster simulation device and an acoustic and light environment system. The secondary disaster simulation device includes a vibration system, a smoke and dust system, and a gas simulation module. The vibration system simulates earthquake waveforms of magnitude 3-8 on the Richter scale by installing a 6-DOF electric vibration platform with a platform size of 6m×6m. The acoustic and light environment system includes adjustable color temperature LED light strips, an infrared thermal imager, and a high-power speaker to play environmental sound effects such as earthquake roars and cries for help. The infrared thermal imager is linked with a visible light camera for monitoring and data acquisition in low-light / dark environments.

6. A method for simulating rubble for earthquake rescue according to claim 5, characterized in that: The environmental simulation equipment also includes a human-computer interaction system, which includes an AR auxiliary terminal that uses customized AR glasses to overlay virtual information in real time, such as virtual markers of trapped personnel, structural stress cloud maps, and rescue operation guidelines. It also supports gesture recognition and voice command control to achieve contactless interaction. The human-computer interaction system also includes a VR pre-show platform that develops VR training sandbox scenarios containing 10 typical earthquake rescue tasks, such as shaft rescue and tunnel collapse. It supports multi-person collaborative training and is equipped with a force feedback vest and vibration foot pedals to simulate physical feedback such as aftershocks and object collisions. The force feedback vest is equipped with 16 tactile feedback units.

7. A method for simulating rubble for earthquake rescue according to claim 1, characterized in that: The intelligent sensing network includes positioning and motion capture as well as environmental monitoring. The positioning and motion capture uses the deployment of UWB positioning base stations and rescue personnel wearing wristbands with integrated IMUs. The positioning accuracy of the UWB positioning base stations is ±10cm. The positioning and motion capture also includes setting up optical motion capture cameras in key rescue areas to capture millimeter-level motion details. The key rescue areas include narrow passages. The resolution of the optical motion capture cameras is 1920×1080, and the frame rate is 120fps.

8. A method for simulating rubble for earthquake rescue according to claim 7, characterized in that: The environmental monitoring employs temperature and humidity sensors and dust concentration sensors deployed every 10 square meters. The temperature and humidity sensors have an accuracy of ±0.5℃ and ±2%RH, while the dust concentration sensors have a measurement range of 0-1000 μg / m³. 3 The data from the temperature and humidity sensor and the dust concentration sensor are transmitted wirelessly to the central control room via LoRa. The environmental monitoring also includes setting up a vital signs monitoring gate at the entrance and exit of the simulated ruins to detect personnel's heart rate and body temperature in a non-contact manner.

9. A method for simulating rubble for earthquake rescue according to claim 1, characterized in that: The data management platform includes a real-time database and an intelligent analysis module. The real-time database uses the time-series database InfluxDB to store sensor data and supports millisecond-level data queries and outlier alerts. The real-time database also establishes a personnel behavior database to record information such as operation steps, decision-making time, and interactive dialogues.

10. A method for simulating rubble for earthquake rescue according to claim 9, characterized in that: The intelligent analysis module uses a decision tree algorithm to evaluate the rationality of rescue priorities, analyzes team collaboration efficiency through Bayesian networks, and then generates a visual training report, including a heat map and a risk assessment radar chart. The heat map represents the area of ​​personnel activity, and the risk assessment radar chart includes data on decision-making, operation, communication, and other dimensions.

Images