A mine accident simulation training method, device and medium based on VR technology

Abstract

The application discloses a mine accident simulation training method and device based on VR technology and a medium, relates to the technical field of mine safety intelligent training, and comprises the following steps: collecting typical mine accident types, constructing accident causality, mapping accident elements in the accident causality with risk nodes in a mine semantic twin scene, and forming an accident causality structure; comparing accident perception delay data with a perception delay threshold, outputting a delay level judgment result, adjusting an accident evolution state according to the delay level judgment result, and outputting an adaptive accident evolution state; comprehensively analyzing the adaptive accident evolution state and a student operation track, reconstructing the accident causality structure, and obtaining a personalized accident retraining scene. The application realizes a VR safety training system by constructing a mine digital twin scene that integrates spatial semantics and accident causality logic, and improves the authenticity, teaching pertinence and training effectiveness of mine accident simulation.

Description

Technical Field

[0001] This invention relates to the field of intelligent training technology for mine safety, and in particular to a method, equipment and medium for mine accident simulation training based on VR technology. Background Technology

[0002] In the field of mine safety production, Virtual Reality (VR) technology has been gradually introduced into safety training and emergency drill systems in recent years. Existing technologies typically use 3D modeling software to construct static roadway geometric models, and combine these with pre-set animations or script-driven methods to simulate typical accident scenarios such as gas explosions, roof falls, and water inrushes, providing trainees with an immersive experience and operational training. This approach enhances the interactivity and immediacy of traditional paper-based or video-based teaching to some extent, showing particular advantages in spatial cognition and familiarity with disaster avoidance routes. Some methods also incorporate head tracking and controller interaction devices, enabling trainees to complete simple operational tasks in a virtual environment. With the development of the digital twin concept, some research attempts to initially map the physical space and information space of mines, using sensor data to drive updates to some dynamic elements, thereby enhancing the realism of the simulation.

[0003] Existing systems typically set accident symptoms, evolution nodes, and response measures as a fixed sequence, lacking semantic association modeling between accident elements and scenario risk nodes. This results in a lack of flexibility in accident evolution based on physical or logical causality. Furthermore, in terms of training and evaluation, existing technologies focus primarily on the completion rate and time recorded for operational steps, while lacking continuous quantitative monitoring of the "perception delay" between trainees' perception of accident symptoms and their first effective response. Nor has a dynamic difficulty adjustment mechanism based on delay data been established. This makes it difficult for the training process to truly reflect the cognitive and reaction characteristics of personnel under stress, and also limits the system's ability to provide personalized retraining programs for individual differences. Summary of the Invention

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

[0005] Therefore, this invention provides a mine accident simulation training method based on VR technology to solve the problems of disconnect between the accident evolution process and the trainee's behavior and insufficient training personalization.

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

[0007] Firstly, this invention provides a method for simulating mine accidents based on VR technology. The method includes: collecting data on mine roadway structure, equipment and facility locations, and evacuation routes; performing unified coordinate transformation and semantic annotation to construct a semantic twin scene of the mine; collecting typical mine accident types, constructing accident causal relationships, and mapping accident elements in the causal relationships to risk nodes in the semantic twin scene of the mine to form an accident causal relationship structure; loading scene resources according to the accident causal relationship structure and initiating VR real-time rendering and posture synchronization to capture the behavior of trainees, driving the dynamic evolution of the accident to form a real-time accident evolution scene; continuously recording the trainees' gaze movement and operational behavior in the real-time accident evolution scene, recording the time of accident signs and the time of the trainees' first effective response to obtain accident perception delay data; comparing the accident perception delay data with a perception delay threshold, outputting a delay level judgment result, and adjusting the accident evolution state according to the delay level judgment result to output an adaptive accident evolution state; and comprehensively analyzing the adaptive accident evolution state and the trainees' operational trajectory to reconstruct the accident causal relationship structure and obtain a personalized accident retraining scene.

[0008] As a preferred embodiment of the VR-based mine accident simulation training method described in this invention, the steps for collecting mine roadway structure, equipment and facility locations, and evacuation route data, performing unified coordinate transformation and semantic annotation, and constructing a semantic twin mine scene are as follows.

[0009] Collect mine roadway data, equipment and facility locations, and disaster avoidance route data; extract the roadway orientation, dimensions, slope, and spatial elevation relationships; and convert them into the mine's three-dimensional coordinate system to form spatial structure data.

[0010] Based on spatial structure data, establish the topological relationship between roadways to form spatial topological information, spatially locate equipment and facilities, output equipment location information, and perform risk semantic and interactive semantic annotation on areas prone to gas anomalies and areas with roof fractures, outputting risk semantic information and interactive semantic information.

[0011] By integrating spatial topology information, equipment positioning information, risk semantic information, and interaction semantic information into a unified semantic data structure, a semantic twin scenario of the mine is output.

[0012] As a preferred embodiment of the VR-based mine accident simulation training method described in this invention, the steps of collecting typical mine accident types, constructing accident causal relationships, and mapping the accident elements in the accident causal relationships to risk nodes in the mine semantic twin scenario to form an accident causal relationship structure are as follows.

[0013] Collect typical mine accident types and extract accident causative factors, environmental conditions, and human behavior factors to form an accident factor set;

[0014] Based on the set of accident elements, rules for accident occurrence conditions are established, and the state associations between accident elements are transformed into accident causal relationships through these rules.

[0015] By associating accident-causing factors, environmental conditions, and personnel behavior factors with risk nodes in the semantic twin scenario of the mine, a causal-risk mapping relationship is constructed and integrated to form an accident causal relationship structure.

[0016] As a preferred embodiment of the VR-based mine accident simulation training method of the present invention, the steps of loading scene resources according to the accident causal relationship structure and starting VR real-time rendering and posture synchronization, capturing the behavior and actions of trainees, driving the dynamic evolution of the accident, and forming a real-time accident evolution scene are as follows.

[0017] Based on the cause-and-effect relationship structure of the accident, scene resources are invoked, and the VR rendering engine is started to render the scene resources in real time, while simultaneously collecting head-mounted display posture data and spatial position information, and outputting the real-time rendered scene.

[0018] Collect the movement paths and actions of trainees in a real-time rendered scene, and output the trainee behavior data;

[0019] The system matches the behavioral data of trainees with the rules governing accident occurrence, outputs updated accident status, adjusts the accident performance in the real-time rendering scene based on the updated accident status, and outputs the real-time accident evolution scene.

[0020] As a preferred embodiment of the VR-based mine accident simulation training method of the present invention, the specific steps for obtaining the accident perception delay data are as follows:

[0021] In real-time accident evolution scenarios, behavioral record data of trainees are collected, and the time point when the accident signs first appear is detected during the accident evolution process to generate accident sign occurrence time data;

[0022] Identify the time point when trainees first perform an effective response action based on behavioral record data, and generate initial response time data;

[0023] The time difference between the data on the time when accident signs appear and the data on the time of the first effective response is calculated to form accident perception delay data.

[0024] As a preferred embodiment of the VR-based mine accident simulation training method of the present invention, the steps of comparing the accident perception delay data with the perception delay threshold, outputting the delay level determination result, adjusting the accident evolution state according to the delay level determination result, and outputting an adaptive accident evolution state are as follows.

[0025] The accident perception delay data is compared with the perception delay threshold to generate a delay level determination result. When the delay level determination result is the first level, the accident spread speed is automatically reduced and key node prompts are added, and the first accident evolution state is output.

[0026] When the delay level is determined to be Level 2, the smoke diffusion intensity is increased, the environmental visibility is reduced, and secondary risks are added, and the second accident evolution state is output.

[0027] The real-time accident evolution scenario is updated based on the first and second accident evolution states to form an adaptive accident evolution state.

[0028] As a preferred embodiment of the mine accident simulation training method based on VR technology described in this invention, the perception delay threshold is generated by statistically analyzing the time difference between the occurrence of accident signs and the first effective response time in different accident scenarios for multiple batches of trainees, calculating the average perception time and standard deviation under each accident type, and dividing it into delay level intervals.

[0029] As a preferred embodiment of the VR-based mine accident simulation training method of the present invention, the specific steps for comprehensively analyzing the adaptive accident evolution state and the trainee's operation trajectory to reconstruct the causal relationship structure of the accident and obtain personalized accident retraining scenarios are as follows.

[0030] Collect operational trajectory data and behavioral record data of trainees under adaptive accident evolution state to form training behavior analysis data;

[0031] Based on training behavior analysis data, identify operational delays and misoperations by trainees during accident handling, and generate analysis data on training weaknesses.

[0032] Based on the analysis data of training weaknesses, the accident triggering conditions and environmental state parameters in the accident causal relationship structure are adjusted to generate an updated accident causal relationship structure;

[0033] Based on the updated accident causal relationship structure, the corresponding scenario resources are loaded to form a personalized accident retraining scenario.

[0034] In a second aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the mine accident simulation training method based on VR technology as described in the first aspect of the present invention.

[0035] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the mine accident simulation training method based on VR technology as described in the first aspect of the present invention.

[0036] The beneficial effects of this invention are as follows: by constructing a mine digital twin scenario that integrates spatial semantics and accident causal logic, and combining it with an adaptive accident evolution and personalized retraining mechanism based on perception delay, a high-fidelity, dynamically responsive, and personalized VR safety training system is realized, which improves the realism of mine accident simulation, the relevance of teaching, and the effectiveness of training. Attached Figure Description

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

[0038] Figure 1 This is a flowchart of a mine accident simulation training method based on VR technology.

[0039] Figure 2 A flowchart for constructing a semantic twin scenario for a mine.

[0040] Figure 3 This is a flowchart mapping the causal relationship structure of an accident to risk nodes.

[0041] Figure 4 A flowchart for generating personalized accident retraining scenarios.

[0042] Figure 5 A diagram illustrating the comparison of accident perception delays for trainees.

[0043] Figure 6 This diagram illustrates the comparison of accident response times under different training modes. Detailed Implementation

[0044] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0045] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0046] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0047] Reference Figures 1-6 As one embodiment of the present invention, this embodiment provides a mine accident simulation training method based on VR technology, including the following steps:

[0048] S1. Collect data on mine roadway structure, equipment and facility locations, and disaster avoidance routes, and perform unified coordinate transformation and semantic annotation to construct a semantic twin scenario of the mine.

[0049] S1.1. Collect mine roadway data, equipment and facility locations, and disaster avoidance route data; extract the roadway orientation, dimensions, slope, and spatial elevation relationships; and convert them to the mine's three-dimensional coordinate system to form spatial structure data.

[0050] Specifically, the data collected includes mine roadway data, equipment and facility locations, and evacuation route data. Mine roadway data includes the three-dimensional coordinates of the roadway centerline, roadway length, cross-sectional dimensions, slope, and elevation information. It may also include the geological attributes of the roadway section, the stability level of the surrounding rock, and risk zone markers for gas and water hazards. Equipment and facility locations include the equipment's three-dimensional coordinates, the roadway number it belongs to, its installation direction, and coverage area. It also includes equipment type identification, functional attributes, operating status, monitoring parameters, and fault alarm information. Evacuation route data includes the sequence of evacuation route nodes, the three-dimensional coordinates of the route, the route length, and the direction of travel, as well as associated refuge chambers, compressed air self-rescue points, emergency exits, and other facilities. The data includes the location of the intersection point, as well as attributes such as path capacity, number of people allowed to pass, and risk-blocking sections. Geometric elements are extracted from the mine roadway data to obtain the roadway centerline orientation, roadway cross-sectional dimensions, roadway slope changes, and spatial elevation relationships of roadway nodes. Location elements are extracted from the equipment and facility locations and evacuation route data to obtain the relative positional relationships of equipment and facility locations in the mine roadway data and the continuous path relationships of evacuation routes in the mine roadway data. Subsequently, a unified coordinate transformation method is used to transform the mine roadway data, equipment and facility location data, and evacuation route data from the source coordinates to the mine's three-dimensional coordinate system, and consistency checks and deviation corrections are performed to form spatial structure data.

[0051] S1.2. Establish the topological relationship between roadways based on spatial structure data to form spatial topological information, spatially locate equipment and facilities, output equipment location information, and perform risk semantic and interactive semantic annotation on gas-prone areas and roof fracture areas, outputting risk semantic information and interactive semantic information.

[0052] Specifically, based on the location of lane nodes, lane connectivity, and lane centerline orientation information in the spatial structure data, topological association processing is performed on the connection paths between adjacent lane intersection nodes. The start and end points of each lane centerline, as well as the locations where they intersect with other lanes, are extracted as lane nodes, and each node is assigned a unique node number. According to the spatial connectivity of lane centerline segments, the centerline segments between adjacent nodes are defined as lane edges, and the corresponding attribute information such as length, slope, and cross-sectional dimensions are recorded. Through node coordinate matching, the intersection relationships of each lane edge at the nodes are identified, establishing the connectivity topology between lanes and generating spatial topology information. Subsequently, the coordinates of equipment and facilities locations in the spatial structure data are matched with the corresponding lane node locations to obtain the spatial location relationships of equipment and facilities in the lane network, forming equipment positioning information.

[0053] Further collection of mine geological survey records, gas monitoring records, and roof stability test records; identification processing of roadway sections corresponding to gas anomaly-prone areas and roof fracture areas in spatial topology information to form risk semantic information (risk semantic information refers to a data set formed by structurally identifying the risk attributes of roadway sections or nodes with potential safety risks in the mine spatial topology, mainly used to describe risk type, risk level, and their spatial location relationship); and based on the association between equipment positioning information and spatial topology information, interactive semantic annotation of node positions involving equipment operation and personnel interaction, outputting interactive semantic information.

[0054] S1.3. Integrate spatial topology information, equipment positioning information, risk semantic information and interaction semantic information into a unified semantic data structure to output a semantic twin scenario of the mine.

[0055] Specifically, the roadway node numbers, roadway connection relationships, and roadway segment attributes in the spatial topology information are associated and integrated with the equipment location coordinates and their respective roadway numbers in the equipment positioning information, establishing a unified identification relationship between roadway nodes and equipment locations. Subsequently, the roadway segment numbers corresponding to the gas anomaly-prone areas and roof fracture areas identified in the risk semantic information are linked to the equipment operation node and personnel interaction node numbers identified in the interaction semantic information, respectively, into the same roadway node identification system, forming a unified semantic data structure that includes roadway spatial connectivity relationships, equipment spatial location relationships, risk area semantic identification relationships, and interaction node semantic identification relationships. The unified semantic data structure is then associated and mapped with the corresponding 3D scene resources to output a mine semantic twin scene.

[0056] S2. Collect typical mine accident types, construct accident causal relationships, and map the accident elements in the accident causal relationships to the risk nodes in the mine semantic twin scenario to form an accident causal relationship structure.

[0057] S2.1. Collect typical mine accident types and extract accident causative factors, environmental conditions and human behavior factors to form an accident factor set.

[0058] Specifically, after collecting typical mine accident types, the formation process of each typical mine accident type is analyzed based on mine accident records, accident investigation reports, and safe operating procedures. Direct triggering factors (such as high-temperature frictional ignition sources generated during the operation of mining equipment) and indirect influencing factors (insufficient ventilation capacity in roadways leading to gradual gas accumulation) are identified. Accident triggering factors related to the operating status of equipment and facilities, environmental state factors related to changes in the roadway environment and the development process of disasters, and personnel behavior factors related to personnel operation processes are extracted. Accident triggering factors, environmental state factors, and personnel behavior factors are then categorized and organized according to the correspondence between typical mine accident types to form a set of accident elements.

[0059] It should be noted that accident triggering factors refer to factors that are directly related to the operating status of equipment, energy release conditions, or the state of hazardous sources, and may directly trigger the occurrence of accidents. They are mainly used to describe potential accident triggering sources in the system, such as abnormal operation of electrical equipment, equipment overload or failure, and shutdown of ventilation equipment.

[0060] Environmental status elements refer to status information related to the spatial environmental conditions of the roadway and the development process of disasters. They are used to describe the changes in environmental parameters during the occurrence or development of an accident, such as changes in gas concentration in the roadway, smoke diffusion, temperature changes, stability of the surrounding rock, and deformation of the roof.

[0061] Personnel behavior elements refer to the operational behaviors and action states of trainees or operators during the formation or handling of accidents. They are used to describe the impact of personnel behavior on the development or control of accidents, such as equipment start-up and shutdown operations, emergency response operations, inspection behaviors, and personnel movement paths.

[0062] S2.2. Establish accident occurrence condition rules based on the accident element set, and transform the state association between accident elements into accident causal relationship through the accident occurrence condition rules.

[0063] Specifically, based on the categorized accident causative factors, environmental state factors, and personnel behavior factors in the accident factor set, the conditional relationships of the accident formation process corresponding to typical mine accident types are organized. The synergistic change relationship between accident causative factors and environmental state factors, the triggering relationship between personnel behavior factors and accident causative factors, and the mutual influence relationship between environmental state factors are combined and associated. Accident occurrence condition rules are established according to the order of state changes before and after the accident (used to describe under what combination of factors may trigger the occurrence of a specific accident type). Subsequently, the accident occurrence condition rules are used to logically associate the state change paths between accident causative factors, environmental state factors, and personnel behavior factors. The accident formation path corresponding to the simultaneous satisfaction of multiple factor conditions is defined as a causal relationship path, thereby transforming the state association between accident factors into accident causal relationships.

[0064] S2.3. Associate the accident-causing factors, environmental state factors, and personnel behavior factors with the risk nodes in the semantic twin scenario of the mine, construct a causal-risk mapping relationship, and integrate them to form an accident causal relationship structure.

[0065] Specifically, based on the established causal relationship of the accident, the accident inducing factors are locationally associated with the equipment and facility risk nodes in the mine semantic twin scenario according to the correspondence of equipment and facility operating status. The environmental state factors are spatially associated with the regional risk nodes in the mine semantic twin scenario according to the correspondence of gas anomaly prone areas and roof fracture areas. The personnel behavior factors are behaviorally associated with the interactive risk nodes in the mine semantic twin scenario according to the correspondence of personnel operation positions and equipment interaction positions. Subsequently, the correspondence between the accident inducing factors, environmental state factors, and personnel behavior factors and their respective associated risk nodes is uniformly numbered and linked. The causal relationship path in the accident causal relationship is integrated and expressed with the corresponding risk node number, forming an accident causal relationship structure that includes the correlation of accident elements and the spatial positional relationship of risk nodes.

[0066] S3. Load scene resources according to the cause-and-effect relationship structure of the accident and start VR real-time rendering and posture synchronization to capture the behavior and actions of the trainees, drive the dynamic evolution of the accident, and form a real-time accident evolution scene.

[0067] S3.1. Based on the cause-and-effect relationship structure of the accident, call the scene resources and start the VR rendering engine to render the scene resources in real time, synchronously collect head-mounted display posture data and spatial position information, and output the real-time rendered scene.

[0068] Specifically, based on the risk node number recorded in the accident causal relationship structure and the accident element association path in the accident causal relationship structure, the lane scene resources, equipment and facility scene resources, and accident performance effect scene resources corresponding to the risk node number are called and loaded into the virtual reality display environment; then the virtual reality rendering engine is started to render the loaded lane scene resources, equipment and facility scene resources, and accident performance effect scene resources in real time. At the same time, head-mounted display device is used to collect head-mounted display posture data and spatial position information and update the view control parameters of the virtual reality rendering engine, outputting a real-time rendered scene that matches the head-mounted display posture data and spatial position information in real time.

[0069] It should be noted that scene resources refer to the collection of various three-dimensional object data and effect data used to construct a virtual mine environment and present it during VR training. These include three-dimensional objects of tunnel structure, three-dimensional objects of equipment and facilities, geological structure objects, risk area identification objects, disaster avoidance facility objects, and accident performance effect resources (such as dynamic performance resources of smoke, fire source, water flow, slag falling, roof cracks, etc.). They also include material information, texture information, lighting parameters, animation control parameters, and interactive attribute data associated with the three-dimensional objects.

[0070] A VR rendering engine is a program component used to perform real-time graphics calculations and image generation processing on virtual scene resources. Its main functions include 3D object display processing, lighting and shadow calculation, particle effect generation, dynamic effect performance, viewpoint synchronization processing, and interactive event response.

[0071] S3.2. Collect the movement paths and actions of trainees in the real-time rendering scene, and output the trainee behavior data.

[0072] Specifically, a head-mounted display device is used to continuously record changes in the spatial position of trainees in the tunnel scenario, and the trainees' movement path information is generated based on the sequence of spatial position changes. At the same time, the interactive control device records the input of equipment operation commands, operation trigger time, and operation target, thereby obtaining the trainees' behavioral information during the interaction with the equipment. Subsequently, the trainees' movement path information and behavioral information are uniformly time-stamped and linked in chronological order to form trainees' behavioral data.

[0073] S3.3. Match the behavioral data of trainees with the accident occurrence condition rules, output the updated accident status, adjust the accident performance in the real-time rendering scene according to the updated accident status, and output the real-time accident evolution scene.

[0074] Specifically, the movement path information and action information of trainees recorded in the trainee behavior data are compared item by item with the accident occurrence condition rules in the accident causal relationship structure according to the time sequence. It is determined whether the changes in the operating status of equipment and facilities corresponding to the trainee's action information and the changes in the spatial position corresponding to the trainee's movement path information meet the trigger conditions in the accident occurrence condition rules. When the trigger conditions are met, the accident status is updated according to the accident status change path recorded in the accident causal relationship structure to obtain the updated accident status. Then, based on the updated accident status, the corresponding accident performance scene resources in the real-time rendering scene are called to adjust the state of the accident performance in the real-time rendering scene and output the real-time accident evolution scene.

[0075] S4. Continuously record the trainees' line of sight movement and operational behavior in real-time accident evolution scenarios, and record the time when accident signs appear and the time when the trainees take their first effective action to obtain accident perception delay data.

[0076] S4.1. Collect behavioral record data of trainees in real-time accident evolution scenarios, and detect the first occurrence time of accident signs during the accident evolution process to generate accident sign occurrence time data.

[0077] Specifically, a head-mounted display device continuously records changes in the trainees' line of sight in the tunnel scene, as well as changes in their spatial position within the tunnel scene, to generate trainees' movement path information. An interactive control device records the input of equipment operation commands, operation trigger times, and the objects of operation to obtain equipment interaction operation data. The line of sight change information, trainees' movement path information, and equipment interaction operation data are then linked and organized according to a unified time stamp to form behavioral record data. During the accident evolution process, based on the state change indicators corresponding to accident symptoms in the accident performance scene resources, the time when the accident performance effect first reaches the accident symptom performance state is recorded (for example, when the tunnel roof shows crack expansion, localized spalling, or significant deformation of the support components, the first appearance of the roof failure symptom performance state can be identified), generating accident symptom occurrence time data.

[0078] S4.2. Identify the time point when the trainee first performs an effective handling operation based on the behavior record data, and generate the first handling time data.

[0079] Specifically, based on the device interaction operation data in the behavior record data, the target object, operation type, and operation trigger time of the trainees' operations are extracted. The target object is matched with the relevant risk node number recorded in the accident causal relationship structure, and the operation type is matched with the relevant action requirements recorded in the accident causal relationship structure. The criteria for selecting candidate operations are that the risk node number corresponding to the target object belongs to the set of target nodes pointed to by the accident handling requirements, the operation type belongs to the set of handling action types of the target node, and the accident evolution state is in the stage where the handling action can be executed when the operation occurs. When the above conditions of object matching, action matching, and stage matching are met simultaneously, the corresponding operation is marked as a candidate operation. Then, the candidate operations are sorted according to the time order of the behavior record data, and the operation trigger time corresponding to the earliest appearing candidate operation is selected as the time point when the trainee first executes a valid handling operation, generating the first handling time data.

[0080] S4.3. Calculate the time difference between the data on the time when accident signs appear and the data on the time of the first effective response to form accident perception delay data.

[0081] Specifically, a unified time base calibration is performed on the data of the time when accident signs appear and the data of the first effective response. The time stamps in the data of the time when accident signs appear and the data of the first effective response are converted to the same time measurement unit, and the data of the time when accident signs appear and the data of the first effective response are matched in chronological order. The time difference between the time value in the data of the first effective response and the time value in the data of the time when accident signs appear is calculated to form accident perception delay data. The accident perception delay data includes the data of the time when accident signs appear, the data of the time when the trainees make their first effective response, and the time difference between the two.

[0082] S5. Compare the accident perception delay data with the perception delay threshold, output the delay level determination result, adjust the accident evolution state according to the delay level determination result, and output the adaptive accident evolution state.

[0083] S5.1. Compare the accident perception delay data with the perception delay threshold to generate a delay level determination result. When the delay level determination result is the first level, automatically reduce the accident spread speed and add key node prompts, and output the first accident evolution state.

[0084] Specifically, the time difference values ​​in the accident perception delay data are compared item by item with the perception delay threshold. A delay level determination result is generated based on the position of the accident perception delay data within the perception delay threshold range. When the time difference value in the accident perception delay data is greater than the perception delay threshold (Level 1), the accident spread rate parameter in the accident environment state parameters is adjusted downward according to the accident state change path recorded in the accident causal relationship structure, and the prompt information display parameters corresponding to the key risk nodes in the accident causal relationship structure are adjusted upward. This increases the display frequency and intensity of the prompt information at the key risk node locations in the real-time accident evolution scenario. Then, the accident state is updated based on the adjusted accident environment state parameters and prompt information display parameters, and the first accident evolution state is output.

[0085] It should be noted that in a gas accident scenario, key risk nodes include main roadway intersections along the accident propagation path, corresponding area power cut-off switch nodes, and local ventilation fan control nodes. In a roof accident scenario, roadway nodes located in weak support sections and operation nodes corresponding to support reinforcement equipment can also be considered key risk nodes. When trainees react slowly, increasing the frequency or intensity of the prompts can make it easier for them to identify and take correct action.

[0086] Furthermore, the perception delay threshold is generated by statistically analyzing the time difference between the occurrence of accident signs and the first effective response time in different accident scenarios for multiple batches of trainees, calculating the average perception time and standard deviation for each accident type, and dividing them into delay level intervals.

[0087] It should be noted that the expression for calculating the average perception time under each accident type is as follows:

[0088] ;

[0089] in, This represents the average perception time for each accident type. This indicates the number of training samples corresponding to the accident type. No. Accident perception delay data from the second training session. This indicates the training sample number.

[0090] The expression for calculating the standard deviation of perceived time is:

[0091] ;

[0092] in, This represents the standard deviation of accident perception delay data under different accident types.

[0093] S5.2. When the delay level determination result is the second level, increase the smoke diffusion intensity, reduce environmental visibility and add secondary risks, and output the second accident evolution state.

[0094] Specifically, when the time difference value in the accident perception delay data is less than the perception delay threshold (Level 2), the smoke diffusion speed parameter in the accident environmental state parameters is adjusted upwards and the environmental visibility parameter is adjusted downwards based on the accident state change path recorded in the accident causal relationship structure. Simultaneously, based on the secondary risk triggering conditions recorded in the accident causal relationship structure (referring to the state judgment conditions corresponding to new dangerous events further triggered by changes in environmental state, equipment state, or personnel behavior during the accident development process after the initial accident event), corresponding secondary risk performance parameters are added at risk nodes that meet the secondary risk triggering conditions, so that the secondary risk performance effect is presented in the real-time accident evolution scenario. Subsequently, the accident state is updated based on the adjusted smoke diffusion speed parameter, environmental visibility parameter, and secondary risk performance effect parameter, and the second accident evolution state is output.

[0095] S5.3. Update the real-time accident evolution scenario based on the first accident evolution state and the second accident evolution state to form an adaptive accident evolution state.

[0096] Specifically, based on the accident environment state parameters, risk node state parameters, and accident performance effect parameters recorded in the first and second accident evolution states, the corresponding parameters are written into the scene state control table in the real-time accident evolution scenario. The smoke diffusion range, environmental visibility, and accident impact area display effect in the real-time accident evolution scenario are updated according to the accident environment state parameters. The display status of the prompt information at key risk node locations is updated according to the risk node state parameters. The performance effect of the fire source, smoke, and secondary risks in the real-time accident evolution scenario is updated according to the accident performance effect parameters. This allows the real-time accident evolution scenario to present the corresponding accident change process according to the first or second accident evolution state, forming an adaptive accident evolution state. The adaptive accident evolution state continuously acquires information on the trainees' operational behavior, accident perception delay, and environmental state changes according to the update cycle. Based on the update results, the accident diffusion speed, environmental visibility, risk warning intensity, and secondary risk presentation are adjusted in stages, so that the accident development process changes with the trainees' reaction speed and handling effect, thus forming an accident evolution performance that can dynamically adjust over time and continuously respond to changes in the training process.

[0097] It should be noted that, as Figure 5As shown, the diagram illustrates the changes in accident perception delay between the control group and the adaptive group during accident training in a semantic twin scenario of a mine. The upper figure shows the distribution curve of accident perception delay for all trainees, with the blue curve representing the control group and the orange curve representing the adaptive group; the red dashed rectangle marks the selected local comparison area; the lower figure is a magnified view of the corresponding area, showing that the overall accident perception delay of the adaptive group is lower than that of the control group, indicating that the adaptive accident evolution mechanism driven by accident perception delay data can shorten the trainees' reaction time to accident signs, thereby improving the dynamic response capability and training relevance of the training system. The control group adopted an accident evolution training scheme driven by a fixed accident script. Before the training started, parameters such as the accident occurrence time, accident propagation path, accident spread speed, and environmental performance effect were set. During the training process, the accident evolution process was not dynamically adjusted according to the trainees' behavioral data or accident perception delay, and the accident development process was always executed in the order of the accident script.

[0098] S6. Conduct a comprehensive analysis of the adaptive accident evolution state and the trainee's operation trajectory, reconstruct the accident causal relationship structure, and obtain personalized accident retraining scenarios.

[0099] S6.1. Collect operational trajectory data and behavioral record data of trainees under adaptive accident evolution state to form training behavior analysis data. Based on the training behavior analysis data, identify operational delay nodes and misoperation nodes that trainees have in the accident handling process and generate training weakness analysis data.

[0100] Specifically, a head-mounted display device continuously records the sequence of trainees' position changes in the tunnel space and generates operational trajectory data. Simultaneously, an interactive control device records the trigger time of trainees' equipment and facility operations, the operation object number, and the operation type information, forming behavioral record data. The operational trajectory data and behavioral record data are then linked and organized according to a unified time stamp to form training behavior analysis data. Subsequently, based on the accident response sequence requirements and response time requirements recorded in the accident causal relationship structure, the time difference between the operation trigger time in the training behavior analysis data and the time of accident symptom appearance is compared to identify operation delay nodes. Operational behaviors with mismatched operation objects in the behavioral record data are identified as misoperation nodes, generating training weakness analysis data. The training weakness analysis data is a dataset reflecting operational deficiencies and behavioral defects. It mainly includes operational delay node identification data, misoperation node identification data, records of non-execution of critical handling operations, records of deviations in the sequence of critical handling operations, and corresponding time and spatial location information of the operations. Operation delay node identification data is used to record risk nodes or equipment nodes where trainees react slowly after the appearance of accident signs. Misoperation node identification data is used to record the location nodes corresponding to trainees' operation behaviors that do not match the accident handling requirements. Records of non-execution of critical handling operations are used to identify important operation steps that were not completed in the prescribed handling process. Records of deviations in the sequence of critical handling operations are used to describe the differences between the trainees' actual operation sequence and the standard handling process.

[0101] S6.2. Based on the analysis data of training weaknesses, adjust the accident triggering conditions and environmental state parameters in the accident causal relationship structure to generate an updated accident causal relationship structure. Based on the updated accident causal relationship structure, load the corresponding scene resources to form a personalized accident retraining scenario.

[0102] Specifically, based on the operational delay nodes and misoperation nodes identified in the training weakness analysis data, the accident triggering conditions of the corresponding nodes in the accident causal relationship structure are adjusted. The accident triggering time conditions associated with operational delay nodes are pre-set, and the accident triggering sequence conditions associated with misoperation nodes are strengthened. At the same time, the environmental state parameters of the corresponding risk nodes in the accident causal relationship structure are adjusted to improve the sensitivity of accident performance, and an updated accident causal relationship structure is generated. Subsequently, based on the risk node number and accident performance effect association path recorded in the updated accident causal relationship structure, the corresponding roadway scene resources, equipment and facility scene resources, and accident performance effect scene resources are called and loaded to generate a personalized accident retraining scenario corresponding to the training weakness in the virtual reality display environment.

[0103] It should be noted that, as Figure 6As shown, the comparison results of the average accident handling time of trainees in the control group (accident evolution training scheme driven by fixed accident script) and the adaptive group are presented. The average handling time of the adaptive group is significantly lower than that of the control group, indicating that after enabling adaptive accident evolution and personalized accident retraining scenarios based on accident perception delay data, the reaction efficiency of trainees in the accident handling process is significantly improved, which verifies the technical effect of the present invention in improving the effectiveness of VR safety training.

[0104] This embodiment also provides a computer device applicable to the mine accident simulation training method based on VR technology, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the mine accident simulation training method based on VR technology proposed in the above embodiment.

[0105] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0106] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the VR-based mine accident simulation training method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0107] In summary, this invention achieves a high-fidelity, dynamically responsive, and personalized VR safety training system by constructing a mine digital twin scenario that integrates spatial semantics and accident causal logic, and combining it with an adaptive accident evolution and personalized retraining mechanism based on perception delay. This enhances the realism of mine accident simulation, the relevance of teaching, and the effectiveness of training.

[0108] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A mine accident simulation training method based on VR technology, characterized in that: include, Data on mine roadway structure, equipment and facility locations, and disaster avoidance routes are collected, and unified coordinate transformation and semantic annotation are performed to construct a semantic twin scenario of the mine. Typical mine accident types are collected, accident causal relationships are constructed, and accident elements in the accident causal relationships are mapped to risk nodes in the mine semantic twin scenario to form an accident causal relationship structure; Based on the causal relationship structure of the accident, scene resources are loaded and VR real-time rendering and posture synchronization are started to capture the behavior and actions of the trainees, drive the dynamic evolution of the accident, and form a real-time accident evolution scene. The trainees' line of sight movement and operational behavior in real-time accident evolution scenarios are continuously recorded, and the time when accident signs appear and the time when the trainees take their first effective action are recorded to obtain accident perception delay data. The accident perception delay data is compared with the perception delay threshold, the delay level judgment result is output, and the accident evolution state is adjusted according to the delay level judgment result to output the adaptive accident evolution state. By comprehensively analyzing the adaptive accident evolution state and the trainee's operation trajectory, the causal relationship structure of the accident is reconstructed, and personalized accident retraining scenarios are obtained.

2. The mine accident simulation training method based on VR technology as described in claim 1, characterized in that: The process involves collecting data on mine roadway structure, equipment and facility locations, and evacuation routes, performing unified coordinate transformation and semantic annotation, and constructing a semantic twin scenario for the mine. The specific steps are as follows: Collect mine roadway data, equipment and facility locations, and disaster avoidance route data; extract the roadway orientation, dimensions, slope, and spatial elevation relationships; and convert them into the mine's three-dimensional coordinate system to form spatial structure data. Based on spatial structure data, establish the topological relationship between roadways to form spatial topological information, spatially locate equipment and facilities, output equipment location information, and perform risk semantic and interactive semantic annotation on areas prone to gas anomalies and areas with roof fractures, outputting risk semantic information and interactive semantic information. By integrating spatial topology information, equipment positioning information, risk semantic information, and interaction semantic information into a unified semantic data structure, a semantic twin scenario of the mine is output.

3. The mine accident simulation training method based on VR technology as described in claim 2, characterized in that: The process involves collecting typical mine accident types, constructing accident causal relationships, and mapping the accident elements in these causal relationships to risk nodes in a mine semantic twin scenario to form an accident causal relationship structure. The specific steps are as follows: Collect typical mine accident types and extract accident causative factors, environmental conditions, and human behavior factors to form an accident factor set; Based on the set of accident elements, rules for accident occurrence conditions are established, and the state associations between accident elements are transformed into accident causal relationships through these rules. By associating accident-causing factors, environmental conditions, and personnel behavior factors with risk nodes in the semantic twin scenario of the mine, a causal-risk mapping relationship is constructed and integrated to form an accident causal relationship structure.

4. The mine accident simulation training method based on VR technology as described in claim 3, characterized in that: The process involves loading scene resources based on the accident causal relationship structure, initiating real-time VR rendering and posture synchronization, capturing the behavior and actions of trainees, driving the dynamic evolution of the accident, and forming a real-time accident evolution scenario. The specific steps are as follows. Based on the cause-and-effect relationship structure of the accident, scene resources are invoked, and the VR rendering engine is started to render the scene resources in real time, while simultaneously collecting head-mounted display posture data and spatial position information, and outputting the real-time rendered scene. Collect the movement paths and actions of trainees in a real-time rendered scene, and output the trainee behavior data; The system matches the behavioral data of trainees with the rules governing accident occurrence, outputs updated accident status, adjusts the accident performance in the real-time rendering scene based on the updated accident status, and outputs the real-time accident evolution scene.

5. The mine accident simulation training method based on VR technology as described in claim 1, characterized in that: The specific steps for obtaining the accident perception delay data are as follows: In real-time accident evolution scenarios, behavioral record data of trainees are collected, and the time point when the accident signs first appear is detected during the accident evolution process to generate accident sign occurrence time data; Identify the time point when trainees first perform an effective response action based on behavioral record data, and generate initial response time data; The time difference between the data on the time when accident signs appear and the data on the time of the first effective response is calculated to form accident perception delay data.

6. The mine accident simulation training method based on VR technology as described in claim 5, characterized in that: The steps for comparing accident perception delay data with a perception delay threshold, outputting a delay level determination result, adjusting the accident evolution state based on the delay level determination result, and outputting an adaptive accident evolution state are as follows. The accident perception delay data is compared with the perception delay threshold to generate a delay level determination result. When the delay level determination result is the first level, the accident spread speed is automatically reduced and key node prompts are added, and the first accident evolution state is output. When the delay level is determined to be Level 2, the smoke diffusion intensity is increased, the environmental visibility is reduced, and secondary risks are added, and the second accident evolution state is output. The real-time accident evolution scenario is updated based on the first and second accident evolution states to form an adaptive accident evolution state.

7. The mine accident simulation training method based on VR technology as described in claim 6, characterized in that: The perception delay threshold is generated by statistically analyzing the time difference between the occurrence of accident signs and the first effective response time in different accident scenarios for multiple batches of trainees, calculating the average perception time and standard deviation for each accident type, and dividing it into delay level intervals.

8. The mine accident simulation training method based on VR technology as described in claim 1, characterized in that: The process involves comprehensively analyzing the adaptive accident evolution state and the trainee's operational trajectory to reconstruct the accident causal relationship structure and obtain personalized accident retraining scenarios. The specific steps are as follows. Collect operational trajectory data and behavioral record data of trainees under adaptive accident evolution state to form training behavior analysis data. Based on the training behavior analysis data, identify the operational delay nodes and misoperation nodes of trainees in the accident handling process and generate training weakness analysis data. Based on the analysis data of training weaknesses, the accident triggering conditions and environmental state parameters in the accident causal relationship structure are adjusted to generate an updated accident causal relationship structure. Based on the updated accident causal relationship structure, the corresponding scene resources are loaded to form a personalized accident retraining scenario.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the mine accident simulation training method based on VR technology as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the mine accident simulation training method based on VR technology as described in any one of claims 1 to 8.

Images