Coal mine accident drilling method, device, equipment, medium and product based on VR

Abstract

The present application relates to the technical field of coal mine accident simulation based on VR, and particularly relates to a coal mine accident simulation method, device, equipment, medium and product based on VR, which comprises the following steps: performing scene analysis on collected coal mine accident cases to obtain a scene script containing a scene evolution path and multiple flexible modules; combining the scene evolution path and the flexible modules of the scene script according to a simulation instruction to generate a VR simulation scene including time nodes; determining a feedback model corresponding to a simulation time node selected by a user from a feedback model library, and receiving accident emergency measure information selected by the user; and interacting based on the accident emergency measure information and the feedback model to realize interactive operation of accident emergency measure information implementation behavior of the VR simulation scene under the influence of the current accident emergency measure information. The present application can carry out targeted emergency simulation, and provides technical means and scientific approaches for improving emergency rescue capability and level.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coal mine accident simulation based on VR, and particularly relates to a coal mine accident simulation method, device, equipment, medium and product based on VR. BACKGROUND

[0002] Traditional coal mine accident emergency rescue simulation mainly adopts sand table deduction, video demonstration and on-site simulation, etc. These methods can improve the understanding of safety knowledge of the staff to a certain extent, but have obvious deficiencies in simulating real complex accident scenes and practicing emergency response measures. The rescue drill has the problems of weak reality of the simulation scene, difficulty in simulating real disaster situations, difficulty in covering various potential accident scenes by the related simulation means, lack of interactivity and immersion, etc., resulting in poor simulation effect and inability to effectively improve the actual emergency response ability of the staff. In addition, due to the particularity and danger of the mine environment, simulation in the real environment not only has high cost, but also has safety risks. Therefore, it is urgent to develop scientific accident emergency rescue simulation technology, which is of great significance to effectively improve the emergency response ability under the premise of ensuring safety and reducing cost. SUMMARY

[0003] In view of the above problems of the prior art, the present application provides a coal mine accident simulation method, device, equipment, medium and product based on VR, which can effectively solve the problems faced by current accident emergency simulation, provide a scientific approach for targeted emergency simulation and improve emergency response ability and level.

[0004] In a first aspect, the present application provides a coal mine accident simulation method based on VR, comprising: performing scene analysis on collected coal mine accident cases to obtain scene scripts of each coal mine accident case, wherein the scene script comprises a scene evolution path and a plurality of flexible modules, and the flexible modules are used to combine the scene evolution path to generate a VR simulation scene according to a simulation instruction; obtaining a simulation instruction of a user, combining the scene evolution path and the flexible modules of the scene script according to the simulation instruction, and generating a VR simulation scene, wherein the VR simulation scene comprises a plurality of time nodes; obtaining a simulation time node selected by the user from different time nodes; determining a feedback model corresponding to the simulation time node from a feedback model library, wherein the feedback model library comprises: disaster state information of each time node, accident emergency measure information corresponding to the disaster state information, and accident emergency measure information implementation behavior corresponding to the accident emergency measure information; showing a plurality of selected accident emergency measure information to the current user, receiving the accident emergency measure information selected by the current user; and interacting with the feedback model based on the accident emergency measure information selected by the current user, and realizing interactive operation of the accident emergency measure information implementation behavior of the VR simulation scene under the influence of the current accident emergency measure information.

[0005] In some embodiments, the drill instruction includes: a site selection instruction, a rescue type selection instruction, a personnel type selection instruction, and a disaster type selection instruction; and the step of combining the scene evolution path and the flexible module of the scene script according to the drill instruction to generate the VR drill scene includes: combining the scene scripts of each coal mine accident case according to the site selection instruction, the rescue type selection instruction, the personnel type selection instruction, and / or the disaster type selection instruction to obtain a determined scene evolution path and a flexible module corresponding to the drill instruction; combining the scene evolution path and the flexible module to obtain scene pheromones; obtaining rendering elements corresponding to the scene pheromones, rendering each scene pheromone according to the rendering elements, and obtaining the VR drill scene.

[0006] In some embodiments, the step of rendering each scene pheromone according to the rendering elements to obtain the VR drill scene includes: constructing an index table according to the rendering elements; integrating each scene pheromone according to the index table to obtain an initial three-dimensional grid graph; and assigning values to the initial three-dimensional grid graph according to each rendering element to obtain the VR drill scene.

[0007] In some embodiments, the step of integrating each scene pheromone according to the index table to obtain an initial three-dimensional grid graph includes: establishing a first initial three-dimensional grid graph at the initial operation; converting the coordinates M of the scene pheromones corresponding to the index table through a reality conversion matrix F1 to obtain real coordinates M1 of the scene pheromones; M1 = MxF1; converting the real coordinates M1 through a display conversion matrix F2 to obtain display coordinates M2; M2 = M1xF2; converting the display coordinates M2 through a mapping conversion matrix F3 to obtain mapping coordinates M3; M3 = F3XM2; and matching the mapping coordinates with each virtual coordinate in the first initial three-dimensional grid graph to obtain the initial three-dimensional grid graph.

[0008] In some embodiments, matching the mapping coordinates with the respective virtual coordinates in the first initial three-dimensional grid map to obtain the initial three-dimensional grid map comprises: calculating the correlation between the respective mapping coordinates; initializing an iteration parameter, the iteration parameter comprising: a first matching coordinate of the respective mapping coordinates, a number of the mapping coordinates, a position of the respective mapping coordinates initially matched, and a matching step length; traversing the respective mapping coordinates, and calculating a first correlation between the respective mapping coordinates under the current first matching coordinate; determining whether the first correlation is a preset multiple of the correlation; if the first correlation is the preset multiple of the correlation, outputting the first matching coordinate as a final target matching coordinate, mapping the pheromone of the scene corresponding to the respective mapping coordinates to the target matching coordinate according to the target matching coordinate to obtain the initial three-dimensional grid map; if the first correlation is not the preset multiple of the correlation, updating the matching step length according to a matching degree between the first correlation and the correlation to obtain a new matching step length; taking the new matching step length as the matching step length and returning to the above operation until the first correlation is the preset multiple of the correlation, and outputting the initial three-dimensional grid map.

[0009] In some embodiments, updating the matching step length according to the matching degree between the first correlation and the correlation to obtain a new matching step length comprises: calculating a gradient value of the current matching degree according to a preset gradient loss function; calculating the new matching step length according to the gradient value; Bnew=B-λ×▽f(x); wherein, λ is a learning rate; B is the matching step length; Bnew is the new matching step length; and ▽f(x) is the gradient value.

[0010] In a second aspect, the present application provides a coal mine accident drilling device based on VR, the device comprising: a decomposition module for analyzing the collected coal mine accident cases to obtain the scene scripts of each coal mine accident case, wherein the scene script comprises a scene evolution path and a flexible module; a combination module for obtaining the drilling instruction of the user, combining the scene evolution path and the flexible module of the scene script according to the drilling instruction, and generating a VR drilling scene, wherein the VR drilling scene comprises a plurality of time nodes; a selection module for obtaining the drilling time node selected by the user from different time nodes; determining the feedback model corresponding to the drilling time node from the feedback model library, wherein the feedback model library comprises: disaster state information of each time node, accident emergency measure information for the disaster state information, and accident emergency measure information implementation behavior corresponding to the accident emergency measure information; and a display module for displaying a plurality of selected accident emergency measure information to the current user, receiving the accident emergency measure information selected by the current user, and interacting with the feedback model based on the accident emergency measure information selected by the current user to realize the interactive operation of the accident emergency measure information implementation behavior of the VR drilling scene under the influence of the current accident emergency measure information.

[0011] In a third aspect, the present application provides a computer device comprising a memory, a processor and a computer program stored in the memory, the processor executing the computer program to implement the steps of the VR-based coal mine accident drilling method of any one of the above aspects.

[0012] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the steps of the VR-based coal mine accident drilling method of any one of the above aspects.

[0013] In a fifth aspect, the present application provides a computer program product comprising a computer program, the computer program being executed by a processor to implement the steps of the VR-based coal mine accident drilling method of any one of the above aspects.

[0014] The present application provides a VR-based coal mine accident drilling method, device, equipment, medium and product, wherein the method comprises: performing scenario analysis on collected coal mine accident cases to obtain scenario scripts of each coal mine accident case, wherein the scenario script comprises a scenario evolution path and a flexible module; obtaining a drilling instruction of a user, combining the scenario evolution path and the flexible module of the scenario script according to the drilling instruction, and generating a VR drilling scene, wherein the VR drilling scene comprises multiple time nodes; obtaining a drilling time node selected by the user from different time nodes; determining a feedback model corresponding to the drilling time node from a feedback model library, wherein the feedback model library comprises: disaster state information of each time node, accident emergency measure information for the disaster state information, and accident emergency measure information implementation behavior corresponding to the accident emergency measure information; displaying multiple selected accident emergency measure information to the current user, and receiving accident emergency measure information selected by the current user; interacting with the feedback model based on the accident emergency measure information selected by the current user, and realizing interactive operation of the accident emergency measure information implementation behavior of the VR drilling scene under the influence of the current accident emergency measure information; and the mine accident safety training can be efficiently and economically performed.

[0015] The technical scheme of the present application generates a VR practice scene according to the practice instruction of a user, simulates a real complex accident scene, and in subsequent operation, based on the disaster state information and the accident emergency measure information in the selected VR practice scene, simulates the implementation behavior of the user for the disaster state information in the current VR practice scene, realizes immersive training of multi-role internal emergency and external rescue through the combination of'macroscopic view whole-process deduction' and 'on-site operation deduction', cooperatively integrates models, animations, audio, interaction, questions and answers and other contents by using dynamic scene display technology, completes virtualized demonstration production of the whole process scene script of accidents such as gas explosion, thereby improving the emergency response capability of the staff for coal mine accidents, and through the immersive practice of VR, improving the interactivity and the authenticity of the practice, greatly improving the learning effect, and can ensure the safety of the staff in the accident practice. BRIEF DESCRIPTION OF DRAWINGS

[0016] The present application will be described in more detail below based on the embodiments and with reference to the accompanying drawings:

[0017] Figure 1 A scene combination selection process schematic diagram is provided for the present application.

[0018] Figure 2 A coal mine accident emergency handling overall process schematic diagram is provided for the present application.

[0019] Figure 3 A coal mine accident emergency scene combination schematic diagram is provided for the present application.

[0020] Figure 4 A total process schematic diagram of a coal mine accident emergency practice method based on virtual reality application is provided for the present application.

[0021] Figure 5 An overall operation process schematic diagram of a coal mine accident emergency practice method based on virtual reality application is provided for the present application.

[0022] Figure 6 A VR practice process schematic diagram in a coal mine accident emergency practice method based on virtual reality application is provided for the present application.

[0023] Figure 7 A gas accident typical scene list schematic diagram in a coal mine accident emergency practice method based on virtual reality application is provided for the present application.

[0024] Figure 8 A fire accident typical scene list schematic diagram in a coal mine accident emergency practice method based on virtual reality application is provided for the present application.

[0025] Figure 9A top plate accident typical scene list schematic diagram in a coal mine accident emergency drilling method based on virtual reality application is provided in the present application;

[0026] Figure 10 A dangerous source and accident site combination list schematic diagram in a coal mine accident emergency drilling method based on virtual reality application is provided in the present application;

[0027] Figure 11 An operation step schematic diagram of a VR drilling scene rendered in a coal mine accident emergency drilling method based on virtual reality application is provided in the present application;

[0028] Figure 12 An operation flow simulation schematic diagram of an initial three-dimensional grid map matched in a coal mine accident emergency drilling method based on virtual reality application is provided in the present application.

[0029] In the drawings, the same components are designated by the same reference numerals, and the drawings are not drawn according to the actual scale. DETAILED DESCRIPTION

[0030] In order to better understand the technical personnel in the art of the present application scheme, and to the application of technical means to solve the technical problems, and to achieve the corresponding technical effect process can be fully understood and implemented, the following will be combined with the drawings in the embodiments of the present application, the technical scheme in the embodiments of the present application is described clearly and completely, obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The embodiments of the present application and the features of the embodiments can be combined with each other without conflict, and the technical scheme formed thereby is within the scope of protection of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0031] It should be noted that the terms "first", "second" and the like in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0032] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that here.

[0033] In the field of mine safety training, the related technology mainly relies on traditional teacher lectures, video learning, and on-site simulation drills. Although these methods have improved the staff's understanding of safety knowledge to some extent, they have obvious shortcomings in simulating real complex accident scenes and practicing emergency response measures (mine rescue team daily rescue drills have difficulties in simulating real disaster situations). Related training methods are difficult to cover all potential accident scenarios comprehensively, and lack interactivity and immersion, resulting in poor learning effect and inability to effectively improve the actual emergency response ability of the staff. In addition, due to the particularity and danger of the mine environment, the safety and economy of the related training methods also face challenges. Simulation training in a real environment not only has high costs, but also has safety risks. There is a technical problem in the field of how to efficiently and economically conduct mine accident safety training.

[0034] In order to solve the above technical problem of how to efficiently and economically conduct mine accident safety training, the present application proposes a coal mine accident drilling method, device, equipment, medium and product based on VR, and the implementation details of the present application are described below. The following content provides implementation details for easy understanding, and is not necessary for implementing the present solution.

[0035] Example One

[0036] In the technical scheme of the embodiment, a coal mine accident drilling method based on VR is provided, which comprises: performing scene analysis on collected coal mine accident cases to obtain scene scripts of each coal mine accident case, wherein the scene script comprises a scene evolution path and a plurality of flexible modules, wherein the flexible modules are used to combine the scene evolution path to generate a VR drilling scene according to drilling instructions; obtaining drilling instructions of a user, combining the scene evolution path and the flexible modules of the scene script according to the drilling instructions, and generating a VR drilling scene, wherein the VR drilling scene comprises a plurality of time nodes; obtaining a drilling time node selected by the user from different time nodes; determining a feedback model corresponding to the drilling time node from a feedback model library, wherein the feedback model library comprises: disaster state information of each time node, accident emergency measure information for the disaster state information, and accident emergency measure information implementation behavior corresponding to the accident emergency measure information; displaying a plurality of selected accident emergency measure information to the current user, and receiving the accident emergency measure information selected by the current user; and interacting with the feedback model based on the accident emergency measure information selected by the current user, to realize interactive operation of the accident emergency measure information implementation behavior of the VR drilling scene under the influence of the current accident emergency measure information.

[0037] Specifically, the generation of the above-mentioned scene script is through the combination of modular design and neural network model. In the script generation model, the corresponding emergency drilling script is generated based on the input modular script elements, and the generated emergency drilling script is automatically evaluated, which can quickly find and modify the syntax and logical errors and other problems in the emergency drilling script, and guide the efficient performance of the accident emergency drilling. Through modular design, the generated emergency drilling script is more scientific and reasonable, and can fully cover possible accident scenes. Specifically:

[0038] Taking the scene library and the measure library as credible data sources, according to the theory of "enterprise independent emergency and social joint rescue", and according to the emergency rescue procedures of different accident subjects, the mine emergency rescue drilling script is modularly split: vertically, based on the time process of accident occurrence, the script is divided into four modules of accident incubation stage A, mine internal disposal B, mine external rescue C, and emergency end D, and each first-level module has its own second-level module, which includes two categories of flexible replacement module and fixed module; horizontally, based on the multi-level scene structure of "dimension-element-attribute", the accident script is structured layer by layer, and then the emergency drilling script library is constructed;

[0039] The target accident scene of the pre-rehearsal is divided into multiple target task modules by using a task decomposition method, target script elements of each of the target task modules are determined, the target script elements include accident cause elements (risk, etc.), accident location elements (including equipment, personnel, distress situation, etc.), accident stage elements (determining accident size, emergency object, etc.), and accident consequence elements (poisoning and burns caused by high temperature, suffocation caused by damage to ventilation facilities, and personnel trapped caused by impact damage to support, etc.);

[0040] The multiple target task modules and the corresponding multiple target script elements thereof are referenced to script content corresponding to a scene module in a learning emergency rehearsal script library, input into a trained script generation model to obtain an initial emergency rehearsal script output by the script generation model, the content including emergency element sentences such as rescue personnel, rescue tools, rescue supplies, and rescue methods, and also including descriptive sentences such as language and action, the initial emergency rehearsal script is evaluated, and if the evaluation result indicates that the initial emergency rehearsal script has a preset error, the initial emergency rehearsal script is corrected to obtain a corrected emergency rehearsal script, and the corrected emergency rehearsal script is used as a target emergency rehearsal script for rehearsal, the preset error includes at least one of a syntax error, a logic error, an information omission, and a format error, and the efficient performance of the accident emergency rehearsal is guided;

[0041] In the method, a new accident case is reconstructed by fitting an accident case of an accident scene, a new script is generated for the new accident case, and the new accident case and the new script are used as training data to train the script generation model, so that the script generation model is trained more specifically in the case of limited accident cases and scenes;

[0042] On the other hand, when rendering each scene information element according to the rendering element to obtain the VR rehearsal scene, the first step is to construct an index table according to the rendering element, which is used to represent the vector of the endpoint element, the vector of the vertical vector element, the vector of the texture mapping element, and the vector of the visible surface element. Vectorize various rendering elements and construct the corresponding index table according to their coordinates. Then, according to the index table, integrate each scene information element. For example, for the mine tunnel scene information element, the corresponding coordinates and other information in the index table are sorted and integrated to obtain an initial three-dimensional grid graph. Then, according to each rendering element, the initial three-dimensional grid graph is assigned values. For example, the RGB value or brightness value of the pixel of the wall 2D image in the visible surface element is assigned to the current mine tunnel wall image corresponding position on the initial three-dimensional grid graph. Through such operations, the VR rehearsal scene is finally obtained. For example, in the simulation of a fire rehearsal scene in a certain section of the coal mine underground roadway, through such a detailed rendering process, the scene can be presented more realistically.

[0043] The technical solution of the embodiment generates a VR drill scene by using rendering elements to construct an index table, integrating scene information, and assigning values, so that the simulated drill scene is more realistic and has more details. In actual mine accident emergency drills, a realistic scene can make the participants feel as if they are in the scene and more truly experience the situation when the accident occurs. For example, when simulating a rescue scene after a gas explosion, the smoke effect and the damage to the roadway in the scene can be accurately presented through rendering, and rescue personnel can make more accurate emergency measures such as reconnaissance, fire extinguishing, and rescuing trapped personnel, thereby improving their operation skills. Moreover, such a high-quality simulation scene can attract more people to participate in the drill, repeatedly practice, and thus improve the cooperation and response capabilities of the entire mine emergency response team, effectively reduce the harm caused by accidents, and ensure the safe and orderly production of the mine.

[0044] Specifically, the flexible module includes a first-level module and a second-level module. More specifically, the first-level module is a site selection module, which is the top-level module in the entire hierarchical structure and is used to determine the geographical location and environmental background of the drill scene. For example, in a coal mine accident, the following options can be set: underground mine area, ground mine portal, equipment distribution area, etc. According to the site selected by the user, the module returns scene environment information corresponding to the site, such as mine structure, wall texture, ceiling strata structure, and lighting conditions, which will provide basic data for the subsequent rendering stage.

[0045] The second-level module is a nested sub-module, including a rescue type selection module, a personnel type selection module, and a disaster type selection module.

[0046] Specifically, after the site selection module is determined, the nested sub-modules are automatically activated by the system, which can further refine the scene requirements according to the selection of the first level, mainly including:

[0047] The rescue type selection module responds to user instructions and determines the applicable rescue mode according to the selected site, such as self-rescue in the mine, internal rescue, or external collaborative rescue, and returns information such as corresponding rescue procedures, emergency equipment models, and emergency instruction texts.

[0048] The personnel type selection module limits the personnel roles that can participate in the rescue according to the site characteristics and accident environment, such as miners, rescue personnel, and command dispatchers, and returns posture models, interactive action instructions, and behavior feedback information corresponding to each personnel role.

[0049] The disaster type selection module determines the disaster type according to the actual coal mine accident risk, such as gas explosion, fire, roof collapse, etc., and outputs information such as disaster emergency measures, dangerous area identification, and accident development evolution logic, and provides corresponding animations and audio instructions in combination with the emergency measure information.

[0050] Each of the above flexible modules is independently designed, containing complete context elements and data of itself, and can be reused in different scenarios.

[0051] All modules directly respond to the user-issued drill instructions. The user first establishes the basic environment of the scene through the location selection module, and then the system automatically calls the sub-modules to further refine the rescue mode, personnel configuration, and accident type information suitable for the environment.

[0052] In the specific execution process, the system receives the user's input drill instructions (such as the selection of location, rescue, personnel, and disaster type), then responds by the location selection module, and extracts the basic flexible modules matching the selected location from the scenario script library. Further, after the basic module is determined, the system calls the corresponding rescue type, personnel type, and disaster type modules, which are nested and combined according to the predetermined tree structure to generate a final scenario script that comprehensively describes the accident scene and emergency measures information.

[0053] Then, according to the final scenario script, VR rendering operations are performed;

[0054] Suppose a user's selected instructions are:

[0055] Location: underground mine area

[0056] Rescue type: internal rescue

[0057] Personnel type: miners and rescue team

[0058] Disaster type: gas explosion

[0059] After the system receives and analyzes the user's selected instructions, it determines the environment data of the underground mine area through the location selection module and outputs basic information such as mine wall, passage structure, etc.

[0060] The system then calls the rescue type selection module to load the emergency procedures and equipment models related to internal rescue in the underground mine;

[0061] At the same time, the personnel type selection module is called to obtain 3D models and interactive actions of miners and rescue personnel;

[0062] The disaster type selection module provides gas explosion accident evolution dynamics, alarm animations, and emergency measures documents.

[0063] The scenario elements generated by each of the above modules are hierarchically nested to form a complete scenario script. According to the rendering steps described in the document, the data is mapped to the initial three-dimensional grid graph using the transformation matrix, and finally a VR drill scene that truly restores the gas explosion accident scene in the underground mine is generated.

[0064] In the implementation process, first, the flexible module level is divided and refined;

[0065] Specifically, it includes the first level module - the location selection module

[0066] The location selection module is the basis and starting point of the entire scenario script construction, used to determine the geographical location and environmental background of the drill scene.

[0067] Examples:

[0068] In coal mine accidents, you can set options such as underground mining area, ground mine, equipment distribution area, etc.

[0069] Output content:

[0070] According to user selection, the location selection module outputs environment element data related to the scene, such as mine structure, wall texture, ceiling structure, lighting conditions, etc., providing basic environmental information for subsequent modules.

[0071] The second level module - nested sub-module

[0072] After determining the location, the system automatically activates the following sub-modules, which further refine the accident emergency scene:

[0073] Rescue type selection module

[0074] In response to user instructions, determine the applicable rescue mode according to the selected location, such as self-rescue, internal rescue, or external collaborative rescue.

[0075] Output content:

[0076] Return the emergency process, equipment model, instruction text, etc. information corresponding to the rescue mode.

[0077] Personnel type selection module

[0078] According to the characteristics of the location and the situation of the accident site, limit the personnel roles participating in the rescue, such as miners, rescue personnel, command dispatchers, etc.

[0079] Output content:

[0080] Output the 3D model, motion interaction data, and behavior feedback information of each role.

[0081] Disaster type selection module

[0082] Determine the disaster type according to the accident risk, such as gas explosion, fire, roof collapse, etc.

[0083] Output content:

[0084] Emergency measures, dangerous area identification, accident evolution logic, emergency animation and audio instructions in corresponding disaster state.

[0085] Further, the implementation process also includes the implementation of hierarchical nesting and time evolution

[0086] First, construct a tree-like hierarchical structure

[0087] Parent-child node organization:

[0088] The location selection module is the root node of the tree structure, and all subordinate modules (rescue type, personnel type, disaster type) are its child nodes.

[0089] Data transmission:

[0090] After the root node (location selection module) determines the basic scene, its output data is used by the subordinate modules, and each subordinate module provides more detailed scene information based on the basic data, forming a hierarchical and combined scene script.

[0091] Further, the implementation of hierarchical nesting and time evolution also includes the implementation of time evolution and Bayesian / Markov model:

[0092] Dynamic transition modeling:

[0093] During the evolution of the accident scene, Bayesian networks or Markov chain models can be used to describe the transition probabilities between different module states. For example, the gas explosion state output by the disaster type selection module can be combined with the internal rescue state output by the rescue type module to form the time sequence probability of accident evolution, so that the system not only generates hierarchical static scene information, but also dynamically adjusts the scene script according to the time sequence and the probability of accident evolution, making the VR simulation more in line with the development law of real accidents.

[0094] After the implementation of hierarchical nesting and time evolution, VR scene rendering and data mapping are performed:

[0095] First, combine the output data of each module obtained by hierarchical nesting to form the final scene script.

[0096] Example flow:

[0097] The user selects "Location: Underground Mine Area", "Rescue Type: Internal Rescue", "Personnel Type: Miner and Rescue Team", and "Disaster Type: Gas Explosion";

[0098] The location selection module returns the structure and environment data of the underground mine area;

[0099] The rescue type module returns the internal rescue process and equipment;

[0100] The personnel type module provides 3D models of relevant roles and interactive actions;

[0101] The disaster type module outputs accident evolution animation, alarm and emergency measure documents.

[0102] These data are spliced through tree nesting to form a complete scenario script, wherein transition probabilities based on Bayesian / Markov chain calculation are embedded between modules, so that the accident drill has both static hierarchical structure and dynamic evolution trend.

[0103] Then, a rendering generation stage is performed to combine scene pheromones in the scenario script with rendering elements:

[0104] The output data of each module are mapped to an initial three-dimensional grid graph through conversion matrices (real conversion matrix F1, display conversion matrix F2 and mapping conversion matrix F3).

[0105] An index table is established to organize various rendering elements (such as endpoint elements, vertical vector elements, texture mapping elements and visible surface elements), and scene assignment is performed using these data to finally form a VR drill scene that truly restores an accident scene (such as a gas explosion in a coal mine).

[0106] The technical problem to be solved by the embodiment is how to efficiently and economically conduct mine accident safety training. In the technical scheme of the embodiment, first, a collected coal mine accident case is subjected to scenario analysis, the scenario analysis of the embodiment is to tease out a scenario script of each coal mine accident case, and the scenario script covers a scenario evolution path and flexible modules. Next, a drill instruction of a user is acquired, the drill instruction includes a location selection instruction, a rescue type selection instruction, a personnel type selection instruction and a disaster type selection instruction, and the scenario evolution path and the flexible modules in the scenario script are reasonably combined according to these instructions, so as to generate a VR drill scene with multiple time nodes. Then, a drill time node selected by the user from different time nodes is acquired, and a feedback model corresponding to the drill time node is determined from a feedback model library containing disaster state information of each time node, accident emergency measure information for the disaster state information and accident emergency measure information corresponding to the accident emergency measure information. Subsequently, the current user is shown a plurality of to-be-selected accident emergency measure information, and a selection of the user is received, and finally, the user-selected accident emergency measure information and the feedback model are interacted to realize interactive operation of the corresponding accident emergency measure information implementation behavior in the VR drill scene. For example, in the simulation of a coal mine gas explosion accident, through such a process, a scene that meets actual training needs can be combined according to different instructions, so that training no longer depends on high-cost and risky real environment drills.

[0107] The technical scheme of the embodiment can simulate diversified disaster situations through the series of operations, and solves the problems of weak reality, difficulty in simulating real disaster situations, high cost, safety risks and the like in previous mine safety training. With the aid of the VR technology, the rescue team members can repeatedly practice in a risk-free environment, such as simulating rescue operations at different stages after a gas explosion for multiple times, familiarize with measures to be taken under various conditions, strengthen decision-making and operation skills, and reduce injury and accident risks in real training. Meanwhile, for the mine workers, the realistic simulation environment and specific operation training can significantly improve their emergency and self-help abilities when encountering an emergency disaster, improve the cooperation and response abilities of the entire mine emergency response team, effectively reduce the threat of accidents to the safety of the workers, reduce the influence range and harm degree of the disaster, and improve the overall safety management level of the mine, so that the mine accident safety training becomes efficient and economical.

[0108] Example Two

[0109] On the basis of the above embodiment, the drill instruction includes a location selection instruction, a rescue type selection instruction, a personnel type selection instruction, and a disaster type selection instruction; the step of generating the VR drill scene by combining the scene evolution path and the flexible module of the scene script according to the drill instruction includes: combining the scene scripts of each coal mine accident case according to the location selection instruction, the rescue type selection instruction, the personnel type selection instruction, and / or the disaster type selection instruction to obtain the determined scene evolution path and the flexible module corresponding to the drill instruction; combining the scene evolution path and the flexible module to obtain scene pheromones; obtaining rendering elements corresponding to the scene pheromones, rendering each scene pheromone according to the rendering elements, and obtaining the VR drill scene.

[0110] The technical problem to be solved by the embodiment is how to combine the scene evolution path and flexible module of the scene script according to the drill instruction to generate a VR drill scene. In the technical solution of the embodiment, the drill instruction includes a site selection instruction, a rescue type selection instruction, a personnel type selection instruction, and a disaster type selection instruction. When the scene evolution path and flexible module of the scene script are combined according to the drill instruction to generate the VR drill scene, first, the scene scripts of various coal mine accident cases are combined according to the site selection instruction, the rescue type selection instruction, the personnel type selection instruction, and / or the disaster type selection instruction, and then the determined scene evolution path and flexible module corresponding to the drill instruction are obtained. For example, in training, if a fire accident occurring in a specific site is to be simulated, the corresponding site can be selected through the site selection instruction, the fire can be selected through the disaster type selection instruction, and then the relevant combination is performed. Then, the determined scene evolution path and flexible module are combined to obtain scene pheromones, which contain, for example, scene environment information in the current drill site, such as mine wall, mine ground, and other related information. The rendering elements corresponding to the scene pheromones are obtained, which include endpoint elements, vertical vector elements, texture mapping elements, and visible surface elements, and the various scene pheromones are rendered, and finally the VR drill scene is obtained.

[0111] The technical solution of the embodiment can enable users to flexibly customize the drill scene according to actual training needs by combining the scene script according to different drill instructions and generating the VR drill scene through a series of processes. For example, a mine enterprise wants to focus on training the ability of employees to respond to roof accidents in a specific area underground, and can accurately select corresponding instructions such as site and disaster type to generate the corresponding scene. This makes the drill scene no longer single and rigid, but can be generated on demand, greatly improving the pertinence and effectiveness of the drill. At the same time, because the virtual scene is generated based on VR technology, the high cost and safety risks of simulation training in a real environment are avoided, enabling rescue personnel and mine workers to participate in simulation training more frequently, constantly familiarizing themselves with response measures in various accident scenarios, improving emergency and rescue capabilities, enhancing the collaboration level of the entire mine emergency response team, and ensuring mine production safety.

[0112] Example Three

[0113] On the basis of the above embodiment, the VR drill scene is obtained by rendering each scene pheromone according to the rendering element, including: constructing an index table according to the rendering element; integrating each scene pheromone according to the index table to obtain an initial three-dimensional mesh graph; and assigning values to the initial three-dimensional mesh graph according to each rendering element to obtain the VR drill scene.

[0114] The technical problem to be solved by the embodiment is how to render each scene information element according to the rendering element to obtain a VR rehearsal scene. In the technical solution of the embodiment, when rendering each scene information element according to the rendering element to obtain a VR rehearsal scene, the first step is to construct an index table according to the rendering element, which is used to represent the vector of the endpoint element, the vector of the normal vector element, the vector of the texture mapping element, and the vector of the visible surface element, vectorize various rendering elements, and construct the corresponding index table according to the coordinates. Then, according to the index table, each scene information element is integrated, for example, for the mine tunnel scene information element, the corresponding coordinates and other information in the index table are sorted and integrated, so as to obtain an initial three-dimensional grid graph. Then, the initial three-dimensional grid graph is valued according to each rendering element, for example, the RGB value or the brightness value of the pixel of the wall surface 2D image in the visible surface element is valued to the position corresponding to the current mine tunnel wall image on the initial three-dimensional grid graph. Through such operation, the VR rehearsal scene is finally obtained. For example, in the simulation of the rehearsal scene of a fire in a certain section of a coal mine tunnel, through such detailed rendering process, the scene can be more realistically presented.

[0115] The technical solution of the embodiment generates a VR rehearsal scene by constructing an index table, integrating scene information elements, and valuing, so that the simulated rehearsal scene is more realistic and detailed. In actual mine accident emergency rehearsal, a realistic scene can make the participants feel as if they are in the scene, and more realistically feel the situation when the accident occurs. For example, in the simulation of the rescue scene after a gas explosion, the smoke effect and the damage of the tunnel in the scene can be accurately presented through rendering, and the rescuers can make more accurate emergency measures such as reconnaissance, fire extinguishing, and rescue of trapped personnel, thereby improving their operation skills. Moreover, such high-quality simulation scene can attract more people to participate in the rehearsal, repeatedly practice, and thus improve the cooperation and response ability of the whole mine emergency response team, effectively reduce the harm caused by the accident, and ensure the safe and orderly production of the mine.

[0116] Example Four

[0117] On the basis of the above embodiment, the various scene pheromones are integrated according to the index table to obtain an initial three-dimensional grid map, including: initially, a first initial three-dimensional grid map is established; coordinates M of the scene pheromones corresponding to the index table are converted to real coordinates M1 of the scene pheromones through a real conversion matrix F1; M1 = MxF1; display coordinates M2 are obtained through coordinate conversion based on the real coordinates M1 through a display conversion matrix F2; M2 = M1xF2; mapping coordinates M3 are obtained through coordinate conversion based on the display coordinates M2 through a mapping conversion matrix F3; M3 = F3XM2; the mapping coordinates are matched with each virtual coordinate in the first initial three-dimensional grid map to obtain the initial three-dimensional grid map.

[0118] The technical problem to be solved by the embodiment is how to integrate the various scene pheromones according to the index table to obtain an initial three-dimensional grid map. In the technical scheme of the embodiment, at the initial operation, a first initial three-dimensional grid map is established, which is empty at the beginning and only contains a grid structure, and is gradually filled with content subsequently. Figure One is empty at the beginning and only contains a grid structure, and is gradually filled with content subsequently. Then, coordinates M of the scene pheromones corresponding to the index table are converted to real coordinates M1 of the scene pheromones through a real conversion matrix F1, which is calculated as M1 = MxF1, that is, two-dimensional image coordinates of scene pheromones such as mine wall are converted to three-dimensional world coordinates. Then, display coordinates M2 are obtained through coordinate conversion based on the real coordinates M1 through a display conversion matrix F2, which is calculated as M2 = M1xF2, and the real coordinates are converted to three-dimensional coordinates in the initial three-dimensional scene, that is, a VR rehearsal scene. Then, mapping coordinates M3 are obtained through coordinate conversion based on the display coordinates M2 through a mapping conversion matrix F3, which is calculated as M3 = F3XM2, and the three-dimensional coordinates are mapped to the mapping coordinates displayed on the VR glasses. Finally, the mapping coordinates are matched with each virtual coordinate in the first initial three-dimensional grid map, for example, the mapping coordinates corresponding to the scene pheromones at different positions of the mine are matched with the virtual coordinates in the grid map, so as to obtain the initial three-dimensional grid map.

[0119] The technical scheme of the embodiment integrates the initial three-dimensional grid map by a series of coordinate conversion and matching operations, thereby laying a foundation for generating a high-quality VR practice scene. For example, when simulating a complex roof accident scene in a coal mine, through accurate coordinate conversion and matching, the roof, support equipment and other elements in the scene can be accurately presented in the three-dimensional grid map, and the accident scene can be accurately restored in the final VR practice scene. Thus, the personnel participating in the practice can more accurately take measures such as emergency escape and rescue organization when facing simulated roof collapse and personnel being trapped, thereby strengthening their emergency operation skills. At the same time, high-quality scene simulation helps the entire mine emergency response team to practice and adapt multiple times, thereby improving their cooperation and response capabilities, reducing losses when a real accident occurs, and protecting the life safety of mine workers and the normal development of mine production.

[0120] Example Five

[0121] On the basis of the above embodiment, the mapping coordinates are matched with each virtual coordinate in the first initial three-dimensional grid map to obtain the initial three-dimensional grid map, including: calculating and obtaining the correlation between each mapping coordinate; initializing iteration parameters, the iteration parameters including: the first matching coordinates of each mapping coordinate, the number of mapping coordinates, the position of the initial matching of each mapping coordinate and the matching step; traversing each mapping coordinate, calculating and obtaining the first correlation between each mapping coordinate under the current first matching coordinates; judging whether the first correlation is a preset multiple of the correlation; if the first correlation is the preset multiple of the correlation, outputting the first matching coordinates as the final target matching coordinates, mapping the scene pheromones corresponding to each mapping coordinate to the target matching coordinates according to the target matching coordinates to obtain the initial three-dimensional grid map; if the first correlation is not the preset multiple of the correlation, updating the matching step according to the matching degree between the first correlation and the correlation to obtain a new matching step; taking the new matching step as the matching step and returning to the above operation until the first correlation is the preset multiple of the correlation, and outputting the initial three-dimensional grid map.

[0122] The technical problem to be solved by the embodiment is how to match the mapping coordinates with each virtual coordinate in the first initial three-dimensional grid map to obtain an initial three-dimensional grid map. In the technical solution of the embodiment, the correlation between each mapping coordinate is first calculated and obtained. The correlation in the embodiment can be determined by a Euclidean distance or a Manhattan distance, etc., which serves as a reference basis for subsequent matching operations. Then, iteration parameters are initialized, which include the first matching coordinates of each mapping coordinate, the number of mapping coordinates, the position of the initial matching of each mapping coordinate, and the matching step length, etc. Then, each mapping coordinate is traversed, the first correlation between each mapping coordinate under the current first matching coordinate is calculated and obtained, and it is judged whether the first correlation is a preset multiple of the correlation. For example, in the coordinate matching process of simulating a coal mine fire accident scene, if the first correlation is the preset multiple, the first matching coordinate is output as the final target matching coordinate, and then the scene information corresponding to each mapping coordinate is mapped to the target matching coordinate according to the target matching coordinate, so as to obtain an initial three-dimensional grid map. If the first correlation is not the preset multiple of the correlation, the matching degree (which can be calculated using a mean square error) between the first correlation and the correlation is used to update the matching step length, a new matching step length is obtained, and the new matching step length is used as the matching step length to repeat the above operation until the first correlation is the preset multiple of the correlation, and an initial three-dimensional grid map is output.

[0123] The technical solution of the embodiment can more accurately match the mapping coordinates with the virtual coordinates to obtain an initial three-dimensional grid map through such a rigorous and iterative optimization matching method, and thus ensure that the generated VR rehearsal scene is more consistent with the actual situation. For example, in the simulation of a complex scene of multiple disasters in a coal mine underground, accurate coordinate matching can correctly present each scene element in the three-dimensional grid map, so that the disaster scene, device layout, personnel position, etc. in the final VR rehearsal scene are all real and reasonable. Personnel participating in the rehearsal can better train their ability to cope with complex situations and improve the accuracy of emergency operations in such an accurate scene. Moreover, for the entire mine emergency response team, a high-quality simulation scene helps to repeatedly rehearse and improve collaboration and response capabilities, so that rescue work can be carried out more quickly and effectively when a real accident occurs, reducing the loss caused by the accident and ensuring the safety of production and personnel in the mine.

[0124] Example Six

[0125] On the basis of the above embodiment, the matching step is updated according to the matching degree between the first association relationship and the association relationship, and a new matching step is obtained, including: according to a preset gradient loss function, a gradient value of the current matching degree is calculated; the new matching step is calculated according to the gradient value; Bnew = B - λ ×▽f(x); wherein, λ is a learning rate; B is the matching step; Bnew is the new matching step; and▽f(x) is the gradient value.

[0126] The technical problem to be solved by the embodiment is how to update the matching step according to the matching degree between the first association relationship and the association relationship, and obtain a new matching step. In the technical solution of the embodiment, first, the gradient value▽f(x) of the current matching degree is calculated according to a preset gradient loss function, which reflects the trend of the change of the matching degree. Then, the new matching step is calculated according to the gradient value, and the calculation formula is Bnew = B - λ ×▽f(x), wherein λ is a learning rate, B is the matching step, and Bnew is the new matching step. For example, in the coordinate matching process of simulating a coal mine gas leakage accident scene, the gradient value of the current matching degree is calculated by the gradient loss function, and the matching step is reasonably adjusted in combination with the set learning rate and other parameters. If the change trend of the matching degree is large, the matching step is appropriately increased to speed up the search for the matching coordinate; if the change of the matching degree is relatively gentle, the matching step is correspondingly reduced to make the search more accurate.

[0127] The technical solution of the embodiment dynamically updates the matching step according to the gradient value of the matching degree, and can more efficiently and accurately search for the target matching coordinate corresponding to the mapping coordinate. When constructing a VR drilling scene, for example, when simulating a complex coal mine accident scene involving a large amount of coordinate matching work, this method can avoid the problems caused by a fixed matching step, does not cause data delay when the data volume is extremely large, and does not cause inaccurate search, slow speed and other situations due to excessively large step length when the data volume is extremely small. Accurate coordinate matching makes the generated VR drilling scene more realistic, and personnel participating in the drilling can train in a nearly real scene, better master emergency measures, and improve emergency capabilities. For the entire mine emergency response team, a high-quality simulation scene can facilitate multiple drills, strengthen cooperation and response capabilities, so that when facing a real mine accident, actions can be taken more quickly and effectively, the harm of the accident is reduced, and the safety of production of the mine and the safety of the workers are ensured.

[0128] Example Seven

[0129] In the technical scheme of the embodiment, a coal mine accident drilling device based on VR is provided, which comprises: a decomposition module, configured to analyze a collected coal mine accident case to obtain a scene script of each coal mine accident case, wherein the scene script comprises a scene evolution path and a plurality of flexible modules, and the flexible modules are configured to be combined into the scene evolution path to generate a VR drilling scene according to drilling instructions; a combination module, configured to obtain drilling instructions of a user, combine the scene evolution path and the flexible modules of the scene script according to the drilling instructions, and generate a VR drilling scene, wherein the VR drilling scene comprises a plurality of time nodes; a selection module, configured to obtain a drilling time node selected by the user from different time nodes, and determine a feedback model corresponding to the drilling time node from a feedback model library, wherein the feedback model library comprises disaster state information of each time node, accident emergency measure information for the disaster state information, and accident emergency measure information implementation behavior corresponding to the accident emergency measure information; and a display module, configured to display a plurality of selected accident emergency measure information to a current user, receive accident emergency measure information selected by the current user, and interact with the feedback model based on the accident emergency measure information selected by the current user, to realize interactive operation of the accident emergency measure information implementation behavior in the VR drilling scene under the influence of the current accident emergency measure information.

[0130] The technical problem to be solved by the embodiment is how to efficiently and economically conduct mine accident safety training. In the technical scheme of the embodiment, first, a collected coal mine accident case is analyzed, and the scene analysis of the embodiment is to sort out a scene script of each coal mine accident case, and the scene script covers a scene evolution path and flexible modules. Then, drilling instructions of a user are obtained, the drilling instructions include location selection instructions, rescue type selection instructions, personnel type selection instructions, disaster type selection instructions, and the like, and the scene evolution path and the flexible modules in the scene script are reasonably combined according to the instructions, thereby generating a VR drilling scene with a plurality of time nodes. Then, a drilling time node selected by the user from different time nodes is obtained, and a feedback model corresponding to the drilling time node is determined from a feedback model library comprising disaster state information of each time node, accident emergency measure information for the disaster state information, and accident emergency measure information implementation behavior corresponding to the accident emergency measure information. Then, a plurality of selected accident emergency measure information is displayed to a current user, and the selection of the user is received. Finally, the feedback model is interacted with based on the accident emergency measure information selected by the user, to realize interactive operation of the accident emergency measure information implementation behavior in the VR drilling scene. For example, in the simulation of a coal mine gas explosion accident, through such a process, a scene meeting the actual training needs can be combined according to different instructions, so that training no longer depends on high-cost and risky real environment drilling.

[0131] The technical scheme of the embodiment can simulate diversified disaster situations through the series of operations, and solves the problems of low reality, difficulty in simulating real disaster situations, high cost, safety risks and the like in previous mine safety training. With the aid of the VR technology, the rescue team members can repeatedly practice in a risk-free environment, such as simulating rescue operations at different stages after a gas explosion for multiple times, familiarize with measures to be taken under various conditions, strengthen decision-making and operation skills, and reduce injury and accident risks in real training. Meanwhile, for the mine workers, the realistic simulation environment and specific operation training can significantly improve their emergency and self-help abilities when encountering an emergency disaster, improve the cooperation and response abilities of the entire mine emergency response team, effectively reduce the threat of accidents to the life safety of the workers, reduce the influence range and harm degree of the disaster, and improve the overall safety management level of the mine, so that the mine accident safety training becomes efficient and economical.

[0132] Example Eight

[0133] In the technical scheme of the embodiment, a computer device is provided, which includes a memory, a processor and a computer program stored on the memory, and the processor executes the computer program to implement the steps of the VR-based coal mine accident drilling method of any one of the above embodiments.

[0134] In the technical scheme of the embodiment, a computer readable storage medium is provided, which stores a computer program, and the computer program is executed by a processor to implement the steps of the VR-based coal mine accident drilling method of any one of the above embodiments.

[0135] In the technical scheme of the embodiment, a computer program product is provided, which includes a computer program, and the computer program is executed by a processor to implement the steps of the VR-based coal mine accident drilling method of any one of the above embodiments.

[0136] The processor can include, but is not limited to, for example, one or more processors or microprocessors, etc. Each processor can be an Application Specific Integrated Circuit (ASIC), a Digital Signal Processor (DSP), a Digital Signal Processing Device (DSPD), a Programmable Logic Device (PLD), a Field Programmable Gate Array (FPGA), a controller, a microcontroller, a microprocessor, or other electronic elements implemented for performing the methods in the above-described embodiments. The computer-readable storage medium can be implemented by any type of volatile or non-volatile storage devices or a combination thereof, and can include, but is not limited to, for example, a Random Access Memory (RAM), a Read-Only Memory (ROM), a flash memory, an EPROM memory, an EEPROM memory, a register, a computer storage medium (such as a hard disk, a floppy disk, a solid state disk, a removable disk, a CD-ROM, a DVD-ROM, a Blu-ray disk, etc.).

[0137] The computer-readable storage medium can also store at least one computer-executable program / instruction, such as computer-readable instructions. The computer-readable storage medium includes, but is not limited to, for example, a volatile memory and / or a non-volatile memory. The volatile memory can include, for example, a Random Access Memory (RAM) and / or a cache, etc. The computer-readable storage medium can include, for example, a Read-Only Memory (ROM), a hard disk, a flash memory, etc. For example, the non-transitory computer-readable storage medium can be connected to a computing device, such as a computer, and then, in the case where the computing device runs the computer-readable instructions stored on the computer-readable storage medium, the various methods as described above can be performed.

[0138] In addition, the computer device can also include, but is not limited to, a data bus, an Input / Output (I / O) bus, a display, and an input / output device (for example, a keyboard, a mouse, a speaker, etc.), etc. The processor can communicate with external devices through the I / O bus via a wired or wireless network. In one embodiment, the at least one computer-executable instruction can also be compiled or composed into a software product / computer program product, wherein one or more computer-executable instructions are executed by the processor to perform the steps of the various functions and / or methods in the embodiments described in the present technology.

[0139] Example Nine

[0140] Based on the above-mentioned embodiments, the present embodiment provides an application example.

[0141] The application example provides a coal mine accident emergency drill method based on virtual reality application. The present application relates to the field of coal mine accident emergency rescue drill, and particularly relates to a coal mine accident emergency drill method based on virtual reality application and a storage medium.

[0142] Production safety is crucial in the field of coal mining, and safety training for coal mining is a necessary measure to prevent large-scale accidents.

[0143] In the field of mine safety training, related technologies mainly rely on traditional teacher lectures, video learning, and on-site simulation drills. Although these methods have improved the staff's understanding of safety knowledge to some extent, they have obvious shortcomings in simulating real complex accident scenes and practicing emergency response measures (mine rescue teams have daily rescue drills, but the scene is not realistic, or it is difficult to simulate real disaster situations). Related training methods cannot fully cover various potential accident scenarios, and lack of interactivity and immersion, resulting in poor learning effectiveness and inability to effectively improve the actual emergency response capabilities of staff.

[0144] In addition, due to the particularity and danger of the mine environment, related training methods also face challenges in safety and economy. Simulation training in real environments not only has high costs, but also has safety risks. There is a technical problem in the field of how to efficiently and economically conduct mine accident safety training. Therefore, there is an urgent need in the field for a new training method that can efficiently and comprehensively simulate various mine accident scenarios while ensuring the safety of the training process.

[0145] The present application embodiment proposes a typical coal mine accident scene construction method and system for "quick and accurate construction of typical coal mine accident scenes", specifically, the present application implements the technical solution in the present application according to the "modularization" idea, first extracts scene elements based on collecting coal mine gas explosion accident cases from annual coal mine accident reports, and determines the scene evolution rule and scene evolution path of the scene elements, combines the scene evolution path, adds flexible modules and fixed modules in accordance with the time sequence of the accident occurrence, organically combines the modules, and finally forms a scene script to guide emergency drills and emergency decision-making.

[0146] Further, the scenario script is set and classified, according to the cause class, location class and evolution class of the scenario elements, the granular content such as personnel and equipment of each class of scenario setting is set to form a dynamic scenario node to constitute a scenario evolution path; through the general level model of scenario expression, based on the multi-level scenario structure of "dimension-element-attribute", the scenario unit is analyzed layer by layer to design the module for the prominent accident script.

[0147] The time continuity, action integrity and uniformity of the trigger end condition of each module are comprehensively considered, and the initial action and end action of the scenario of each module are set taking the gas explosion emergency as an example.

[0148] Further, as shown in Figure 1 , the above module setting and combination are comprehensively considered, the fixed module is added, and the whole scenario is formed for module combination by selecting the cause element, selecting the accident location element (including equipment, personnel and distress situation), selecting the accident stage element (determining the accident size and emergency object), selecting the accident consequence element, and adding the fixed module in turn; in the subsequent operation, the granular elements in each module are combined and used to reduce the work task, realize the flexible combination of scenario elements, and achieve the flexible module replacement.

[0149] In the specific operation process, the cause element is selected: different accidents have different causes, and the same accident cause also exists differently. The gas explosion accident is selected by combining the gas accumulation (poor ventilation, abnormal outburst) and the fire source (human open fire, electric spark, open fire of blasting, friction spark, coal spontaneous combustion) two by two.

[0150] The accident location element (including equipment, personnel and distress situation) is selected: the accident location is selected from the preset 8 types of locations, and the various equipment, personnel and distress situation of the personnel after the accident are determined.

[0151] The accident stage element (determining the accident size and emergency object) is selected: the accident size is determined, and the internal forces of the mine party are used for emergency in small range accidents, and the external rescue forces are used for emergency in large range accidents.

[0152] The accident consequence element is selected: the accident consequence is selected from the high-temperature related poisoning and burn, asphyxia caused by damage of ventilation facilities, and personnel trapped caused by impact damage of support, the degree (injury and loss degree) of the accident consequence is different for different accident sizes, and the emergency disposal method is also different.

[0153] The fixed module is added: the fixed modules (alarm, start preplan, alarm receiving, alarm response, in-well investigation, etc.) of each stage are added to the disposal process.

[0154] The whole scenario is formed: the fixed module and the flexible module of the accident emergency disposal are combined to form the whole scenario.

[0155] Further, when flexible module replacement is performed, the particle elements in each module can also be used in combination to reduce the work task. For example, in the initial scene, the scene element state is changed to obtain a hidden danger scene; the hidden danger state in the hidden danger scene can evolve into an accident state; in addition, the environment of the on-site disposal stage adds an emergency rescue information prompt state element. When transitioning to the external rescue stage, the environment state element of the on-site disposal stage can be directly used, the accident state element is changed, and a new scene is formed; in summary, through state evolution, element addition, element reuse, and the like, the reuse of each particle element in the scene module can be obtained, and the workload of scene construction is greatly simplified.

[0156] Taking a coal mining face as an example, the process of forming a script by module combination is described:

[0157] 1) Selecting a cause element

[0158] Selecting gas accumulation a21 caused by poor ventilation;

[0159] Selecting a fire source a22 caused by coal spontaneous combustion: in the goaf, the coal exposed to air, due to inadequate fire prevention measures, slowly oxidizes and heats up, produces carbon monoxide and other toxic and harmful gases, and smokes until a visible fire appears.

[0160] 2) Selecting an accident location element

[0161] Selecting a location a13: fully mechanized caving face;

[0162] Selecting equipment a12: coal mining equipment: coal mining machine, scraper conveyor, hydraulic support;

[0163] Selecting personnel a11: coal mining machine driver, scraper conveyor driver, shield support installer, and shield support dismantler.

[0164] 3) Selecting an accident stage element

[0165] Local gas explosion B: the scope of influence is small, and the internal forces of the mine should be disposed in an emergency.

[0166] Scene: at the tail of the face conveyor, the fire source is floating coal beside the tail, the CO concentration is 30 ppm at 5 meters upwind of the fire source, the temperature is 30°C, the oxygen concentration is 20.1%, and the fire is not large;

[0167] Emergency disposal: the rescue team member holds the dry powder fire extinguisher carried by the rescue team member up and down several times, pulls out the safety pin, holds the spray pipe with one hand, aims at the root of the flame, holds the pressure handle with the other hand, and presses the pressure handle, and the dry powder is sprayed out from the spray pipe to extinguish the fire. The rescue team member uses the fire extinguisher to extinguish the open fire, and then connects the dust removal pipeline and opens the high-pressure water pump to spray high-pressure water on the upper part of the support;

[0168] Large gas explosion C: out of control, with external professional rescue team as emergency force;

[0169] Scenario: the open fire has been extinguished, but the smoke is still very large, and the fire source cannot be completely eliminated. The methane in the scene is rising, and there is a risk of explosion at any time;

[0170] Emergency disposal C4: immediately organize forces to drill from the ground to the upper corner of the working face, and inject CO2 and gel to the possible fire point through the drill hole; send 2 professional rescue teams to close the air and machine lanes of the working face at the same time; before closing the air and machine lanes, first send someone to lay another nitrogen injection pipe from the working face machine lane (the pipe is laid inside the fireproof door), inject nitrogen from the underground to the working face, speed up the inerting of the working face, and prevent the spread of fire outside and speed up the extinguishing.

[0171] 4) Select accident consequence elements

[0172] The accident caused poisoning of personnel.

[0173] Scenario: the injured person is unconscious, and the breathing and pulse are weak.

[0174] The accident caused people to be trapped.

[0175] Scenario: the gas explosion caused the destruction of the roadway, producing a large amount of toxic and harmful gas, the walking route was long, the self-rescue protection time was limited, and it could not be safely evacuated from the disaster area.

[0176] 5) Add fixed modules

[0177] On-site disaster avoidance B1, information reporting B2, pre-plan starting module B3, first alarm B4, police B5, well entry reconnaissance B6, emergency disposal C4, police module C5, well entry reconnaissance C6 are fixed modules, which are added to the accident emergency disposal process in order; for example:

[0178] Alarm: the monitor XXX reports to the dispatcher: the methane sensor at the upper corner of 2421 fully mechanized working face alarms, and the methane concentration rises from 0.5% to 1%;

[0179] Pre-plan starting: the on-duty mine manager realizes the urgency of the situation, so he instructs the dispatcher: ① immediately report the situation to the mine manager; ② start the emergency plan and notify the emergency plan stipulated emergency command members to gather at the dispatch center as soon as possible, especially the ventilation team leader and the part-time rescue team leader to arrive immediately;

[0180] Alarm: the part-time team telephone duty officer XXX answers the phone and records the accident content, and immediately sounds the alarm;

[0181] Police: the part-time mine rescue team is assembled, wearing oxygen respirators, and carrying the required rescue equipment to the wellhead;

[0182] Entry reconnaissance: Establish a ground rescue base, the platoon leader in the command center, others on standby at the base;

[0183] The captain of the follow-up team reports to the dispatch room: an explosion may have occurred at the 2421 fully mechanized coal face, and the belt conveyor head has a shock;

[0184] The rescue commander convenes the members of the rescue command center to analyze the disaster situation and develop a rescue plan;

[0185] The rescue team enters the mine along the auxiliary inclined shaft, meets with the general commander at the transportation roadway and the underground base, establishes an underground rescue base at the lower part of the track, and then carries out disaster area reconnaissance according to the division of labor, the main task is to rescue the trapped and injured personnel, and immediately extinguish the fire source;

[0186] The two team leaders organize their own teams to conduct pre-war inspections (including self-inspections and mutual inspections);

[0187] After the three rescue teams arrive at the lower part of the track, they meet with the captain of the team and choose a suitable location as the rescue base.

[0188] 6) Form an overall scene

[0189] Flexible module:

[0190] Initial scene A1 (location a13, equipment a12, personnel a11), initial scene A2 (gas accumulation a21, fire source a22); mine internal disposal B (disaster area reconnaissance b64, emergency disposal B7); mine external rescue C (disaster area reconnaissance c34, emergency disposal C4).

[0191] Fixed module:

[0192] Mine internal disposal B (on-site disaster avoidance B1, information reporting B2, pre-plan activation module B3, first alarm B4, police B5, entry reconnaissance B6); mine external rescue C (second alarm C1, second police C2, second entry reconnaissance C3); emergency end stage D (personnel hoisting D1, rescue team procedures D2); flexible module and fixed module are combined according to the disposal order to form a gas explosion emergency rescue scene script;

[0193] As Figure 2 shown, in the process of coal mine accident emergency, the overall process will include: accident incubation stage, mine internal disposal stage, mine external rescue stage and emergency end stage;

[0194] Specifically, the accident incubation stage includes two modules of initial scene and hidden danger scene. The initial scene includes two fixed modules of site personnel and equipment and a flexible module of site; the hidden danger scene includes two flexible modules of gas and fire source; at this time, the accident has occurred, normal production and operation, life and property and personal safety have been affected, and emergency measures need to be taken, which is the critical point of emergency (referred to as "emergency point");

[0195] The internal disposal stage of the mine includes seven modules of site evacuation, information reporting, starting of plan, receiving alarm, going to the site, investigation into the mine and emergency disposal. At this time, the scale of the accident has exceeded the capacity of internal disposal of the enterprise, and external rescue forces are needed for rescue, which is the critical point of accident rescue (referred to as "rescue point");

[0196] The external rescue stage of the mine includes four modules of receiving alarm, going to the site, investigation into the mine and emergency disposal;

[0197] The emergency end stage includes two modules of personnel ascending to the surface and return of the rescue team;

[0198] Among them, the accident incubation stage includes three flexible modules of site, gas and fire source, and there are 7 kinds of sites, 2 kinds of gas hazards and 5 kinds of fire sources, so that 20 kinds of early scene combinations can be obtained, as shown in Figure 3 .

[0199] The internal disposal of the mine: the disaster occurs initially, and the mine itself disposes according to the sequence of the modules of site evacuation, information reporting, starting of plan, receiving alarm, going to the site, investigation into the mine and emergency disposal, wherein the modules of disaster area investigation and emergency disposal are flexible modules and need to be replaced according to the actual situation of the accident; if the internal disposal of the mine eliminates the disaster, the emergency ends;

[0200] The external rescue of the mine: if the internal disposal of the mine cannot control the disaster, the disaster expands, at this time, external forces need to be involved in rescue, and the rescue team needs to dispose according to the sequence of the modules of receiving alarm, going to the site, investigation into the mine and emergency disposal, wherein the modules of disaster area investigation and emergency disposal are flexible modules and need to be replaced according to the actual situation of the accident; the disaster is eliminated, and the emergency ends;

[0201] Further, as shown in Figure 4 , a scene script is generated according to the emergency point, and then a VR scene is generated based on the scene script.

[0202] According to the above coal mine accident emergency process, the application provides a coal mine accident emergency drilling method based on virtual reality application. Based on the above coal mine accident emergency process, a scene script is formed by module division and combination through a coal mine accident case, and then a VR drilling scene is generated based on the scene script; the coal mine accident emergency drilling is carried out based on the VR drilling scene, so as to improve the cooperation and response ability of the whole mine emergency response team.

[0203] As Figure 5 shown, the application proposes a coal mine accident emergency drill method based on virtual reality application, comprising the following operation steps:

[0204] Step S10: Collecting a plurality of coal mine accident cases; analyzing the coal mine accident cases to generate a plurality of scenario scripts.

[0205] The above scenario script itself is a text and image formed by a plurality of coal mine accident cases; the scenario script includes disaster state information and disaster state change information of the coal mine accident scene on the time axis; the scenario script is essentially some scene images in the collected case of the coal mine accident scene and the accident emergency measures images in the collected case of the coal mine accident scene and some text information of the accident emergency measures, etc.

[0206] It should be noted that the above embodiment of the application uses the Python-based jieba word segmentation package to perform text analysis and image extraction on the coal mine accident case data, and extracts the scenario script (i.e. extracts the corresponding image and text), so the scenario script not only contains some disaster state information of the case, but also includes some accident emergency measure information (more accident emergency measure information forms a measure database).

[0207] Specifically, the acquisition of the above scenario script is to use a "dimension-element-attribute" scenario expression hierarchical architecture, to interactively integrate the accident evolution process and the emergency rescue process, to construct a "full process, full scenario, full element" accident scenario; based on a large amount of historical accident data, using Chinese word segmentation technology, according to the accident mechanism, a quantifiable basic accident (evolution) scenario library is constructed, and a matching emergency rescue measure library is established; further, an emergency drill model based on dynamic scenarios is established, the accident rescue process is standardized and decomposed according to the rescue process, multi-scenario path construction is performed with the aid of dynamic Bayesian network technology, emergency decision-making is optimized through scenario path matching; finally, according to the established multi-scenario path, a typical accident drill scene is selected for assembly, combined with the emergency rescue measure library, personnel roles and action points are set, and then the scenario script is prepared.

[0208] For example, some accident emergency measure information in the scenario script; alarm emergency measure information: the monitor reports to the dispatcher including methane sensor alarm, methane concentration rising from 0.5% to 1%.

[0209] Pre-arranged emergency measure information: information on answering the call: the on-duty staff answers the phone and records the accident content, and immediately sounds the accident alarm;

[0210] Police behavior information: the part-time mine rescue team assembles, wears an oxygen respirator, carries the required rescue equipment and sets off to the shaft mouth;

[0211] The behavior information of the reconnaissance into the well is that a ground rescue base is set up, and the team leader is in the command department, and the others are on standby at the base;

[0212] The behavior information of the rescue team is that the rescue team meets the general commander of the underground base at the lower part of the track on the shaft, and then sets up an underground rescue base, and then carries out disaster area reconnaissance work according to the division of work, and the main task is to rescue the personnel in distress and disaster, and to find the fire source and extinguish it immediately.

[0213] In the specific rescue process, from the perspective of the rescue team personnel, according to the script content architecture setting, the alarm, the police, the establishment of the rescue base on the well and the like can be presented in a fixed form.

[0214] For example, some disaster state information in the scene script can be a mine collapse accident state image and text, and the like, and will not be described here.

[0215] Step S20: Obtain the drill instruction of the current training personnel (i.e. the current user), and generate a VR drill scene according to the drill instruction and the scene script.

[0216] The drill instruction includes: a location selection instruction, a rescue type selection instruction, a personnel type selection instruction, and a disaster type selection instruction.

[0217] It should be noted that the drill location, the drill rescue type, the drill personnel type, and the drill disaster type contained in the drill scene data information set obtained through the scene script above; the deduction perspective adopts a combination of macroscopic perspective full-process deduction and on-site operation deduction. The main types of accidents are gas accidents, fire accidents, and roof accidents.

[0218] The above embodiment of the present application obtains the drill instruction of the current user, selects the drill location, the rescue type, the drill personnel type, and the disaster type of the drill scene, and generates a VR drill scene through virtual reality technology in combination with the scene script. The coal mine accident emergency rescue drill operation is carried out based on the VR drill scene. The virtual reality technology generates a VR drill scene, which is a simulated environment simulated by computer technology, allowing users to immerse themselves in it and interact with it. Virtual reality technology usually uses a head-mounted display or other VR devices to bring users into a completely computer-generated virtual world, making users feel as if they are in a real environment. After generating a VR drill scene, the drill personnel wears a head display device and a handle to interact with the VR drill scene through the head display device and the handle, thereby realizing the drill operation.

[0219] The above scene script includes a fixed module and a flexible module.

[0220] The fixed module includes a mine interior treatment module B, a mine exterior rescue module C, and an emergency end stage module D.

[0221] The flexible module includes a first initial scene A1, a second initial scene A2, the mine interior treatment B, and the mine exterior rescue C.

[0222] The mine interior treatment module B includes a disaster avoidance B1, an information report B2, a preplan start B3, a first alarm receiving B4, a first alarm leaving B5, and a mine entry reconnaissance B6.

[0223] The mine exterior rescue module C includes a second alarm receiving C1, a second alarm leaving C2, and a second mine entry reconnaissance C3.

[0224] The emergency end stage module D includes a personnel lifting D1 and an ambulance team procedure D2.

[0225] The first initial scene module A1 includes a location a13, equipment a12, and personnel a11.

[0226] The second initial scene module A2 includes a gas accumulation a21 and a fire source a22.

[0227] The mine interior treatment module B includes a disaster area reconnaissance b64 and an emergency treatment B7.

[0228] The mine exterior rescue C includes a disaster area reconnaissance c34 and an emergency treatment C4.

[0229] The embodiment of the present application extracts scenario elements by using a Python-based jieba word segmentation package for text analysis of coal mine accident case data, and constructs a scenario library and a measure library based on the scenario elements; the scenario evolution law and the scenario evolution path are constructed by using a dynamic Bayesian network model and a hidden Markov model through the scenario library and the measure library; after the scenario path is constructed, each scenario is decomposed and classified by hierarchical clustering to analyze the reusability and independence of each scenario element and module, and then each scenario fixed module and element are obtained, and combined with the scenario evolution path, each flexible and fixed module is added in time sequence of accident occurrence, so that the modules are organically combined, and finally a scenario script is formed to guide emergency drills and emergency decision-making.

[0230] The specific application of each fixed module is as follows:

[0231] Alarm: The monitor reports to the dispatcher that the methane sensor at the upper corner of the fully mechanized caving face alarms, and the methane concentration rises from 0.5% to 1%;

[0232] Pre-arranged start: the on-duty mine director realizes the urgency of the situation, and then instructs the dispatcher: ① immediately report the situation to the mine director; ② start the emergency plan and notify the emergency command center members specified in the emergency plan to gather at the dispatch center as soon as possible, especially the ventilation team leader and the part-time rescue team leader to arrive immediately;

[0233] Police call: the part-time team telephone duty officer answers the phone and records the accident content, and immediately sounds the alarm;

[0234] Police: the part-time mine rescue team has completed the assembly, wearing oxygen respirators, carrying the required rescue equipment to the shaft;

[0235] In-Well Reconnaissance: Establish a ground rescue base, with the team leader in the command center and the others on standby at the base;

[0236] The fully mechanized team follow-up team leader reports to the dispatch room: an explosion may have occurred on the fully mechanized working face, and there is a shock at the head of the belt conveyor;

[0237] The rescue commander convenes the rescue command center members to analyze the disaster situation and develop a rescue plan;

[0238] The rescue team enters the shaft along the auxiliary inclined shaft and the main transportation roadway, and meets with the underground base commander at the lower part of the track on the mountain, establishing an underground rescue base, and then carrying out disaster area reconnaissance work according to the division of labor, the main task is to rescue the trapped and injured personnel, and to extinguish the fire immediately;

[0239] The two team leaders organize their own teams for pre-war inspection (including self-inspection and mutual inspection);

[0240] After the three rescue teams arrive at the lower part of the track on the mountain, they meet with the on-duty mine director and choose a suitable location as the rescue base.

[0241] In the specific rescue process, in the VR simulation scene constructed later, from the perspective of the rescue team personnel, according to the script content architecture setting, corresponding to the disaster area reconnaissance module and the emergency disposal module, other modules do not need to be changed, that is, the alarm, police call, police, and the establishment of the rescue base on the shaft can be fixed. Fixed modules do not need user interaction, and make a transition animation to connect the scene.

[0242] The above embodiment of the present application obtains a scenario script by analyzing collected coal mine accident cases, and then generates a VR practice scene according to a user's practice instruction to simulate a real complex accident scene. In subsequent operations (steps S30-S40 described below), based on the user wearing a VR device, disaster state information and accident emergency measure information in the selected VR practice scene are simulated to simulate the user's behavior of implementing the accident emergency measure information for the current VR practice scene disaster state information, thereby improving the emergency response capability of the staff for the coal mine accident, and through the immersive practice of VR, the interactivity and the authenticity of the practice are improved, the learning effect is greatly improved, and the safety of the staff in the accident practice can be ensured.

[0243] The VR technology is adopted to produce a mine rescue team VR practice system, which covers three main disasters of coal mine rescue: coal mine fire accident rescue, coal mine gas explosion rescue, and coal mine roof accident rescue. A real complex accident scene is simulated through VR; and in subsequent operations, based on the user wearing a VR device, the occurrence place, danger source and the like in the selected VR practice scene are simulated to simulate the user's behavior of implementing the accident emergency measure information for the current VR practice scene disaster state information, thereby improving the emergency response capability of the staff for the coal mine accident, and through the immersive practice of VR, the interactivity and the authenticity of the practice are improved, the learning effect is improved, and the safety of the staff in the accident practice can be ensured.

[0244] After step S20, a feedback model library in the form of a combination of text and image is constructed by using a simulation practice scene of the VR practice scene; it can be known from the above scheme that the VR practice scene itself is a VR practice scene simulated by a case scenario script and a practice instruction; similarly, the VR practice scene can also be fed back to the current user for demonstration.

[0245] That is, the VR practice scene (i.e. the feedback model library) serves as the feedback model library in the form of a combination of text and image; then different accident emergency measure information prompt modes at a practice time node of the VR practice scene are presented to the current user for selection and viewing; after the current user selects one of the current disaster state information and one of the accident emergency measure information, a generation effect is clicked; then the VR practice scene simulates the VR practice scene after the behavior of implementing the accident emergency measure information according to the selection result, and then displays the VR practice scene after the behavior in the form of image interaction.

[0246] Step S30: forming a feedback model library based on a data set constituted by the VR practice scene at different time nodes.

[0247] The VR practice scene at each time node is equivalent to a demonstration effect of a current time period.

[0248] It should be noted that, as Figure 8 shown, in the embodiment of the present application, a data set formed based on the VR practice system at different time nodes forms a practice scene and a story line; disaster state information at each time node, accident emergency measure information corresponding to the disaster state information, and accident emergency measure information implementation behavior corresponding to the accident emergency measure information are performed in the VR system.

[0249] The above embodiment of the present application constructs a feedback model library based on a data set formed by disaster state information at different time nodes of the current VR practice scene, accident emergency measure information corresponding to the disaster state information, and accident emergency measure information implementation behavior corresponding to the accident emergency measure information (the accident emergency measure implementation behavior is displayed in the form of picture interaction); the above accident emergency measure implementation information and accident emergency measure implementation behavior are the most efficient processing mode for the above disaster state information.

[0250] That is, the above feedback model library represents disaster state information at each time node, accident emergency measure information corresponding to the disaster state information, and accident emergency measure information implementation behavior corresponding to the accident emergency measure information; the above feedback model library can provide emergency reference data when a user performs accident emergency response processing subsequently.

[0251] After the current disaster state information is confirmed, the accident emergency measure information corresponding to the current disaster state information input by the current user is obtained (the current user needs to intervene in the current disaster state information, so an accident emergency measure information is input); for example, the disaster state information specifically refers to a fire, and the accident emergency measure information corresponding to the current disaster state information is input (for example, alarm emergency measure information), and then the corresponding accident emergency measure information implementation behavior at the time node is matched; the corresponding accident emergency measure information implementation behavior is interacted in the form of an image, and specific contents are shown in step S40.

[0252] Step S40: After the current user wears the VR device to enter the VR training scene, the user determines the training time node to be entered from different time nodes; the training time node determines the corresponding VR training scene according to the feedback model library; the VR device sends the current user a plurality of selected emergency measures information (the selected emergency measures information or the selected emergency instruction can be the use of fire extinguisher measures, or the alarm emergency measures information or the alarm measures), receives the selected emergency measures information (the emergency measures information or the emergency instruction) of the current user; based on the emergency measures information and the VR training scene, the interactive operation of the emergency measures information implementation behavior of the VR training scene under the current emergency measures information is realized.

[0253] Specifically, the above-mentioned emergency instruction refers to the real-time emergency instruction issued by the current user according to the real-time disaster state information in the current VR training scene. For example, if the disaster state type in the current VR training scene is fire, the current user may issue an emergency instruction to use a fire extinguisher. The current VR training scene responds to the emergency instruction and performs the emergency operation of using the fire extinguisher (the emergency measures information implementation behavior of the VR training scene is usually displayed in the form of image interaction);

[0254] For example: a mine rescue team member enters a virtual reality training system and wears a VR device. The system generates a VR training scene containing a fire scene based on the coal mine accident case data analyzed in step S10. The scene simulates an emergency rescue scene of a coal mine fire. The VR system displays the following content:

[0255] Disaster state information: a fire has occurred in a working face of a coal mine. The fire source is near the roof above the working face, surrounded by a large amount of coal dust and gas, and the fire is spreading;

[0256] Training instruction information: at this time, the system prompts the user to select different types of rescue measures, such as "alarm", "fire extinguishing" or "evacuation", etc.

[0257] After the user enters the system, the VR system presents the following two selected emergency measures information according to the user's selection:

[0258] Alarm emergency measures: the monitor reports the fire to the dispatcher and starts the fire emergency plan;

[0259] Fire extinguishing emergency measures: select appropriate fire extinguishing equipment (such as dry powder fire extinguisher) for fire extinguishing operation;

[0260] Further, based on the selected emergency measures information of the user, the VR training scene is updated.

[0261] When the user selects the "fire extinguishing emergency measures" information, the system updates the VR simulation scene based on this selection, with the following specific steps:

[0262] User selects fire extinguisher: In the VR scene, the user sees a simulated coal mine working face with a fire source burning. The user uses a handle or controller to select a fire extinguisher (such as a dry powder fire extinguisher);

[0263] Display fire extinguishing operation prompts: The VR system provides instruction information on the screen to guide the user on how to correctly use the fire extinguisher, such as "press the handle to spray dry powder" and "aim the nozzle at the base of the fire source";

[0264] Start fire extinguishing action: The user presses the controller button on the VR device to activate the fire extinguisher spray; at this time, the user sees that the flames are gradually covered by the dry powder sprayed by the fire extinguisher, and the fire source is gradually extinguished;

[0265] Visual feedback: As the fire extinguishing operation progresses, the VR scene shows the gradual extinguishing of the flames, and the fire scene changes, with smoke gradually dissipating and the on-site environment returning to clarity; the user feels the "spray" feedback from the handle controlling the fire extinguisher, enhancing the sense of immersion;

[0266] Further, interaction and result feedback.

[0267] Fire extinguishing effect: As the fire extinguisher operation is completed, the fire source is extinguished, and the system displays the visual effect of successful fire extinguishing, such as the "fire source extinguished" message and the "fire extinguishing successful" prompt;

[0268] Further operation: After the fire source is extinguished, the VR system updates the new situation and prompts the user to continue subsequent emergency rescue work; the system may prompt the user whether to perform "personnel evacuation" or "accident reporting" operations, and the user can choose to continue to practice other emergency measures;

[0269] Finally, feedback and summary.

[0270] After completing the fire extinguishing operation, the VR system generates a simulation feedback, recording each step and selection of the user in the virtual scene;

[0271] The system updates the feedback model library based on the user's actual operation results (such as success or failure of fire extinguishing, operation accuracy, etc.), providing more personalized reference data for future simulations;

[0272] The entire process described above simulates the real coal mine fire emergency handling process through virtual reality devices. The user not only learns emergency handling skills through interactive operation, but also feels the operation effect in real time, thereby improving the emergency response capability;

[0273] The disaster state accident corresponding to the emergency instruction in the above embodiment of the present application isFigure 9 , Figure 10 , Figure 11 as well as Figure 12 As shown, each type of accident can be combined with the hazard source and the accident location to enter the simulation exercise storyline of that accident combination, and the complete performance of that storyline can be realized in VR:

[0274] like Figure 9 The image shows typical scenarios of gas accidents. Coal mine gas disasters refer to accidents such as explosions and combustion caused by the accumulation or release of gas (mainly methane) during coal mining. Five typical scenarios of gas accidents are listed below:

[0275] 1) Poor ventilation + external fire – a total of 3 types;

[0276] 2) Upper corner + spontaneous combustion of coal – 1 type in total;

[0277] 3) Abnormal surge + electrical spark – 1 type in total;

[0278] like Figure 10 The image shows typical fire accident scenarios. There are six typical fire scenarios:

[0279] 1) There are two types: coal + coal spontaneous combustion - electric spark;

[0280] 2) Explosives + artillery blasting (spontaneous combustion) - 1 type;

[0281] 3) There are three types of sparks: electromechanical equipment + open flame, friction and impact sparks, and electric sparks.

[0282] like Figure 11 As shown, typical scenarios of roof collapse accidents are listed.

[0283] 1) One type of problem is the combination of periodic pressure application and poor support parameter design;

[0284] 2) One type of poor design of support parameters combined with mine seismic activity;

[0285] 3) One type of roof collapse + poor support parameter design;

[0286] like Figure 12 As shown, typical scenarios of roof collapse accidents are listed.

[0287] Specifically, in step S20, a VR training scenario is generated based on the training instructions and the scenario script, including the following steps:

[0288] Step S21: obtaining a set of drilling scene data information based on the drilling instruction and the scenario script; the set of drilling scene data information includes drilling site, drilling rescue type, drilling personnel type, drilling disaster type, and scenario script (the scenario script includes accident emergency measure information and disaster state information and disaster state change information of the coal mine accident scene on a time axis, which will not be described here).

[0289] It should be noted that the drilling site, drilling rescue type, drilling personnel type, and drilling disaster type obtained by the above-mentioned drilling instruction (i.e., set by the above-mentioned drilling instruction) and the disaster state information and disaster state change information of the coal mine accident scene on a time axis obtained by the scenario script can provide a data basis for subsequent extraction of scene information.

[0290] Step S22: obtaining scene information based on the set of drilling scene data information; the scene information includes scene environment information in the current drilling site (such as mine wall, mine ground, mine facility, and disaster instance under mine disaster scene).

[0291] It should be noted that the above-mentioned embodiment of the present application extracts scene information containing scene environment information in the current drilling site by analyzing the set of drilling scene information, which provides a rendering basis for subsequent rendering of the VR drilling scene.

[0292] Step S23: obtaining a rendering element corresponding to the scene information; rendering each of the scene information based on the rendering element to obtain a VR drilling scene.

[0293] The rendering element includes an endpoint element, a vertical vector element, a texture mapping element, and a visible surface element.

[0294] It should be noted that the above-mentioned endpoint element is one of the basic constituent units of the VR drilling scene, which usually represents a point in space, is a feature point of the scene environment information in the current drilling site obtained by extracting extreme points or corner features, and can match the initial three-dimensional grid graph in the subsequent rendering process, thereby realizing the establishment of the VR drilling scene.

[0295] For example, the mine wall in the scene environment information; when performing the step of obtaining the rendering element corresponding to the scene information element (i.e., the mine wall), the collapse point (end point element) of the mine wall can be obtained; then, the rendering of the current scene information element is performed according to the rendering element (collapse point), and the VR practice scene is obtained in combination with other rendering elements; for another example, the visible surface element can be an image photo, for example, when performing the step of obtaining the rendering element corresponding to the scene information element (i.e., the mine wall), the wall surface (i.e., the visible surface element, which can be a wall surface 2D picture or 2D map in essence) of the mine wall can be obtained.

[0296] The above-mentioned normal vector element is a vector tangent to the surface geometry of the VR practice scene (obtained by the camera or the mobile robot), which is perpendicular to the surface of the scene information element. It is used to calculate the shadow effect in the process of lighting and rendering to ensure that the light is correctly reflected and refracted, so that the object looks more real.

[0297] The above-mentioned texture mapping element is the process of two-dimensional texture mapping to the surface of the VR practice scene. In the texture mapping element, each texture has a corresponding mapping coordinate to determine the position in the texture image. Through the texture mapping element, the texture can be accurately pasted on the model surface to achieve the visual effect.

[0298] The above-mentioned visible surface element is the visible surface of the VR practice scene (which can also be called a surface image, which is the surface image of each scene environment information at the current practice scene location obtained by the camera or the mobile robot). For example, a triangle or a quadrilateral can be a face. The connection mode of the face and the arrangement order of the vertex affect the appearance and performance of the model.

[0299] Specifically, as shown in Figure 6 In step S25, each scene information element is rendered according to the rendering element to obtain a VR practice scene, including the following operation steps:

[0300] Step S251: An index table is constructed according to the rendering element, which represents the vector of the end point element, the vector of the normal vector element, the vector of the texture mapping element, and the vector of the visible surface element.

[0301] It should be noted that the above-mentioned index table is first vectorized by the rendering element, and then the index table of the rendering element coordinate, the rendering element, and the rendering element vector is constructed according to the coordinate of each rendering element.

[0302] Step S252: Each scene information element is integrated according to the index table to obtain an initial three-dimensional mesh graph.

[0303] Step S253: according to each of the rendering elements, the initial three-dimensional grid map is valued, and a VR practice scene is obtained (read the value of the rendering element, value each point of the three-dimensional grid map, the above value refers to using the value of the rendering element to value the corresponding position of the initial three-dimensional grid map, for example, the current rendering element is a wall surface 2D picture of a visible surface element, and then the RGB value or brightness value of the pixel of the wall surface 2D picture is used to value the position corresponding to the current cave wall image on the initial three-dimensional grid map, so that the VR practice scene can be obtained).

[0304] It should be noted that the above embodiment of the present application values the initial three-dimensional grid map through each rendering element in the index table, thereby obtaining the VR practice scene.

[0305] The above embodiment of the present application first integrates the initial three-dimensional grid map according to the index table after the construction of the index table, which can simplify the operation of the VR practice scene, and further values the initial three-dimensional grid map through the rendering element, thereby obtaining the complete VR practice scene.

[0306] Specifically, in step S252, each of the scene pheromones is integrated according to the index table to obtain an initial three-dimensional grid map, including the following operation steps:

[0307] Step S2521: initially, a first initial three-dimensional grid map is established;

[0308] It should be noted that the above first initial three-dimensional grid map is an initial empty three-dimensional grid map containing only a grid, and in subsequent operations, the three-dimensional grid map is filled using the index table to obtain the initial three-dimensional grid map.

[0309] Step S2522: according to the coordinates M of the scene pheromones corresponding to the index table, the real coordinates M1 of the scene pheromones are obtained through the reality conversion matrix F1;

[0310] M1=MxF1;

[0311] It should be noted that the above embodiment of the present application converts the coordinates M (two-dimensional image coordinates) of each rendering element of the scene pheromones (such as a wall surface 2D picture) into real coordinates M1 (three-dimensional world coordinates, i.e. wall surface 3D coordinates) through the reality conversion matrix F1.

[0312] Step S2523: based on the real coordinates M1, the display coordinates M2 are obtained through the display conversion matrix F2;

[0313] M2=M1xF2;

[0314] It should be noted that the above embodiment of this application converts the real coordinates M1 (such as the wall 3D coordinates obtained above) into the 3D coordinates in the initial 3D scene (i.e., the VR training scene) through the display transformation matrix F2.

[0315] Step S2524: Based on the displayed coordinates M2, perform coordinate transformation using the mapping transformation matrix F3 to obtain the mapped coordinates M3;

[0316] M3 = F3 × M2;

[0317] It should be noted that the above embodiment of this application transforms the three-dimensional coordinates of the initial three-dimensional scene to the mapped coordinates M3 displayed on the VR glasses through the mapping transformation matrix F3. That is, it maps the 3D coordinates of the wall surface on the three-dimensional virtual scene to the mapped coordinates on the VR glasses worn by the trainees.

[0318] Step S2525: Match the mapped coordinates with each virtual coordinate in the first initial three-dimensional mesh map to obtain the initial three-dimensional mesh map.

[0319] It should be noted that in the above embodiments of this application, the mapped coordinates on the VR glasses are matched with the coordinates of each empty value in the first initial three-dimensional mesh diagram, so as to obtain an initial three-dimensional mesh diagram that can be viewed by any VR glasses.

[0320] In the above-described embodiments of this application, the coordinates of the scene pheromone are transformed into the coordinates of the three-dimensional mesh by sequentially performing the coordinate transformation through the real transformation matrix F1, the display transformation matrix F2, and the mapping transformation matrix F3, thereby transforming the coordinates of the scene pheromone into the mapped coordinates step by step; in a further operation, the initial three-dimensional mesh is obtained by matching and mapping the mapped coordinates with the virtual coordinates of the empty three-dimensional mesh of the first initial three-dimensional mesh.

[0321] Specifically, such as Figure 7 As shown, in step S2525, the mapped coordinates are matched with each virtual coordinate in the first initial 3D mesh map to obtain the initial 3D mesh map, including the following operation steps:

[0322] Step S25251: Calculate and obtain the correlation relationship between each of the mapped coordinates;

[0323] It should be noted that the correlation relationship in the embodiments of the present application can use the Euclidean distance or Manhattan distance; by first calculating the correlation relationship between each mapping coordinate, a reference basis can be provided for subsequent matching, that is, after the mapping coordinates are matched to virtual coordinates, the newly generated correlation relationship of each matched mapping coordinate needs to be confirmed, so as to confirm whether the matching result is accurate; when the matching result is not accurate, it proves that the matching relationship is wrong, and then re-iterative matching is needed.

[0324] Step S25252: initialize the iteration parameters, including the first matching coordinates (matching positions, that is, the positions of the virtual coordinates matched to the first initial three-dimensional grid map) of each mapping coordinate, the number of mapping coordinates, the initial matching positions of each mapping coordinate, and the matching step length;

[0325] Step S25253: traverse each mapping coordinate, calculate and obtain the first correlation relationship between each mapping coordinate under the current first matching coordinate; determine whether the first correlation relationship is a preset multiple of the correlation relationship; if yes, output the first matching coordinate as the final target matching coordinate, map the corresponding scene pheromone of each mapping coordinate to the target matching coordinate according to the target matching coordinate, and obtain an initial three-dimensional grid map; if not, update the matching step length according to the matching degree (the matching degree can be calculated using the mean square error) between the first correlation relationship and the correlation relationship, and obtain a new matching step length;

[0326] Step S25254: take the new matching step length as the matching step length and return to the above operation until the first correlation relationship is the preset multiple of the correlation relationship, and output an initial three-dimensional grid map.

[0327] It should be noted that the embodiments of the present application first calculate the correlation relationship of each mapping coordinate at the initial time, and then in the iterative matching process, the correlation between the management relationship of each matching coordinate and the initially obtained correlation relationship is determined to determine whether the matching coordinate is accurate; when the matching coordinate is determined to be inaccurate, the matching step length is updated to re-iteratively search until the correct matching coordinate is obtained, thereby improving the accuracy of matching.

[0328] It should be noted that the above embodiment of the application first determines the correlation between each mapping coordinate, and then initializes the iteration parameters for iteratively searching the matching coordinates corresponding to each mapping coordinate, including the first matching coordinate of the mapping coordinate (the matching position is the position of the virtual coordinate after matching to the first initial three-dimensional grid map), the number of mapping coordinates, the initial matching position of each mapping coordinate, and the matching step size. In the further operation, the first correlation between each matching coordinate is calculated when each mapping coordinate is in the first matching coordinate, and then the relationship between the first correlation and the initial correlation is determined to obtain the final target matching coordinate. This is because the correlation between each mapping coordinate is fixed at the initial time (that is, on the above wall surface 2D image), and when each mapping coordinate is matched into the first initial three-dimensional grid map, if the first correlation between each matching coordinate changes, it means that the scene environment information constituted by these matching coordinates has been distorted or changed, which means that the current first initial three-dimensional grid map is greatly different from the initial scene. Therefore, using the relationship between the first correlation and the initial correlation to determine whether the matching coordinate is correct can make the scene environment information in the initial three-dimensional grid map more consistent and similar to the original scene environment information, so that the final VR simulation scene is more consistent with the actual situation.

[0329] In the specific implementation process of the above embodiment of the application, the technical personnel found that when the matching coordinate is incorrect, if a fixed matching step size is used to update and search for a new matching coordinate in the case of a large amount of data, it will cause a large data delay, and in the case of a small amount of data, using a fixed matching step size to update and search for a new matching coordinate will also use a too large matching step size to search, which will also make the search of the target matching coordinate not accurate and not fast enough. Therefore, the update of the matching step size also needs to be further judged and calculated to search for the target matching coordinate of the mapping coordinate more quickly and accurately.

[0330] Specifically, in step S25253, the matching step size is updated according to the matching degree between the first correlation and the correlation to obtain a new matching step size, including the following operation steps:

[0331] Step S252531: The gradient value of the current matching degree is calculated according to the preset gradient loss function.

[0332] Step S252532: The new matching step size is calculated according to the gradient value.

[0333] Bnew=B-λ×▽f(x); λ is the learning rate; B is the matching step size; Bnew is the new matching step size.

[0334] It should be noted that the new matching step length is calculated by matching the gradient value of the application, which can be selected more in the virtual coordinate, and the gradient value is judged to select a larger matching step length to search for the matching coordinate, and the learning rate and other hyperparameters are reasonably selected to further improve the efficiency of the algorithm, so as to obtain the optimal solution (i.e. the matching coordinates corresponding to the mapping coordinates) faster.

[0335] In summary, the coal mine accident emergency drilling method and storage medium based on virtual reality application proposed in the present application are analyzed by collecting coal mine accident cases to obtain a scenario script, and then generate a VR drilling scene according to the user's drilling instruction to simulate a real complex accident scene; and in subsequent operations, based on the user wearing VR equipment, the disaster state information and accident emergency measure information in the selected VR drilling scene are simulated to simulate the user's disaster state information and accident emergency measure information implementation behavior for the current VR drilling scene, thereby improving the emergency response capability of the staff for the coal mine accident, and through the immersive drilling of VR, the interactivity and the authenticity of the drilling are improved, the learning effect is greatly improved, and the safety of the staff accident drilling can be guaranteed;

[0336] Further, in the process of generating the VR drilling scene, the drilling site, drilling rescue type, drilling personnel type and drilling disaster type obtained by selecting the site selection instruction, rescue type selection instruction, personnel type selection instruction and disaster type selection instruction of the drilling instruction, and the disaster state information and disaster state change information of the coal mine accident scene on the time axis obtained by the scenario script are extracted as scene information elements, and then an index table is established according to the rendering elements of the scene information elements, and the coordinates of the scene information elements corresponding to the index table are converted by a reality conversion matrix, a display conversion matrix and a mapping conversion matrix to obtain mapping coordinates of the scene information elements; further, based on the mapping coordinates of the scene information elements, the search step length dynamically transformed by the gradient descent algorithm is used to search the coordinates in the initially empty first initial three-dimensional grid graph, and then the initial three-dimensional grid graph is obtained; finally, the corresponding positions of the initial three-dimensional grid graph are valued by using the values of the rendering elements, and the VR drilling scene is obtained.

[0337] In the VR training scene, users can simulate various disaster situations, simulate training situations that cannot be simulated in daily exercises, rescue team members can repeatedly practice in a risk-free environment, and decision-making and operation skills can be enhanced to reduce the risk of injury and accidents in real training; Through the realistic simulation environment and specific operation training, the emergency, self-rescue ability of the mine workers when encountering an emergency disaster, and the rescue ability and rescue efficiency of the rescue team can be significantly improved, the cooperation and response ability of the entire mine emergency response team can be improved, thereby effectively reducing the threat to the life safety of the workers in the accident, effectively reducing the influence range and harm degree of the disaster, and improving the overall safety management level of the mine.

[0338] In the embodiments of the present application, it should be understood that the disclosed device and method can also be implemented in other ways. The device embodiments described above are only illustrative, for example, the flowcharts and block diagrams in the drawings show the possible implementation architecture, function and operation of the device, method and computer program product according to the embodiments of the present application. In this regard, each block in the flowchart or block diagram can represent a module, a program segment or a part of code, which contains one or more executable instructions for implementing the specified logic function. It should also be noted that in some alternative implementations, the functions noted in the block can occur in different order from that noted in the drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and sometimes they can be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for executing the specified function or action, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0339] It should be noted that in the present application, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes elements inherent to such process, method, article or equipment. Without more limitations, the element limited by the statement "including a" does not exclude the presence of another identical element in the process, method, article or equipment including the element.

[0340] Although the embodiments of the present application have been disclosed with reference to the above embodiments, the above embodiments are merely used to facilitate understanding of the present application and are not used to limit the present application. Any person skilled in the art, without departing from the spirit and scope of the present application, can make any modification and change in the form and details of the embodiments, but the patent protection scope of the present application should be subject to the scope defined by the appended claims.

Claims

1. A VR-based coal mine accident drilling method, characterized in that, The method comprises the following steps: carrying out scene analysis on collected coal mine accident cases to obtain scene scripts of the coal mine accident cases, wherein the scene scripts comprise scene evolution paths and flexible modules, and the flexible modules are used to combine the scene evolution paths to generate a VR drilling scene according to drilling instructions; obtaining drilling instructions of a user, combining the scene evolution paths and the flexible modules of the scene scripts according to the drilling instructions, and generating a VR drilling scene, wherein the VR drilling scene comprises multiple time nodes; obtaining a drilling time node selected by the user from different time nodes; determining a feedback model corresponding to the drilling time node from a feedback model library, wherein the feedback model library comprises disaster state information of each time node, accident emergency measure information corresponding to the disaster state information, and accident emergency measure information implementation behaviors corresponding to the accident emergency measure information; showing multiple selected accident emergency measure information to the current user, and receiving accident emergency measure information selected by the current user; interacting with the feedback model based on the accident emergency measure information selected by the current user, and realizing interactive operation of the accident emergency measure information implementation behaviors of the VR drilling scene under the influence of the current accident emergency measure information; wherein the drilling instructions comprise a location selection instruction, a rescue type selection instruction, a personnel type selection instruction, and a disaster type selection instruction; the step of combining the scene evolution paths and the flexible modules of the scene scripts according to the drilling instructions to generate the VR drilling scene comprises: combining the scene scripts of each coal mine accident case according to the location selection instruction, the rescue type selection instruction, the personnel type selection instruction, and / or the disaster type selection instruction to obtain a determined scene evolution path and flexible modules corresponding to the drilling instructions; combining the scene evolution paths and the flexible modules to obtain scene pheromones; obtaining rendering elements corresponding to the scene pheromones, rendering each of the scene pheromones according to the rendering elements, and obtaining a VR drilling scene; wherein the step of rendering each of the scene pheromones according to the rendering elements to obtain the VR drilling scene comprises: constructing an index table according to the rendering elements; integrating each of the scene pheromones according to the index table to obtain an initial three-dimensional grid graph; assigning values to the initial three-dimensional grid graph according to each of the rendering elements to obtain the VR drilling scene; wherein the step of integrating each of the scene pheromones according to the index table to obtain the initial three-dimensional grid graph comprises: establishing a first initial three-dimensional grid graph at the initial operation; converting the coordinates M of the scene pheromones corresponding to the index table through a reality conversion matrix F1 to obtain real coordinates M1 of the scene pheromones; M1=M×F1; converting the real coordinates M1 through a display conversion matrix F2 to obtain display coordinates M2; M2=M1×F2; converting the display coordinates M2 through a mapping conversion matrix F3 to obtain mapping coordinates M3; M3=F3×M2; Matching the mapping coordinates with each virtual coordinate in the first initial three-dimensional grid map to obtain an initial three-dimensional grid map; The matching the mapping coordinates with each virtual coordinate in the first initial three-dimensional grid map to obtain an initial three-dimensional grid map comprises: Calculating a correlation between each mapping coordinate; Initializing an iteration parameter, the iteration parameter comprising: a first matching coordinate of each mapping coordinate, a number of mapping coordinates, a position of each mapping coordinate initially matched, and a matching step length; Traversing each mapping coordinate, calculating a first correlation between each mapping coordinate under a current first matching coordinate; determining whether the first correlation is a preset multiple of the correlation; if the first correlation is the preset multiple of the correlation, outputting the first matching coordinate as a final target matching coordinate, mapping pheromone of a scene corresponding to each mapping coordinate to the target matching coordinate according to the target matching coordinate to obtain an initial three-dimensional grid map; if the first correlation is not the preset multiple of the correlation, updating the matching step length according to a matching degree between the first correlation and the correlation to obtain a new matching step length; Taking the new matching step length as the matching step length and returning to the above operation until the first correlation is the preset multiple of the correlation, and outputting an initial three-dimensional grid map.

2. The VR-based coal mine accident drilling method according to claim 1, characterized in that, The updating the matching step length according to the matching degree between the first correlation and the correlation to obtain a new matching step length comprises: Calculating a gradient value of the current matching degree according to a preset gradient loss function; Calculating a new matching step length according to the gradient value; Bnew = B - λ x ▽f(x); Wherein, λ is a learning rate; B is a matching step length; Bnew is a new matching step length; ▽f(x) is a gradient value.

3. A VR-based coal mine accident drilling device applying the VR-based coal mine accident drilling method according to any one of claims 1 to 2, characterized in that, The device comprises: A decomposition module configured to perform scenario analysis on the collected coal mine accident cases to obtain scenario scripts of the coal mine accident cases, wherein the scenario scripts comprise scenario evolution paths and a plurality of flexible modules, and the flexible modules are configured to be combined into the scenario evolution paths to generate a VR drilling scene according to drilling instructions; A combination module configured to obtain drilling instructions of a user, combine the scenario evolution paths and the flexible modules of the scenario scripts according to the drilling instructions, and generate a VR drilling scene, wherein the VR drilling scene comprises a plurality of time nodes; A selection module configured to obtain a drilling time node selected by the user from different time nodes, and determine a feedback model corresponding to the drilling time node from a feedback model library, wherein the feedback model library comprises: disaster state information of each time node, accident emergency measure information corresponding to the disaster state information, and accident emergency measure information implementation behaviors corresponding to the accident emergency measure information. The display module is configured to display a plurality of candidate accident emergency measure information to the current user, receive accident emergency measure information selected by the current user, and interact with the feedback model based on the accident emergency measure information selected by the current user to realize interactive operation of accident emergency measure information implementation behavior in the VR simulation scene under the influence of the current accident emergency measure information.

4. A computer device comprising a memory, a processor, and a computer program stored on the memory, wherein the computer program comprises instructions that, when executed by the processor, cause the processor to perform the method of any one of claims 1-3. The processor executes the computer program to implement the steps of the VR-based coal mine accident simulation method according to any one of claims 1 to 2.

5. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the VR-based coal mine accident simulation method according to any one of claims 1 to 2.

6. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the VR-based coal mine accident simulation method according to any one of claims 1 to 2. The computer program is executed by the processor to implement the steps of the VR-based coal mine accident simulation method according to any one of claims 1 to 2.

Images