AR-based forest fire terrain simulation training method and system

Abstract

The application relates to the technical field of fire-fighting simulation training, and particularly discloses a forest fire-fighting terrain simulation training method and system based on AR. The application determines a target forest region, constructs a three-dimensional terrain model, carries out dynamic fire simulation, generates a fire-fighting training scene, creates initial guide information, carries out initial guidance of AR fire-fighting training, evaluates a plurality of training monitoring data, selects representative training data one by one, generates representative guide information, and carries out step-by-step guidance and monitoring of AR fire-fighting training. The application can create initial guide information, initially guide, monitor and evaluate a plurality of fire-fighting personnel in the fire-fighting training scene, select representative training data one by one to generate representative guide information, and guide and monitor the plurality of fire-fighting personnel one by one. The application can automatically create guide information that is more in line with actual training, so that the fire-fighting capacity of the fire-fighting personnel is effectively and gradually improved through step-by-step fire-fighting simulation training.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of fire-fighting simulation training, and particularly relates to a forest fire-fighting terrain simulation training method and system based on AR. BACKGROUND

[0002] Fire-fighting simulation training is a process in which a realistic fire scene and rescue environment are constructed by means of virtual reality, augmented reality, simulation modeling and other digital technologies, so that fire-fighters can train their skills in emergency response, fire-fighting operations, and personnel evacuation without real danger. This method breaks through the limitations of traditional training in terms of site, safety, and cost, improves the flexibility and immersion of training, and helps to enhance the real combat response ability, cooperative combat level, and risk prediction ability of fire-fighters.

[0003] In the prior art, fire-fighting simulation training usually only follows a preset process and fixed guidance to conduct AR fire-fighting simulation training for fire-fighters. It cannot effectively monitor and evaluate fire-fighters, and cannot select the monitoring data of the best-performing fire-fighter according to the monitoring data and evaluation results of multiple fire-fighters to recreate guidance that is more in line with actual training. As a result, the fire-fighting ability of fire-fighters cannot be effectively and gradually improved in simulation training. SUMMARY

[0004] The purpose of the embodiments of the present application is to provide a forest fire-fighting terrain simulation training method and system based on AR, which aims to solve the problems raised in the background.

[0005] To achieve the above-mentioned purpose, the embodiments of the present application provide the following technical solutions:

[0006] The forest fire-fighting terrain simulation training method based on AR specifically includes the following steps:

[0007] Determine a target forest area, collect regional terrain data of the target forest area, and construct a three-dimensional terrain model through data fusion and spatial analysis;

[0008] Based on the three-dimensional terrain model, generate a fire-fighting training scene through multi-physical field coupling simulation;

[0009] Create initial guidance information, and conduct initial guidance for AR fire-fighting training for multiple fire-fighters in the fire-fighting training scene, and collect multiple training monitoring data by constructing a cross-modal interaction model;

[0010] Evaluate multiple training monitoring data, select representative training data one by one, generate representative guidance information, and conduct step-by-step guidance and monitoring for AR fire-fighting training for multiple fire-fighters.

[0011] As a further limitation of the technical scheme of the embodiment of the present application, the determining the target forest area, collecting the regional terrain data of the target forest area, and constructing the three-dimensional terrain model through data fusion and spatial analysis specifically include the following steps:

[0012] Determining the target forest area;

[0013] Obtaining regional map data of the target forest area;

[0014] According to the regional map data, planning a detection collection route through data fusion and spatial analysis;

[0015] According to the detection collection route, controlling the unmanned aerial vehicle to fly over the target forest area for detection and collecting regional terrain data;

[0016] Processing the regional terrain data to construct a three-dimensional terrain model.

[0017] As a further limitation of the technical scheme of the embodiment of the present application, the processing the regional terrain data to construct a three-dimensional terrain model specifically includes the following steps:

[0018] Pretreating the regional terrain data to generate standard terrain data;

[0019] According to the standard terrain data, constructing a basic terrain grid;

[0020] From the standard terrain data, identifying and extracting vegetation coverage data, and performing coverage mapping in the basic terrain grid to generate a vegetation structure model;

[0021] From the standard terrain data, identifying and extracting surface detail data, and performing detail supplementation in the vegetation structure model to generate a three-dimensional terrain model.

[0022] As a further limitation of the technical scheme of the embodiment of the present application, the generating a fire training scene through multi-physical field coupling simulation based on the three-dimensional terrain model specifically includes the following steps:

[0023] Obtaining fire source parameter data and obtaining dynamic meteorological data;

[0024] In the three-dimensional terrain model, performing multi-physical field coupling simulation according to the fire source parameter data and the dynamic meteorological data to generate a fire training scene.

[0025] As a further limitation of the technical scheme of the embodiment of the present application, the creating initial guidance information, initially guiding a plurality of firefighters in the fire training scene for AR fire training, and collecting a plurality of training monitoring data through the construction of a cross-modal interaction model specifically includes the following steps:

[0026] Determine a simulation training task and obtain corresponding demonstration training data;

[0027] Create initial guide information according to the demonstration training data;

[0028] In the fire training scene, the initial guide information is used to guide the AR fire training of the plurality of firefighters;

[0029] Real-time capture of fire fighting actions of the plurality of firefighters, corresponding AR effect interaction through construction of a cross-modal interaction model and collection of a plurality of training monitoring data.

[0030] As a further limitation of the technical scheme of the embodiment of the application, the evaluation of the plurality of training monitoring data, the selection of representative training data, the generation of representative guide information, and the successive guidance and monitoring of the AR fire training of the plurality of firefighters specifically include the following steps:

[0031] Behavioral evaluation of the plurality of training monitoring data, recording of behavioral evaluation data;

[0032] Effect evaluation of the plurality of training monitoring data, recording of effect evaluation data;

[0033] Comprehensive evaluation of the behavioral evaluation data and the effect evaluation data, and selection of representative training data;

[0034] Creation of representative guide information according to the representative training data;

[0035] In the fire training scene, the representative guide information is used to guide the AR fire training of the plurality of firefighters.

[0036] An AR-based forest fire terrain simulation training system, the system comprising a terrain model construction unit, a dynamic fire simulation unit, an initial guide processing unit and a successive representative guide unit, wherein:

[0037] The terrain model construction unit is used to determine a target forest area, collect regional terrain data of the target forest area, and construct a three-dimensional terrain model through data fusion and spatial analysis;

[0038] The dynamic fire simulation unit is used to generate a fire training scene through multi-physical field coupling simulation based on the three-dimensional terrain model;

[0039] The initial guide processing unit is used to create initial guide information, guide the AR fire training of the plurality of firefighters in the fire training scene, and collect a plurality of training monitoring data through construction of a cross-modal interaction model;

[0040] Successively representative guiding unit is used for evaluating a plurality of training monitoring data, successively selecting representative training data, generating representative guiding information, and guiding and monitoring AR fire fighting training of a plurality of fire fighters.

[0041] As a further limitation of the technical scheme of the embodiment of the application, the terrain model construction unit specifically comprises:

[0042] A target area determination module is configured to determine a target forest area.

[0043] A map data acquisition module is configured to acquire regional map data of the target forest area.

[0044] A collection route planning module is configured to plan a detection collection route through data fusion and spatial analysis according to the regional map data.

[0045] A flight detection control module is configured to control a UAV to perform flight detection on the target forest area according to the detection collection route and collect regional terrain data.

[0046] A three-dimensional terrain model construction module is configured to process the regional terrain data and construct a three-dimensional terrain model.

[0047] As a further limitation of the technical scheme of the embodiment of the application, the three-dimensional terrain model construction module specifically comprises:

[0048] A data preprocessing submodule is configured to preprocess the regional terrain data and generate standard terrain data.

[0049] A terrain grid construction submodule is configured to construct a basic terrain grid according to the standard terrain data.

[0050] A vegetation coverage mapping submodule is configured to identify and extract vegetation coverage data from the standard terrain data, perform coverage mapping in the basic terrain grid, and generate a vegetation structure model.

[0051] A detail supplement submodule is configured to identify and extract surface detail data from the standard terrain data, perform detail supplement in the vegetation structure model, and generate a three-dimensional terrain model.

[0052] As a further limitation of the technical scheme of the embodiment of the application, the successively representative guiding unit specifically comprises:

[0053] A behavior evaluation module is configured to perform behavior evaluation on a plurality of training monitoring data and record behavior evaluation data.

[0054] An effect evaluation module is configured to perform effect evaluation on a plurality of training monitoring data and record effect evaluation data.

[0055] a representative sequential selection module configured to comprehensively select representative training data from the behavior evaluation data and the effect evaluation data;

[0056] a guide information creation module configured to create representative guide information according to the representative training data;

[0057] a sequential guide monitoring module configured to sequentially guide and monitor the AR fire training of the plurality of firefighters in the fire training scene according to the representative guide information.

[0058] Compared with the prior art, the present application has the following advantages:

[0059] The embodiment of the present application determines the target forest area, constructs a three-dimensional terrain model, performs dynamic fire simulation to generate a fire training scene, creates initial guide information to guide the initial AR fire training, evaluates a plurality of training monitoring data, sequentially selects representative training data to generate representative guide information, and sequentially guides and monitors the AR fire training. The initial guide information can be created to guide, monitor and evaluate the initial AR fire training of the plurality of firefighters in the fire training scene. The representative guide information is generated by sequentially selecting the representative training data to guide and monitor the plurality of firefighters. The guide information that is more suitable for the actual training can be automatically created to effectively and gradually improve the fire fighting ability of the firefighters through the sequential fire simulation training. BRIEF DESCRIPTION OF DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application.

[0061] Figure 1 A flowchart of the method provided by the embodiment of the present application is shown.

[0062] Figure 2 A flowchart of constructing a three-dimensional terrain model in the method provided by the embodiment of the present application is shown.

[0063] Figure 3 A flowchart of processing regional terrain data in the method provided by the embodiment of the present application is shown.

[0064] Figure 4 A flowchart of the initial guide of the fire training in the method provided by the embodiment of the present application is shown.

[0065] Figure 5 A flowchart of the sequential guide of the fire training in the method provided by the embodiment of the present application is shown.

[0066] Figure 6An application architecture diagram of the system provided by the embodiment of the present application is shown.

[0067] Figure 7 A structural block diagram of a terrain model construction unit in the system provided by the embodiment of the present application is shown.

[0068] Figure 8 A structural block diagram of a three-dimensional terrain model construction module in the system provided by the embodiment of the present application is shown.

[0069] Figure 9 A structural block diagram of a successive representative guide unit in the system provided by the embodiment of the present application is shown. DETAILED DESCRIPTION

[0070] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0071] It can be understood that the fire-fighting simulation training in the prior art usually only carries out AR fire-fighting simulation training on fire-fighting personnel according to a preset process and fixed guide, cannot effectively monitor and evaluate the fire-fighting personnel, and cannot select the monitoring data of the fire-fighting personnel with the best performance according to the monitoring data and evaluation results of the multiple fire-fighting personnel to re-create a guide that is more suitable for actual training, so that the fire-fighting ability of the fire-fighting personnel cannot be effectively and gradually improved in the simulation training.

[0072] To solve the above problems, the embodiment of the present application determines a target forest area, collects regional terrain data of the target forest area, processes the regional terrain data, and constructs a three-dimensional terrain model; performs dynamic fire simulation based on the three-dimensional terrain model to generate a fire-fighting training scene; creates initial guide information, performs initial guide on the multiple fire-fighting personnel in the fire-fighting training scene, and collects multiple training monitoring data; evaluates the multiple training monitoring data, selects representative training data successively, generates representative guide information, and performs successive guide and monitoring on the multiple fire-fighting personnel in the AR fire-fighting training. The initial guide information can be created, the multiple fire-fighting personnel can be initially guided, monitored and evaluated in the fire-fighting training scene, the representative training data can be selected successively to generate the representative guide information, and the multiple fire-fighting personnel can be successively guided and monitored, so that the guide information that is more suitable for actual training can be automatically created, and thus the fire-fighting ability of the fire-fighting personnel can be effectively and gradually improved through successive fire-fighting simulation training.

[0073] Figure 1 A flowchart of the method provided by the embodiment of the present application is shown.

[0074] Specifically, the AR-based forest fire terrain simulation training method specifically comprises the following steps:

[0075] In step S101, a target forest area is determined, regional terrain data of the target forest area is collected, and a three-dimensional terrain model is constructed through data fusion and spatial analysis.

[0076] Specifically, according to the regional map data, a detection collection route is planned through data fusion and spatial analysis, and the specific steps are as follows:

[0077] The elevation grid, slope distribution, and vegetation type vector are obtained from the regional map data. Based on the elevation grid, slope distribution, and vegetation type vector, the spatial feature analysis method is used to evaluate the terrain elevation change intensity, slope dynamic trend, and vegetation spatial blocking effect to generate gridded terrain classification data.

[0078] The historical fire point spatial distribution data is obtained. Based on the terrain classification data and the historical fire point spatial distribution data, the spatial correlation analysis is used to identify the superimposed regions of high complexity regions and historical fire point adjacent ranges to obtain an identification result. Based on the identification result, a multi-factor weight evaluation is performed in combination with the ground combustible feature to generate a key region distribution map.

[0079] Based on the key region distribution map and the unmanned aerial vehicle performance parameter library, a dynamic matching relationship between the detection level and the flight parameters is established through a priority driving mechanism, and a dynamic matching result is obtained. Based on the dynamic matching result, a photo parameter adaptive adjustment strategy is used to scan and gradientize the high-priority region and the low-priority region to generate a flight parameter configuration table.

[0080] Based on the gridded terrain classification data, the key region distribution map, and the flight parameter configuration table, a path optimization model is constructed through a multi-target A* algorithm. Based on the path optimization model, in combination with the terrain complexity gradient change limitation and the endurance time requirement, a shortest distance calculation is performed to determine a preliminary detection route.

[0081] Based on the preliminary detection route and the real-time obstacle database, a three-dimensional dynamic safety corridor model is constructed. Based on the three-dimensional dynamic safety corridor model, through spatial conflict detection and B-spline curve flight path smoothing optimization algorithm, under the condition of meeting the obstacle avoidance constraint, a detection collection route is generated.

[0082] The real-time obstacle database includes high-voltage line tower and communication base station spatial coordinates. The detection collection route integrates flight path continuity, safety margin guarantee, and terrain adaptability.

[0083] Further, the application makes the detection route focus on covering high fire risk areas by spatial correlation of terrain complexity classification and historical fire points, and improves the data collection density of the target area; and the dynamic flight parameter configuration mechanism is used to improve the detection efficiency of a single flight while ensuring the data accuracy of the key area.

[0084] In the embodiment of the application, a target forest area is determined, regional map data of the target forest area is acquired, a detection and collection route is planned according to the regional map data, then the unmanned aerial vehicle is controlled to fly and detect the target forest area according to the detection and collection route, and regional terrain data collected by the unmanned aerial vehicle is acquired, after that, the regional terrain data is preprocessed, the point cloud data is voxel filtered, outlier points are removed, and smoothing processing is performed, and all data is converted into a unified spatial reference coordinate system to generate standard terrain data, then a basic terrain grid is constructed according to the standard terrain data, and vegetation coverage data such as coniferous, broad-leaved, shrub and grassland is identified and extracted from the standard terrain data, the vegetation coverage data is mapped in the basic terrain grid to generate a vegetation structure model, at the same time, surface detail data such as soil, rock, road and river is identified and extracted from the standard terrain data, the surface detail data is supplemented in the vegetation structure model to generate a three-dimensional terrain model.

[0085] Specifically, Figure 2 A flowchart of constructing a three-dimensional terrain model in the method provided by the embodiment of the application is shown.

[0086] In the preferred embodiment provided by the application, the determination of the target forest area and the collection of the regional terrain data of the target forest area, and the construction of the three-dimensional terrain model through data fusion and spatial analysis specifically include the following steps:

[0087] Step S1011, determining a target forest area;

[0088] Step S1012, acquiring regional map data of the target forest area;

[0089] Step S1013, planning a detection and collection route through data fusion and spatial analysis according to the regional map data;

[0090] Step S1014, controlling an unmanned aerial vehicle to fly and detect the target forest area according to the detection and collection route, and collecting regional terrain data;

[0091] Step S1015, processing the regional terrain data to construct a three-dimensional terrain model.

[0092] Specifically, Figure 3 A flowchart of processing regional terrain data in the method provided by the embodiment of the application is shown.

[0093] In the preferred embodiments provided by the present application, the processing of the regional terrain data to construct a three-dimensional terrain model specifically includes the following steps:

[0094] Step S10151, preprocessing the regional terrain data to generate standard terrain data;

[0095] Step S10152, constructing a basic terrain grid according to the standard terrain data;

[0096] Step S10153, identifying and extracting vegetation coverage data from the standard terrain data, performing coverage mapping in the basic terrain grid to generate a vegetation structure model;

[0097] Step S10154, identifying and extracting surface detail data from the standard terrain data, performing detail supplementation in the vegetation structure model to generate a three-dimensional terrain model.

[0098] Specifically, the vegetation coverage data is identified and extracted from the standard terrain data, and the coverage mapping is performed in the basic terrain grid to generate a vegetation structure model, and the specific steps are as follows:

[0099] Obtain satellite remote sensing spectral features and ground measured vegetation plot data, and based on the standard terrain data, satellite remote sensing spectral features and ground measured vegetation plot data, construct a three-dimensional correlation matrix related to terrain elevation, spectral reflectance and vegetation type; based on the three-dimensional correlation matrix, perform registration processing on the multivariate data space through a feature matching algorithm, and generate a vegetation feature data set;

[0100] Based on the vegetation feature data set and the forest ecological database, a deep convolution classification model is constructed, the input layer of the deep convolution classification model integrates multi-band spectral features and terrain slope data, and then the attention mechanism in the deep convolution classification model is used to strengthen feature extraction, and the forest ecological parameters in the forest ecological database are combined to calculate the vegetation type probability distribution, and a vegetation distribution heat map is output;

[0101] Obtain terrain height distribution parameters from the basic terrain grid, use the hierarchical difference value coverage density in the vegetation distribution heat map as the basis for the horizontal dimension, use the terrain height distribution parameters as the basis for the vertical dimension, and then perform three-dimensional density field space mapping to construct a vegetation network model;

[0102] Obtain the crown leaf water content dynamic attenuation parameter, the tree trunk wood heat value change parameter and the ground litter layer humidity diffusion coefficient from the forest ecological database, use the vegetation network model to perform hierarchical parameterization mapping processing on the crown leaf water content dynamic attenuation parameter, the tree trunk wood heat value change parameter and the ground litter layer humidity diffusion coefficient, and generate a static structure model.

[0103] The acquired real-time ecological parameter library is used for simulation through a static structure model to obtain a vegetation structure model.

[0104] Further, the application improves the mountain vegetation remote sensing recognition precision through the topographic correction multi-source data fusion mechanism; and the virtual vegetation has real physical properties through the dynamic parameter binding mechanism, thereby laying a foundation for multi-physical field coupling simulation.

[0105] Further, the AR-based forest fire terrain simulation training method further comprises the following steps:

[0106] In step S102, a fire training scene is generated through multi-physical field coupling simulation based on the three-dimensional terrain model.

[0107] In the embodiment of the application, the fire source parameter data is acquired, and the dynamic meteorological data is acquired, and then based on a Firenz simulation platform, a FireCloud simulation platform and the like, a controllable spreading path and speed parameter are constructed in the three-dimensional terrain model according to the fire source parameter data and the dynamic meteorological data, and a smoke diffusion algorithm such as FireCloud and Fire3D is introduced, concentration change simulation is performed according to vegetation types, wind speed, temperature and the like, dynamic fire simulation is realized, spatial projection is realized on an AR device, a fusion view of a real physical space and a virtual fire scene is formed, and a fire training scene is generated.

[0108] Specifically, in the preferred embodiment provided by the application, generating the fire training scene through multi-physical field coupling simulation based on the three-dimensional terrain model specifically comprises the following steps:

[0109] The fire source parameter data is acquired, and the dynamic meteorological data is acquired;

[0110] In the three-dimensional terrain model, multi-physical field coupling simulation is performed according to the fire source parameter data and the dynamic meteorological data, and a fire training scene is generated.

[0111] Specifically, in the three-dimensional terrain model, multi-physical field coupling simulation is performed according to the fire source parameter data and the dynamic meteorological data, and a fire training scene is generated, and the specific steps are as follows:

[0112] Slope direction data is obtained based on the three-dimensional terrain model, burning attribute parameters are obtained based on the vegetation structure model, and an association matrix of fire propagation direction and terrain features is constructed based on the slope direction data and the burning attribute parameters; fire line advance rates of different vegetation types are calculated based on the association matrix of fire propagation direction and terrain features, so as to generate a fire propagation vector diagram;

[0113] The wind speed and wind direction parameters are obtained through dynamic meteorological data, and an energy exchange model of heat convection and air flow is established in a manner of spatial superposition of a meteorological vector field and a fire spread field based on the fire spread vector diagram and the wind speed and wind direction parameters;

[0114] The three-dimensional combustion surface data and the spatial distribution characteristics of the vegetation are obtained through the vegetation structure model, the flame surface radiation flux is obtained through a heat conduction equation based on the three-dimensional fire dynamic distribution field and the three-dimensional combustion surface data, the thermal radiation gradient field is constructed by combining the spatial distribution characteristics of the vegetation and the flame surface radiation flux, and the dynamic thermal radiation distribution cloud picture is generated through the thermal radiation gradient field;

[0115] The airflow channel data and the terrain elevation shielding parameters are obtained through the three-dimensional terrain model, the simulation result is obtained by simulating the smoke rising track through the thermal buoyancy effect based on the dynamic thermal radiation distribution cloud picture and the airflow channel data, and the dynamic smoke particle motion model is generated through the simulation result and the terrain elevation shielding parameters;

[0116] Based on the dynamic thermal radiation distribution cloud picture and the dynamic smoke particle motion model, the corresponding relationship between the thermal radiation intensity and the color temperature value is established to obtain an AR visual color gradient field, and the fire training scene is generated based on the AR visual color gradient field and in combination with the smoke transparency dynamic mapping layer construction technology.

[0117] Further, the present application realizes the visual conversion of physical parameters through the correlation of the thermal radiation gradient field and the AR color temperature mapping, and enhances the authenticity of the training scene through the dynamic binding of the smoke particle motion model and the transparency map.

[0118] Further, the AR-based forest fire terrain simulation training method further includes the following steps:

[0119] Step S103, initial guidance information is created, initial guidance of AR fire training of the plurality of firefighters is performed in the fire training scene, and a plurality of training monitoring data are collected by constructing a cross-modal interaction model.

[0120] In the embodiment of the present application, the simulation training task is determined, the corresponding demonstration training data are acquired, the initial guidance information is created according to the demonstration training data, and then the initial guidance of the action, route, voice, direction, etc. of the AR fire training of the plurality of firefighters is performed in the fire training scene according to the initial guidance information, the fire-fighting actions of the plurality of firefighters are captured in real time, the corresponding AR effect interaction is performed through the construction of the cross-modal interaction model, and the training monitoring data of the position change, operation path, response time delay, fire extinguishing coverage range, fire extinguishing frequency, etc. of the plurality of firefighters are collected during the corresponding AR effect interaction.

[0121] Specifically, Figure 4 A flowchart of initial guidance of fire training in the method provided by the embodiment of the application is shown.

[0122] In the preferred embodiment provided by the application, the creating of the initial guidance information and the initial guidance of the AR fire training of the plurality of firefighters in the fire training scene by collecting the plurality of training monitoring data through the construction of the cross-modal interaction model specifically includes the following steps:

[0123] Step S1031, determining a simulation training task and obtaining corresponding demonstration training data;

[0124] Step S1032, creating initial guidance information according to the demonstration training data;

[0125] Step S1033, performing initial guidance of AR fire training of a plurality of firefighters in the fire training scene according to the initial guidance information;

[0126] Step S1034, capturing fire actions of the plurality of firefighters in real time, performing corresponding AR effect interaction through the construction of the cross-modal interaction model and collecting a plurality of training monitoring data.

[0127] Specifically, the capturing of the fire actions of the plurality of firefighters in real time, the performing of the corresponding AR effect interaction through the construction of the cross-modal interaction model and the collecting of the plurality of training monitoring data specifically include the following steps:

[0128] The motion trajectory of the firefighter, the behavior feature of the person, the terrain constraint parameter, the fire spread vector and the smoke obscuration degree parameter are obtained through the AR device, based on the motion trajectory of the firefighter, the spatial coordinates of the three-dimensional terrain model and the dynamic fire spread parameter are utilized, the multi-element data synchronization technology is adopted to eliminate the space-time deviation, and the cross-modal interaction model is constructed; based on the cross-modal interaction model, the behavior feature of the person, the terrain constraint parameter and the fire spread vector are fused and processed to generate a comprehensive feature vector;

[0129] Based on the vegetation structure model, the combustible material density distribution and the real-time meteorological data are obtained, based on the combustible material density distribution and the real-time meteorological data, the comprehensive feature vector is combined to construct a fire field thermodynamic response model; the physical response parameter is calculated based on the fire field thermodynamic response model;

[0130] According to the physical response parameter, the three-dimensional flame particle configuration is constructed by establishing a dynamic mapping of the burning rate and the combustible material heat value based on the basic terrain grid; based on the training target difficulty coefficient in the simulation training task, the smoke obscuration degree parameter is adjusted through the Lagrange particle model to obtain an adjustment result; the three-dimensional flame particle configuration and the adjustment result are combined to generate a multi-channel AR feedback instruction;

[0131] The surface detail data is obtained through a three-dimensional terrain model, a spatial relationship between a firefighter movement trajectory and a detection and collection route is dynamically analyzed based on the surface detail data and a multi-channel AR feedback instruction to obtain an analysis result, an interactive signal is generated based on the analysis result, and a plurality of training is collected in the process of corresponding AR effect interaction through the interactive signal.

[0132] Further, the application solves the action distortion problem caused by the asynchronous data of the wearable device through a multi-source data space-time interpolation algorithm, and realizes physical safety protection in virtual training through a dynamic generation method of a safety envelope line based on terrain characteristics.

[0133] Further, the AR-based forest fire terrain simulation training method further includes the following steps:

[0134] In step S104, the plurality of training monitoring data is evaluated, the representative training data is selected, the representative guide information is generated, and the plurality of firefighters are guided and monitored in the AR fire training.

[0135] In the embodiment of the application, the plurality of training monitoring data is behaviorally evaluated in terms of position trajectory, fire extinguishing action, response time, etc., the behavior evaluation data is recorded, the plurality of training monitoring data is effectually evaluated in terms of fire change and final extinguishing efficiency, the effect evaluation data is recorded, and then the behavior evaluation data and the effect evaluation data are comprehensively evaluated, the best training monitoring data is selected, the representative training data is determined, the corresponding representative guide information is created according to the representative training data, and then in the subsequent fire training scene, the plurality of firefighters are guided and monitored in the AR fire training according to the corresponding representative guide information, the subsequent representative training data is selected, and the subsequent representative training data is evaluated.

[0136] Specifically, Figure 5 A flowchart of the step-by-step guidance of the fire training in the method provided by the embodiment of the application is shown.

[0137] In the preferred embodiment provided by the application, the evaluation of the plurality of training monitoring data, the selection of the representative training data, the generation of the representative guide information, and the step-by-step guidance and monitoring of the plurality of firefighters in the AR fire training specifically include the following steps:

[0138] In step S1041, the plurality of training monitoring data is behaviorally evaluated, and the behavior evaluation data is recorded.

[0139] In step S1042, the plurality of training monitoring data is effectually evaluated, and the effect evaluation data is recorded.

[0140] Step S1043, comprehensively integrating the behavior evaluation data and the effect evaluation data, sequentially selecting representative training data;

[0141] Step S1044, creating representative guide information according to the representative training data;

[0142] Step S1045, sequentially guiding and monitoring the AR fire-fighting training of the plurality of fire-fighting personnel according to the representative guide information in the fire-fighting training scene.

[0143] Further, Figure 6 The application architecture diagram of the system provided by the embodiment of the application is shown.

[0144] In another preferred embodiment provided by the application, the AR-based forest fire-fighting terrain simulation training system comprises:

[0145] The terrain model construction unit 101 is configured to determine a target forest region, collect regional terrain data of the target forest region, and construct a three-dimensional terrain model through data fusion and spatial analysis.

[0146] In the embodiment of the application, the terrain model construction unit 101 determines a target forest region, acquires regional map data of the target forest region, plans a detection and collection route according to the regional map data, controls a UAV to fly over the target forest region according to the detection and collection route, and acquires regional terrain data collected by the UAV, pre-processes the regional terrain data, performs voxel filtering, outlier rejection, and smoothing processing on point cloud data, converts all data into a unified spatial reference coordinate system, generates standard terrain data, constructs a basic terrain grid according to the standard terrain data, identifies and extracts vegetation coverage data such as coniferous, broad-leaved, shrub, and grassland from the standard terrain data, performs coverage mapping in the basic terrain grid according to the vegetation coverage data, generates a vegetation structure model, identifies and extracts surface detail data such as soil, rock, road, and river from the standard terrain data, performs detail supplementation in the vegetation structure model according to the surface detail data, and generates a three-dimensional terrain model.

[0147] Specifically, Figure 7 The structural block diagram of the terrain model construction unit 101 in the system provided by the embodiment of the application is shown.

[0148] In the preferred embodiment provided by the application, the terrain model construction unit 101 specifically comprises:

[0149] The target region determination module 1011 is configured to determine a target forest region.

[0150] The map data acquisition module 1012 is configured to acquire regional map data of the target forest area.

[0151] The collection route planning module 1013 is configured to plan a detection collection route by data fusion and spatial analysis according to the regional map data.

[0152] The flight detection control module 1014 is configured to control the unmanned aerial vehicle to perform flight detection on the target forest area according to the detection collection route, and collect regional terrain data.

[0153] The three-dimensional terrain model construction module 1015 is configured to process the regional terrain data and construct a three-dimensional terrain model.

[0154] Specifically, Figure 8 The structure block diagram of the three-dimensional terrain model construction module 1015 in the system provided by the embodiment of the application is shown.

[0155] In the preferred embodiment provided by the application, the three-dimensional terrain model construction module 1015 specifically includes:

[0156] The data preprocessing submodule 10151 is configured to preprocess the regional terrain data and generate standard terrain data.

[0157] The terrain grid construction submodule 10152 is configured to construct a basic terrain grid according to the standard terrain data.

[0158] The vegetation coverage mapping submodule 10153 is configured to identify and extract vegetation coverage data from the standard terrain data, perform coverage mapping in the basic terrain grid, and generate a vegetation structure model.

[0159] The detail supplement submodule 10154 is configured to identify and extract surface detail data from the standard terrain data, perform detail supplement in the vegetation structure model, and generate a three-dimensional terrain model.

[0160] Further, the AR-based forest fire terrain simulation training system further includes:

[0161] The dynamic fire simulation unit 102 is configured to generate a fire training scene by multi-physical field coupling simulation based on the three-dimensional terrain model.

[0162] In the embodiment of the present application, the dynamic fire simulation unit 102 obtains fire source parameter data and dynamic meteorological data, and then constructs controllable spread path and speed parameters in a three-dimensional terrain model based on a simulation platform such as Firenz or FireCloud, introduces smoke diffusion algorithms such as FireCloud or Fire3D, and simulates concentration changes according to vegetation types, wind speeds, temperatures, and the like to achieve dynamic fire simulation, thereby achieving spatial projection on an AR device to form a fusion view of a real physical space and a virtual fire scene and generate a fire training scene.

[0163] The initial guidance processing unit 103 is configured to create initial guidance information for initial guidance of AR fire training of the plurality of firefighters in the fire training scene by constructing a cross-modal interaction model to collect a plurality of training monitoring data.

[0164] In the embodiment of the present application, the initial guidance processing unit 103 determines a simulation training task, obtains corresponding demonstration training data, creates initial guidance information according to the demonstration training data, and then performs initial guidance of actions, routes, voices, directions, and the like of AR fire training of the plurality of firefighters in the fire training scene according to the initial guidance information, and captures fire actions of the plurality of firefighters in real time, performs corresponding AR effect interaction by constructing a cross-modal interaction model, and collects training monitoring data such as position changes, operation paths, response time delays, fire extinguishing coverage ranges, and fire extinguishing frequencies of the plurality of firefighters in the process of performing the corresponding AR effect interaction.

[0165] The successive representative guidance unit 104 is configured to evaluate the plurality of training monitoring data, select representative training data successively, generate representative guidance information, and perform successive guidance and monitoring of AR fire training of the plurality of firefighters.

[0166] In the embodiment of the present application, the successive representative guidance unit 104 performs behavior evaluation on the plurality of training monitoring data in terms of position trajectories, fire extinguishing actions, and response times, records behavior evaluation data, performs effect evaluation on the plurality of training monitoring data in terms of fire changes and final extinguishing efficiency, records effect evaluation data, and then comprehensively evaluates the behavior evaluation data and the effect evaluation data, selects the best training monitoring data successively, determines the training monitoring data as representative training data, creates corresponding representative guidance information according to the representative training data, and then performs successive guidance and monitoring of AR fire training of the plurality of firefighters in subsequent fire training scenes according to the corresponding representative guidance information, evaluates and selects subsequent representative training data.

[0167] Specifically, Figure 9 A structural block diagram of the successive representative guidance unit 104 in the system provided by the embodiment of the present application is shown.

[0168] wherein, in the preferred embodiments provided by the present application, the representative progressive guiding unit 104 specifically comprises:

[0169] a behavior evaluation module 1041 configured to perform behavior evaluation on the plurality of training monitoring data and record behavior evaluation data;

[0170] an effect evaluation module 1042 configured to perform effect evaluation on the plurality of training monitoring data and record effect evaluation data;

[0171] a representative progressive selection module 1043 configured to comprehensively select representative training data from the behavior evaluation data and the effect evaluation data;

[0172] a guiding information creation module 1044 configured to create representative guiding information according to the representative training data;

[0173] a progressive guiding and monitoring module 1045 configured to perform progressive guiding and monitoring on the plurality of firefighters in the AR firefighting training according to the representative guiding information in the firefighting training scene.

[0174] It should be understood that, although each step in the flowchart of each embodiment of the present application is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other sequences. Moreover, at least a part of the steps in each embodiment can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these sub-steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least a part of other steps or sub-steps or stages of other steps.

[0175] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer readable storage medium, and when the program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0176] Any combination of the technical features of the above-mentioned embodiments can be combined. In order to make the description simple, all possible combinations of the technical features in the above-mentioned embodiments are not described, however, as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present application.

[0177] The above-mentioned embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it should not be understood as limiting the scope of the present application. It should be noted that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

[0178] The above-mentioned embodiments are only the preferred embodiments of the present application, and are not used to limit the present application. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. An AR-based forest fire terrain simulation training method, characterized by, The method specifically comprises the following steps: determining a target forest area, collecting regional terrain data of the target forest area, and constructing a three-dimensional terrain model through data fusion and spatial analysis; based on the three-dimensional terrain model, generating a fire training scene through multi-physical field coupling simulation; creating initial guidance information, and initially guiding multiple firefighters in AR fire training in the fire training scene, and collecting multiple training monitoring data by constructing a cross-modal interaction model, specifically comprising the following steps: determine the simulation training task and obtain the corresponding demonstration training data; create initial guidance information according to the demonstration training data; in the fire training scene, according to the initial guidance information, initially guide multiple firefighters in AR fire training; real-time capture of multiple firefighters' fire fighting actions, corresponding AR effect interaction through the construction of a cross-modal interaction model and collection of multiple training monitoring data, the specific steps are as follows: obtain the fire personnel motion trajectory, personnel behavior characteristics, terrain constraint parameters, fire spread vector and smoke obscuration degree parameter through the AR device, based on the fire personnel motion trajectory, use the spatial coordinates of the three-dimensional terrain model and the dynamic fire spread parameter, adopt the multi-element data synchronization technology to eliminate the space-time deviation, and construct the cross-modal interaction model; based on the cross-modal interaction model, the personnel behavior characteristics, the terrain constraint parameters and the fire spread vector are fused and processed to obtain the comprehensive feature vector; based on the vegetation structure model, obtain the combustible material density distribution and real-time meteorological data, based on the combustible material density distribution and real-time meteorological data, combine the comprehensive feature vector to construct the fire field thermodynamic response model; based on the fire field thermodynamic response model, calculate the physical response parameter; according to the physical response parameter, on the basis of the basic terrain grid, by establishing the dynamic mapping of the burning rate and the combustible material heat value to construct the three-dimensional flame particle configuration; based on the training target difficulty coefficient in the simulation training task, adjust the smoke obscuration degree parameter through the Lagrange particle model to obtain the adjustment result; combine the three-dimensional flame particle configuration and the adjustment result to generate multi-channel AR feedback instructions; obtain the ground detail data through the three-dimensional terrain model, based on the ground detail data and the multi-channel AR feedback instructions, dynamically analyze the spatial relationship between the fire personnel motion trajectory and the detection collection route to obtain the analysis result; based on the analysis result, generate an interaction signal, and through the interaction signal, the corresponding AR effect interaction is carried out, and multiple training monitoring data are collected during the corresponding AR effect interaction; evaluate the multiple training monitoring data, select representative training data one by one, generate representative guidance information, and guide and monitor multiple firefighters in AR fire training one by one; the determination of the target forest area, the collection of the regional terrain data of the target forest area, and the construction of the three-dimensional terrain model through data fusion and spatial analysis specifically comprises the following steps: determine the target forest area; obtain the regional map data of the target forest area; according to the regional map data, plan the detection collection route through data fusion and spatial analysis; According to the detection collection route, the unmanned aerial vehicle is controlled to fly over the target forest area to collect regional terrain data; The regional terrain data is processed to construct a three-dimensional terrain model, specifically including the following steps: The regional terrain data is preprocessed to generate standard terrain data; According to the standard terrain data, a basic terrain grid is constructed; From the standard terrain data, vegetation coverage data is identified and extracted, and coverage mapping is performed in the basic terrain grid to generate a vegetation structure model; From the standard terrain data, surface detail data is identified and extracted, and detail supplementation is performed in the vegetation structure model to generate a three-dimensional terrain model. 2.The AR-based forest firefighting topography simulation training method of claim 1, wherein, According to the regional map data, a detection collection route is planned through data fusion and spatial analysis, specifically as follows: Through regional map data, an elevation grid, slope distribution, and vegetation type vector are obtained. Based on the elevation grid, slope distribution, and vegetation type vector, the terrain elevation change intensity, slope dynamic trend, and vegetation spatial blocking effect are evaluated through spatial feature analysis to generate gridded terrain classification data; Obtain historical fire point spatial distribution data. Based on the terrain classification data and the historical fire point spatial distribution data, the superimposed region of the high complexity region and the historical fire point adjacent range is identified through spatial correlation analysis to obtain an identification result. Based on the identification result, multi-factor weight evaluation is performed in combination with the surface combustible material characteristics to generate a key region distribution map; Based on the key region distribution map and the unmanned aerial vehicle performance parameter library, a dynamic matching relationship between the detection level and the flight parameters is established through a priority driving mechanism, and a dynamic matching result is obtained. Based on the dynamic matching result, a scanning and gradient processing is performed on the high priority region and the low priority region through an aerial photography parameter adaptive adjustment strategy to generate a flight parameter configuration table; Based on the gridded terrain classification data, the key region distribution map, and the flight parameter configuration table, a path optimization model is constructed through a multi-objective A* algorithm; Based on the path optimization model, in combination with the terrain complexity gradient change limitation and the endurance time requirement, a shortest distance calculation is performed to determine a preliminary detection route; Based on the preliminary detection route and the real-time obstacle database, a three-dimensional dynamic safety corridor model is constructed; Based on the three-dimensional dynamic safety corridor model, through spatial conflict detection and B-spline curve flight path smoothing optimization algorithm, a detection collection route is generated under the condition of meeting the obstacle avoidance constraint. 3.The AR-based forest firefighting topography simulation training method of claim 1, wherein, From the standard terrain data, vegetation coverage data is identified and extracted, and coverage mapping is performed in the basic terrain grid to generate a vegetation structure model, specifically as follows: Satellite remote sensing spectral features and ground measured vegetation sample data are obtained. Based on the standard terrain data, satellite remote sensing spectral features, and ground measured vegetation sample data, a three-dimensional correlation matrix related to terrain elevation, spectral reflectance, and vegetation type is constructed. Based on the three-dimensional correlation matrix, a feature matching algorithm is used to perform registration processing on the multi-dimensional data space, and a vegetation feature data set is generated. A deep convolution classification model is constructed based on the vegetation feature dataset and the forest ecological database, multi-band spectral features and terrain slope data are integrated through the input layer of the deep convolution classification model, the attention mechanism in the deep convolution classification model is used to strengthen feature extraction, and the forest ecological parameters in the forest ecological database are combined to calculate the vegetation type probability distribution, and a vegetation distribution heat map is output; A terrain height distribution parameter is obtained through a basic terrain grid, a hierarchical difference value coverage density in the vegetation distribution heat map is taken as a basis for a horizontal dimension, the terrain height distribution parameter is taken as a basis for a vertical dimension, and a vegetation network model is constructed through three-dimensional density field space mapping; A canopy leaf water content dynamic attenuation parameter, a tree trunk lignin heat value change parameter, and a ground litter layer humidity diffusion coefficient are obtained from the forest ecological database, the canopy leaf water content dynamic attenuation parameter, the tree trunk lignin heat value change parameter, and the ground litter layer humidity diffusion coefficient are subjected to hierarchical parameterization mapping processing by using the vegetation network model, and a static structure model is generated; Real-time ecological parameter library is obtained, and a vegetation structure model is generated by simulation through the static structure model. 4.The AR-based forest firefighting terrain simulation training method of claim 3, wherein, The three-dimensional terrain model is used to generate a fire training scene through multi-physical field coupling simulation, specifically including the following steps: Obtain fire source parameter data and dynamic meteorological data; In the three-dimensional terrain model, multi-physical field coupling simulation is performed according to the fire source parameter data and the dynamic meteorological data to generate a fire training scene. 5.The AR-based forest firefighting topography simulation training method of claim 4, wherein, In the three-dimensional terrain model, multi-physical field coupling simulation is performed according to the fire source parameter data and the dynamic meteorological data to generate a fire training scene, specifically as follows: Obtain slope direction data based on the three-dimensional terrain model, obtain burning attribute parameters based on the vegetation structure model, and construct an association matrix of fire propagation direction and terrain characteristics based on the slope direction data and the burning attribute parameters; calculate the fire line advance rate of different vegetation types based on the association matrix of fire propagation direction and terrain characteristics to generate a fire propagation vector diagram; Obtain wind speed and direction parameters from the dynamic meteorological data, and establish an energy exchange model of thermal convection and air flow based on the fire propagation vector diagram and the wind speed and direction parameters in a manner of spatial superposition of meteorological vector field and fire propagation field; generate a three-dimensional fire dynamic distribution field based on the energy exchange model; Obtain three-dimensional burning surface data and vegetation spatial distribution characteristics through the vegetation structure model, and obtain flame surface radiation flux by solving a heat conduction equation based on the three-dimensional fire dynamic distribution field and the three-dimensional burning surface data; construct a thermal radiation gradient field based on the vegetation spatial distribution characteristics and the flame surface radiation flux; generate a dynamic thermal radiation distribution cloud image based on the thermal radiation gradient field; Obtain airflow channel data and terrain elevation shielding parameters through the three-dimensional terrain model, simulate a smoke rising trajectory through thermal buoyancy effect based on the dynamic thermal radiation distribution cloud image and the airflow channel data to obtain simulation results; and generate a dynamic smoke particle motion model based on the simulation results and the terrain elevation shielding parameters. Based on the dynamic thermal radiation distribution cloud map and the dynamic smoke particle motion model, the corresponding relationship between the thermal radiation intensity and the color temperature value is established to obtain the AR visual color gradient field; based on the AR visual color gradient field, combined with the smoke transparency dynamic map layer construction technology, the fire training scene is generated. 6.The AR-based forest firefighting terrain simulation training method of claim 1, wherein, The evaluation of the plurality of training monitoring data, the selection of representative training data, the generation of representative guidance information, and the step-by-step guidance and monitoring of the AR fire training of the plurality of firefighters specifically include the following steps: behavior evaluation of the plurality of training monitoring data, recording behavior evaluation data; effect evaluation of the plurality of training monitoring data, recording effect evaluation data; comprehensive behavior evaluation data and effect evaluation data, and selection of representative training data; creating representative guidance information according to the representative training data; in the fire training scene, according to the representative guidance information, the step-by-step guidance and monitoring of the AR fire training of the plurality of firefighters.

7. An AR-based forest fire topography simulation training system, characterized by, The system applies the AR-based forest fire terrain simulation training method according to any one of claims 1 to 6, and the system comprises: a terrain model construction unit for determining a target forest area, collecting regional terrain data of the target forest area, and constructing a three-dimensional terrain model through data fusion and spatial analysis; a dynamic fire simulation unit for generating a fire training scene through multi-physical field coupling simulation based on the three-dimensional terrain model; an initial guidance processing unit for creating initial guidance information, initially guiding the AR fire training of the plurality of firefighters in the fire training scene, and collecting a plurality of training monitoring data through the construction of a cross-modal interaction model; a step-by-step representative guidance unit for evaluating the plurality of training monitoring data, selecting representative training data, generating representative guidance information, and step-by-step guiding and monitoring the AR fire training of the plurality of firefighters.

Images