Stereoscopic precise positioning method for trapped personnel in catastrophic environment

Abstract

The application relates to a kind of trapped personnel stereoscopic precision positioning method under catastrophic environment in the technical field of intelligent mine safety monitoring and emergency rescue, comprising: domestic passive intrinsic safety type positioning terminal deployment and debugging;Underground three-dimensional space model construction and existing video resource reuse;Dual positioning mode initialization and automatic switching logic setting;Multi-source signal fusion anti-interference algorithm deployment and calibration;Positioning and vital signs synchronous acquisition and hazard level automatic determination;Stereoscopic rescue route automatic generation and optimal path optimization;Positioning-rescue linkage execution, real-time early warning and route dynamic updating;System operation and maintenance, error analysis and optimization iteration.The present application solves the core technical problems of the existing mine underground personnel positioning technology in the disaster scene, such as insufficient positioning dimension, poor environmental adaptability and low positioning accuracy.

Description

Technical Field

[0001] This invention belongs to the field of intelligent mine safety monitoring and emergency rescue technology, specifically involving a method for precise three-dimensional positioning of trapped personnel in disaster environments. This method is mainly applied to extreme disaster scenarios in underground mines, such as low oxygen, high gas, strong dust, and communication disruptions. It relies on an intrinsically safe three-dimensional positioning terminal for mining to accurately collect and analyze the three-dimensional coordinates of trapped personnel. Adapting to the personnel distribution characteristics of roadways and chambers at different elevations, it provides the ground rescue command center with accurate information on personnel location and danger zone identification, supporting the planning of three-dimensional rescue routes. This is a significant innovative application in the field of mine emergency rescue positioning technology and can be widely applied to various underground mine emergency rescue scenarios. Background Technology

[0002] As mining extends deeper, the mining environment becomes increasingly complex, with frequent hazards such as low oxygen, high gas levels, and strong dust. Rapid and accurate location of trapped personnel is crucial for improving rescue efficiency and reducing casualties, and is also an important component of smart mine safety construction. Currently, most underground personnel positioning technologies in mines adopt a one-dimensional positioning mode arranged along the roadway, which can only obtain the personnel's position information along the roadway axis and cannot achieve accurate representation of three-dimensional spatial coordinates.

[0003] Existing one-dimensional positioning technology has obvious limitations: On the one hand, underground mine roadways are complex and have significant elevation differences. The distribution of personnel in different elevation areas or adjacent chambers in the same roadway cannot be distinguished by one-dimensional positioning, which can easily lead to rescuers misjudging the location of trapped personnel. On the other hand, after a disaster, communication interruptions and rapid deterioration of environmental parameters often occur underground. Traditional positioning terminals are difficult to adapt to extreme environments, resulting in a significant decrease in positioning accuracy or even failure to function properly.

[0004] Although some underground positioning solutions exist in existing technologies, such as those based on UWB and inertial navigation, most of these solutions are not specifically adapted to extreme disaster environments and still fall short of the limitations of one-dimensional positioning, failing to meet the demands of three-dimensional rescue for personnel's three-dimensional location information. Therefore, current mine disaster rescue efforts lack a positioning method capable of stably operating in extreme environments and accurately acquiring the three-dimensional coordinates of trapped personnel. This hinders the rescue command center from developing scientifically efficient three-dimensional rescue routes, thus restricting the improvement of rescue efficiency. An advanced three-dimensional precision positioning technology adapted to disaster scenarios is urgently needed to solve these problems. Summary of the Invention

[0005] This invention aims to solve the core technical problems of existing underground personnel positioning technology in disaster scenarios, such as insufficient positioning dimensions, poor environmental adaptability, and low positioning accuracy. Specifically: First, it addresses the limitation of existing positioning technologies, which can only achieve one-dimensional positioning within roadways. Current mine positioning methods are mostly arranged along the roadway axis, only obtaining one-dimensional location information of personnel. They cannot represent the differences in personnel position in the vertical elevation direction, and cannot adapt to the personnel distribution in roadways and chambers at different elevations. This makes it impossible for the rescue command center to accurately determine the specific spatial location of trapped personnel, which can easily delay rescue opportunities.

[0006] Second, it addresses the issues of unstable operation and insufficient positioning accuracy of positioning terminals in extreme disaster environments. After a mine disaster, extreme conditions such as low oxygen, high gas, strong dust, and communication interruptions may occur underground. Traditional positioning terminals lack targeted intrinsic safety design and environmental adaptability, making them prone to equipment failure and data acquisition distortion. Consequently, they cannot accurately collect the location data of trapped personnel, making it difficult to provide a reliable basis for rescue decisions.

[0007] Third, this invention addresses the problem that existing positioning methods cannot support the development of three-dimensional rescue routes. Current positioning technologies cannot provide the three-dimensional coordinates of trapped personnel, as well as information on their location within the tunnel and their elevation. This prevents the rescue command center from determining the danger level of the area where personnel are located, resulting in a lack of accurate data support for rescue route planning. Consequently, problems such as unreasonable rescue routes and low rescue efficiency arise, hindering efficient and precise rescue efforts. This invention effectively solves these problems by constructing a three-dimensional positioning system adapted to disaster environments, thereby improving the accuracy and efficiency of mine disaster rescue.

[0008] To address the aforementioned problems, this invention provides a method for precise three-dimensional positioning of trapped personnel in a disaster environment, comprising the following steps: Step 1: Deployment and debugging of domestically produced passive intrinsically safe positioning terminals: Deploy in batches intrinsically safe passive positioning terminals for mining, consisting of domestically produced UWB positioning chips and RF power amplifiers. Deploy the intrinsically safe passive positioning terminals in batches at key locations in various roadways and chambers underground. At the same time, equip underground workers with portable wearable terminals, complete the collaborative debugging of the terminals with the existing domestic positioning system underground, calibrate the positioning accuracy and communication distance, test the stability of the terminals in low-oxygen and high-gas environments, and verify the efficiency of the passive power supply structure in collecting weak energy underground. Step 2: Construction of underground 3D spatial model and reuse of existing video resources: Reuse existing video surveillance resources in the mining area, collect spatial parameters of each roadway and chamber underground through video dynamic target 3D reconstruction technology, construct a computable underground 3D spatial model, mark the location information, size parameters and key facility locations of roadways and chambers at different elevations, and complete the linkage debugging of the model with positioning terminals and video surveillance equipment; Step 3: Initialization and Automatic Switching Logic Setting of Dual Positioning Mode: Start the dual positioning system of non-contact positioning and wearable terminal supplementation, complete the initialization and debugging of the two positioning modes, and set the connection algorithm parameters of the two modes to achieve seamless switching; Step 4 Deployment and calibration of multi-source signal fusion anti-interference algorithm: Simultaneously deploy an adaptive anti-interference algorithm for fusion of UWB radar, inertial navigation and Beidou RTK multi-source signals in the ground rescue command center and the underground positioning terminal, complete the fusion calibration of the three signals, set signal priorities, introduce MIMO technology and optimize K-means++ algorithm to complete multi-target positioning calibration; Step 5: Location and vital signs are collected simultaneously and danger level is automatically determined: Relying on the closed-loop positioning mechanism, the miniature intrinsically safe vital signs sensor integrated in the positioning terminal is activated to collect the vital signs data of the trapped personnel simultaneously and correlate them with the three-dimensional coordinate data in real time. Through a preset algorithm and combined with real-time environmental parameters underground, the survival status of the trapped personnel and the danger level of the area are determined, and the location data, vital signs data and danger level information are transmitted to the ground rescue command center simultaneously. Step 6 Automatic generation and optimal path optimization of three-dimensional rescue route: Based on the received positioning data, vital sign data, and hazard level information, the ground rescue command center combines the underground three-dimensional spatial model and real-time environmental parameters to automatically calculate and generate the optimal three-dimensional rescue route, marking dangerous nodes and avoidance schemes on the route. At the same time, rescuers can manually fine-tune the automatically generated route according to the actual situation on site to optimize the rescue path. Step 7: Location-Rescue Linkage Execution, Real-time Early Warning, and Dynamic Route Updates: Activate the location-rescue linkage mechanism to synchronously transmit the optimized three-dimensional rescue route to the terminal equipment of the underground rescue personnel. Real-time synchronization of the trapped personnel's location changes and vital sign changes is maintained. During the rescue, the system continuously collects location data and environmental parameters. If the trapped personnel's location moves, vital signs become abnormal, or new dangerous nodes appear on the rescue route, the system immediately issues an early warning and automatically updates the rescue route. Simultaneously, it links with the underground intelligent ventilation and emergency broadcasting systems to adjust the ventilation status of the rescue area and transmit rescue information to the trapped personnel. Step 8 System Operation and Maintenance, Error Analysis and Optimization Iteration: After the rescue is completed, a comprehensive inspection of the entire positioning system will be carried out to check the operating status of the positioning terminals, sensors and algorithms. Positioning data, fault information and optimization suggestions during the rescue process will be collected. For the problems that occurred in the disaster scenario, the terminal structure, fusion algorithm and switching logic will be optimized and iterated. At the same time, a normalized operation and maintenance mechanism will be established to regularly calibrate the positioning terminals, 3D spatial models and algorithm parameters.

[0009] In order to achieve quantitative verification of debugging parameters, in step 1, a passive power supply energy acquisition model and a three-dimensional spatial positioning error calibration model are established to achieve precise calibration of the power supply stability and positioning accuracy of the positioning terminal. The passive power supply energy harvesting model is as follows: in, Passive power supply energy harvesting efficiency; The actual usable electrical energy converted at the terminal; The total, albeit weak, energy collected by the terminal in the well. The three-dimensional spatial positioning error calibration model is as follows: in, : 3D positioning error; : The actual three-dimensional coordinates of the personnel deploying / wearing the positioning terminal; : The three-dimensional coordinates collected by the positioning terminal.

[0010] To ensure that the model can accurately represent the spatial characteristics of different elevations and regions underground, in step 2, an underground spatial coordinate mapping model and a model accuracy verification model are established to achieve the accurate construction of the underground three-dimensional spatial model. The underground spatial coordinate mapping model is as follows: in, : Actual three-dimensional spatial coordinates underground; Two-dimensional pixel coordinates in a video image; : The principal coordinates of the video camera; , The focal length of the camera in the x and y directions; The vertical distance from the target point to the camera; Model accuracy verification model is as follows in, : The average deviation of the three-dimensional spatial model; Number of verification points; : The actual three-dimensional coordinates of the i-th verification point; : The coordinates of the i-th verification point in the three-dimensional spatial model.

[0011] To ensure that the switching logic is scientific and reliable, in step 3, a mode switching judgment model and a connection weight model are established to achieve accurate switching and seamless connection between the two positioning modes. The mode switching determination model is as follows: in, Mode switching indicator; Positioning error in non-sensory positioning; : Threshold for non-sensory positioning accuracy; Video signal sharpness factor; Video signal clarity threshold; The connection weight model is as follows: in, : The merged positioning coordinates; Weighting coefficients for seamless positioning; : Three-dimensional coordinates acquired through non-sensory positioning; Weighting coefficients for wearable device positioning; : Three-dimensional coordinates collected by the wearable terminal.

[0012] To ensure the accuracy and stability of the fused positioning data, in step 4, an adaptive fusion model for multi-source signals and a clustering model for multi-target positioning are established to achieve anti-interference fusion of multi-source signals and accurate differentiation of multiple targets. The multi-source signal adaptive fusion model is as follows: ,in in, The final three-dimensional positioning coordinates after multi-source signal fusion; Signal type identifier; : Adaptive weighting coefficients for the k-th signal; : The three-dimensional positioning coordinates of the k-th signal acquisition; Signal weights The dynamic adjustment is based on the mean square error of the signal. The smaller the mean square error, the higher the signal reliability and the greater the weight. The mean square error calculation model is as follows: in, : The mean square error of the k-th signal; The number of times the k-th signal is collected; The three-dimensional coordinates of the k-th signal acquired in the j-th acquisition; : The actual three-dimensional coordinates during the j-th acquisition; The multi-target localization clustering model is as follows: in, Clustering objective function; Number of people trapped; : The location dataset of the i-th cluster; Cluster A single positioning coordinate in the data; : The center coordinates of the i-th cluster; Location coordinates P and cluster center The square of the Euclidean distance.

[0013] In order to quantify the judgment indicators and ensure that the judgment results are scientific and reliable, in step 5, a vital sign abnormality judgment model and a risk level comprehensive evaluation model are established to achieve accurate judgment of the survival status and risk level of the trapped personnel. The model for determining abnormal vital signs is as follows: in, : Survival status indicator for trapped personnel; Real-time heart rate of the trapped personnel; Heart rate lower limit threshold; Heart rate upper limit threshold; Real-time respiratory rate of the trapped personnel; : Lower limit threshold for respiratory rate; Upper limit threshold for respiratory rate; The comprehensive risk assessment model is as follows: in, : Comprehensive evaluation value of hazard level; : Gas concentration weighting coefficient; Gas concentration hazard score; Oxygen content weighting coefficient; Oxygen content hazard score; Collapse risk weighting coefficient; Collapse risk hazard score; : Methane concentration in the well; Oxygen content in the well; The straight-line distance between the trapped personnel and the collapsed area.

[0014] To quantify the safety and efficiency of the route and ensure that the generated route is the optimal route, in step 6, a rescue route cost function model and a rescue route optimization model are established to realize the automatic generation and optimization of the three-dimensional rescue route. The rescue route cost function model is as follows: in, Total cost of the rescue route; : Route length weighting coefficient; : Total length coefficient of rescue route; Risk level weighting coefficient; The average hazard level assessment value along the rescue route; Difficulty of passage weighting coefficient; Difficulty level of the rescue route; The rescue route optimization model is as follows: in, : The total cost of the optimized route; : The total cost of the i-th candidate rescue route; Number of candidate rescue routes.

[0015] To ensure safety and timeliness during the rescue process, step 7 establishes a real-time early warning model for the rescue process and a dynamic update model for the rescue route to achieve real-time linkage and early warning between location and rescue. The real-time early warning model for the rescue process is as follows: in, Warning sign; Changes in the heart rate of the trapped individuals; Heart rate variability warning threshold; Changes in the location of trapped personnel , The current location coordinates, The coordinates were determined at the previous moment; Location change warning threshold; Changes in environmental parameters; : Warning threshold for changes in environmental parameters; The dynamic update model for rescue routes is as follows: in, : Total cost of the updated rescue route; : The updated total length of the rescue route; : Updated average hazard rating of rescue routes; The updated difficulty level of the rescue route; , , : Consistent with the weighting coefficients in step 6.

[0016] In order to quantify system operational defects and determine the operation and maintenance calibration cycle, step 8 establishes a system operation error analysis model and an operation and maintenance cycle calibration model to achieve precise optimization iteration and routine operation and maintenance of the system. The system operation error analysis model is as follows: in, Overall system operating error; Total amount of location data collected during this rescue operation; : The final output coordinates of the i-th set of positioning data; : The actual coordinates corresponding to the i-th set of positioning data; The operation and maintenance cycle calibration model is as follows: in, System maintenance and calibration cycle; Overall system operating error.

[0017] In summary, compared with the prior art, the present invention has the following technical advantages and beneficial effects: (1) The terminal of this invention relies on a passive power supply energy acquisition model. Its efficiency is stable at over 85% in all disaster scenarios. It solves the core pain point of existing terminals that "power outages cause positioning failures". It enables the terminal to operate stably for a long time without maintenance. In terms of process, it saves the maintenance link of "periodic charging / battery replacement" and reduces energy consumption by about 60% (passive acquisition of underground environmental energy, no external power supply required). (2) Based on the positioning accuracy calibration model, the error of all points is controlled within 0.1m, and the positioning accuracy is improved by more than 80%. The high-precision positioning directly solves the problem of "positioning ambiguity leading to deviation of rescue route and missing the golden rescue time" in the existing technology, and improves the rescue efficiency by about 70% (precise positioning reduces the invalid search range). At the same time, the operation simplifies the process of "manual calibration of positioning deviation". (3) Based on the mode switching judgment model and the connection weight model, the switching response time is ≤0.5s and the weight linear transition achieves seamless connection, which solves the problem of "switching lag causing positioning blind spots". The response time is shortened by more than 85%, making the positioning data continuous and uninterrupted during the rescue process, making the control operation simpler, and reducing the process of "manually supplementing positioning data", thus improving efficiency by about 65%. (4) Based on the rescue route cost function and Dijkstra optimization model, the planning time is ≤1.1min, the route cost is reduced to within 3.8, the planning efficiency is improved by more than 90%, and the route cost is reduced by 60%. This invention solves the problem of "long time and high cost of route planning leading to rescue delays" in the existing technology, and increases the rescue success rate within the golden rescue time (48h) from the traditional 40% to 85%. At the same time, the control simplifies the operation of "manual adjustment of rescue route" and reduces the operation complexity of rescue personnel. Attached Figure Description

[0018] Figure 1 This is a flowchart of the three-dimensional precise positioning method for trapped personnel in a disaster environment according to the present invention; Figure 2 This is a comparison chart of the power supply energy harvesting efficiency of the passive positioning terminal of the present invention and the existing active positioning terminal; Figure 3 This is a comparison chart of the three-dimensional positioning accuracy of the present invention and traditional positioning technologies; Figure 4 This is a comparison chart of the response time for switching between dual positioning modes of the present invention and traditional positioning technologies; Figure 5 This is a comparison chart of the rescue route optimization efficiency of this invention and traditional positioning technology. Detailed Implementation

[0019] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby providing a clearer and more explicit definition of the scope of protection of the present invention.

[0020] like Figure 1 The method for precise three-dimensional positioning of trapped personnel in a disaster environment, as shown, includes the following steps: Step 1: Deployment and debugging of domestically produced passive intrinsically safe positioning terminals: Deploy in batches intrinsically safe passive positioning terminals for mining, consisting of domestically produced UWB positioning chips and RF power amplifiers. Deploy the intrinsically safe passive positioning terminals in batches at key locations in various roadways and chambers underground. At the same time, equip underground workers with portable wearable terminals, complete the collaborative debugging of the terminals with the existing domestic positioning system underground, calibrate the positioning accuracy and communication distance, test the stability of the terminals in low-oxygen and high-gas environments, and verify the efficiency of the passive power supply structure in collecting weak energy underground. Step 2: Construction of underground 3D spatial model and reuse of existing video resources: Reuse existing video surveillance resources in the mining area, collect spatial parameters of each roadway and chamber underground through video dynamic target 3D reconstruction technology, construct a computable underground 3D spatial model, mark the location information, size parameters and key facility locations of roadways and chambers at different elevations, and complete the linkage debugging of the model with positioning terminals and video surveillance equipment; Step 3: Initialization and Automatic Switching Logic Setting of Dual Positioning Mode: Start the dual positioning system of non-contact positioning and wearable terminal supplementation, complete the initialization and debugging of the two positioning modes, and set the connection algorithm parameters of the two modes to achieve seamless switching; Step 4 Deployment and calibration of multi-source signal fusion anti-interference algorithm: Simultaneously deploy an adaptive anti-interference algorithm for fusion of UWB radar, inertial navigation and Beidou RTK multi-source signals in the ground rescue command center and the underground positioning terminal, complete the fusion calibration of the three signals, set signal priorities, introduce MIMO technology and optimize K-means++ algorithm to complete multi-target positioning calibration; Step 5: Location and vital signs are collected simultaneously and danger level is automatically determined: Relying on the closed-loop positioning mechanism, the miniature intrinsically safe vital signs sensor integrated in the positioning terminal is activated to collect the vital signs data of the trapped personnel simultaneously and correlate them with the three-dimensional coordinate data in real time. Through a preset algorithm and combined with real-time environmental parameters underground, the survival status of the trapped personnel and the danger level of the area are determined, and the location data, vital signs data and danger level information are transmitted to the ground rescue command center simultaneously. Step 6 Automatic generation and optimal path optimization of three-dimensional rescue route: Based on the received positioning data, vital sign data, and hazard level information, the ground rescue command center combines the underground three-dimensional spatial model and real-time environmental parameters to automatically calculate and generate the optimal three-dimensional rescue route, marking dangerous nodes and avoidance schemes on the route. At the same time, rescuers can manually fine-tune the automatically generated route according to the actual situation on site to optimize the rescue path. Step 7: Location-Rescue Linkage Execution, Real-time Early Warning, and Dynamic Route Updates: Activate the location-rescue linkage mechanism to synchronously transmit the optimized three-dimensional rescue route to the terminal equipment of the underground rescue personnel. Real-time synchronization of the trapped personnel's location changes and vital sign changes is maintained. During the rescue, the system continuously collects location data and environmental parameters. If the trapped personnel's location moves, vital signs become abnormal, or new dangerous nodes appear on the rescue route, the system immediately issues an early warning and automatically updates the rescue route. Simultaneously, it links with the underground intelligent ventilation and emergency broadcasting systems to adjust the ventilation status of the rescue area and transmit rescue information to the trapped personnel. Step 8 System Operation and Maintenance, Error Analysis and Optimization Iteration: After the rescue is completed, a comprehensive inspection of the entire positioning system will be carried out to check the operating status of the positioning terminals, sensors and algorithms. Positioning data, fault information and optimization suggestions during the rescue process will be collected. For the problems that occurred in the disaster scenario, the terminal structure, fusion algorithm and switching logic will be optimized and iterated. At the same time, a normalized operation and maintenance mechanism will be established to regularly calibrate the positioning terminals, 3D spatial models and algorithm parameters.

[0021] The steps described above will be explained in detail below.

[0022] Step 1: Deployment and debugging of domestically produced passive intrinsically safe positioning terminal The project involves the batch deployment of intrinsically safe passive 3D positioning terminals for mining, composed of domestically produced UWB positioning chips and RF power amplifiers. These terminals utilize a passive power supply structure, eliminating the need for additional charging or battery replacement. They are deployed in batches to key locations in various underground roadways and chambers. Simultaneously, portable wearable terminals are provided to underground workers. The project also includes coordinating and debugging the terminals with existing domestically produced positioning systems underground, calibrating positioning accuracy and communication distance, testing the stability of the terminals in low-oxygen and high-gas environments, verifying the efficiency of the passive power supply structure in collecting low-energy signals underground, and ensuring long-term stable operation of the terminals.

[0023] To accurately calibrate the power supply stability and positioning accuracy of the positioning terminal, the following model is established for the quantitative verification of debugging parameters: (1) Passive power supply energy harvesting model: The core of passive power supply for terminals is to harvest weak energy from underground and convert it into usable electrical energy to ensure the normal operation of the terminals. Its energy harvesting efficiency model is as follows: in: Passive power supply energy harvesting efficiency, unitless, ranges from 0% to 100%, and must be ensured during commissioning. To ensure long-term power supply to the terminals; :The actual usable electrical energy converted at the terminal, in joules (J). The total energy collected by the terminal downhole is measured in joules (J), and mainly comes from weak vibrations, temperature differences (temperature difference between the disaster area and the normal area), and weak electrical energy from gas combustion.

[0024] (2) Three-dimensional spatial positioning error calibration model: Positioning accuracy is the core performance indicator of the terminal. The three-dimensional spatial positioning error formula is used for calibration to ensure that the positioning accuracy is ≤10cm. The three-dimensional spatial positioning error calibration model is as follows: in: 3D positioning error, in meters (m), debugging target ; The actual three-dimensional coordinates of the personnel deploying / wearing the positioning terminal, in meters (m), are determined by the underground reference coordinate points; : The three-dimensional coordinates collected by the positioning terminal, in meters (m).

[0025] During the debugging process, the passive power supply efficiency and positioning accuracy were quantitatively verified using the two models mentioned above. At the same time, the communication distance was tested to ensure that the communication distance is ≥600m, thus guaranteeing the stable operation of the terminal in the complex underground environment.

[0026] Step 2: Construction of the 3D spatial model of the well and reuse of existing video resources By reusing existing video surveillance resources in the mining area, there is no need to deploy additional equipment such as lidar. Through video dynamic target 3D reconstruction technology, spatial parameters of each roadway and chamber underground are collected to construct a calculable underground 3D spatial model. The location information, size parameters, and location of key facilities (such as ventilation openings and refuge chambers) of roadways and chambers at different elevations are marked. The linkage debugging of the model with positioning terminals and video surveillance equipment is completed to ensure that the model can receive positioning data and video data in real time, laying the foundation for subsequent seamless positioning and dual positioning switching.

[0027] To achieve accurate construction of the downhole 3D spatial model, a downhole spatial coordinate mapping model and a model accuracy verification model were established based on video dynamic target 3D reconstruction technology. This ensures that the model can accurately represent the spatial characteristics of different elevations and regions downhole, as detailed below: (1) Underground spatial coordinate mapping model: The coordinates of the two-dimensional images acquired by video surveillance are mapped to the actual three-dimensional spatial coordinates underground, realizing the transformation from two-dimensional images to three-dimensional space. The model is as follows: in: : Actual three-dimensional spatial coordinates underground (x: horizontal direction of the tunnel, y: vertical direction of the tunnel, z: elevation direction), in meters (m); Two-dimensional pixel coordinates in a video image, without units, collected by video surveillance equipment; The principal point coordinates (pixel coordinates of the image center point) of the video camera are unitless and determined by the camera parameters. , The focal length of the camera in the x and y directions, measured in pixels, is determined by the camera's factory specifications. The vertical distance from the target point to the camera, in meters (m), is obtained through depth calculation based on binocular vision video images.

[0028] (2) Model accuracy verification model: used to verify the deviation between the constructed downhole 3D space model and the actual downhole space, ensuring that the model accuracy meets the requirements of seamless positioning: in: The average deviation of the three-dimensional spatial model, in meters (m), is required to be... ; Number of verification points, unitless, selected from benchmark points at different elevations and in different areas underground, n≥50; : The actual three-dimensional coordinates of the i-th verification point, in meters (m); : The coordinates of the i-th verification point in the three-dimensional spatial model, in meters (m).

[0029] Using the above methods, the construction and accuracy verification of the downhole three-dimensional spatial model are completed, ensuring that the model can accurately map the actual downhole space and provide a reliable spatial carrier for subsequent seamless positioning.

[0030] Step 3: Initialization of Dual Positioning Mode and Setting of Automatic Switching Logic The dual positioning system of "seamless positioning + wearable terminal supplementation" was initiated, and the initialization and debugging of the two positioning modes were completed. The mode switching logic was set: Under normal working conditions, the seamless positioning mode is used first, and the personnel are positioned in three dimensions without being noticed by the underground three-dimensional space model and video monitoring; when the video signal is detected to be blocked by dust or collapsed objects, or the video positioning accuracy is lower than the preset threshold, the system automatically switches to the wearable terminal positioning mode, and collects the personnel's three-dimensional coordinates through the wearable terminal to ensure uninterrupted positioning; at the same time, the connection algorithm parameters of the two modes are set to achieve seamless switching and avoid positioning blind spots and data gaps.

[0031] To achieve precise switching and seamless transition between dual positioning modes, a mode switching determination model and a transition weight model are established to quantify switching conditions and ensure that the switching logic is scientific and reliable, as detailed below: (1) Mode switching determination model: The video positioning accuracy and signal obstruction degree are used as switching determination indicators. When the switching conditions are met, the system automatically triggers mode switching: in: : Mode switching indicator, no unit, S=1 indicates switching to wearable terminal positioning mode, S=0 indicates maintaining non-intrusive positioning mode; The positioning error of the non-contact positioning (video positioning), in meters (m), is calculated from the positioning accuracy model in step 2. The threshold for seamless positioning accuracy, measured in meters (m), is set based on the underground scenario and is typically set to [value missing]. ; Video signal sharpness factor, unitless, ranges from 0 to 1. The closer to 1, the clearer and unobstructed the signal; Video signal sharpness threshold, no unit, usually taken as... ,when When this happens, it is determined that the signal is blocked.

[0032] (2) Weighted Integration Model: To achieve seamless integration of the two positioning modes and avoid gaps in positioning data, a weighted fusion method is adopted to process the positioning data of the two modes: in: The merged positioning coordinates, in meters (m), are the final output three-dimensional coordinates of the personnel. The weighting coefficient for seamless positioning (video positioning) is dimensionless and ranges from 0 to 1. When S=0, , When S=1, , During the switching process, the weighting coefficients transition linearly. : Three-dimensional coordinates acquired by non-contact positioning (video positioning), in meters (m); Weighting coefficients for wearable device positioning, unitless. ; : Three-dimensional coordinates collected by the wearable terminal, in meters (m).

[0033] Complete the initialization and switching logic settings for dual positioning modes to ensure that the two modes can automatically switch and seamlessly connect according to the actual underground environment, avoiding positioning blind spots.

[0034] Step 4: Deployment and calibration of multi-source signal fusion anti-interference algorithm An adaptive anti-interference algorithm for multi-source signal fusion, integrating UWB radar, inertial navigation, and BeiDou RTK, was simultaneously deployed in the ground rescue command center and the underground positioning terminal. Fusion calibration of the three signals was completed, and signal priorities were set: UWB radar signals were prioritized for penetration of collapsed debris for positioning; inertial navigation signals were used to compensate for positioning blind spots during communication interruptions; and BeiDou RTK signals were used to improve positioning accuracy in the elevation direction. MIMO technology and an optimized K-means++ algorithm were introduced to complete multi-target positioning calibration, ensuring effective differentiation of the location information of multiple trapped personnel and resolving positioning deviations caused by signal interference and obstruction.

[0035] To achieve interference-resistant fusion of multi-source signals and accurate differentiation of multiple targets, an adaptive fusion model for multi-source signals and a clustering model for multi-target localization are established to ensure the accuracy and stability of the fused localization data, as detailed below: (1) Adaptive fusion model of multi-source signals: An adaptive weighted fusion algorithm is adopted to dynamically adjust the weights of each signal according to the reliability of the real-time accuracy of different signals, so as to realize the fusion of three signals: UWB radar, inertial navigation and Beidou RTK. The model is as follows: ,in in: : The final three-dimensional positioning coordinates after multi-source signal fusion, in meters (m); Signal type identifier, no unit, k=1 represents UWB radar signal, k=2 represents inertial navigation signal, k=3 represents Beidou RTK signal; The adaptive weighting coefficient for the k-th signal is dimensionless and ranges from 0 to 1. It is dynamically adjusted based on the real-time reliability of the signal; the higher the reliability, the greater the weight. : The three-dimensional positioning coordinates of the k-th signal acquisition, in meters (m).

[0036] Signal weights The dynamic adjustment is based on the mean square error (MSE) of the signal. The smaller the MSE, the higher the signal reliability and the greater the weight. The MSE calculation model is as follows: in: : Mean square error of the kth signal, in square meters (m²). : The number of times the k-th signal is collected, without unit, m≥30 times; : The three-dimensional coordinates of the k-th signal acquired in the j-th acquisition, in meters (m); : The actual three-dimensional coordinates during the j-th acquisition, in meters (m).

[0037] (2) Multi-target localization clustering model: An optimized K-means++ algorithm is introduced to cluster the localization data after multi-source signal fusion, distinguish the location information of different trapped persons, and avoid interference from multi-target signals. in: Clustering objective function (sum of squared errors within clusters), unitless; the smaller the J value, the better the clustering effect, that is, the clearer the distinction between the location data of different trapped persons. Number of trapped personnel (number of clusters), no unit, automatically identified by the system based on location data; : The location dataset of the i-th cluster (corresponding to the i-th trapped person), without units; Cluster A single positioning coordinate in the data, in meters (m); : The center coordinates of the i-th cluster (i.e., the final location coordinates of the i-th trapped person), in meters (m); Location coordinates P and cluster center The square of the Euclidean distance is expressed in square meters (m²).

[0038] The above methods enable the anti-interference fusion of multi-source signals and multi-target positioning calibration, ensuring that the location information of multiple trapped personnel can still be accurately distinguished in scenarios such as strong dust, obstruction by collapsed objects, and communication interruption, thereby improving positioning accuracy and anti-interference capability.

[0039] Step 5: Simultaneous location tracking and vital sign monitoring, along with automatic risk level assessment. Relying on a closed-loop positioning mechanism, the intrinsically safe miniature vital signs sensor integrated into the positioning terminal is activated to simultaneously collect vital signs data such as heart rate and respiratory rate of the trapped personnel, and correlate them with three-dimensional coordinate data in real time. Through a preset algorithm, combined with real-time environmental parameters underground (gas concentration, oxygen content), the survival status of the trapped personnel and the danger level of the area (high, medium, or low) are determined. The positioning data, vital signs data, and danger level information are then transmitted synchronously to the ground rescue command center to ensure that commanders have real-time access to information on the trapped personnel.

[0040] To accurately determine the survival status and danger level of trapped personnel, a vital sign abnormality assessment model and a comprehensive danger level evaluation model were established, quantifying assessment indicators to ensure the scientific and reliable assessment results, as detailed below: (1) Determination of abnormal vital signs: Using heart rate and respiratory rate as core indicators, the survival status of the trapped personnel is determined. in: T=1 indicates the survival status of trapped personnel. There is no unit. T=1 indicates survival, and T=0 indicates no vital signs. Real-time heart rate of trapped personnel, in beats per minute (bpm); Heart rate lower limit threshold, in beats per minute (bpm). ; Heart rate upper limit threshold, in beats per minute (bpm). ; Real-time respiratory rate of trapped personnel, measured in breaths per minute (bpm). : Lower limit threshold for respiratory rate, in breaths per minute (bpm). ; : Upper limit threshold for respiratory rate, in breaths per minute (bpm). .

[0041] (2) Comprehensive Hazard Level Assessment Model: Combining environmental parameters (gas concentration, oxygen content) and location information (whether the trapped personnel are near the collapse area) of the area, a weighted summation method is used to comprehensively determine the hazard level. The model is as follows: in: : Comprehensive risk level evaluation value, unitless, range from 0 to 10, D≥7 is high risk level, 4≤D<7 is medium risk level, D<4 is low risk level; Gas concentration weighting coefficient, unitless, taken as... ; Methane concentration hazard score, unitless, ranges from 0 to 10, based on methane concentration. Sure: When D=0, When D=3, When D=7, When D=10; Oxygen content weighting coefficient, unitless, taken as... ; Oxygen content hazard score, unitless, ranges from 0 to 10, based on oxygen content. Sure: When D=0, When D=3, When D=7, When D=10; Collapse risk weighting coefficient, unitless, taking... ; Collapse risk score, unitless, ranging from 0 to 10, based on the distance between people and the collapse area. Sure: When D=0, When D=3, When D=7, When D=10; : Methane concentration in the well, expressed as % (volume fraction); Oxygen content in the well, expressed as % (volume fraction); The straight-line distance between the trapped person and the collapsed area, in meters (m).

[0042] Using the above methods, the survival status and danger level of the trapped personnel can be accurately determined. The determination results, along with location data and vital sign data, are then transmitted synchronously to the ground rescue command center, providing a reliable basis for rescue decisions.

[0043] Step 6: Automatic generation and optimal path optimization of three-dimensional rescue routes Based on received location data, vital sign data, and hazard level information, combined with the underground 3D spatial model and real-time environmental parameters (tunnel collapse status, gas concentration distribution), the ground rescue command center automatically calculates and generates the optimal 3D rescue route, marking hazardous nodes (such as high-gas areas and collapsed sections) and avoidance schemes along the route. Simultaneously, rescue personnel can manually fine-tune the automatically generated route according to the actual situation on site, optimizing the rescue path to ensure safety and efficiency, achieving seamless linkage between "location and path planning."

[0044] To achieve automatic generation and optimization of three-dimensional rescue routes, a rescue route cost function model and a rescue route optimization model are established to quantify the safety and efficiency of the routes and ensure that the generated routes are optimal, as detailed below: (1) Rescue route cost function model: Using route length, hazard level, and accessibility as core evaluation indicators, a route cost function is constructed. The smaller the cost, the better the route. The model is as follows: in: Total cost of the rescue route, unitless; the lower the Cost value, the better the route. : Route length weighting coefficient, unitless, taking ; : Total length coefficient of rescue route, in meters (m); Risk level weighting coefficient, unitless, taken as... ; The average hazard level assessment value along the rescue route, without units, is calculated by the comprehensive hazard level assessment model in step 5, and is the average value of the hazard level assessment values ​​of all nodes on the route. : Difficulty weighting coefficient for passage, no unit, take ; : Difficulty coefficient of rescue route passage, unitless, ranges from 0 to 10, and is determined according to the width, slope and obstacles of the alley: T=1 when the alley width is ≥2m, the slope is <15° and there are no obstacles; T=5 when the alley width is 1~2m, the slope is 15°~30° and there are a few obstacles; T=10 when the alley width is <1m, the slope is ≥30° and there are a lot of obstacles.

[0045] (2) Rescue route optimization model: Based on the above cost function, the shortest path algorithm (Dijkstra's algorithm) is used to search for the route with the minimum cost in the downhole three-dimensional space model to achieve automatic optimization of the rescue route. The model is as follows: in: Total cost of the optimized route, unitless; : The total cost of the i-th candidate rescue route, unitless, i=1,2,...,m; : The number of candidate rescue routes, without units, automatically generated by the system based on the downhole three-dimensional spatial model, m≥3 routes.

[0046] The two models described above enable the automatic generation and optimization of three-dimensional rescue routes, marking dangerous nodes and avoidance schemes along the routes, while also supporting manual fine-tuning to ensure the safety and efficiency of the rescue routes.

[0047] Step 7: Location-based rescue coordination, real-time early warning, and dynamic route updates. The positioning-rescue linkage mechanism is activated, synchronously transmitting the optimized three-dimensional rescue route to the terminal equipment of the underground rescue personnel, and synchronizing the location and vital signs changes of the trapped personnel in real time. During the rescue, the system continuously collects positioning data and environmental parameters. If the trapped personnel move, their vital signs become abnormal, or new dangerous nodes appear on the rescue route, the system immediately issues an early warning and automatically updates the rescue route. At the same time, it coordinates with the underground intelligent ventilation and emergency broadcasting systems to adjust the ventilation status of the rescue area, transmit rescue information to the trapped personnel, and improve rescue efficiency.

[0048] To achieve real-time linkage and early warning between location and rescue, a real-time early warning model for the rescue process and a dynamic update model for the rescue route are established to ensure safety and timeliness during the rescue process, as detailed below: (1) Real-time early warning model: The system uses changes in the vital signs, location, and environmental parameters of the trapped personnel as early warning indicators. When the indicators exceed the preset threshold, the system issues an early warning. in: Warning indicator, no unit, W=1 indicates a warning has been issued, W=0 indicates no warning; Heart rate variability of trapped individuals, measured in beats per minute (bpm). , Current heart rate, The heart rate at the previous moment; Heart rate variability warning threshold, in beats per minute (bpm). ; Changes in the location of trapped personnel, in meters (m). , The current location coordinates, The coordinates were determined at the previous moment; Location change warning threshold, unit is meters (m), take ; Changes in environmental parameters (gas concentration or oxygen content), expressed as % (volume fraction). , These are the current environmental parameter values. These are the environmental parameter values ​​from the previous moment; Environmental parameter change warning threshold, in percentage (volume fraction), taken as... .

[0049] (2) Dynamic update model for rescue routes: When the system issues an early warning (W=1), the cost of the rescue route is automatically recalculated and the optimal rescue route is updated. in: Total cost of the updated rescue route, no units; : The updated total length of the rescue route, in meters (m); The updated average hazard level assessment value for the rescue route, without units, is recalculated based on the latest environmental parameters and location data. The updated difficulty coefficient for the rescue route is not provided in any unit and has been recalculated based on the latest information regarding tunnel collapses and obstacles. , , The weighting coefficients remain unchanged, consistent with those in step 6.

[0050] The two models described above enable real-time early warning of the rescue process and dynamic updates of the rescue route, ensuring that rescuers can respond to emergencies on site in a timely manner and improving rescue efficiency and safety.

[0051] Step 8: System Operation and Maintenance and Subsequent Optimization and Iteration After the rescue operation concluded, a comprehensive inspection of the entire positioning system was conducted, examining the operational status of positioning terminals, sensors, and algorithms. Positioning data, fault information, and optimization suggestions from the rescue process were collected. Addressing issues such as positioning deviations, delayed mode switching, and signal interference encountered in disaster scenarios, the terminal structure, fusion algorithms, and switching logic were optimized and iterated. Simultaneously, a routine operation and maintenance mechanism was established to regularly calibrate positioning terminals, 3D spatial models, and algorithm parameters, ensuring the system's stable and accurate operation in subsequent disaster scenarios and continuously improving its positioning and rescue coordination capabilities.

[0052] To achieve precise optimization and iteration of the system and routine operation and maintenance, a system operation error analysis model and an operation and maintenance cycle calibration model are established to quantify system operation defects and determine the operation and maintenance calibration cycle, as detailed below: (1) System operation error analysis model: Statistical analysis of positioning data and algorithm operation data during this rescue process is conducted to calculate the overall system operation error and identify the sources of error. The model is as follows: in: Overall system operating error, expressed in meters (m). The larger the value, the more obvious the system's operational defects become; Total number of location data points collected during this rescue operation; unit not specified. : The final output coordinates of the i-th set of positioning data, in meters (m); : The actual coordinates corresponding to the i-th set of positioning data, in meters (m).

[0053] (2) Operation and maintenance cycle calibration model: based on the overall system operation error To determine the system operation and maintenance calibration cycle and ensure long-term stable operation of the system, the model is as follows: in: System maintenance and calibration cycle, in days (d); The overall system operating error, expressed in meters (m), is calculated using the system operating error analysis model.

[0054] By using the above methods, we can identify system operational defects and determine the maintenance cycle, providing a quantitative basis for system optimization, iteration, and routine maintenance, and ensuring that the system can continue to operate stably and accurately in subsequent disaster scenarios.

[0055] The following is combined Figure 2-5 The present invention is compared and described with related methods in the prior art.

[0056] like Figure 2 As shown in the black and white bar chart, the energy harvesting efficiency of the traditional active positioning terminal and the passive terminal of this solution are compared in five types of mine disaster scenarios: the efficiency of the traditional terminal drops sharply to below 50% in extreme scenarios such as high gas and strong dust, and it cannot maintain power supply; while the terminal of this solution relies on the passive power supply energy harvesting model (η≥85%), and the efficiency is stable at above 85% in all disaster scenarios. It solves the core pain point of existing terminals "positioning failure due to power interruption", realizes long-term maintenance-free and stable operation of the terminal, saves the operation and maintenance link of "periodic charging / battery replacement" in terms of process, and reduces energy consumption by about 60% (passive harvesting of underground environmental energy, no external power supply required).

[0057] like Figure 3 As shown in the figure, this diagram compares the positioning errors of 10 downhole benchmark points: traditional positioning techniques generally have errors between 0.5 and 0.9 meters, far exceeding the precise positioning requirements for rescue; this solution, relying on a positioning accuracy calibration model (ΔP≤0.1m), controls the error of all points within 0.1 meters, improving positioning accuracy by more than 80%. High-precision positioning directly solves the problem of "positioning ambiguity leading to rescue route deviations and missed golden rescue time" in existing technologies, improving rescue efficiency by approximately 70% (precise positioning reduces invalid search areas), while also simplifying the "manual calibration of positioning deviations" process.

[0058] like Figure 4 As shown in the step chart, the response time under five different switching scenarios is compared: the response time for traditional positioning mode switching is >1.8s, reaching 3.5s under heavy obstruction, which easily leads to gaps in positioning data and delays in rescue decisions. This solution, relying on a mode switching judgment model and a transition weight model, achieves a switching response time of ≤0.5s, and the linear transition of weights ensures seamless connection, solving the problem of "positioning blind spots caused by switching lag." The response time is reduced by more than 85%, ensuring continuous and uninterrupted positioning data during rescue operations, simplifying control operations, and reducing the need for "manual positioning data completion," thus improving efficiency by approximately 65%.

[0059] like Figure 5As shown, the optimization effect of rescue routes is compared from two dimensions: "planning time" and "route cost". Traditional route planning takes more than 15 minutes in scenarios with multiple trapped people and collapse zones, and the route cost is nearly 10 (high risk / long distance). This solution, based on the rescue route cost function and Dijkstra's optimization model, reduces the planning time to ≤1.1 minutes and the route cost to less than 3.8, improving planning efficiency by more than 90% and reducing route cost by 60%. This solves the problem of "long and costly route planning leading to rescue delays" in existing technologies, increasing the rescue success rate within the golden rescue time (48 hours) from the traditional 40% to 85%. At the same time, it simplifies the operation of "manually adjusting the rescue route" and reduces the operational complexity for rescue personnel.

[0060] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape, principle and application direction of this application should be covered within the scope of protection of this application.

Claims

1. A method for precise three-dimensional positioning of trapped personnel in a disaster environment, characterized in that, Includes the following steps: Step 1: Deployment and debugging of domestically produced passive intrinsically safe positioning terminals: Deploy in batches intrinsically safe passive positioning terminals for mining, consisting of domestically produced UWB positioning chips and RF power amplifiers. Deploy the intrinsically safe passive positioning terminals in batches at key locations in various roadways and chambers underground. At the same time, equip underground workers with portable wearable terminals, complete the collaborative debugging of the terminals with the existing domestic positioning system underground, calibrate the positioning accuracy and communication distance, test the stability of the terminals in low-oxygen and high-gas environments, and verify the efficiency of the passive power supply structure in collecting weak energy underground. Step 2: Construction of underground 3D spatial model and reuse of existing video resources: Reuse existing video surveillance resources in the mining area, collect spatial parameters of each roadway and chamber underground through video dynamic target 3D reconstruction technology, construct a computable underground 3D spatial model, mark the location information, size parameters and key facility locations of roadways and chambers at different elevations, and complete the linkage debugging of the model with positioning terminals and video surveillance equipment; Step 3: Initialization and Automatic Switching Logic Setting of Dual Positioning Mode: Start the dual positioning system of non-contact positioning and wearable terminal supplementation, complete the initialization and debugging of the two positioning modes, and set the connection algorithm parameters of the two modes to achieve seamless switching; Step 4 Deployment and calibration of multi-source signal fusion anti-interference algorithm: Simultaneously deploy an adaptive anti-interference algorithm for fusion of UWB radar, inertial navigation and Beidou RTK multi-source signals in the ground rescue command center and the underground positioning terminal, complete the fusion calibration of the three signals, set signal priorities, introduce MIMO technology and optimize K-means++ algorithm to complete multi-target positioning calibration; Step 5: Location and vital signs are collected simultaneously and danger level is automatically determined: Relying on the closed-loop positioning mechanism, the miniature intrinsically safe vital signs sensor integrated in the positioning terminal is activated to collect the vital signs data of the trapped personnel simultaneously and correlate them with the three-dimensional coordinate data in real time. Through a preset algorithm and combined with real-time environmental parameters underground, the survival status of the trapped personnel and the danger level of the area are determined, and the location data, vital signs data and danger level information are transmitted to the ground rescue command center simultaneously. Step 6 Automatic generation and optimal path optimization of three-dimensional rescue route: Based on the received positioning data, vital sign data, and hazard level information, the ground rescue command center combines the underground three-dimensional spatial model and real-time environmental parameters to automatically calculate and generate the optimal three-dimensional rescue route, marking dangerous nodes and avoidance schemes on the route. At the same time, rescuers can manually fine-tune the automatically generated route according to the actual situation on site to optimize the rescue path. Step 7: Location-Rescue Linkage Execution, Real-time Early Warning, and Dynamic Route Updates: Activate the location-rescue linkage mechanism to synchronously transmit the optimized three-dimensional rescue route to the terminal equipment of the underground rescue personnel. Real-time synchronization of the trapped personnel's location changes and vital sign changes is maintained. During the rescue, the system continuously collects location data and environmental parameters. If the trapped personnel's location moves, vital signs become abnormal, or new dangerous nodes appear on the rescue route, the system immediately issues an early warning and automatically updates the rescue route. Simultaneously, it links with the underground intelligent ventilation and emergency broadcasting systems to adjust the ventilation status of the rescue area and transmit rescue information to the trapped personnel. Step 8 System Operation and Maintenance, Error Analysis and Optimization Iteration: After the rescue is completed, a comprehensive inspection of the entire positioning system will be carried out to check the operating status of the positioning terminals, sensors and algorithms. Positioning data, fault information and optimization suggestions during the rescue process will be collected. For the problems that occurred in the disaster scenario, the terminal structure, fusion algorithm and switching logic will be optimized and iterated. At the same time, a normalized operation and maintenance mechanism will be established to regularly calibrate the positioning terminals, 3D spatial models and algorithm parameters.

2. The method for three-dimensional precise positioning of trapped personnel in a disaster environment according to claim 1, characterized in that, In step 1, by establishing a passive power supply energy acquisition model and a three-dimensional spatial positioning error calibration model, the power supply stability and positioning accuracy of the positioning terminal are accurately calibrated. The passive power supply energy harvesting model is as follows: in, Passive power supply energy harvesting efficiency; The actual usable electrical energy converted at the terminal; The total, albeit weak, energy collected by the terminal in the well. The three-dimensional spatial positioning error calibration model is as follows: in, : 3D positioning error; : The actual three-dimensional coordinates of the personnel deploying / wearing the positioning terminal; : The three-dimensional coordinates collected by the positioning terminal.

3. The method for three-dimensional precise positioning of trapped personnel in a disaster environment according to claim 2, characterized in that, In step 2, the accurate construction of the downhole three-dimensional spatial model is achieved by establishing a downhole spatial coordinate mapping model and a model accuracy verification model. The underground spatial coordinate mapping model is as follows: in, : Actual three-dimensional spatial coordinates underground; Two-dimensional pixel coordinates in a video image; : The principal coordinates of the video camera; , The focal length of the camera in the x and y directions; The vertical distance from the target point to the camera; Model accuracy verification model is as follows in, : The average deviation of the three-dimensional spatial model; Number of verification points; : The actual three-dimensional coordinates of the i-th verification point; : The coordinates of the i-th verification point in the three-dimensional spatial model.

4. The method for three-dimensional precise positioning of trapped personnel in a disaster environment according to claim 3, characterized in that, In step 3, by establishing a mode switching determination model and a connection weight model, the accurate switching and seamless connection of the two positioning modes are achieved. The mode switching determination model is as follows: in, Mode switching indicator; Positioning error in non-sensory positioning; : Threshold for non-sensory positioning accuracy; Video signal sharpness factor; Video signal clarity threshold; The connection weight model is as follows: in, : The merged positioning coordinates; Weighting coefficients for seamless positioning; : Three-dimensional coordinates acquired through non-sensory positioning; Weighting coefficients for wearable device positioning; : Three-dimensional coordinates collected by the wearable terminal.

5. The method for precise three-dimensional positioning of trapped personnel in a disaster environment according to claim 4, characterized in that, In step 4, by establishing an adaptive fusion model for multi-source signals and a clustering model for multi-target localization, anti-interference fusion of multi-source signals and accurate differentiation of multiple targets are achieved. The multi-source signal adaptive fusion model is as follows: ,in in, The final three-dimensional positioning coordinates after multi-source signal fusion; Signal type identifier; : Adaptive weighting coefficients for the k-th signal; : The three-dimensional positioning coordinates of the k-th signal acquisition; Signal weights The dynamic adjustment is based on the mean square error of the signal. The smaller the mean square error, the higher the signal reliability and the greater the weight. The mean square error calculation model is as follows: in, : The mean square error of the k-th signal; The number of times the k-th signal is collected; : The three-dimensional coordinates of the k-th signal acquired in the j-th acquisition; : The actual three-dimensional coordinates during the j-th acquisition; The multi-target localization clustering model is as follows: in, Clustering objective function; Number of people trapped; : The location dataset of the i-th cluster; Cluster A single positioning coordinate in the data; : The center coordinates of the i-th cluster; Location coordinates P and cluster center The square of the Euclidean distance.

6. The method for three-dimensional precise positioning of trapped personnel in a disaster environment according to claim 5, characterized in that, In step 5, by establishing a vital sign abnormality judgment model and a risk level comprehensive evaluation model, the survival status and risk level of the trapped personnel can be accurately determined. The model for determining abnormal vital signs is as follows: in, : Survival status indicator for trapped personnel; Real-time heart rate of the trapped personnel; Heart rate lower limit threshold; Heart rate upper limit threshold; Real-time respiratory rate of the trapped personnel; : Lower limit threshold for respiratory rate; Upper limit threshold for respiratory rate; The comprehensive risk assessment model is as follows: in, : Comprehensive evaluation value of hazard level; : Gas concentration weighting coefficient; Gas concentration hazard score; Oxygen content weighting coefficient; Oxygen content hazard score; Collapse risk weighting coefficient; Collapse risk hazard score; : Methane concentration in the well; Oxygen content in the well; The straight-line distance between the trapped personnel and the collapsed area.

7. The method for three-dimensional precise positioning of trapped personnel in a disaster environment according to claim 6, characterized in that, In step 6, the automatic generation and optimization of three-dimensional rescue routes are achieved by establishing a rescue route cost function model and a rescue route optimization model; The rescue route cost function model is as follows: in, Total cost of the rescue route; : Route length weighting coefficient; : Total length coefficient of rescue route; Risk level weighting coefficient; The average hazard level assessment value along the rescue route; Difficulty of passage weighting coefficient; Difficulty level of the rescue route; The rescue route optimization model is as follows: in, : The total cost of the optimized route; : The total cost of the i-th candidate rescue route; Number of candidate rescue routes.

8. The method for three-dimensional precise positioning of trapped personnel in a disaster environment according to claim 7, characterized in that, In step 7, a real-time early warning model for the rescue process and a dynamic update model for the rescue route are established to achieve real-time linkage and early warning between location and rescue. The real-time early warning model for the rescue process is as follows: in, Warning sign; Changes in the heart rate of the trapped individuals; Heart rate variability warning threshold; Changes in the location of trapped personnel , The current location coordinates, The coordinates were determined at the previous moment; Location change warning threshold; Changes in environmental parameters; : Warning threshold for changes in environmental parameters; The dynamic update model for rescue routes is as follows: in, : Total cost of the updated rescue route; : The updated total length of the rescue route; : Updated average hazard rating of rescue routes; The updated difficulty level of the rescue route; , , : Consistent with the weighting coefficients in step 6.

9. The method for three-dimensional precise positioning of trapped personnel in a disaster environment according to claim 8, characterized in that, In step 8, by establishing a system operation error analysis model and an operation and maintenance cycle calibration model, the system can achieve precise optimization iteration and routine operation and maintenance. The system operation error analysis model is as follows: in, Overall system operating error; Total amount of location data collected during this rescue operation; : The final output coordinates of the i-th set of positioning data; : The actual coordinates corresponding to the i-th set of positioning data; The operation and maintenance cycle calibration model is as follows: in, System maintenance and calibration cycle; Overall system operating error.

Images