5g-based coal mine underground personnel positioning system and method

Abstract

The present application relates to the field of wireless communication and mine safety technology, and discloses a coal mine underground personnel positioning system and method based on 5G. The method establishes a node space coordinate database of the underground roadway, and directly substitutes the node constraint model matched with the current area in the positioning calculation for collaborative calculation, so that the calculation process naturally conforms to the physical structure of the roadway, and the generation of invalid position coordinates is avoided from the source. The speed distribution pattern and the steering probability matrix extracted from the historical movement of the personnel are used to evaluate the trajectory continuity of multiple geometric feasible candidate positions, so as to select the real-time position most consistent with the individual behavior habit. The method overcomes the problems of signal multipath interference and path ambiguity in the complex underground environment, and improves the physical rationality and trajectory continuity of the positioning result.

Description

Technical Field

[0001] This invention relates to the fields of wireless communication and mine safety technology, specifically to a 5G-based underground personnel positioning system and method for coal mines. Background Technology

[0002] In the complex underground tunnels of coal mines, achieving high-precision, continuous, and reliable personnel positioning is crucial for safety management. Existing technologies often employ 5G-based time-of-arrival (TOA) positioning methods, estimating location by measuring the time delay difference between the arrival of signals from personnel-carried identification cards at multiple base stations. However, the narrow structure and limited space of underground tunnels exacerbate non-line-of-sight electromagnetic signal propagation and multipath effects, leading to significant ranging errors. Traditional positioning methods typically perform unconstrained geometric calculations directly after obtaining a set of distance difference equations for signal propagation. The resulting location points often lie outside the physical boundaries of the tunnel, such as in rock walls or solid structures, lacking a direct connection to the actual physical space. This necessitates subsequent map matching steps for correction, resulting in unreliable positioning results in physical space. Furthermore, multiple ambiguous solutions are prone to appear in areas with tunnel intersections or similar structures.

[0003] After obtaining initial location points, conventional methods often employ filtering algorithms based on general motion models such as uniform velocity and uniform acceleration for trajectory smoothing and correlation. However, the movement of underground personnel is influenced by work tasks, tunnel slope, obstacles, and personal habits, resulting in specific and random movement patterns that deviate significantly from idealized physical models. When multiple candidate location points exist, general models struggle to effectively distinguish which one represents a continuation of the personnel's actual behavior, easily leading to trajectory jumps or misjudgments. Existing solutions fail to effectively utilize the inherent tunnel topology information and the behavioral patterns formed by personnel over long-term operations to address the problems of inaccurate signal measurement and location ambiguity. Summary of the Invention

[0004] The purpose of this invention is to provide a 5G-based underground personnel positioning system and method in coal mines to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a 5G-based method for locating personnel underground in coal mines, the method comprising:

[0006] A spatial coordinate database of the underground roadway environment is established in the positioning computing center. The spatial coordinate database contains the three-dimensional coordinates of all roadway nodes and the topological associations between nodes.

[0007] The carrier signal transmitted by the 5G positioning identification card carried by the personnel is received in real time through the underground 5G positioning beacon. The carrier signal contains the unique number of the identification card and the initial timestamp.

[0008] The carrier signal is subjected to multipath separation processing to obtain the arrival time sequence of the direct path signal; based on the arrival time sequence of the direct path signal and the known coordinates of the 5G positioning beacon, the difference in signal propagation distance between the personnel and multiple 5G positioning beacons is calculated.

[0009] The node constraint model matching the current roadway partition is retrieved from the spatial coordinate database, and the signal propagation distance difference is substituted into the node constraint model for collaborative solution to obtain the initial position coordinate set of the personnel.

[0010] The initial set of position coordinates is filtered by roadway connectivity to obtain multiple candidate position points that meet the roadway structure constraints;

[0011] Extracting historical movement trajectory features of personnel, including speed distribution patterns and turning probability matrices;

[0012] The historical movement trajectory features are used to evaluate the trajectory continuity of the multiple candidate location points to determine the final real-time location of the personnel.

[0013] Preferably, performing multipath separation processing on the carrier signal to obtain the arrival time sequence of the direct path signal specifically includes:

[0014] Receive raw waveform data of carrier signals synchronously uploaded by multiple 5G positioning beacons;

[0015] An adaptive equalization filter is used to suppress channel interference in the original waveform data of the carrier signal to obtain a pre-equalized signal waveform.

[0016] Multipath component search is performed on the pre-equalized signal waveform to identify all resolvable paths whose signal energy exceeds a preset threshold.

[0017] Based on the arrival angle and arrival time difference of the distinguishable path, and combined with the preset tunnel reflection surface model, the signal component that best matches the straight-line propagation characteristics is selected from all the distinguishable paths, and the signal component is marked as the straight-line signal.

[0018] The arrival times of the signal peaks of the direct path signal are accurately extracted and arranged in chronological order to generate a sequence of arrival times of the direct path signal.

[0019] Preferably, the calculation of the signal propagation distance difference between the operator and multiple 5G positioning beacons specifically includes:

[0020] Extract the individual arrival time corresponding to each 5G positioning beacon from the arrival time sequence of the direct path signal;

[0021] Read the base time of each 5G positioning beacon when it sends its own synchronization beacon;

[0022] Calculate the time difference between the individual arrival time and the reference time to obtain the one-way propagation time of the signal from the personnel's location to each 5G positioning beacon;

[0023] Multiplying the one-way propagation time by a preset signal propagation speed constant yields the one-way spatial distance from the personnel's location to each 5G positioning beacon.

[0024] A 5G positioning beacon is selected as the distance reference benchmark. The one-way spatial distance from the person's location to each other 5G positioning beacon is subtracted from the one-way spatial distance from the person's location to the distance reference benchmark to generate the signal propagation distance difference.

[0025] Preferably, retrieving the node constraint model matching the current roadway partition from the spatial coordinate database specifically includes:

[0026] Based on the most recent calculated personnel location, determine the possible lane section number where the personnel are currently located;

[0027] Based on the lane partition number, the three-dimensional coordinates of all lane nodes within the lane partition number are indexed from the spatial coordinate database;

[0028] Based on the three-dimensional coordinates of all roadway nodes within the roadway partition number, construct an adjacency matrix describing the connection relationships between nodes;

[0029] Based on the physical width parameters of the tunnel, a location feasible domain space is generated with each tunnel node as the center;

[0030] By integrating the adjacency matrix with the location feasible domain space of all lane nodes, a node constraint model is formed that describes the possible distribution of personnel locations within the lane partition.

[0031] Preferably, substituting the signal propagation distance difference into the node constraint model for collaborative solution to obtain the initial set of personnel position coordinates specifically includes:

[0032] Within the entire feasible domain space defined by the node constraint model, a system of equations is established regarding the difference in signal propagation distance, where the unknowns in the system of equations are the three-dimensional coordinates of the personnel.

[0033] A global search algorithm is used to generate a large number of random location sampling points in the location feasible region space;

[0034] Substitute the coordinates of each random sampling point into the system of equations to calculate the difference in theoretical signal propagation distance corresponding to the random sampling point.

[0035] Compare the theoretical signal propagation distance difference corresponding to each random sampling point with the actual measured signal propagation distance difference, and calculate the sum of squared residuals between the two.

[0036] Select several random sampling points with the smallest sum of squared residuals, and use the set of their coordinates as the initial set of coordinates of the personnel.

[0037] Preferably, performing roadway connectivity filtering on the initial set of position coordinates to obtain multiple candidate position points that meet the roadway structure constraints specifically includes:

[0038] Read each coordinate point in the initial position coordinate set;

[0039] Calculate the vertical distance from each coordinate point to the centerline of all roadway nodes in the node constraint model;

[0040] Determine whether the vertical distance is less than or equal to half the physical width of the tunnel; if it is greater, mark the coordinate point as an invalid point and discard it.

[0041] For coordinate points with valid vertical distance, further check whether there is a continuous path formed by connecting roadway nodes between the coordinate points and the valid personnel positions confirmed at the previous moment.

[0042] If no continuous path exists, the coordinate points will also be marked as invalid and removed.

[0043] The remaining coordinate points after both vertical distance and connectivity are used to determine multiple candidate locations that meet the constraints of the roadway structure.

[0044] Preferably, the extracted historical movement trajectory features of personnel specifically include:

[0045] Retrieve all successful location coordinates of the person's location identification card within a preset historical time period from the location history database;

[0046] Connect all the successfully located coordinate points in chronological order to form a continuous historical movement trajectory line of the person;

[0047] The continuous historical movement trajectory is segmented, the average movement speed of the person on each trajectory segment is calculated, and the distribution of all average movement speeds is statistically analyzed to obtain the speed distribution pattern.

[0048] Identify all turning points on the continuous historical movement trajectory line, calculate the number and frequency of turns made by personnel at each roadway node to each possible branch roadway, and form the turning probability matrix.

[0049] Preferably, evaluating the trajectory continuity of the plurality of candidate location points using the historical movement trajectory features specifically includes:

[0050] Obtain the real-time location of the last confirmed person at the previous moment; calculate the straight-line distance from the real-time location of the last confirmed person at the previous moment to each of the multiple candidate location points at the current moment;

[0051] Based on the speed distribution pattern in the historical movement trajectory features, the maximum distance that a person may move within the positioning time interval is estimated.

[0052] Candidate location points whose straight-line distance exceeds the maximum distance range are initially screened out;

[0053] For the remaining candidate locations, calculate the directional change required to move from the previous location to the candidate location, evaluate the probability of the directional change based on the turning probability matrix, and generate a continuous likelihood value for the trajectory of each remaining candidate location.

[0054] Preferably, determining the final real-time location of personnel specifically includes:

[0055] Compare the continuous likelihood values ​​of the trajectories of all remaining candidate locations, and select the candidate location with the highest continuous likelihood value.

[0056] Check whether the continuous likelihood value of the candidate position point with the highest continuous likelihood value exceeds a preset confidence threshold;

[0057] If the number of candidate locations exceeds the limit, the coordinates of the candidate locations will be determined as the final real-time location of the personnel.

[0058] If the time limit is not exceeded, the real-time location of the last confirmed person will remain unchanged, and the current time will be marked as a location ambiguity state, waiting for the location data of the next time moment to be processed together.

[0059] Preferably, when the processor executes the computer program, it implements the steps of the 5G-based underground personnel positioning method for coal mines as described in any of the above-mentioned methods.

[0060] Compared with the prior art, the beneficial effects of the present invention are:

[0061] The tunnel network is pre-abstracted into a node constraint model containing three-dimensional coordinates and connectivity relationships. During the localization solution stage, a model matching the current area is directly retrieved from the database and used as the core mathematical constraint condition in conjunction with the signal arrival time difference equation for solution. This ensures that the localization solution process is strictly confined to the effective tunnel space, avoiding the generation of a large number of invalid location solutions located inside solid walls. It eliminates physically unreasonable locations caused by signal errors, directly outputting an initial set of locations conforming to the tunnel topology. This improves the fundamental reliability and spatial rationality of the localization results in complex tunnel environments and reduces reliance on post-processing error correction.

[0062] This method extracts and quantifies the speed distribution patterns and turning probability matrices at roadway nodes during a person's historical movement, forming a personalized behavioral feature model. During localization, when faced with multiple candidate locations that may be valid under various spatial constraints, this feature model is used to evaluate the trajectory assumptions from the previous location to each candidate point. The evaluation criteria are whether the moving speed value conforms to the person's habitual speed distribution in such roadways, and whether the moving direction matches their historical turning probability at similar nodes. Through this evaluation based on actual behavioral logic, the location point that best matches the target individual's behavioral habits can be selected from multiple geometrically feasible solutions as the final result. This overcomes the difficulty of making accurate judgments based solely on signal strength and geometric relationships in similar path environments such as roadway intersections and loops, improving trajectory continuity and the accuracy of location determination. Attached Figure Description

[0063] Figure 1 This is a schematic diagram illustrating the working principle of the 5G-based underground personnel positioning method in coal mines as described in this invention.

[0064] Figure 2 A flowchart for processing direct path signals in multipath separation;

[0065] Figure 3 To retrieve the flowchart of the node constraint model;

[0066] Figure 4 Grouped bar charts showing the distribution of positioning accuracy for personnel underground in coal mines;

[0067] Figure 5 This is a heat map showing the probability of turning at nodes in underground coal mine roadways. Detailed Implementation

[0068] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0069] Please see Figure 1 This invention provides a 5G-based method for locating personnel underground in coal mines. The method includes establishing a spatial coordinate database of the underground roadway environment at a positioning computing center. This database contains the three-dimensional coordinates of all roadway nodes and the topological relationships between nodes. Carrier signals transmitted by 5G positioning identification cards carried by personnel are received in real time via underground 5G positioning beacons. These carrier signals contain the unique identification card number and an initial timestamp. Multipath separation processing is performed on the carrier signals to obtain the arrival time sequence of direct path signals. Based on the arrival time sequence of the direct path signals and the known coordinates of the 5G positioning beacons, the signal propagation distance difference between the personnel and multiple 5G positioning beacons is calculated. A node constraint model matching the current roadway partition is retrieved from the spatial coordinate database. The signal propagation distance difference is substituted into the node constraint model for collaborative solution to obtain the initial set of personnel position coordinates. Roadway connectivity filtering is applied to the initial set of position coordinates to obtain multiple candidate position points that meet the roadway structure constraints. Extract the historical movement trajectory features of the personnel, including speed distribution patterns and turning probability matrices; use the historical movement trajectory features to evaluate the trajectory continuity of multiple candidate location points to determine the final real-time location of the personnel.

[0070] In one embodiment of the present invention, see [reference] Figure 2The system receives raw carrier signal waveform data synchronously uploaded by multiple 5G positioning beacons. An adaptive equalization filter is used to suppress channel interference in the raw carrier signal waveform data, resulting in a pre-equalized signal waveform. Multipath component search is performed on the pre-equalized signal waveform to identify all resolvable paths whose signal energy exceeds a preset threshold. Based on the arrival angle and time difference of the resolvable paths, and combined with a preset tunnel reflection surface model, the signal component that best matches the straight-line propagation characteristics is selected from all resolvable paths and marked as the direct path signal. The arrival time of the signal peak of the direct path signal is accurately extracted and arranged in chronological order to generate a sequence of arrival times for the direct path signals. Extract the individual arrival time corresponding to each 5G positioning beacon from the arrival time sequence of the direct path signal; read the reference time of each 5G positioning beacon when it sends its own synchronization beacon; calculate the time difference between the individual arrival time and the reference time to obtain the one-way propagation time of the signal from the person's location to each 5G positioning beacon; multiply the one-way propagation time by a preset signal propagation speed constant to obtain the one-way spatial distance from the person's location to each 5G positioning beacon; select a 5G positioning beacon as the distance reference reference, and subtract the one-way spatial distance from the person's location to the distance reference reference from the one-way spatial distance from the person's location to each other 5G positioning beacon to generate the signal propagation distance difference.

[0071] In practical implementation, multiple 5G positioning beacons synchronously upload the received raw carrier signal waveform data to the positioning computing center. This raw carrier signal waveform data includes sampling points of the radio frequency signal emitted by the 5G positioning identification card carried by the personnel. In this implementation, an adaptive equalization filter is used to suppress channel interference in the raw carrier signal waveform data. The adaptive equalization filter pre-configures filtering coefficients based on the channel characteristics of the underground roadway and performs convolution operations on the raw carrier signal waveform data to suppress multipath interference and noise, obtaining a pre-equalized signal waveform. In some embodiments, a multipath component search is performed on the pre-equalized signal waveform. A sliding correlation or matched filtering algorithm is used to identify all resolvable paths whose signal energy exceeds a preset threshold. Each resolvable path corresponds to a signal propagation path with significant energy. Optionally, path discrimination is performed based on the arrival angle and arrival time difference of the distinguishable path, combined with a preset roadway reflective surface model. The roadway reflective surface model defines the wall geometry parameters and electromagnetic reflection characteristics of the main underground roadways. The signal component that best matches the straight-line propagation characteristics is selected from all distinguishable paths. This signal component has the earliest arrival time and the arrival angle is consistent with the theoretical direct path direction of the 5G positioning beacon and the 5G positioning tag carried by the personnel. The signal component is then marked as the direct path signal.

[0072] In specific implementation, a single arrival time corresponding to each 5G positioning beacon is extracted from the arrival time sequence of the direct path signal, with each 5G positioning beacon corresponding to a time value in the arrival time sequence. The reference time of each 5G positioning beacon when it sends its own synchronization beacon is read; this reference time is guaranteed by the precision clock synchronization mechanism of the underground 5G. The time difference between the single arrival time and the reference time is calculated to obtain the one-way propagation time of the signal from the personnel's location to each 5G positioning beacon. It can be understood that multiplying the one-way propagation time by a preset signal propagation speed constant yields the one-way spatial distance from the personnel's location to each 5G positioning beacon; the signal propagation speed constant is the speed of light in a vacuum. In some embodiments, a 5G positioning beacon is selected as a distance reference reference, and the one-way spatial distance from the personnel's location to each other 5G positioning beacon is subtracted from the one-way spatial distance from the personnel's location to the distance reference reference, generating a set of signal propagation distance differences. The calculation of the signal propagation distance differences can be expressed as follows:

[0073]

[0074] in: This represents the difference in signal propagation distance between the i-th 5G positioning beacon and the 5G positioning beacon used as a distance reference. This represents the one-way spatial distance from the location of a person to the i-th 5G positioning beacon. This indicates the one-way spatial distance from a person's location to a 5G positioning beacon that serves as a distance reference. This represents the preset signal propagation speed constant. This represents the one-way propagation time of the signal from the location of a person to the i-th 5G positioning beacon. This indicates the one-way propagation time of a signal from a person's location to a 5G positioning beacon that serves as a distance reference.

[0075] In one embodiment of the present invention, see [reference] Figure 3Based on the most recently calculated personnel location, the current lane section number of the personnel is determined. Using the lane section number, the 3D coordinates of all lane nodes within that section are indexed from the spatial coordinate database. Based on the 3D coordinates of all lane nodes within the lane section number, an adjacency matrix describing the connectivity between nodes is constructed. Combining the physical width parameter of the lane, a location feasible domain space is generated centered on each lane node. Integrating the adjacency matrix and the location feasible domain spaces of all lane nodes forms a node constraint model describing the personnel location distribution within the lane section. Within the entire feasible location space defined by the node constraint model, a system of equations is established regarding the signal propagation distance difference, where the unknowns in the system are the three-dimensional coordinates of the personnel. A global search algorithm is used to generate a large number of random location sampling points within the feasible location space. The coordinates of each random location sampling point are substituted into the system of equations to calculate the theoretical signal propagation distance difference corresponding to the random location sampling point. The theoretical signal propagation distance difference corresponding to each random location sampling point is compared with the actual measured signal propagation distance difference, and the sum of squared residuals is calculated. Several random location sampling points with the smallest sum of squared residuals are selected, and their coordinate set is used as the initial location coordinate set of the personnel.

[0076] In practical implementation, based on the most recently calculated personnel location coordinates, the positioning calculation center determines the current roadway partition number of the personnel according to a preset roadway partition mapping relationship. The roadway partition mapping relationship defines the correspondence rules between underground three-dimensional spatial coordinates and logical partition numbers. Based on the roadway partition number, the three-dimensional coordinates of all roadway nodes within that roadway partition number are indexed from the spatial coordinate database. The spatial coordinate database stores the geometric and topological information of the corresponding roadway nodes using the roadway partition number as the primary key. In some embodiments, based on the three-dimensional coordinates of all roadway nodes within the roadway partition number, an adjacency matrix describing the connection relationships between nodes is constructed. The adjacency matrix is ​​a two-dimensional array, where the values ​​of the matrix elements indicate whether there is a direct roadway connection between two indexed roadway nodes. Combining the roadway's physical width parameter, a location feasible domain space is generated with each roadway node as the center. The location feasible domain space is typically a three-dimensional volume space centered on the roadway node coordinates, extending along the roadway direction and covering the roadway's physical width. By integrating the adjacency matrix with the location feasible domain space of all lane nodes, a node constraint model is formed to describe the distribution of personnel positions within a lane section. This node constraint model essentially defines a location solution search space constrained by the physical structure of the lane. The specific formation of the node constraint model begins by determining the current lane section number based on the most recently calculated personnel positions, and then indexing the three-dimensional coordinates of all lane nodes within the section from a spatial coordinate database based on this number. Subsequently, the system constructs an adjacency matrix describing the connectivity relationships between nodes based on these coordinates. This matrix records the direct connectivity of each node in the lane network in the form of a two-dimensional array. Simultaneously, combined with pre-measured lane physical width parameters, a location feasible domain space is generated centered on each lane node. This space is a three-dimensional volume extending along the lane direction and covering the physical width surrounding the node, ensuring that personnel positions do not exceed the actual lane boundaries. The integration process maps the topological connectivity expressed by the adjacency matrix to all location feasible domains, making each location feasible domain a node in the graph structure. Based on the connection relationships in the adjacency matrix, accessibility constraints between spaces are established, thus forming a node constraint model that fully describes the location distribution of people within the lane partition.

[0077] It is understandable that incorporating the signal propagation distance difference into the node constraint model for collaborative solution requires searching and matching within the entire feasible location space defined by the node constraint model. Within the feasible location space, a system of equations is established regarding the signal propagation distance difference, where the unknowns are the three-dimensional coordinates of the person, and the equations are constructed based on the geometric relationship between the signal arrival time difference and the coordinates. A global search algorithm is used to generate a large number of random location sampling points within the feasible location space. Global search algorithms, such as the Monte Carlo method, ensure that the sampling points cover all location areas within the feasible location space. The coordinates of each random location sampling point are substituted into the system of equations to calculate the theoretical signal propagation distance difference corresponding to the random location sampling point. The theoretical signal propagation distance difference is the expected measured difference in signal propagation distance between the person and each 5G positioning beacon, assuming the person is located at that sampling point coordinate.

[0078] Compare the difference between the theoretical signal propagation distance and the actual measured signal propagation distance at each random sampling point, and calculate the sum of squared residuals. The sum of squared residuals characterizes the degree of agreement between the sampling point location and the actual measurement data. Optionally, select several random sampling points with the smallest sum of squared residuals, and use the set of their coordinates as the initial set of personnel position coordinates. The calculation of the sum of squared residuals during collaborative solution can be expressed as follows:

[0079]

[0080] in: Indicates sampling points at random locations. The sum of squared residuals at the point, This indicates the number of 5G positioning beacon pairs used for the calculation. Indicates the actual measurement obtained from the first The difference in the propagation distance of each signal Indicates based on sampling points The three-dimensional coordinates of the first A function representing the difference in the theoretical signal propagation distance.

[0081] In one embodiment of the present invention, each coordinate point in the initial set of position coordinates is read; the vertical distance from each coordinate point to the centerline of all roadway nodes in the node constraint model is calculated. It is determined whether the vertical distance is less than or equal to half the physical width of the roadway; if it is greater, the coordinate point is marked as invalid and removed. For coordinate points with valid vertical distances, it is further checked whether there is a continuous path formed by roadway nodes connecting the coordinate point and the previously confirmed valid personnel position; if no continuous path exists, the coordinate point is also marked as invalid and removed. The remaining coordinate points after the dual judgment of vertical distance and connectivity constitute multiple candidate position points that meet the roadway structure constraints.

[0082] In specific implementation, each coordinate point in the initial position coordinate set is read, and each coordinate point represents a spatial location of a person obtained by collaborative solution. The vertical distance from each coordinate point to the centerline of all roadway nodes in the node constraint model is calculated. The centerline of the roadway nodes is defined by the connection line of the three-dimensional coordinates of adjacent roadway nodes stored in the spatial coordinate database. In some embodiments, it is determined whether the vertical distance is less than or equal to half of the physical width of the roadway. The physical width of the roadway is the roadway cross-sectional width parameter that is pre-measured and entered into the spatial coordinate database. If the vertical distance is greater than half of the physical width of the roadway, the coordinate point is marked as invalid and discarded. This step ensures that the person's location does not appear outside the solid wall of the roadway. For coordinate points with valid vertical distances, it is further checked whether there is a continuous path formed by connecting roadway nodes between the coordinate point and the valid person location confirmed at the previous moment. The continuous path check is performed by graph traversal search based on the adjacency matrix and the location feasible domain space in the node constraint model.

[0083] It is understandable that if there is no continuous path between the coordinate point and the previously confirmed valid personnel position, the coordinate point is also marked as invalid and eliminated. This step, based on the tunnel topology, eliminates impossible positions where personnel can move instantaneously in discontinuous space. After both vertical distance and connectivity are checked, the remaining coordinate points constitute multiple candidate positions that meet the tunnel structure constraints. The process of calculating the vertical distance can be described as follows:

[0084]

[0085] in: This indicates the coordinates of the point to be judged from the roadway node. With lane nodes The vertical distance of the determined centerline segment of the tunnel. This represents the three-dimensional coordinate vector of the point to be determined. and These represent the three-dimensional coordinate vectors of two adjacent roadway nodes that constitute the centerline segment of the roadway, with symbols... Represents the cross product of vectors. It represents the magnitude of the vector.

[0086] In some embodiments, for a roadway consisting of multiple sequentially connected nodes, it is necessary to calculate the vertical distance from the coordinate point to each centerline segment of the roadway, and take the minimum value as the final judgment criterion. Optionally, the connectivity check uses a breadth-first or depth-first algorithm to search in the graph consisting of roadway nodes and connection relationships for whether there is a path between the node where the valid personnel position was located at the previous time and the nearest neighbor node associated with the current coordinate point.

[0087] In one embodiment of the present invention, all successful positioning coordinates of the positioning identification card carried by the person within a preset historical time period are retrieved from the positioning history database; all successful positioning coordinates are connected in chronological order to form a continuous historical movement trajectory line of the person. The continuous historical movement trajectory line is segmented, the average movement speed of the person on each trajectory segment is calculated, and the distribution of all average movement speeds is statistically analyzed to obtain a speed distribution pattern; all turning points on the continuous historical movement trajectory line are identified, and the number and frequency of turning from each roadway node to each branch roadway are calculated to form a turning probability matrix.

[0088] In specific implementation, the previously confirmed real-time location of the final personnel is obtained. This location is output by the positioning system in the previous complete positioning process and stored in a cache. The straight-line distance from the previously confirmed real-time location of the final personnel to each of the multiple candidate locations at the current time is calculated using the Euclidean distance formula for three-dimensional spatial coordinates. Based on the velocity distribution pattern in the historical movement trajectory characteristics, the maximum distance range of personnel movement within the positioning time interval is estimated. The velocity distribution pattern provides statistical information on the typical movement rate of personnel in different underground tunnel sections. Candidate locations whose straight-line distance exceeds the maximum distance range are initially screened out. This initial screening is performed based on a geometric threshold set for the maximum distance range. In some embodiments, for the remaining candidate locations, the directional change required to move from the previous location to the candidate location is calculated. The directional change is obtained by calculating the vector angle. The probability of a change in direction is assessed based on the turning probability matrix, which defines the prior probability distribution of personnel turning into various connected lanes at a specific lane node. The continuous likelihood value of the trajectory for each remaining candidate location point is generated. The continuous likelihood value of the trajectory is a quantitative score that combines the reasonableness of distance and the probability of direction.

[0089] It is understandable that determining the final real-time location of a person requires comparing the continuous likelihood values ​​of the trajectories of all remaining candidate locations, selecting the candidate location with the highest continuous likelihood value as the primary location at the current moment. The continuous likelihood value of the candidate location with the highest value is checked to see if it exceeds a preset confidence threshold, which is a pre-set threshold used to judge the reliability of the positioning result. If the continuous likelihood value exceeds the confidence threshold, the coordinates of the candidate location are determined as the final real-time location of the person, and the system output and historical records are updated. Optionally, if the continuous likelihood value does not exceed the confidence threshold, the final real-time location of the person confirmed in the previous moment remains unchanged, and the current moment is marked as a positioning ambiguity state, awaiting the positioning data from the next moment for joint processing. Joint processing involves cross-moment trajectory smoothing or filtering with the candidate point set from the next moment. The calculation of the continuous likelihood value can be combined with the distance factor and the direction factor, expressed as:

[0090]

[0091] in: Represents the continuous likelihood value of the trajectory. and These are the weighting coefficients for the distance factor and the direction factor, respectively. It is a straight-line distance between the current candidate point and the position at the previous time step. The function of distance The smaller the value, the larger the function value. It is about the directional change required to move from the previous position to the current candidate point. The function, direction change In the steering probability matrix, the higher the corresponding probability, the larger the function value. In some embodiments, the steering probability matrix is ​​stored in tabular form, as shown in Table 1.

[0092] Table 1: Segment of the Turning Probability Matrix

[0093] Current node number The previous alleyway leads to Next possible node number Steering angle description Historical turning probability N001 From node N000 N002 straight 0.85 N001 From node N000 N005 Turn left approximately 90 degrees 0.10 N001 From node N000 N008 Turn right approximately 90 degrees 0.05 N002 From node N001 N003 straight 0.70 N002 From node N001 N010 Turn left 45 degrees from the branch 0.30

[0094] See Figure 4 This is a bar chart showing the distribution of positioning accuracy for personnel in underground coal mines, illustrating the number of positioning attempts within different accuracy ranges. The comparison between "total positioning attempts" and "effective positioning attempts" reflects the reliability of the positioning results; higher positioning accuracy (smaller range) and more positioning attempts indicate more stable system performance in the high-precision range. For underground coal mine scenarios, a high proportion of high-precision positioning supports personnel safety management, while the proportion of effective positioning can serve as a quantitative indicator of system availability. The chart visually reflects the concentration of positioning system accuracy, with the 0-0.5m range having the highest proportion, indicating more stable system performance in the high-precision range, and serving as a core basis for verifying system reliability.

[0095] In one embodiment of the present invention, the final real-time location of the person confirmed at the previous moment is obtained; the straight-line distance from the final real-time location of the person confirmed at the previous moment to each of the multiple candidate location points at the current moment is calculated. Based on the speed distribution pattern in the historical movement trajectory features, the maximum distance range of the person's movement within the positioning time interval is estimated; candidate location points whose straight-line distance exceeds the maximum distance range are initially screened out; for the remaining candidate location points, the direction change required to move from the previous location to the candidate location point is calculated, and the probability of the direction change is evaluated based on the turning probability matrix to generate the trajectory continuous likelihood value of each remaining candidate location point. The trajectory continuous likelihood values ​​of all remaining candidate location points are compared, and the candidate location point with the highest trajectory continuous likelihood value is selected; it is checked whether the trajectory continuous likelihood value of the candidate location point with the highest trajectory continuous likelihood value exceeds a preset confidence threshold; if it exceeds, the coordinates of the candidate location point are determined as the final real-time location of the person; if it does not exceed, the final real-time location of the person confirmed at the previous moment is kept unchanged, and the current moment is marked as a positioning ambiguity state, waiting for the positioning data of the next moment for joint processing.

[0096] In practice, the previously confirmed real-time location of the person is obtained; this location is the three-dimensional coordinates output by the positioning system in the most recent successful positioning cycle. The straight-line Euclidean distance from the previously confirmed real-time location to each of the multiple candidate locations at the current moment is calculated, generating a distance list. In some embodiments, the maximum distance range of the person's movement within the positioning time interval is estimated based on the velocity distribution pattern in the historical movement trajectory characteristics. The velocity distribution pattern is the distribution of different velocity intervals and their frequencies statistically derived from the person's historical movement data. It can be understood that candidate locations whose straight-line Euclidean distance exceeds the maximum distance range are initially screened out; this step eliminates unreasonable candidate locations based on the constraint of physical motion continuity.

[0097] For the remaining candidate locations, calculate the direction change vector required to move from the previously confirmed final real-time location of the person to each candidate location. Evaluate the probability of this direction change vector based on the turning probability matrix, which records the statistical probability of a person turning into different connected lanes at each node in the lane network. Generate a continuous likelihood value for the trajectory of each remaining candidate location, which combines distance plausibility and directional probability. Compare the continuous likelihood values ​​of all remaining candidate locations and select the candidate location with the highest continuous likelihood value as the primary candidate location for the current moment. Check whether the continuous likelihood value of the candidate location with the highest continuous likelihood value exceeds a preset confidence threshold, which is a pre-set numerical threshold based on the accuracy requirements of the positioning system.

[0098] Optionally, if the trajectory continuous likelihood value exceeds the confidence threshold, the coordinates of the candidate location point are determined as the final real-time location of the person, and the positioning output is updated. In some embodiments, if the trajectory continuous likelihood value does not exceed the confidence threshold, the final real-time location of the person confirmed at the previous moment is kept unchanged, and the current moment is marked as a positioning ambiguity state, waiting for the positioning data from the next moment to be jointly processed. The trajectory continuous likelihood value can be calculated using an evaluation function, expressed as:

[0099]

[0100] in: Let represent the continuous likelihood value of the trajectory of the i-th candidate location point. Let represent the linear Euclidean distance from the previous position to the i-th candidate position. This represents the reasonable maximum distance to move within the positioning time interval, estimated based on the velocity distribution pattern. It is a distance attenuation coefficient. This represents the roadway node where a given person was at the previous moment. Under the given conditions, the required directional change to move to the current candidate location point The conditional probability corresponding to the turning probability matrix, the function This represents the natural exponential function.

[0101] See Figure 5 This is a heatmap showing the turning probability distribution between nodes in underground coal mine roadways. Combined with a person's current node, this map can predict the high-probability target node for their next turn, assisting the positioning system in selecting candidate locations. High-probability turning paths can serve as prior information for the positioning algorithm, improving the accuracy of assessing the continuity of personnel trajectories. The overall turning probability between nodes is relatively low, consistent with the relatively fixed movement paths of underground personnel. Some nodes have the potential for multi-directional turning and are key diversion nodes in the underground path.

[0102] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0103] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A 5G-based method for locating personnel underground in coal mines, characterized in that: The method includes: A spatial coordinate database of the underground roadway environment is established in the positioning computing center. The spatial coordinate database contains the three-dimensional coordinates of all roadway nodes and the topological associations between nodes. The carrier signal transmitted by the 5G positioning identification card carried by the personnel is received in real time through the underground 5G positioning beacon. The carrier signal contains the unique number of the identification card and the initial timestamp. The carrier signal is subjected to multipath separation processing to obtain the arrival time sequence of the direct path signal; based on the arrival time sequence of the direct path signal and the known coordinates of the 5G positioning beacon, the difference in signal propagation distance between the personnel and multiple 5G positioning beacons is calculated. The node constraint model matching the current roadway partition is retrieved from the spatial coordinate database, and the signal propagation distance difference is substituted into the node constraint model for collaborative solution to obtain the initial position coordinate set of the personnel. The initial set of position coordinates is filtered by roadway connectivity to obtain multiple candidate position points that meet the roadway structure constraints; Extracting historical movement trajectory features of personnel, including speed distribution patterns and turning probability matrices; The historical movement trajectory features are used to evaluate the trajectory continuity of the multiple candidate location points to determine the final real-time location of the personnel.

2. The 5G-based underground personnel positioning method in coal mines as described in claim 1, characterized in that, The process of performing multipath separation on the carrier signal to obtain the arrival time sequence of the direct path signal specifically includes: Receive raw waveform data of carrier signals synchronously uploaded by multiple 5G positioning beacons; An adaptive equalization filter is used to suppress channel interference in the original waveform data of the carrier signal to obtain a pre-equalized signal waveform. Multipath component search is performed on the pre-equalized signal waveform to identify all resolvable paths whose signal energy exceeds a preset threshold. Based on the arrival angle and arrival time difference of the distinguishable path, and combined with the preset tunnel reflection surface model, the signal component that best matches the straight-line propagation characteristics is selected from all the distinguishable paths, and the signal component is marked as the straight-line signal. The arrival times of the signal peaks of the direct path signal are accurately extracted and arranged in chronological order to generate a sequence of arrival times of the direct path signal.

3. The 5G-based underground personnel positioning method in coal mines as described in claim 2, characterized in that, The specific differences in signal propagation distance between the operator and multiple 5G positioning beacons include: Extract the individual arrival time corresponding to each 5G positioning beacon from the arrival time sequence of the direct path signal; Read the base time of each 5G positioning beacon when it sends its own synchronization beacon; Calculate the time difference between the individual arrival time and the reference time to obtain the one-way propagation time of the signal from the personnel's location to each 5G positioning beacon; Multiplying the one-way propagation time by a preset signal propagation speed constant yields the one-way spatial distance from the personnel's location to each 5G positioning beacon. A 5G positioning beacon is selected as the distance reference benchmark. The one-way spatial distance from the person's location to each other 5G positioning beacon is subtracted from the one-way spatial distance from the person's location to the distance reference benchmark to generate the signal propagation distance difference.

4. The 5G-based underground personnel positioning method in coal mines as described in claim 1, characterized in that, Retrieving the node constraint model matching the current roadway partition from the spatial coordinate database specifically includes: Based on the most recent calculated personnel location, determine the possible lane section number where the personnel are currently located; Based on the lane partition number, the three-dimensional coordinates of all lane nodes within the lane partition number are indexed from the spatial coordinate database; Based on the three-dimensional coordinates of all roadway nodes within the roadway partition number, construct an adjacency matrix describing the connection relationships between nodes; Based on the physical width parameters of the tunnel, a location feasible domain space is generated with each tunnel node as the center; By integrating the adjacency matrix with the location feasible domain space of all lane nodes, a node constraint model is formed that describes the possible distribution of personnel locations within the lane partition.

5. The 5G-based underground personnel positioning method in coal mines as described in claim 4, characterized in that, Substituting the signal propagation distance difference into the node constraint model for collaborative solution, the initial set of personnel position coordinates is obtained, specifically including: Within the entire feasible domain space defined by the node constraint model, a system of equations is established regarding the difference in signal propagation distance, where the unknowns in the system of equations are the three-dimensional coordinates of the personnel. A global search algorithm is used to generate a large number of random location sampling points in the location feasible region space; Substitute the coordinates of each random sampling point into the system of equations to calculate the difference in theoretical signal propagation distance corresponding to the random sampling point. Compare the theoretical signal propagation distance difference corresponding to each random sampling point with the actual measured signal propagation distance difference, and calculate the sum of squared residuals between the two. Select several random sampling points with the smallest sum of squared residuals, and use the set of their coordinates as the initial set of coordinates of the personnel.

6. The 5G-based underground personnel positioning method in coal mines as described in claim 5, characterized in that, Performing tunnel connectivity filtering on the initial set of location coordinates to obtain multiple candidate location points that meet the tunnel structure constraints specifically includes: Read each coordinate point in the initial position coordinate set; Calculate the vertical distance from each coordinate point to the centerline of all roadway nodes in the node constraint model; Determine whether the vertical distance is less than or equal to half the physical width of the tunnel; if it is greater, mark the coordinate point as an invalid point and discard it. For coordinate points with valid vertical distance, further check whether there is a continuous path formed by connecting roadway nodes between the coordinate points and the valid personnel positions confirmed at the previous moment. If no continuous path exists, the coordinate points will also be marked as invalid and removed. The remaining coordinate points after both vertical distance and connectivity are used to determine multiple candidate locations that meet the constraints of the roadway structure.

7. The 5G-based underground personnel positioning method in coal mines as described in claim 6, characterized in that, The specific features of the extracted personnel's historical movement trajectory include: Retrieve all successful location coordinates of the person's location identification card within a preset historical time period from the location history database; Connect all the successfully located coordinate points in chronological order to form a continuous historical movement trajectory line of the person; The continuous historical movement trajectory is segmented, the average movement speed of the person on each trajectory segment is calculated, and the distribution of all average movement speeds is statistically analyzed to obtain the speed distribution pattern. Identify all turning points on the continuous historical movement trajectory line, calculate the number and frequency of turns made by personnel at each roadway node to each possible branch roadway, and form the turning probability matrix.

8. The 5G-based underground personnel positioning method in coal mines as described in claim 1, characterized in that, The evaluation of trajectory continuity of the multiple candidate location points using the historical movement trajectory features specifically includes: Obtain the real-time location of the last confirmed person at the previous moment; calculate the straight-line distance from the real-time location of the last confirmed person at the previous moment to each of the multiple candidate location points at the current moment; Based on the speed distribution pattern in the historical movement trajectory features, the maximum distance that a person may move within the positioning time interval is estimated. Candidate location points whose straight-line distance exceeds the maximum distance range are initially screened out; For the remaining candidate locations, calculate the directional change required to move from the previous location to the candidate location, evaluate the probability of the directional change based on the turning probability matrix, and generate a continuous likelihood value for the trajectory of each remaining candidate location.

9. The 5G-based underground personnel positioning method in coal mines as described in claim 8, characterized in that, Determining the final real-time location of personnel specifically includes: Compare the continuous likelihood values ​​of the trajectories of all remaining candidate locations, and select the candidate location with the highest continuous likelihood value. Check whether the continuous likelihood value of the candidate position point with the highest continuous likelihood value exceeds a preset confidence threshold; If the number of candidate locations exceeds the limit, the coordinates of the candidate locations will be determined as the final real-time location of the personnel. If the time limit is not exceeded, the real-time location of the last confirmed person will remain unchanged, and the current time will be marked as a location ambiguity state, waiting for the location data of the next time moment to be processed together.

10. A 5G-based underground personnel positioning system for coal mines, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the 5G-based underground personnel positioning method for coal mines as described in any one of claims 1 to 9.

Images