UWB-based personnel precise positioning and trajectory tracing safety management method and system for utility tunnels

Abstract

This invention provides a method and system for precise UWB personnel positioning and trajectory backtracking safety management in utility tunnels, relating to the field of utility tunnel safety management. The method includes collecting wireless signals and combining them with the linear structural constraints of the utility tunnel for positioning calculation; extracting spatial semantic tags to construct a trajectory representation structure; establishing zone-specific differentiated anomaly identification rules; backtracking historical trajectories and retrieving related personnel trajectories when safety risks exist; and calculating trajectory similarity as a weighting coefficient to trigger collaborative early warning. This invention improves the positioning accuracy of personnel in utility tunnels, enhances the interpretability of trajectory data, and achieves precise risk early warning and source tracing.

Description

Technical Field

[0001] This invention relates to utility tunnel safety management technology, and more particularly to a method and system for precise positioning and trajectory tracing of personnel in utility tunnels using UWB (Ultra-Wideband) technology. Background Technology

[0002] Traditional personnel management systems for utility tunnels primarily rely on access cards and video surveillance, which cannot achieve real-time, accurate personnel location tracking. With the development of IoT technology, UWB (Ultra-Wideband) technology, due to its high positioning accuracy, strong resistance to multipath interference, and good penetration, is widely used in indoor positioning. In the utility tunnel environment, UWB technology can overcome problems such as complex structures and interference from metal equipment, providing technical support for accurate personnel location tracking.

[0003] However, existing technologies primarily focus on acquiring single location information, lacking consideration for the unique environmental factors of utility tunnels. They fail to integrate location information with spatial semantic information such as functional zoning and environmental risk levels, resulting in a lack of targeted and differentiated safety management. This makes it impossible to adjust safety management strategies according to the risk levels of different areas. Existing systems often only provide real-time location monitoring, lacking the ability to effectively record and analyze historical trajectories. After a safety incident, they cannot quickly trace the activity trajectories of relevant personnel, hindering accident cause analysis and accountability, thus reducing the effectiveness of safety management. Furthermore, there is a general lack of personnel behavior pattern analysis and related personnel identification mechanisms, making it impossible to detect potential collaborative work safety risks. When an individual exhibits abnormal behavior, the system cannot automatically identify other personnel in close contact with them, leading to inaccurate safety warnings and hindering effective risk prevention and control, increasing the likelihood of safety incidents. Summary of the Invention

[0004] This invention provides a method and system for precise positioning and trajectory tracing of personnel in utility tunnels using UWB (Ultra-Wideband) technology, which can solve the problems in the prior art.

[0005] A first aspect of this invention provides a method for precise positioning and trajectory tracing of personnel in utility tunnels using UWB (Ultra-Wideband) for safety management, comprising:

[0006] The system collects wireless signals emitted by the target personnel, and performs positioning calculations based on the signal arrival time difference and the linear structural constraints of the utility tunnel to obtain real-time spatial coordinates.

[0007] Extract the utility tunnel functional zoning attributes and environmental risk levels corresponding to the real-time spatial location coordinates as spatial semantic tags, embed them into the mobile trajectory data chain, construct a semantically enhanced trajectory representation structure, and store it in the trajectory database;

[0008] Based on spatial semantic tags, zone-differentiated anomaly identification rules are established, behavioral constraints that match the environmental risk level are set, and safety status assessment results are generated by coupling location change features with spatial semantic tags.

[0009] When the safety status assessment results indicate that there is a safety risk, the historical trajectory of the target personnel is retrieved from the trajectory database, and the trajectories of related personnel in the same functional area that are spatially and temporally associated with the target personnel are retrieved. The trajectory similarity is calculated by comparing the movement path and the stop position.

[0010] Using trajectory similarity as a weighting coefficient, collaborative warnings are triggered for associated personnel whose weighting coefficients exceed a preset threshold. At the same time, the interaction patterns between historical trajectories and the trajectories of associated personnel are extracted as behavioral baseline features, which are used to dynamically update the behavioral constraints of the anomaly identification rules.

[0011] The system collects wireless signals emitted by the target personnel and performs positioning calculations based on the signal arrival time difference and the linear structural constraints of the utility tunnel, obtaining real-time spatial coordinates including:

[0012] By receiving wireless signals transmitted by the target personnel through multiple ultra-wideband base stations, calculating the signal arrival time difference between each ultra-wideband base station and the target personnel, and obtaining multiple sets of spatial distance measurement values ​​based on the signal arrival time difference;

[0013] The central axis and physical boundary of the utility tunnel are extracted to construct an accessibility constraint space. Multiple sets of spatial distance measurements are projected and mapped within the accessibility constraint space. By constraining the intersection of the distance measurement circles to the neighborhood of the central axis, the distance measurement results that deviate from the direction of the utility tunnel are corrected to the corresponding positions on the central axis.

[0014] The vertical offset distance between the corrected position coordinates and the central axis of the utility tunnel is calculated. When the vertical offset distance exceeds the allowable range of the physical boundary of the utility tunnel, the position coordinates are moved back to the inside of the physical boundary in the vertical direction to obtain real-time spatial position coordinates that conform to the spatial topology characteristics of the utility tunnel.

[0015] Extracting the functional zoning attributes and environmental risk levels of the utility tunnel corresponding to real-time spatial location coordinates as spatial semantic tags, embedding them into the movement trajectory data chain, constructing a semantically enhanced trajectory representation structure, and storing it in the trajectory database includes:

[0016] The functional zones of the utility tunnels are determined based on real-time spatial coordinates, and the attributes of the functional zones and environmental risk levels are extracted and combined into spatial semantic tags.

[0017] Spatial semantic labels are bound to real-time spatial location coordinates and timestamps to form trajectory data units, and multiple trajectory data units are connected in sequence according to timestamp order to construct a mobile trajectory data chain.

[0018] Extract continuous trajectory data unit sequences with the same spatial semantic labels from the mobile trajectory data chain, calculate the dispersion of the time span and spatial location coordinates of the continuous trajectory data unit sequences, and embed the product of the time span and dispersion as the partition dwell feature into the spatial semantic label.

[0019] For the location where the spatial semantic label is switched in the mobile trajectory data chain, the difference in the partition dwell feature before and after the switch is extracted, and the difference in the partition dwell feature is used as the partition transition strength marker to the spatial semantic label switching location;

[0020] The mobile trajectory data chain, which includes partition dwell features and partition transition intensity, is constructed into a semantically enhanced trajectory representation structure and stored in the trajectory database.

[0021] Based on spatial semantic tags, zone-differentiated anomaly identification rules are established, and behavioral constraints matching environmental risk levels are set. Safety status assessment results are generated by coupling location change features with spatial semantic tags, including:

[0022] Based on the functional zoning attributes of the utility tunnel in the spatial semantic tags, establish zoning-differentiated anomaly identification rules, and set behavioral constraints according to the environmental risk level in the spatial semantic tags.

[0023] Extract the real-time spatial location coordinate change sequence of the target person from the semantically enhanced trajectory representation structure, and calculate the position change features of the real-time spatial location coordinate change sequence;

[0024] Extract the historical location change characteristics of target personnel in the mobile trajectory data chain under the same utility tunnel functional zoning attributes but different environmental risk levels, construct the inverse constraint mapping relationship between environmental risk level and location change characteristic tolerance threshold, and convert the current environmental risk level into a dynamic tolerance threshold for location change characteristics based on the inverse constraint mapping relationship.

[0025] The location change features are compared with the dynamic tolerance threshold. When the location change features exceed the dynamic tolerance threshold, the anomaly identification rule is triggered. The extent of the location change features exceeding the dynamic tolerance threshold is extracted, and the extent of the exceedance is normalized with the dynamic tolerance threshold to generate a safety status assessment result.

[0026] When the safety status assessment indicates a safety risk, the historical trajectory of the target personnel is retrieved from the trajectory database, and the trajectories of related personnel within the same functional area who have spatiotemporal associations with the target personnel are searched. The trajectory similarity is calculated by comparing the movement path and the stopping position, including:

[0027] When the safety status assessment results indicate that there is a safety risk, the historical trajectory of the target personnel is retrieved from the trajectory database.

[0028] Retrieve the trajectories of related personnel within the same functional area from the trajectory database, and determine whether there is a spatiotemporal correlation between the trajectories of related personnel and the historical trajectories of the target personnel;

[0029] Extract the spatial semantic tag sequences contained in the historical trajectory of the target personnel and the trajectory of related personnel, and use the environmental risk level in the spatial semantic tag sequences as a weighting factor to construct a risk-weighted movement path;

[0030] By comparing risk-weighted movement paths, when the historical trajectory of the target person overlaps with the risk-weighted movement path of the associated person, the spatial semantic labels corresponding to the overlapping area are extracted. The environmental risk level of the overlapping area is multiplied by the spatial range of the overlapping area to generate the risk cross intensity. The risk cross intensity is used as a correction term for trajectory similarity to calculate trajectory similarity.

[0031] The interaction patterns between historical trajectories and the trajectories of associated individuals are extracted as behavioral baseline features. These features are used to dynamically update the behavioral constraints of anomaly identification rules.

[0032] Extract temporal correlation features of spatial location coordinate changes from the historical trajectory of the target personnel and the trajectory of related personnel, and decompose the temporal correlation features into synchronous movement patterns and alternating movement patterns;

[0033] For synchronous movement mode, the consistency of movement direction between target personnel and related personnel within the same time window is extracted; for alternating movement mode, the continuity of spatial position between target personnel and related personnel within adjacent time windows is extracted. The consistency of movement direction and the continuity of spatial position are integrated to construct the interaction pattern as the behavioral baseline feature.

[0034] The behavior baseline features are compared with the behavior constraints in the partitioned differential anomaly identification rules. When the consistency of movement direction or the continuity of spatial position in the behavior baseline features exceeds the preset range of the behavior constraints, the offset exceeding the preset range is extracted. The boundary parameters of the behavior constraints are expanded and adjusted according to the offset, and the behavior constraints of the anomaly identification rules are dynamically updated.

[0035] A second aspect of this invention provides a UWB-based personnel precise positioning and trajectory tracing safety management system for utility tunnels, comprising:

[0036] The first unit is used to collect the wireless signals emitted by the target personnel, and to perform positioning calculations based on the signal arrival time difference and the linear structural constraints of the utility tunnel to obtain the real-time spatial coordinates.

[0037] The second unit is used to extract the utility tunnel functional zoning attributes and environmental risk levels corresponding to the real-time spatial location coordinates as spatial semantic tags, embed them into the mobile trajectory data chain, construct a semantically enhanced trajectory representation structure, and store it in the trajectory database.

[0038] The third unit is used to establish zone-differentiated anomaly identification rules based on spatial semantic labels, set behavioral constraints that match the environmental risk level, and generate safety status assessment results by coupling location change features with spatial semantic labels.

[0039] The fourth unit is used to retrieve the historical trajectory of the target personnel from the trajectory database when the safety status assessment results indicate that there is a safety risk, and to retrieve the trajectory of related personnel in the same functional area who have a spatiotemporal relationship with the target personnel, and to calculate the trajectory similarity by comparing the movement path and the stop position.

[0040] The fifth unit is used to use trajectory similarity as a weighting coefficient to trigger collaborative warnings for associated personnel whose weighting coefficients exceed a preset threshold. At the same time, it extracts the interaction patterns between historical trajectories and the trajectories of associated personnel as behavioral baseline features to dynamically update the behavioral constraints of the anomaly identification rules.

[0041] A third aspect of the present invention provides an electronic device, comprising:

[0042] processor;

[0043] Memory used to store processor-executable instructions;

[0044] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0045] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0046] This invention utilizes wireless signal time-of-arrival (TOA) positioning combined with linear structural constraints to calculate the location of target personnel. It extracts the corresponding utility tunnel functional zone attributes and environmental risk levels as semantic tags, effectively improving positioning accuracy in complex utility tunnel environments and achieving the fusion of location information and environmental semantics. This significantly enhances the spatial perception capability of the utility tunnel personnel positioning system. The semantically enhanced trajectory representation structure constructed in this invention, combined with differentiated anomaly identification rules and risk-matching behavioral constraints, can dynamically adjust safety monitoring standards according to the characteristics of different functional zones. This enables accurate assessment of personnel behavior and early risk detection, significantly improving the intelligence level and timeliness of utility tunnel safety management. Through trajectory backtracking and correlation analysis mechanisms, it can quickly identify potentially risk-related personnel and achieve collaborative early warning based on trajectory similarity. Simultaneously, it dynamically updates behavioral constraints using historical interaction patterns, forming an adaptive safety rule system. This effectively solves the problems of isolated monitoring and static rules in traditional utility tunnel safety management, which cannot cope with complex scenarios, thus improving the systematic nature and adaptability of safety management. Attached Figure Description

[0047] Figure 1 This is a flowchart illustrating the safety control method for precise positioning and trajectory tracing of personnel in utility tunnels according to an embodiment of the present invention.

[0048] Figure 2 This is a flowchart of the pipe gallery anomaly identification operation according to an embodiment of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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.

[0050] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0051] Figure 1 This is a flowchart illustrating the safety control method for precise positioning and trajectory tracing of personnel in utility tunnels according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0052] The system collects wireless signals emitted by the target personnel, and performs positioning calculations based on the signal arrival time difference and the linear structural constraints of the utility tunnel to obtain real-time spatial coordinates.

[0053] Extract the utility tunnel functional zoning attributes and environmental risk levels corresponding to the real-time spatial location coordinates as spatial semantic tags, embed them into the mobile trajectory data chain, construct a semantically enhanced trajectory representation structure, and store it in the trajectory database;

[0054] Based on spatial semantic tags, zone-differentiated anomaly identification rules are established, behavioral constraints that match the environmental risk level are set, and safety status assessment results are generated by coupling location change features with spatial semantic tags.

[0055] When the safety status assessment results indicate that there is a safety risk, the historical trajectory of the target personnel is retrieved from the trajectory database, and the trajectories of related personnel in the same functional area that are spatially and temporally associated with the target personnel are retrieved. The trajectory similarity is calculated by comparing the movement path and the stop position.

[0056] Using trajectory similarity as a weighting coefficient, collaborative warnings are triggered for associated personnel whose weighting coefficients exceed a preset threshold. At the same time, the interaction patterns between historical trajectories and the trajectories of associated personnel are extracted as behavioral baseline features, which are used to dynamically update the behavioral constraints of the anomaly identification rules.

[0057] In one optional implementation, the wireless signal emitted by the target personnel is collected, and the positioning solution is performed based on the signal arrival time difference combined with the linear structural constraints of the utility tunnel, to obtain the real-time spatial coordinates, including:

[0058] By receiving wireless signals transmitted by the target personnel through multiple ultra-wideband base stations, calculating the signal arrival time difference between each ultra-wideband base station and the target personnel, and obtaining multiple sets of spatial distance measurement values ​​based on the signal arrival time difference;

[0059] The central axis and physical boundary of the utility tunnel are extracted to construct an accessibility constraint space. Multiple sets of spatial distance measurements are projected and mapped within the accessibility constraint space. By constraining the intersection of the distance measurement circles to the neighborhood of the central axis, the distance measurement results that deviate from the direction of the utility tunnel are corrected to the corresponding positions on the central axis.

[0060] The vertical offset distance between the corrected position coordinates and the central axis of the utility tunnel is calculated. When the vertical offset distance exceeds the allowable range of the physical boundary of the utility tunnel, the position coordinates are moved back to the inside of the physical boundary in the vertical direction to obtain real-time spatial position coordinates that conform to the spatial topology characteristics of the utility tunnel.

[0061] In a utility tunnel environment, multiple ultra-wideband (UWB) base stations are deployed along the tunnel to receive wireless signals emitted by location tags carried by personnel. These tags periodically emit UWB pulse signals, characterized by a wide frequency range and narrow time-domain pulses, providing nanosecond-level time resolution. As personnel move within the tunnel, the signals emitted by their location tags are received by UWB base stations at different locations. In a utility tunnel environment, base stations are typically installed every 50 meters along the central axis of the tunnel to ensure continuous and reliable signal coverage.

[0062] The timestamps of signals received by each ultra-wideband (UWB) base station are collected. By calculating the time difference between different base stations receiving the same signal, the relative distance difference between the target person and each base station is obtained. For example, four UWB base stations A, B, C, and D are deployed in the utility tunnel, located at different positions within the tunnel passage. When the positioning tag carried by the target person transmits a signal, base station A receives the signal at time t1, base station B at time t2, base station C at time t3, and base station D at time t4. The time differences between the base stations Δt(AB) = t2 - t1, Δt(AC) = t3 - t1, and Δt(AD) = t4 - t1 are calculated and multiplied by the speed of light to obtain the distance differences ΔDAB, ΔDAC, and ΔDAD.

[0063] Based on the obtained spatial distance difference measurements, a system of hyperbolic equations is constructed. Under normal circumstances, the intersection of these hyperbolas is the location of the target personnel. However, due to interference from factors such as signal reflection and multipath effects, especially in narrow and enclosed spaces like utility tunnels, the directly calculated location often has a large error, which may result in the positioning result deviating from the physical space of the utility tunnel.

[0064] To address the aforementioned issues, this method introduces topological constraints on the utility tunnel passage. First, the geometric model of the utility tunnel passage is obtained through laser scanning or high-precision measurement, and its central axis coordinate sequence and passage width parameters are extracted. The central axis of the utility tunnel passage can be represented by a series of three-dimensional coordinate points, with adjacent points forming a continuous axis through linear interpolation. The passage width parameter defines the maximum walkable range extending to both sides of the central axis.

[0065] After obtaining the preliminary positioning results, project them into the constrained space of the utility tunnel. Find the point on the central axis closest to the initial positioning point and calculate the vertical distance from the initial positioning point to that point. If the vertical distance is less than half the width of the tunnel, retain the original positioning results; if the vertical distance is greater than half the width of the tunnel, move the positioning point vertically to the inside of the tunnel boundary.

[0066] For example, the width of the utility tunnel is 3 meters, and the initial coordinates obtained from a certain positioning calculation are (x0, y0, z0). The nearest point on the central axis (xc, yc, zc) is found, and the calculated vertical distance d = 2.1 meters, which exceeds half the tunnel width (1.5 meters). The positioning point is then corrected to (x1, y1, z1), so that the corrected point is located inside the physical boundary of the tunnel, with a vertical distance of 1.5 meters. This positioning result not only satisfies the distance measurement equation constraints but also conforms to the physical spatial limitations of the utility tunnel.

[0067] When a target person is located at a corner or fork in the utility tunnel, the system determines their path based on the movement trajectories of multiple consecutive positioning points and the tunnel's topology. If multiple possible paths exist, a Bayesian filtering algorithm is used to comprehensively consider historical movement trajectories and the probabilities of each path, selecting the most probable path to further improve positioning accuracy. Simultaneously, optimizations are made to address the multipath effect in the unique environment of the utility tunnel. When ultra-wideband signals propagate within the tunnel, they are reflected and scattered by walls, pipes, and other structures, resulting in multiple propagation paths. By analyzing signal waveform characteristics, the system identifies and extracts the earliest arriving direct signal, filtering out subsequent multipath components to reduce the impact of multipath interference on positioning accuracy. A dynamic threshold is set to detect the signal leading edge, and a pulse shape matching algorithm is used to accurately capture the arrival time of the direct signal.

[0068] This invention effectively solves the technical problem of inaccurate positioning caused by multipath effects and signal attenuation in closed linear spaces using traditional positioning methods. By utilizing the spatial topological characteristics of the utility tunnel as prior knowledge, the original ranging results are effectively constrained and corrected, ensuring that the positioning results always conform to the physical spatial limitations of the utility tunnel and avoiding the generation of unreasonable positioning points. By combining signal processing technology with spatial topological constraints, high-precision positioning is achieved while maintaining the simplicity of system deployment. This provides reliable technical support for the safety management of personnel within the utility tunnel and has significant value in improving the efficiency of utility tunnel operation and maintenance, ensuring personnel safety, and assisting emergency rescue. It can be widely applied in various utility tunnel safety management scenarios.

[0069] In one optional implementation, extracting the utility tunnel functional zoning attributes and environmental risk levels corresponding to real-time spatial location coordinates as spatial semantic tags, embedding them into the movement trajectory data chain, constructing a semantically enhanced trajectory representation structure, and storing it in the trajectory database includes:

[0070] The functional zones of the utility tunnels are determined based on real-time spatial coordinates, and the attributes of the functional zones and environmental risk levels are extracted and combined into spatial semantic tags.

[0071] Spatial semantic labels are bound to real-time spatial location coordinates and timestamps to form trajectory data units, and multiple trajectory data units are connected in sequence according to timestamp order to construct a mobile trajectory data chain.

[0072] Extract continuous trajectory data unit sequences with the same spatial semantic labels from the mobile trajectory data chain, calculate the dispersion of the time span and spatial location coordinates of the continuous trajectory data unit sequences, and embed the product of the time span and dispersion as the partition dwell feature into the spatial semantic label.

[0073] For the location where the spatial semantic label is switched in the mobile trajectory data chain, the difference in the partition dwell feature before and after the switch is extracted, and the difference in the partition dwell feature is used as the partition transition strength marker to the spatial semantic label switching location;

[0074] The mobile trajectory data chain, which includes partition dwell features and partition transition intensity, is constructed into a semantically enhanced trajectory representation structure and stored in the trajectory database.

[0075] First, by obtaining the real-time spatial coordinates of moving objects within the utility tunnel, the functional zone of the tunnel to which those coordinates belong is determined. These functional zones can include different sections such as power supply, water supply, gas supply, and communication. Each functional zone has a pre-defined environmental risk level, such as low-risk, medium-risk, and high-risk areas. The functional zone database is then searched using the spatial coordinates to obtain the corresponding functional zone attributes and environmental risk level. For example, at coordinates (120.52, 30.68, 15.3), the location is found to belong to "Gas Supply - High-Risk Zone". "Gas Supply - High-Risk Zone" is then combined to form the spatial semantic tag for that location.

[0076] Spatial semantic tags are bound to real-time spatial location coordinates and timestamps to form trajectory data units. Each trajectory data unit contains three parts: location coordinates, timestamp, and spatial semantic tag. For example, a trajectory data unit might be {Location: (120.52, 30.68, 15.3), Timestamp: 2023-08-10 14:30:25, Semantic Tag: "Gas Chamber - High-Risk Area"}. Multiple trajectory data units are chained together according to the order of their timestamps to construct a complete movement trajectory data chain. For example, a trajectory data chain containing 1000 trajectory data units records the complete movement trajectory of a moving object within the utility tunnel over one hour.

[0077] Extract continuous trajectory data unit sequences with the same spatial semantic labels from the mobile trajectory data chain. For each trajectory data unit in the mobile trajectory data chain, check if its spatial semantic label is the same as the previous unit. If they are the same, add the unit to the currently processed continuous sequence; if they are different, end the processing of the current sequence and start a new continuous sequence. For each extracted continuous trajectory data unit sequence, calculate its time span and spatial position coordinate dispersion. The time span is calculated by subtracting the timestamp of the first unit from the timestamp of the last unit in the continuous sequence, obtaining the time length covered by the continuous sequence in seconds. The spatial position coordinate dispersion is calculated by the standard deviation of all position coordinates in the continuous sequence, representing the degree of positional change. The product of the time span and the dispersion is used as the zone dwell feature. For example, for a continuous trajectory data unit sequence in the "Power Cabin - Medium Risk Zone" with a time span of 300 seconds and a spatial position coordinate dispersion of 2.5 meters, the zone dwell feature is 300 × 2.5 = 750. The resident feature of this partition is embedded into the spatial semantic label corresponding to the continuous sequence to form an enhanced spatial semantic label: "Power Module - Medium Risk Zone - Resident Feature: 750".

[0078] For locations where spatial semantic labels switch in the mobile trajectory data chain, the difference in partition dwell feature before and after the switch is extracted. In the trajectory data chain, when two adjacent trajectory data units are detected to have different spatial semantic labels, it indicates that the moving object has moved from one functional partition to another. The partition dwell feature values ​​of the continuous sequence before and after the switch are extracted, and the difference between the two is calculated. For example, when switching from "Power Cabin - Medium Risk Zone" to "Communication Cabin - Low Risk Zone", the partition dwell feature of the former is 750, and the partition dwell feature of the latter is 450. The difference in partition dwell feature is 750-450=300. This difference in partition dwell feature is used as the partition transition strength and marked at the location of the spatial semantic label switch. Specifically, a partition transition strength field with a value of 300 is added to the trajectory data unit at the switch location, indicating the transition strength from the previous partition to the current partition.

[0079] The mobile trajectory data chain, including partition dwell features and partition transition strength, is constructed into a semantically enhanced trajectory representation structure. Each trajectory data unit is enhanced to include the following fields: location coordinates, timestamp, spatial semantic label, partition dwell feature, and partition transition strength (if switching locations). For example, an enhanced trajectory data unit is {location: (120.55, 30.70, 15.8), timestamp: 2023-08-10 14:35:42, semantic label: "communication cabin - low-risk area", dwell feature: 450, transition strength: 300}. The entire semantically enhanced trajectory representation structure is stored in a trajectory database for subsequent trajectory analysis and mining. The trajectory data in the database can be indexed and retrieved according to multiple dimensions such as mobile object ID, time period, and functional partition, supporting complex trajectory queries and analyses.

[0080] This invention successfully extracted the functional zoning attributes and environmental risk levels of utility tunnels corresponding to real-time spatial location coordinates as spatial semantic tags, embedded them into the movement trajectory data chain, constructed a semantically enhanced trajectory representation structure, and stored it in the trajectory database. This semantically enhanced trajectory representation structure not only records the location and time information of the moving object, but also contains rich semantic information and behavioral characteristics, providing strong support for the safety monitoring and management of utility tunnels.

[0081] like Figure 2 The diagram illustrates the operation process for identifying abnormalities in the utility tunnel in this embodiment.

[0082] In one optional implementation, a zone-differentiated anomaly identification rule is established based on spatial semantic tags, behavioral constraints matching the environmental risk level are set, and a safety status assessment result is generated by coupling location change features with spatial semantic tags.

[0083] Based on the functional zoning attributes of the utility tunnel in the spatial semantic tags, establish zoning-differentiated anomaly identification rules, and set behavioral constraints according to the environmental risk level in the spatial semantic tags.

[0084] Extract the real-time spatial location coordinate change sequence of the target person from the semantically enhanced trajectory representation structure, and calculate the position change features of the real-time spatial location coordinate change sequence;

[0085] Extract the historical location change characteristics of target personnel in the mobile trajectory data chain under the same utility tunnel functional zoning attributes but different environmental risk levels, construct the inverse constraint mapping relationship between environmental risk level and location change characteristic tolerance threshold, and convert the current environmental risk level into a dynamic tolerance threshold for location change characteristics based on the inverse constraint mapping relationship.

[0086] The location change features are compared with the dynamic tolerance threshold. When the location change features exceed the dynamic tolerance threshold, the anomaly identification rule is triggered. The extent of the location change features exceeding the dynamic tolerance threshold is extracted, and the extent of the exceedance is normalized with the dynamic tolerance threshold to generate a safety status assessment result.

[0087] In this embodiment, a differentiated anomaly identification rule base is first established based on the functional zoning attributes of the utility tunnel. The rule base sets corresponding abnormal behavior judgment criteria for the characteristics of different functional zones. In the power compartment, where equipment is densely packed and energized, the anomaly identification rules mainly focus on situations such as personnel remaining stationary for extended periods or abnormally rapid movement. In the water supply compartment, where pipes crisscross, the rules focus on monitoring behaviors such as personnel frequently turning back and forth or prolonged stays outside maintenance points. In the communication compartment, where equipment is sophisticated, the rules focus on situations such as unauthorized contact and excessive vibration. Environmental risk levels are also considered for each functional zone, translating them into behavioral constraints. Risk levels are divided into four levels: low, medium, high, and extremely high, each corresponding to different levels of behavioral restrictions. Low-risk areas have relaxed behavioral constraints, allowing for greater freedom of movement; as the risk level increases, behavioral constraints gradually become stricter, with extremely high-risk areas imposing strict restrictions on personnel movement speed, dwell time, and activity range.

[0088] The real-time spatial location coordinate change sequence of the target person is extracted from the semantically enhanced trajectory representation structure. This sequence consists of multiple coordinate points arranged in ascending order of timestamps. Position change features are calculated for the extracted coordinate change sequence, including four dimensions: velocity change rate, direction change frequency, dwell time proportion, and spatial fluctuation amplitude. The velocity change rate is obtained by removing the position value between adjacent time points and dividing the time interval to obtain the instantaneous velocity, then calculating the degree of change in the instantaneous velocity; the direction change frequency is obtained by calculating the number of angle changes formed by three consecutive coordinate points; the dwell time proportion refers to the proportion of time when the velocity is close to zero to the total observation time; and the spatial fluctuation amplitude is obtained by calculating the variance of the distance the coordinate point deviates from the average direction of movement.

[0089] Taking the trajectory of a maintenance worker in the power compartment as an example, the coordinate sequence over a 10-second period was extracted, and the average moving speed was calculated to be 0.5 m / s, with a maximum speed change rate of 0.2 m / s. 2 The direction changed 3 times, the dwell time accounted for 20%, and the spatial fluctuation amplitude was 0.3 meters. These values ​​together constitute the feature vector of the person's positional change within the current time window.

[0090] To establish an inverse constraint mapping relationship between environmental risk levels and tolerance thresholds for location change characteristics, historical data needs to be analyzed. Historical trajectory data of target personnel within the same utility tunnel functional zone but at different environmental risk levels are selected from the movement trajectory data chain, and the statistical distribution of location change characteristics under each risk level is extracted. By analyzing the differences in location change characteristics under different risk levels, an inverse constraint mapping model is constructed. The inverse constraint mapping model is a key technology for converting environmental risk levels into tolerance thresholds for location change characteristics. This model is built based on the analysis of a large amount of historical data, achieved by analyzing the statistical characteristics of normal personnel behavior patterns under different functional zones and risk levels. The construction process includes three main steps: data classification and statistics, mapping function fitting, and mapping table generation.

[0091] Specifically, for each dimension of location change characteristics, the tolerance threshold T corresponding to the risk level R can be expressed as the product of a base threshold and a risk adjustment factor. The base threshold is determined by the normal operational characteristics of that functional area, while the risk adjustment factor decreases as the risk level increases. For example, for the velocity change rate characteristic, the base threshold for the power compartment area is 0.5 m / s. 2 The risk adjustment factor is 1.0 for low risk, 0.8 for medium risk, 0.6 for high risk, and 0.4 for very high risk. When the power compartment area is classified as high risk, the tolerance threshold for the rate of change of velocity is 0.5 × 0.6 = 0.3 m / s. 2 .

[0092] By using a reverse constraint mapping relationship, the current environmental risk level is converted into dynamic tolerance thresholds for each dimension of the location change characteristics. For the example above, the current environmental risk level of the power compartment is high risk, and the corresponding tolerance threshold vector is: velocity change rate 0.3 m / s. 2 The direction change frequency is 4 times / 10 seconds, the dwell time ratio is 30%, and the spatial fluctuation amplitude is 0.4 meters.

[0093] The calculated location change characteristics are compared with a dynamic tolerance threshold. If any dimension exceeds the corresponding tolerance threshold, an anomaly detection rule is triggered. In the example, the maintenance personnel's velocity change rate is 0.2 m / s. 2 Below the threshold of 0.3 m / s 2 The frequency of directional changes was 3 times per 10 seconds, which is lower than the threshold of 4 times per 10 seconds; the dwell time ratio was 20%, which is lower than the threshold of 30%; the spatial fluctuation amplitude was 0.3 meters, which is lower than the threshold of 0.4 meters. All dimensions did not exceed the tolerance threshold and were judged to be in a normal state.

[0094] If the location change characteristics exceed the dynamic tolerance threshold, the exceedance magnitude is calculated. The exceedance magnitude equals the actual characteristic value minus the tolerance threshold, then divided by the tolerance threshold, representing the relative magnitude of the exceedance. The exceedance magnitudes of each dimension are weighted and averaged to obtain the comprehensive exceedance magnitude. The comprehensive exceedance magnitude is compared with the preset safety status level threshold to determine the safety status assessment result. The safety status assessment result is divided into four levels: safe, alert, warning, and danger, each corresponding to different response measures, such as normal monitoring, administrator prompts, audible and visual warnings, and emergency intervention.

[0095] If the space fluctuation range of the maintenance personnel suddenly increases to 0.6 meters, exceeding the tolerance threshold of 0.4 meters, the calculated over-limit range is (0.6-0.4) / 0.4=0.5. Assuming that the weight of this dimension is 0.3, and other dimensions do not exceed the limit, the comprehensive over-limit range is 0.5×0.3=0.15. This value is between the attention level threshold of 0.1 and the warning level threshold of 0.3. Therefore, the safety status assessment result is "warning," and the system will send a warning message to the administrator, prompting them to pay attention to the status of this personnel.

[0096] This invention achieves refined monitoring of personnel behavior within utility tunnels by establishing a dynamic coupling mechanism between spatial semantic tags and location change features. Compared to traditional fixed threshold judgment methods, this method can dynamically adjust the tolerance threshold based on different regional characteristics and real-time environmental risk levels, improving the accuracy and sensitivity of anomaly identification. In particular, through a reverse constraint mapping model, the system can intelligently associate risk levels with behavioral constraints, enabling behavioral constraints to adaptively adjust with changes in environmental risk. Furthermore, by analyzing multiple dimensions of location change features, it can capture richer patterns of abnormal behavior, reducing false alarm rates.

[0097] In one optional implementation, when the security status assessment indicates a security risk, the historical trajectory of the target personnel is retrieved from the trajectory database, and the trajectories of related personnel within the same functional area who are spatiotemporally associated with the target personnel are searched. Trajectory similarity is calculated by comparing movement paths and stopping locations, including:

[0098] When the safety status assessment results indicate that there is a safety risk, the historical trajectory of the target personnel is retrieved from the trajectory database.

[0099] Retrieve the trajectories of related personnel within the same functional area from the trajectory database, and determine whether there is a spatiotemporal correlation between the trajectories of related personnel and the historical trajectories of the target personnel;

[0100] Extract the spatial semantic tag sequences contained in the historical trajectory of the target personnel and the trajectory of related personnel, and use the environmental risk level in the spatial semantic tag sequences as a weighting factor to construct a risk-weighted movement path;

[0101] By comparing risk-weighted movement paths, when the historical trajectory of the target person overlaps with the risk-weighted movement path of the associated person, the spatial semantic labels corresponding to the overlapping area are extracted. The environmental risk level of the overlapping area is multiplied by the spatial range of the overlapping area to generate the risk cross intensity. The risk cross intensity is used as a correction term for trajectory similarity to calculate trajectory similarity.

[0102] When a safety assessment indicates a safety risk to a target personnel, their historical trajectory data must be retrieved from the trajectory database. The time window for retrieval can be adaptively adjusted according to the risk level. Generally, for risks at the "attention" level, trajectories within the past 2 hours are retrieved; for "warning" levels, within the past 8 hours; and for "hazard" levels, within the past 24 hours. The trajectory data includes timestamps, spatial coordinates, functional zones, and environmental risk levels. During retrieval, the target personnel ID and time range are indexed to filter eligible trajectory data units from the trajectory database, and these are then sorted by timestamp to form a complete historical trajectory sequence.

[0103] When retrieving the trajectories of other individuals potentially associated with the target individual from the trajectory database, a spatiotemporal overlap filtering mechanism is employed. This mechanism first determines overlapping time ranges, selecting other individuals who also appeared in the system within the target individual's historical trajectory time window; secondly, it filters for overlapping spatial ranges, meaning the trajectories of these individuals must intersect with the target individual's trajectory within the same functional area. Spatiotemporal overlap filtering can be achieved through rapid retrieval using the spatiotemporal index of the trajectory database.

[0104] When determining whether there is a spatiotemporal correlation between the trajectory of a related person and the historical trajectory of a target person, a multi-dimensional correlation determination standard is adopted. In the temporal dimension, the temporal overlap of the two trajectories is calculated, i.e., the proportion of the overlapping time period to the total time period. In the spatial dimension, the spatial proximity of the two trajectories is calculated, i.e., the average spatial distance between the two trajectories within the overlapping time period. In the functional dimension, it is determined whether the two individuals belong to the same work group or have similar job responsibilities. A spatiotemporal correlation is determined when the temporal overlap exceeds 30%, the spatial proximity is less than 10 meters, or the two individuals belong to the same work group.

[0105] Extract the spatial semantic tag sequences contained in the historical trajectory of the target personnel and the trajectories of related personnel. The spatial semantic tags include functional zoning attributes and environmental risk levels, arranged in chronological order to form a sequence. The spatial semantic tag sequence reflects the activity patterns of personnel in different functional areas and their exposure to different risk environments. In practical applications, the spatial semantic tag sequence of a maintenance personnel might be "Power compartment - Medium risk → Power compartment - High risk → Communication compartment - Low risk → Power compartment - Medium risk," representing the personnel's trajectory in these areas in chronological order.

[0106] Risk-weighted movement paths are constructed by using environmental risk levels from spatial semantic label sequences as weighting factors. Risk levels are categorized into four levels: low, medium, high, and very high, with corresponding weighting factors of 1, 2, 3, and 4, respectively. By weighting each point on the movement path according to the risk level weighting factor, areas with higher risk levels are given greater weight in trajectory similarity calculations, highlighting the importance of high-risk areas. The risk-weighted movement path can be represented as a series of three-dimensional coordinate points combined with their corresponding risk weights, forming a weighted set of trajectory points.

[0107] By comparing risk-weighted movement paths, the overlapping areas between the target individual's historical trajectory and the trajectories of related individuals are identified. The overlapping area refers to the region where two trajectories are spatially close and temporally overlapping. The method for determining the overlapping area involves pairing and comparing the coordinates of the two trajectories within the same time window. When the spatial distance between two coordinate points is less than a preset threshold (e.g., 3 meters), the areas where these two points are located are considered to overlap. All point pairs that meet the condition are extracted; the spatial regions corresponding to these point pairs are the overlapping areas.

[0108] Extract spatial semantic labels corresponding to overlapping areas and analyze the functional zone attributes and environmental risk levels at which the overlap occurs. Spatial semantic label extraction is based on coordinate points within the overlapping areas, obtaining the corresponding functional zone attributes and environmental risk levels through spatial mapping. If the overlapping area spans multiple functional zones, spatial semantic labels for each functional zone must be extracted separately.

[0109] The risk cross intensity is generated by multiplying the environmental risk level of the overlapping area by the spatial extent of the overlapping area. The environmental risk level is represented numerically: 1 for low risk, 2 for medium risk, 3 for high risk, and 4 for extremely high risk. The spatial extent is obtained by calculating the spatial distribution of coordinate points within the overlapping area, and can be represented as the volume or area of ​​the overlapping area. The risk cross intensity reflects the degree to which two individuals are jointly active in a high-risk area; a higher value indicates a higher potential collaborative risk.

[0110] The risk crossover intensity is used as a correction term for trajectory similarity to calculate the final trajectory similarity. The base trajectory similarity value is calculated comprehensively from three dimensions: trajectory shape similarity, velocity pattern similarity, and stop point consistency. Trajectory shape similarity uses the longest common subsequence algorithm to discretize the two trajectories into a series of path segments and calculate the proportion of common path segments to the total path segments; velocity pattern similarity is obtained by comparing the velocity distribution characteristics of the two trajectories at corresponding time points; stop point consistency is calculated by comparing the overlap of key stop point positions of the two trajectories. The base trajectory similarity is added to the correction value of the risk crossover intensity to obtain the final trajectory similarity.

[0111] For example, a power maintenance engineer's safety status assessment result is "warning." The system reviews the engineer's historical trajectory over the past 8 hours and retrieves three related personnel who appeared in the power compartment during the same time period. For one of these related personnel, spatiotemporal correlation analysis reveals a 43% temporal overlap between the two, with an average spatial distance of 6.8 meters, indicating a spatiotemporal correlation. Spatial semantic tag sequences are extracted from the two individuals to construct risk-weighted movement paths. Path comparison reveals significant overlap between the two individuals in the "power compartment - high-risk" area, primarily located near two high-voltage distribution cabinets, covering a spatial area of ​​approximately 15 square meters, with an environmental risk level of 3. The risk cross-similarity is calculated as 3 × 15 = 45. The baseline trajectory similarity is 0.68. After adding a correction value of 0.09 for the risk cross-similarity, the final trajectory similarity is 0.77, exceeding the warning threshold of 0.75, triggering a collaborative risk warning.

[0112] This invention's trajectory similarity calculation method, by introducing spatial semantic labels and environmental risk levels, achieves precise analysis of the correlation of personnel activities in utility tunnel environments. It not only considers the spatiotemporal characteristics of trajectories but also emphasizes the overlap of activities in high-risk areas, providing in-depth insights for utility tunnel safety management. Through retrospective analysis of the historical trajectories of target personnel and related personnel, it can effectively identify potential collaborative violations and collective risk exposures, enabling the tracking of safety risk diffusion and correlation transmission analysis. Based on a semantically enhanced trajectory representation structure, it can distinguish behavioral patterns under different functional zones and risk levels, improving the accuracy and interpretability of abnormal behavior identification. By calculating the risk crossover intensity as a correction term for trajectory similarity, it highlights the importance of shared activities in high-risk areas, making risk assessment more aligned with actual utility tunnel safety management needs.

[0113] In one optional implementation, the interaction patterns between historical trajectories and associated personnel trajectories are extracted as behavioral baseline features. These features are used to dynamically update the behavioral constraints of the anomaly identification rules, including:

[0114] Extract temporal correlation features of spatial location coordinate changes from the historical trajectory of the target personnel and the trajectory of related personnel, and decompose the temporal correlation features into synchronous movement patterns and alternating movement patterns;

[0115] For synchronous movement mode, the consistency of movement direction between target personnel and related personnel within the same time window is extracted; for alternating movement mode, the continuity of spatial position between target personnel and related personnel within adjacent time windows is extracted. The consistency of movement direction and the continuity of spatial position are integrated to construct the interaction pattern as the behavioral baseline feature.

[0116] The behavior baseline features are compared with the behavior constraints in the partitioned differential anomaly identification rules. When the consistency of movement direction or the continuity of spatial position in the behavior baseline features exceeds the preset range of the behavior constraints, the offset exceeding the preset range is extracted. The boundary parameters of the behavior constraints are expanded and adjusted according to the offset, and the behavior constraints of the anomaly identification rules are dynamically updated.

[0117] When extracting temporal correlation features of spatial coordinate changes from the historical trajectory of a target person and the trajectories of related persons, the two trajectories need to be segmented and aligned according to time windows. The length of the time window can be set according to the characteristics of personnel activities in the utility tunnel environment, typically ranging from 5 to 30 seconds. During the alignment process, a dynamic time warping algorithm is used to handle the time asynchrony problem between the two trajectories, ensuring that trajectory segments within the same time window are comparable. The extracted temporal correlation features include three dimensions: relative distance change, relative direction change, and relative speed change. Relative distance change describes the trend of distance between two people changing over time, relative direction change describes the degree of consistency in the movement directions of the two people, and relative speed change describes the matching of the speeds of the two people.

[0118] Temporal correlation features can be decomposed into two basic types: synchronous movement patterns and alternating movement patterns. Synchronous movement patterns refer to the coordinated movement characteristics of the target person and related persons within the same time window, such as moving simultaneously in the same or opposite directions, or maintaining a fixed distance. Alternating movement patterns refer to the sequential movement characteristics of two people within adjacent time windows, such as one person leaving an area and the other immediately entering, or one person completing an operation and the other continuing the operation. Pattern recognition technology is used in the decomposition process, based on a combination of characteristics including the rate of change of relative distance, the rate of change of relative direction, and the rate of change of relative speed. When the relative distance remains relatively stable or changes regularly, and the rate of change of relative direction is low, it is determined to be a synchronous movement pattern; when the relative distance first increases and then decreases, and there is a clear sequential order in the time points when the two people appear in similar positions, it is determined to be an alternating movement pattern.

[0119] For synchronized movement, it is necessary to extract the consistency feature of movement direction. Movement direction consistency is obtained by calculating the similarity of the direction vectors of the movement trajectories of two individuals within the same time window. Specifically, for each trajectory segment within a time window, the direction vector from the starting point to the ending point is extracted, and the cosine of the angle between the two direction vectors is calculated as the direction consistency index. This index ranges from -1 to 1; a value closer to 1 indicates a more consistent direction, closer to -1 indicates opposite directions, and closer to 0 indicates orthogonal directions. For the synchronized movement trajectory segment of maintenance personnel A and B, a sequence of direction consistency values ​​within 10 consecutive time windows is extracted: [0.95, 0.92, 0.96, 0.93, 0.91, 0.94, 0.97, 0.92, 0.95, 0.93], with an average value of 0.94. This indicates that the two individuals maintained a highly consistent movement direction during this period, possibly moving together along the utility tunnel.

[0120] For alternating movement patterns, spatial location continuity features are extracted. Spatial location continuity describes the degree to which one person leaves a certain location and another person subsequently arrives at a similar location. The location continuity index is obtained by calculating the spatial distance between the ending position of person A in the previous time window and the starting position of person B in the next time window. The smaller the distance, the stronger the continuity. The time interval between the two time windows is also considered to form a spatiotemporal continuity coefficient. For maintenance personnel C and D, who move alternately, the extracted spatial location continuity distances within five consecutive alternating intervals are [2.1 m, 1.8 m, 2.3 m, 1.9 m, 2.0 m], with an average continuity distance of 2.02 m. The time interval is 40 seconds, and the calculated spatiotemporal continuity coefficient is 0.85, indicating strong alternating operation characteristics, possibly indicating alternating operations at equipment maintenance points.

[0121] The interaction pattern is constructed by integrating the consistency of movement direction and the continuity of spatial location as the behavioral baseline feature. The integration process considers the typical proportions of synchronous and alternating movement modes in the activities of personnel in the utility tunnel, assigning appropriate weights. For purely synchronous movement, the behavioral baseline feature is mainly determined by the consistency of direction; for purely alternating movement, the behavioral baseline feature is mainly determined by the continuity of spatial location; for mixed modes, a weighted integration is performed based on the proportions of the two modes. In practical applications, the two members of a maintenance team typically exhibit 70% synchronous movement and 30% alternating movement. Therefore, their behavioral baseline feature is calculated as: 0.7 × 0.94 (consistency of direction) + 0.3 × 0.85 (continuity of location) = 0.913, indicating that the two individuals have a high degree of collaborative work characteristics.

[0122] The extracted behavioral baseline features are compared with the behavioral constraints in the zone-differentiated anomaly identification rules. The behavioral constraints define the normal range of personnel interaction behavior within different functional zones, including the upper and lower limits of directional consistency in synchronous movement mode and the upper and lower limits of spatial position continuity in alternating movement mode. For example, the behavioral constraints for the power compartment area specify that the normal range for directional consistency in synchronous movement is [0.7, 0.9], and the normal range for spatial position continuity is [0.6, 0.8]. When the directional consistency (0.94) in the behavioral baseline features exceeds the upper limit of 0.9, the calculated offset is 0.04; when the position continuity (0.85) exceeds the upper limit of 0.8, the calculated offset is 0.05.

[0123] The boundary parameters of the behavioral constraints are expanded and adjusted based on the offset. A gradual adjustment strategy is adopted to avoid overly relaxing the constraints due to short-term behavioral fluctuations. For cases where directional consistency exceeds the upper limit, the new upper limit is calculated as the original upper limit plus the product of the offset and the learning rate. The learning rate is typically set between 0.3 and 0.5, representing the degree of acceptance of newly observed behavioral patterns. For the aforementioned power compartment area, using a learning rate of 0.4, the new upper limit for directional consistency is calculated to be 0.9 + 0.04 × 0.4 = 0.916, and the new upper limit for position continuity is 0.8 + 0.05 × 0.4 = 0.82. The adjusted behavioral constraints better reflect the actual working patterns of the maintenance team, reducing false alarms.

[0124] To prevent behavioral constraints from expanding indefinitely, an adjustment cap and a decay mechanism are implemented. The adjustment cap ensures that behavioral constraints do not exceed reasonable limits; for example, adjustments for directional consistency do not exceed 0.95, and adjustments for positional continuity do not exceed 0.9. The decay mechanism gradually restores constraints corresponding to behavioral patterns that have not been observed for a long time to their default values, achieving dynamic contraction of constraints. The decay period is typically set to 7 days; if no behavior close to the boundary is observed each day, the constraint boundary shrinks by 5% towards the default value.

[0125] The dynamically updated anomaly detection rule constraints are applied to subsequent security status assessments, enabling adaptive optimization of the rules. The updated rules take into account the work habits and interaction characteristics of specific personnel groups, reducing false alarms caused by individual differences while retaining sensitivity to genuinely anomalous behavior.

[0126] This invention achieves adaptive optimization of utility tunnel safety supervision rules by extracting interaction patterns from personnel trajectories as behavioral baseline features. This method abandons the shortcomings of traditional fixed thresholds and introduces a dynamic learning mechanism, enabling anomaly identification rules to continuously adjust according to actual working patterns, thus improving the accuracy and adaptability of safety supervision. By distinguishing between synchronous and alternating movement modes, the method can comprehensively capture different types of personnel collaborative behaviors, adapting to diverse utility tunnel maintenance operation scenarios. The dynamic update mechanism of behavioral constraints considers both the adaptation to new observed behaviors and prevents excessive relaxation through upper limit settings and decay mechanisms, maintaining the rigor of safety supervision. This method endows the utility tunnel safety supervision system with "learning" capabilities, enabling it to gradually adapt to the working habits of different maintenance teams, reduce false alarm interference, improve the reliability of early warnings, and maintain sensitivity to abnormal behaviors, comprehensively enhancing the intelligent level of utility tunnel safety management.

[0127] A second aspect of the present invention provides a UWB-based personnel precise positioning and trajectory tracing safety management system for utility tunnels, the system comprising:

[0128] The first unit is used to collect the wireless signals emitted by the target personnel, and to perform positioning calculations based on the signal arrival time difference and the linear structural constraints of the utility tunnel to obtain the real-time spatial coordinates.

[0129] The second unit is used to extract the utility tunnel functional zoning attributes and environmental risk levels corresponding to the real-time spatial location coordinates as spatial semantic tags, embed them into the mobile trajectory data chain, construct a semantically enhanced trajectory representation structure, and store it in the trajectory database.

[0130] The third unit is used to establish zone-differentiated anomaly identification rules based on spatial semantic labels, set behavioral constraints that match the environmental risk level, and generate safety status assessment results by coupling location change features with spatial semantic labels.

[0131] The fourth unit is used to retrieve the historical trajectory of the target personnel from the trajectory database when the safety status assessment results indicate that there is a safety risk, and to retrieve the trajectory of related personnel in the same functional area who have a spatiotemporal relationship with the target personnel, and to calculate the trajectory similarity by comparing the movement path and the stop position.

[0132] The fifth unit is used to use trajectory similarity as a weighting coefficient to trigger collaborative warnings for associated personnel whose weighting coefficients exceed a preset threshold. At the same time, it extracts the interaction patterns between historical trajectories and the trajectories of associated personnel as behavioral baseline features to dynamically update the behavioral constraints of the anomaly identification rules.

[0133] A third aspect of the present invention provides an electronic device, comprising:

[0134] processor;

[0135] Memory used to store processor-executable instructions;

[0136] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0137] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0138] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0139] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for precise positioning and trajectory tracing of personnel in utility tunnels using UWB, characterized in that: include: The system collects wireless signals emitted by the target personnel, and performs positioning calculations based on the signal arrival time difference and the linear structural constraints of the utility tunnel to obtain real-time spatial coordinates. Extract the utility tunnel functional zoning attributes and environmental risk levels corresponding to the real-time spatial location coordinates as spatial semantic tags, embed them into the mobile trajectory data chain, construct a semantically enhanced trajectory representation structure, and store it in the trajectory database; Based on spatial semantic tags, zone-differentiated anomaly identification rules are established, behavioral constraints that match the environmental risk level are set, and safety status assessment results are generated by coupling location change features with spatial semantic tags. When the safety status assessment results indicate that there is a safety risk, the historical trajectory of the target personnel is retrieved from the trajectory database, and the trajectories of related personnel in the same functional area that are spatially and temporally associated with the target personnel are retrieved. The trajectory similarity is calculated by comparing the movement path and the stop position. Using trajectory similarity as a weighting coefficient, collaborative warnings are triggered for associated personnel whose weighting coefficients exceed a preset threshold. At the same time, the interaction patterns between historical trajectories and the trajectories of associated personnel are extracted as behavioral baseline features, which are used to dynamically update the behavioral constraints of the anomaly identification rules.

2. The method according to claim 1, characterized in that, The system collects wireless signals emitted by the target personnel and performs positioning calculations based on the signal arrival time difference and the linear structural constraints of the utility tunnel, obtaining real-time spatial coordinates including: By receiving wireless signals transmitted by the target personnel through multiple ultra-wideband base stations, calculating the signal arrival time difference between each ultra-wideband base station and the target personnel, and obtaining multiple sets of spatial distance measurement values ​​based on the signal arrival time difference; The central axis and physical boundary of the utility tunnel are extracted to construct an accessibility constraint space. Multiple sets of spatial distance measurements are projected and mapped within the accessibility constraint space. By constraining the intersection of the distance measurement circles to the neighborhood of the central axis, the distance measurement results that deviate from the direction of the utility tunnel are corrected to the corresponding positions on the central axis. The vertical offset distance between the corrected position coordinates and the central axis of the utility tunnel is calculated. When the vertical offset distance exceeds the allowable range of the physical boundary of the utility tunnel, the position coordinates are moved back to the inside of the physical boundary in the vertical direction to obtain real-time spatial position coordinates that conform to the spatial topology characteristics of the utility tunnel.

3. The method according to claim 1, characterized in that, Extracting the functional zoning attributes and environmental risk levels of the utility tunnel corresponding to real-time spatial location coordinates as spatial semantic tags, embedding them into the movement trajectory data chain, constructing a semantically enhanced trajectory representation structure, and storing it in the trajectory database includes: The functional zones of the utility tunnels are determined based on real-time spatial coordinates, and the attributes of the functional zones and environmental risk levels are extracted and combined into spatial semantic tags. Spatial semantic labels are bound to real-time spatial location coordinates and timestamps to form trajectory data units, and multiple trajectory data units are connected in sequence according to timestamp order to construct a mobile trajectory data chain. Extract continuous trajectory data unit sequences with the same spatial semantic labels from the mobile trajectory data chain, calculate the dispersion of the time span and spatial location coordinates of the continuous trajectory data unit sequences, and embed the product of the time span and dispersion as the partition dwell feature into the spatial semantic label. For the location where the spatial semantic label is switched in the mobile trajectory data chain, the difference in the partition dwell feature before and after the switch is extracted, and the difference in the partition dwell feature is used as the partition transition strength marker to the spatial semantic label switching location; The mobile trajectory data chain, which includes partition dwell features and partition transition intensity, is constructed into a semantically enhanced trajectory representation structure and stored in the trajectory database.

4. The method according to claim 1, characterized in that, Based on spatial semantic tags, zone-differentiated anomaly identification rules are established, and behavioral constraints matching environmental risk levels are set. Safety status assessment results are generated by coupling location change features with spatial semantic tags, including: Based on the functional zoning attributes of the utility tunnel in the spatial semantic tags, establish zoning-differentiated anomaly identification rules, and set behavioral constraints according to the environmental risk level in the spatial semantic tags. Extract the real-time spatial location coordinate change sequence of the target person from the semantically enhanced trajectory representation structure, and calculate the position change features of the real-time spatial location coordinate change sequence; Extract the historical location change characteristics of target personnel in the mobile trajectory data chain under the same utility tunnel functional zoning attributes but different environmental risk levels, construct the inverse constraint mapping relationship between environmental risk level and location change characteristic tolerance threshold, and convert the current environmental risk level into a dynamic tolerance threshold for location change characteristics based on the inverse constraint mapping relationship. The location change features are compared with the dynamic tolerance threshold. When the location change features exceed the dynamic tolerance threshold, the anomaly identification rule is triggered. The extent of the location change features exceeding the dynamic tolerance threshold is extracted, and the extent of the exceedance is normalized with the dynamic tolerance threshold to generate a safety status assessment result.

5. The method according to claim 1, characterized in that, When the safety status assessment indicates a safety risk, the historical trajectory of the target personnel is retrieved from the trajectory database, and the trajectories of related personnel within the same functional area who have spatiotemporal associations with the target personnel are searched. The trajectory similarity is calculated by comparing the movement path and the stopping position, including: When the safety status assessment results indicate that there is a safety risk, the historical trajectory of the target personnel is retrieved from the trajectory database. Retrieve the trajectories of related personnel within the same functional area from the trajectory database, and determine whether there is a spatiotemporal correlation between the trajectories of related personnel and the historical trajectories of the target personnel; Extract the spatial semantic tag sequences contained in the historical trajectory of the target personnel and the trajectory of related personnel, and use the environmental risk level in the spatial semantic tag sequences as a weighting factor to construct a risk-weighted movement path; By comparing risk-weighted movement paths, when the historical trajectory of the target person overlaps with the risk-weighted movement path of the associated person, the spatial semantic labels corresponding to the overlapping area are extracted. The environmental risk level of the overlapping area is multiplied by the spatial range of the overlapping area to generate the risk cross intensity. The risk cross intensity is used as a correction term for trajectory similarity to calculate trajectory similarity.

6. The method according to claim 1, characterized in that, The interaction patterns between historical trajectories and the trajectories of associated individuals are extracted as behavioral baseline features. These features are used to dynamically update the behavioral constraints of anomaly identification rules. Extract temporal correlation features of spatial location coordinate changes from the historical trajectory of the target personnel and the trajectory of related personnel, and decompose the temporal correlation features into synchronous movement patterns and alternating movement patterns; For synchronous movement mode, the consistency of movement direction between target personnel and related personnel within the same time window is extracted; for alternating movement mode, the continuity of spatial position between target personnel and related personnel within adjacent time windows is extracted. The consistency of movement direction and the continuity of spatial position are integrated to construct the interaction pattern as the behavioral baseline feature. The behavior baseline features are compared with the behavior constraints in the partitioned differential anomaly identification rules. When the consistency of movement direction or the continuity of spatial position in the behavior baseline features exceeds the preset range of the behavior constraints, the offset exceeding the preset range is extracted. The boundary parameters of the behavior constraints are expanded and adjusted according to the offset, and the behavior constraints of the anomaly identification rules are dynamically updated.

7. A UWB-based personnel precise positioning and trajectory tracking safety control system for utility tunnels, used to implement the method described in any one of claims 1-6, characterized in that, include: The first unit is used to collect the wireless signals emitted by the target personnel, and to perform positioning calculations based on the signal arrival time difference and the linear structural constraints of the utility tunnel to obtain the real-time spatial coordinates. The second unit is used to extract the utility tunnel functional zoning attributes and environmental risk levels corresponding to the real-time spatial location coordinates as spatial semantic tags, embed them into the mobile trajectory data chain, construct a semantically enhanced trajectory representation structure, and store it in the trajectory database. The third unit is used to establish zone-differentiated anomaly identification rules based on spatial semantic labels, set behavioral constraints that match the environmental risk level, and generate safety status assessment results by coupling location change features with spatial semantic labels. The fourth unit is used to retrieve the historical trajectory of the target personnel from the trajectory database when the safety status assessment results indicate that there is a safety risk, and to retrieve the trajectory of related personnel in the same functional area who have a spatiotemporal relationship with the target personnel, and to calculate the trajectory similarity by comparing the movement path and the stop position. The fifth unit is used to use trajectory similarity as a weighting coefficient to trigger collaborative warnings for associated personnel whose weighting coefficients exceed a preset threshold. At the same time, it extracts the interaction patterns between historical trajectories and the trajectories of associated personnel as behavioral baseline features to dynamically update the behavioral constraints of the anomaly identification rules.

8. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 6.

Images