Booster station barrier-free channel digital twin evacuation early warning method

Abstract

The invention discloses a booster station barrier-free channel digital twinborn evacuation early warning method. The problem of path planning misalignment caused by insufficient multi-source sensing data fusion is effectively solved. Through spatial data integration of the digital twin model, the available state of the barrier-free facility and the disaster influence range can be accurately identified. A dynamic path planning algorithm continuously optimizes an evacuation route in the disaster evolution process, and people are prevented from entering an affected dangerous area. The multi-modal instruction generation mechanism ensures that people with different physiological characteristics can obtain adaptive evacuation guidance, the retention risk caused by a single information transmission mode is reduced, the response speed and reliability of the emergency evacuation system are further improved through the real-time state monitoring and path updating functions, the geometric dimension and space coordinates of facilities are accurately collected, and the emergency evacuation system is more accurate in real-time state monitoring and path updating. The consistency of a virtual model and an actual environment is ensured, and data support is provided for facility availability evaluation in a disaster environment by configuring material characteristics and accessibility parameters.

Description

Technical Field

[0001] This application relates to the field of early warning and evacuation technology, and in particular to a digital twin evacuation early warning method for barrier-free access in a booster station. Background Technology

[0002] In the field of power system safety and emergency management, with the improvement of digitalization and intelligence, the means of ensuring the operational safety of critical infrastructure such as substations and step-up substations are gradually shifting from manual inspection to information-based monitoring. Existing technologies generally employ multimodal sensing devices such as building information modeling, video surveillance, environmental sensors, and infrared detectors to monitor personnel distribution, environmental parameters, and passageway status in real time. In the context of step-up substations, these technologies are mainly used to identify whether there are obstacles in the passageways, detect smoke and temperature anomalies, and locate the positions of personnel on site, thereby providing basic information support for emergency management and evacuation decisions.

[0003] However, in practical applications, existing multimodal sensing technologies generally suffer from data fragmentation and information redundancy. Different types of sensors often lack uniformity in terms of acquisition frequency, spatial reference, and data semantics. This leads to situations where, although the system can detect "foreign objects in the passage" or "abnormal temperature rise," it struggles to accurately distinguish whether the situation is "personnel passing by" or "burning debris." Furthermore, the lack of effective fusion and filtering mechanisms for multi-source information means it is often presented directly to the command center in its raw form, exposing maintenance personnel to information overload in disaster scenarios and making it difficult to promptly extract key contradictions. This can potentially cause delays in evacuation warnings or inaccurate route planning. Therefore, it is evident that existing technologies still lack sufficient ability to fuse and intelligently understand multimodal data in complex environments, limiting their effectiveness in unimpeded evacuation warnings at substations. Summary of the Invention

[0004] The main objective of this application is to provide a digital twin evacuation early warning method for barrier-free access in substations to solve the problems mentioned in the background.

[0005] To achieve the above objectives, this application provides the following technical solution: A digital twin evacuation early warning method for barrier-free access in substations, comprising the following specific steps: S1. Establish a digital twin model of the substation, collect building information, construct a digital twin model including barrier-free access and barrier-free facilities, and set geometric attributes, spatial location relationships and accessibility parameters for the facilities; S2. Collect sensor data deployed in and around the barrier-free facilities. The sensor data includes video surveillance, infrared detection, pressure detection and door magnetic detection data. Integrate the collected data into the digital twin model. S3. Perform time synchronization, spatial coordinate transformation and data fusion on multi-source sensor data, determine the availability status of barrier-free facilities based on the fusion results, and update the availability status to the digital twin model; S4. Receive disaster data, including fire alarm, smoke diffusion and abnormal temperature and humidity data. Overlay the disaster data onto the digital twin model to generate a dangerous area map and identify the risk range. S5. Based on the availability of accessibility facilities and the mapping of dangerous areas, use path planning algorithms to generate evacuation routes that meet accessibility constraints and avoid dangerous areas, and update the evacuation routes when the disaster situation changes. S6. Output evacuation instructions based on personnel characteristic information. Evacuation instructions include vibration prompts, light strips or image prompts, voice broadcasts and LED ground indicators to guide different personnel to evacuate. S7. During the evacuation process, use positioning devices and behavior recognition devices to monitor the evacuation status of personnel, determine whether there is congestion, deviation or stagnation. If there is an abnormality, repeat steps S5 and S6 to update the evacuation route and evacuation instructions, and record the information of personnel arriving at the safe area after the evacuation is completed.

[0006] Preferably, step S1 is performed in the following manner: S1.1 Collect data on the building structure, passage layout and accessibility facilities of the substation. The data includes geometric dimensions, spatial coordinates, hierarchical relationships and material properties, which are used to generate the basic dataset required for digital twin modeling. S1.2. Based on the data collected step by step, construct a three-dimensional digital twin model containing barrier-free access and barrier-free facilities, and configure geometric attributes, spatial location relationships and accessibility parameters for each barrier-free facility to form a virtual model corresponding to the actual substation environment.

[0007] Preferably, step S2 is performed as follows: S2.1. Deploy video surveillance devices, infrared detectors, pressure sensors and door magnetic sensors in and around the accessible facilities to collect real-time data on the occupancy status, structural status and surrounding environmental parameters of the accessible facilities and generate sensor data. S2.2 The collected sensor data is connected to the digital twin platform via the data transmission interface, the sensor data is formatted, and the processed data is mapped to the corresponding facility nodes in the digital twin model.

[0008] Preferably, step S3 is performed as follows: S3.1 Perform time synchronization processing and spatial coordinate transformation processing on the multi-source sensor data accessed in step S2, unify different types of data to the same time reference and the same spatial coordinate system, and generate a standardized dataset; S3.2. The standardized dataset obtained in step S3.1 is fused. The fusion process includes target recognition and contour matching of video surveillance data and infrared detection data to determine whether there are moving obstacles in the accessible passage area; threshold comparison of pressure detection data and door magnetic detection data to determine whether the passage is occupied and whether the relevant facilities are in the open state; on this basis, the judgment results of different types of sensor data are cross-validated to generate the occupancy status, integrity status and accessibility status of the accessible facilities, and the status results are updated to the corresponding facility nodes in the digital twin model.

[0009] Preferably, step S4 is performed as follows: S4.1 Receive fire alarm data, smoke diffusion data, and abnormal temperature and humidity data; perform format conversion and parameter correction on data from different sources; and generate a disaster dataset in a unified format. S4.2. Overlay the disaster dataset obtained in step S4.1 onto the digital twin model, perform disaster correlation calculations on the model's spatial regions. The disaster correlation calculations include determining the fire source location based on fire alarm data, simulating the diffusion trend based on smoke diffusion data, determining environmental thresholds based on abnormal temperature and humidity data, and spatially overlaying the calculation results to generate a hazardous area mapping. In the hazardous area mapping, the risk range is divided and marked according to the hazard level, and the division results are updated to the digital twin model.

[0010] Preferably, step S5 is performed as follows: S5.1 Based on the accessibility facility availability status obtained in step S3 and the hazardous area mapping obtained in step S4, the access network of the digital twin model is modeled, and the accessibility path nodes and hazardous area boundaries are set as constraints. Path planning algorithms are used to search for paths, including weighted graph shortest path search, heuristic search and Dijkstra's algorithm. The accessibility, path length and obstacle avoidance conditions of candidate paths are comprehensively calculated to generate an initial evacuation path that meets the accessibility constraints and avoids the hazardous area. S5.2 When disaster data changes, based on the updated availability of accessibility facilities and the mapping of dangerous areas, the effectiveness of existing evacuation routes is checked. The check includes determining whether path nodes fall into dangerous areas and whether path edges meet accessibility conditions. When a part of the path fails, the path planning algorithm is called to locally replan the failed path segment. When the overall path is not feasible, the global path is recalculated and the updated evacuation route is written into the digital twin model.

[0011] Preferably, in step S6.1, personnel feature information is collected based on a positioning device, identity recognition system, or sensor network. The feature information includes the personnel's real-time location, movement speed, auditory or visual sensitivity, and dependence on assistive devices. The collected data is classified and labeled, and personnel are grouped according to their mobility and sensory characteristics. The evacuation needs of various types of personnel are analyzed in conjunction with the evacuation paths in the digital twin model to form differentiated evacuation instruction generation conditions. S6.2 Based on the generation conditions formed in step S6.1, the instruction mapping module is invoked to match different groups of people with corresponding prompting methods. Vibration prompts are generated for people with mobility impairments, voice broadcasts are generated for visually impaired people, light strips or image prompts are generated for hearing impaired people, and LED ground indicators are generated for ordinary people. The generated evacuation instructions are sent to the corresponding prompting devices through the network communication interface, and the evacuation instructions are updated in real time according to changes in the location of people or changes in the environmental conditions during the evacuation process.

[0012] Preferably, step S7 is performed as follows: S7.1. The real-time location of personnel is collected through the positioning device, and the movement speed, movement direction and behavior pattern of personnel are obtained by combining the inertial measurement unit, video monitoring and behavior recognition algorithm. The trajectory analysis and group density calculation are used to determine whether congestion occurs, the path deviation detection algorithm is used to determine whether personnel deviate from the predetermined evacuation route, and the dwell time threshold is used to determine whether personnel are stranded, so as to realize real-time monitoring and anomaly detection of evacuation status. S7.2 When an abnormal situation is detected, different processing strategies are triggered based on the type of abnormality. When local congestion or path blockage occurs, the path planning module is called to dynamically replan the local path. When large-scale congestion or danger spread occurs, global path planning is executed. Simultaneously, the evacuation instruction generation module is called to issue updated evacuation instructions to the affected personnel and to provide prompts through a multimodal prompting device. After the personnel reach the safe area, the arrival time and location coordinates are recorded using a positioning device and the data is stored in the evacuation database to form an evacuation process record. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating the steps of the method described in this application. Detailed Implementation

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0015] The terms "first," "second," and "third" in this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this application are only used to explain the relative positional relationships and movements between components in a specific orientation (as shown in the figures). If the specific orientation changes, the directional indications also change accordingly. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0016] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0017] Example 1: Please refer to Figure 1 A digital twin evacuation early warning method for barrier-free access in substations, the specific steps of which are as follows: S1. Establish a digital twin model of the substation, collect building information, construct a digital twin model including barrier-free access and barrier-free facilities, and set geometric attributes, spatial location relationships and accessibility parameters for the facilities; S2. Collect sensor data deployed in and around the barrier-free facilities. The sensor data includes video surveillance, infrared detection, pressure detection and door magnetic detection data. Integrate the collected data into the digital twin model. S3. Perform time synchronization, spatial coordinate transformation and data fusion on multi-source sensor data, determine the availability status of barrier-free facilities based on the fusion results, and update the availability status to the digital twin model; S4. Receive disaster data, including fire alarm, smoke diffusion and abnormal temperature and humidity data. Overlay the disaster data onto the digital twin model to generate a dangerous area map and identify the risk range. S5. Based on the availability of accessibility facilities and the mapping of dangerous areas, use path planning algorithms to generate evacuation routes that meet accessibility constraints and avoid dangerous areas, and update the evacuation routes when the disaster situation changes. S6. Output evacuation instructions based on personnel characteristic information. Evacuation instructions include vibration prompts, light strips or image prompts, voice broadcasts and LED ground indicators to guide different personnel to evacuate. S7. During the evacuation process, use positioning devices and behavior recognition devices to monitor the evacuation status of personnel, determine whether there is congestion, deviation or stagnation. If there is an abnormality, repeat steps S5 and S6 to update the evacuation route and evacuation instructions, and record the information of personnel arriving at the safe area after the evacuation is completed.

[0018] In this embodiment: In the prior art, emergency evacuation of critical power system facilities mainly relies on multimodal sensing devices for environmental monitoring and personnel location. However, sensor data lacks uniformity in terms of time reference, spatial coordinates, and semantics, making it difficult for the system to accurately distinguish the types of obstacles in passageways and the evolution trend of disasters. For example, when a pressure sensor detects passageway occupancy, it cannot quickly determine whether it is a temporary occupant or a fixed obstacle; when abnormal temperature and humidity and smoke alarms are triggered simultaneously, it is difficult to assess the actual range of hazard spread. This data fragmentation problem exposes emergency decision-making to the risk of information overload, potentially delaying the timely adjustment of evacuation routes.

[0019] To address these issues, researchers recognized the need to construct a unified data processing framework that deeply correlates multi-source sensor information with building spatial characteristics. Analysis revealed that existing technologies failed to spatially overlay the status of accessibility facilities with dynamic changes in disaster conditions, causing route planning to overlook changes in passage availability. Further consideration was given to using digital twin models as a data fusion carrier, eliminating differences between multi-source data through time synchronization and coordinate transformation, and achieving dynamic optimization of evacuation routes based on real-time status updates.

[0020] This application proposes the following technical solutions: establishing a digital twin model of the substation and collecting building information; deploying sensors to collect data on barrier-free facilities and the surrounding environment; performing time synchronization, spatial conversion, and fusion processing on multi-source data to determine the facility status; overlaying disaster data to generate a hazardous area mapping; generating barrier-free evacuation routes based on facility status and hazardous areas; outputting multimodal evacuation instructions based on personnel characteristics; and monitoring the evacuation status in real time and dynamically updating the routes.

[0021] Digital twin models refer to virtual environments constructed using 3D modeling technology. Specifically, this can be achieved by fusing BIM with point cloud scanning data, accurately reflecting the geometric attributes and spatial topology of accessible pathways. Multi-source sensor data fusion involves spatiotemporal alignment of outputs from multiple sensor types, such as video, infrared, and pressure sensors. This can be achieved using timestamp synchronization and coordinate system transformation algorithms to eliminate data fragmentation and improve the accuracy of status assessments. Hazard area mapping involves spatially overlaying disaster data, such as fire and smoke diffusion data, with 3D models. This can be achieved using heatmap rendering and regional diffusion simulation algorithms to dynamically identify risk areas. Path planning algorithms are path search methods based on accessibility constraints. Specifically, this can be achieved using the A* algorithm combined with accessibility network weight adjustment to generate evacuation routes that avoid hazardous areas while meeting accessibility requirements. Multimodal evacuation instructions refer to prompts adapted to different people's perceptual abilities. This can be achieved using a combination of vibration devices, voice broadcasts, and LED indicators to cover the differentiated needs of people with mobility impairments, visual impairments, and hearing impairments.

[0022] First, a digital twin model incorporating accessibility facilities is constructed using 3D scanning and BIM modeling to ensure spatial consistency between the virtual environment and the actual building. A sensor network is deployed at key locations to collect real-time data on passageway occupancy and facility operation. Asynchronous data such as video frames and pressure detection are aligned using a time synchronization module, and coordinate transformation maps the sensor data to a unified 3D space. Based on the fused data, obstacle types are identified, for example, pressure sensors continuously detect and determine if an obstacle is fixed, while infrared detection eliminates false alarms. When a fire alarm is received, the smoke diffusion path is simulated in the digital twin model, and the topological relationship between areas with abnormal temperature and humidity and the accessibility passage is overlaid and analyzed. The path planning module calculates the shortest safe path using a heuristic search algorithm based on real-time updated obstacle distribution and hazardous areas. For different user characteristics, such as prioritizing ramp paths for wheelchair users and providing directional prompts via vibration devices, and providing voice navigation for visually impaired individuals, the system monitors personnel movement trajectories during evacuation. When path congestion is detected, detour routes are recalculated and guidance information is updated.

[0023] This solution achieves spatial correlation of multi-source data through a digital twin model, solving the problem of fragmented sensor data. Traditional methods struggle to accurately assess the impact of passageway obstructions on evacuation when processing various alarm information individually; however, this solution, by fusing pressure detection and video analysis, can distinguish between temporary personnel and fixed obstacles. Existing route planning is typically based on static maps, while this solution combines real-time disaster spread simulation to dynamically adjust the boundaries of danger zones, ensuring that evacuation routes always avoid the latest risk areas. Furthermore, while traditional evacuation alerts use a unified broadcast method, this solution matches multimodal instructions based on personnel's mobility and perception characteristics, significantly improving the evacuation efficiency for vulnerable groups.

[0024] This application effectively solves the problem of inaccurate route planning caused by insufficient fusion of multi-source sensing data. Through spatial data integration using a digital twin model, the availability of accessibility facilities and the scope of disaster impact can be accurately identified. The dynamic route planning algorithm continuously optimizes evacuation routes as the disaster evolves, preventing people from entering already affected danger zones. A multimodal instruction generation mechanism ensures that people with different physiological characteristics receive appropriate evacuation guidance, reducing the risk of being stranded due to limited information delivery methods. Real-time status monitoring and route update functions further improve the response speed and reliability of the emergency evacuation system.

[0025] Example 2: Please refer to Figure 1 The specific method for step S1 is as follows: S1.1 Collect data on the building structure, passage layout and accessibility facilities of the substation. The data includes geometric dimensions, spatial coordinates, hierarchical relationships and material properties, which are used to generate the basic dataset required for digital twin modeling. S1.2. Based on the data collected step by step, construct a three-dimensional digital twin model containing barrier-free access and barrier-free facilities, and configure geometric attributes, spatial location relationships and accessibility parameters for each barrier-free facility to form a virtual model corresponding to the actual substation environment.

[0026] In this embodiment: This application further proposes a specific implementation method for establishing a digital twin model of a substation, including data collection on the building structure, passage layout, and accessibility facilities of the substation. The data includes geometric dimensions, spatial coordinates, hierarchical relationships, and material properties, generating a basic dataset required for digital twin modeling. Based on the collected data, a three-dimensional digital twin model containing accessibility passages and facilities is constructed, and geometric attributes, spatial positional relationships, and accessibility parameters are configured for each accessibility facility, forming a virtual model corresponding to the actual substation environment.

[0027] Geometric dimensions refer to physical parameters such as the width of accessible pathways and the inclination angle of ramps, obtained through laser scanning or 3D imaging technology. These can be achieved using 3D laser scanners or photogrammetry equipment to accurately recreate the actual spatial form of the facility. Spatial coordinates refer to the facility's location information established based on the Global Positioning System (GPS) or indoor positioning technology, such as through BeiDou positioning modules or UWB base stations, used to determine the facility's relative position in the model. Hierarchical relationships refer to the connection logic between pathways and facilities established through topological analysis. For example, graph theory algorithms are used to network-model stairs, ramps, and platforms to describe the passage sequence between facilities. Material properties refer to attributes such as fire resistance and load-bearing capacity of the facility, obtained through material sensors or database queries. For example, infrared spectroscopy is used to detect material composition to assess the facility's reliability in disaster environments. A 3D digital twin model refers to a virtual environment built using BIM software or game engines, such as Revit or Unity platforms, used to dynamically map the real-time state of the physical space. Geometric attributes refer to the mathematical descriptions of the shape and size of facilities in the model. For example, defining ramp curvature using NURBS surfaces or voxel meshes supports spatial calculations for path planning algorithms. Spatial location relationships refer to the orientation and distance parameters between facilities. For example, determining the connection points between elevators and corridors using coordinate system transformation matrices verifies the connectivity of evacuation routes. Accessibility parameters are indices that quantify the passageway's traversability. For example, defining wheelchair accessibility conditions using width and slope thresholds constrains the rules for generating evacuation routes.

[0028] First, a 3D laser scanner was used to collect point cloud data of the accessible pathways in the substation, measuring parameters such as ramp length and handrail height with millimeter-level precision. Simultaneously, indoor positioning devices were used to record the latitude and longitude coordinates of each facility. The collected geometric dimensions and hierarchical relationships were then imported into BIM modeling software, and parametric modeling tools were used to generate a 3D model including ramps, tactile paving, and elevators. During model construction, attributes such as material fire resistance rating and load-bearing capacity were added to each facility node; for example, fire door material was correlated with fire resistance time. Furthermore, topological connections between pathways and rooms were established based on spatial coordinates. Accessibility parameters, such as setting ramp inclination angles not exceeding 8% and pathway widths not less than 1.2 meters, were further optimized to ensure the model accurately reflects the accessibility needs of people with disabilities in the actual environment. The final virtual model maintains geometric consistency and attribute correlation with the actual substation, providing a precise spatial reference for subsequent sensor data mapping and path planning.

[0029] Traditional methods typically only collect two-dimensional planar layout or local structural data, resulting in models that cannot accurately reflect the three-dimensional spatial relationships of accessible facilities. For example, existing technologies may import floor plans from CAD drawings, but lack key geometric parameters such as ramp inclination and handrail height, making it impossible to determine the actual feasibility of wheelchair access during path planning. This solution, however, integrates multi-dimensional data such as geometric dimensions and material properties to construct a three-dimensional model with physical attributes and spatial topology. This allows evacuation route calculations to simultaneously consider structural safety and accessibility, avoiding path planning errors caused by model simplification.

[0030] This application addresses the problem of low reliability of evacuation routes due to insufficient model accuracy in existing technologies. By accurately collecting facility geometry and spatial coordinates, it ensures consistency between the virtual model and the actual environment; by configuring material properties and accessibility parameters, it provides data support for facility availability assessment in disaster environments; and by establishing hierarchical relationships and topological connections, it eliminates the risk of path connectivity misjudgment present in traditional two-dimensional models. The resulting digital twin model provides a high-precision spatial reference for subsequent sensor data fusion, hazardous area mapping, and dynamic path planning, thereby improving the overall reliability of the barrier-free evacuation early warning system.

[0031] Example 3: Please refer to Figure 1 The specific method for step S2 is as follows: S2.1. Deploy video surveillance devices, infrared detectors, pressure sensors and door magnetic sensors in and around the accessible facilities to collect real-time data on the occupancy status, structural status and surrounding environmental parameters of the accessible facilities and generate sensor data. S2.2 The collected sensor data is connected to the digital twin platform via the data transmission interface, the sensor data is formatted, and the processed data is mapped to the corresponding facility nodes in the digital twin model.

[0032] In this embodiment: This application further proposes to deploy video surveillance devices, infrared detectors, pressure sensors and door magnetic sensors in and around the barrier-free facility to collect real-time data on the occupancy status, structural status and surrounding environmental parameters of the barrier-free facility, and generate sensor data; the collected sensor data is connected to the digital twin platform through the data transmission interface, the sensor data is formatted, and the processed data is mapped to the corresponding facility node in the digital twin model.

[0033] Video surveillance devices, specifically network cameras with night vision capabilities, capture real-time image data of accessible pathway areas to identify moving obstacles or unusual objects. Infrared detectors detect the presence of objects by receiving thermal radiation signals; passive infrared sensors are used to assist in determining pathway occupancy in low-light conditions. Pressure sensors detect pressure changes to determine occupancy; piezoelectric sensors, for example, can be installed on the floor or surface of accessible pathways to monitor wheelchair or pedestrian positions. Door magnetic sensors detect the opening / closing status of facilities using magnetic induction; magnetic switch sensors, for example, can be installed on accessible pathway doors to determine whether the door is open or closed. Data transmission interfaces are communication modules used to upload sensor data to the digital twin platform; Ethernet, RS-485, or wireless transmission protocols can be used to ensure real-time data transmission. Formatting converts sensor data from different protocols or structures into a unified format; data parsing and standardized coding rules are used to eliminate data heterogeneity and improve model compatibility.

[0034] By deploying multiple sensors in and around accessible pathways, facility status data can be acquired in real time from different dimensions. For example, pressure sensors and door magnetic sensors work together to distinguish whether a pathway is blocked due to normal pedestrian traffic or equipment malfunction; the fusion of video surveillance and infrared detection data can accurately identify obstacle types under complex lighting conditions. The collected raw data is transmitted to the digital twin platform via a communication interface, undergoes format conversion and standardization, and is mapped to the corresponding 3D facility nodes in the model, forming a data layer that is synchronously updated with the physical environment. Thus, the model can reflect pathway occupancy, facility opening / closing status, and changes in the surrounding environment in real time, providing a complete data foundation for subsequent status determination.

[0035] Traditional methods typically rely on single sensors for status monitoring, such as using video surveillance to determine if a passage is clear. This is prone to misjudgments due to changes in lighting or obstructed viewing angles. This application, however, employs a combination of multiple sensor types and data formatting processing to eliminate the limitations of a single data source. For example, when video surveillance fails due to smoke interference, infrared and pressure sensors can still provide redundant data, preventing misjudgments caused by a single sensor malfunction. Furthermore, in existing technologies, sensor data is often stored independently, lacking dynamic correlation with the 3D model. This application, by mapping data to model nodes, achieves synchronous updates of spatial attributes and status parameters.

[0036] This application addresses the integration difficulties caused by the heterogeneity of multi-source sensor data in existing technologies, improving the accuracy and reliability of accessibility facility status monitoring. For example, cross-validation of data from pressure sensors and door magnetic sensors effectively distinguishes between temporary occupancy and long-term blockage of passageways; formatting eliminates data protocol differences, ensuring that multi-source data can be efficiently parsed by the model. This provides high-precision, low-latency data support for subsequent facility status determination and evacuation route planning, avoiding early warning delays or route planning errors caused by data fragmentation.

[0037] Example 4: Please refer to Figure 1 The specific method for step S3 is as follows: S3.1 Perform time synchronization processing and spatial coordinate transformation processing on the multi-source sensor data accessed in step S2, unify different types of data to the same time reference and the same spatial coordinate system, and generate a standardized dataset; S3.2. The standardized dataset obtained in step S3.1 is fused. The fusion process includes target recognition and contour matching of video surveillance data and infrared detection data to determine whether there are moving obstacles in the accessible passage area; threshold comparison of pressure detection data and door magnetic detection data to determine whether the passage is occupied and whether the relevant facilities are in the open state; on this basis, the judgment results of different types of sensor data are cross-validated to generate the occupancy status, integrity status and accessibility status of accessible facilities, and the status results are updated to the corresponding facility nodes in the digital twin model.

[0038] In this embodiment, the specific method of step S3 is as follows: Time synchronization and spatial coordinate transformation are performed on the multi-source sensor data accessed in step S2 to unify different types of data to the same time reference and the same spatial coordinate system, generating a standardized dataset; the standardized dataset is fused, including target recognition and contour matching of video surveillance data and infrared detection data to determine whether there are moving obstacles in the accessible passage area; pressure detection data and door magnetic detection data are compared with thresholds to determine whether the passage is occupied and whether the relevant facilities are in the open state; based on this, the judgment results of different types of sensor data are cross-validated to generate the occupancy status, integrity status, and accessibility status of the accessible facilities, and the status results are updated to the corresponding facility nodes in the digital twin model.

[0039] Time synchronization processing refers to aligning data collected by different sensors according to a unified timestamp. This can be achieved using network time protocols or hardware clock synchronization modules to eliminate data timing misalignment caused by differences in acquisition time. Spatial coordinate transformation processing refers to converting spatial data collected by different sensors to the same coordinate system. This can be achieved using coordinate transformation matrices or spatial registration algorithms to solve the problem of inconsistent spatial references caused by differences in sensor installation locations. Data fusion processing refers to integrating multi-source data into a unified judgment result through algorithms. Specifically, target recognition algorithms can be used to match video and infrared data, combined with threshold comparison methods to logically verify pressure and door magnetic data, thereby improving the accuracy of status determination.

[0040] Multi-source sensor data first undergoes time synchronization and spatial coordinate transformation to eliminate temporal deviations and spatial reference differences, forming a standardized dataset with consistent spatiotemporal attributes. Subsequently, video surveillance data and infrared detection data are processed using target recognition algorithms to extract the contour features of moving objects and perform matching verification to determine the presence of temporary obstacles in the accessible passageway. Pressure detection data and door / magnetic sensor data are used to determine passage occupancy and facility open / closed status using preset thresholds. Finally, the judgment results from different data sources are cross-validated. For example, if infrared data detects a moving object but pressure data does not indicate passage occupancy, false positives for fixed obstacles can be ruled out. This accurately generates the occupancy, integrity, and accessibility status of the accessible facilities, and these statuses are updated in real-time to the digital twin model.

[0041] In existing technologies, multimodal sensor data lacks a unified spatiotemporal reference, making it difficult for systems to distinguish between real obstacles and temporary interference. For example, it cannot effectively determine whether a passageway is blocked by pedestrians or debris. This solution establishes data consistency through time synchronization and spatial coordinate transformation. Combined with a multi-level fusion mechanism of target matching and threshold comparison, it can accurately identify the differences between moving and fixed obstacles and eliminate the risk of misjudgment from a single sensor through cross-validation.

[0042] This application solves the problem of misjudgment caused by inconsistencies in spatiotemporal references in multi-source sensor data, improves the accuracy of accessibility facility status determination, and avoids evacuation route planning errors caused by data redundancy or conflict, thereby providing reliable status information support for personnel evacuation in disaster scenarios.

[0043] Example 5: Please refer to Figure 1 The specific method for step S4 is as follows: S4.1 Receive fire alarm data, smoke diffusion data, and abnormal temperature and humidity data; perform format conversion and parameter correction on data from different sources; and generate a disaster dataset in a unified format. S4.2. Overlay the disaster dataset obtained in step S4.1 onto the digital twin model, perform disaster correlation calculations on the model's spatial regions. The disaster correlation calculations include determining the fire source location based on fire alarm data, simulating the diffusion trend based on smoke diffusion data, determining environmental thresholds based on abnormal temperature and humidity data, and spatially overlaying the calculation results to generate a hazardous area mapping. In the hazardous area mapping, the risk range is divided and marked according to the hazard level, and the division results are updated to the digital twin model.

[0044] In this embodiment: This application further proposes receiving fire alarm data, smoke diffusion data, and abnormal temperature and humidity data, performing format conversion and parameter correction on data from different sources to generate a disaster dataset in a unified format; superimposing the disaster dataset onto a digital twin model, performing disaster correlation calculations on the model's spatial regions, including determining the fire source location based on fire alarm data, simulating diffusion trends based on smoke diffusion data, determining environmental thresholds based on abnormal temperature and humidity data, and spatially superimposing the calculation results to generate a hazardous area mapping, dividing and marking the risk range according to the hazard level in the hazardous area mapping, and updating the division results to the digital twin model.

[0045] The risk index is calculated as follows:

[0046] Where T is the abnormal temperature value, S is the abnormal smoke concentration value, H is the abnormal humidity value, and a1, a2 and a3 are weight values, and the values ​​of a1, a2 and a3 are adjusted and set by the user. Format conversion refers to converting data from different communication protocols or data structures into a unified data format. This can be achieved using standardized templates like JSON or XML, and data cleaning and field mapping can eliminate compatibility issues related to heterogeneous data. Parameter calibration refers to correcting errors in raw data collected by sensors. This can be achieved using linear regression or Kalman filtering algorithms, and by compensating for environmental interference or equipment deviations, data reliability can be improved. Hazard area mapping is the process of dynamically associating disaster data with a 3D spatial model. This can be achieved using rasterized spatial overlay or vector layer fusion methods, and by using heatmap rendering and boundary markings, the risk distribution can be visually presented.

[0047] After fire alarm data, smoke diffusion data, and abnormal temperature and humidity data are accessed by the system via an IoT interface, they first undergo a format conversion module to remove redundant fields and standardize timestamps. Then, a parameter correction module eliminates inherent equipment errors and environmental noise. The corrected dataset is input into a digital twin model, triggering a smoke diffusion simulation algorithm based on the fire source coordinates. For example, a computational fluid dynamics model is used to predict the smoke concentration gradient, while temperature and humidity sensor readings are combined to dynamically adjust the hazard threshold. The simulation results are overlaid with real-time data, and a hazard area heat map is generated using a spatial interpolation algorithm. Based on preset risk level thresholds, the risk area is divided into prohibited areas, restricted areas, and safe areas. Finally, the classification results are updated in real-time in the 3D model.

[0048] Existing solutions typically display fire alarm and smoke detection data independently, without establishing a unified spatial correlation. This forces maintenance personnel to manually compare multi-source information to determine the risk range. This solution achieves unified processing of multi-source data through format conversion and parameter correction. Combined with dynamic diffusion simulation and threshold determination, it generates a spatialized hazardous area mapping, which can automatically identify the impact range of fire sources and quantify the risk level, avoiding the delays and misjudgments caused by manual data integration.

[0049] This application can transform discrete disaster data into a spatialized risk distribution model, enabling the system to identify dangerous area boundaries and quantify risk levels in real time. It solves the problem of response lag caused by fragmented disaster information in existing technologies and improves the accuracy and timeliness of evacuation route planning.

[0050] Example 6: Please refer to Figure 1 The specific method for step S5 is as follows: S5.1 Based on the accessibility facility availability status obtained in step S3 and the hazardous area mapping obtained in step S4, the access network of the digital twin model is modeled, and the accessibility path nodes and hazardous area boundaries are set as constraints. Path planning algorithms are used to search for paths, including weighted graph shortest path search, heuristic search and Dijkstra's algorithm. The accessibility, path length and obstacle avoidance conditions of candidate paths are comprehensively calculated to generate an initial evacuation path that meets the accessibility constraints and avoids the hazardous area. S5.2 When disaster data changes, based on the updated availability of accessibility facilities and the mapping of dangerous areas, the effectiveness of existing evacuation routes is checked. The check includes determining whether path nodes fall into dangerous areas and whether path edges meet accessibility conditions. When a part of the path fails, the path planning algorithm is called to locally replan the failed path segment. When the overall path is not feasible, the global path is recalculated and the updated evacuation route is written into the digital twin model.

[0051] In this embodiment: This application further proposes to model the accessibility network of the digital twin model based on the availability status of accessible facilities and the mapping of dangerous areas. Accessible passage nodes and dangerous area boundaries are set as constraints. Path planning algorithms are used for path search, including weighted graph shortest path search, A* heuristic search, or Dijkstra's algorithm. The accessibility, path length, and obstacle avoidance conditions of candidate paths are comprehensively calculated to generate initial evacuation paths that meet accessibility constraints and avoid dangerous areas. When disaster data changes, the effectiveness of existing evacuation paths is checked based on the updated availability status of accessible facilities and dangerous area mapping. The check includes determining whether path nodes fall into dangerous areas and whether path edges meet accessibility conditions. When a part of the path fails, the path planning algorithm is called to locally replan the failed path segment. When the entire path is infeasible, the global path is recalculated, and the updated evacuation path is written into the digital twin model.

[0052] The path safety factor is calculated as follows:

[0053] Where C is the accessibility coefficient of the channel. L is the path length; Accessibility network modeling refers to transforming accessible pathways and facilities in the physical environment into a virtual network topology. This can be achieved using a node-edge model from graph theory, where nodes represent pathway intersections or facility locations, and edges represent pathway segments and their accessibility attributes. This model allows for the quantitative analysis of path feasibility. Constraints include the boundary coordinates of hazardous areas and the state parameters of accessible facilities, which can be implemented through spatial coordinate transformation and attribute mapping to limit the path search range. Weighted graph shortest path search in path planning algorithms is suitable for calculating optimal paths in static environments; A* heuristic search is suitable for scenarios with dynamic obstacles; and Dijkstra's algorithm is suitable for global path traversal. Algorithm switching can adapt to the path planning needs of different disaster scenarios. Validity detection refers to dynamically verifying generated paths, which can be implemented using spatial overlay analysis and state matching algorithms to identify whether path nodes are covered by hazardous areas or whether facility states have changed. Local replanning refers to recalculating only failed path segments, which can be implemented using incremental search algorithms to reduce computational resource consumption; global replanning refers to searching the entire model space, suitable for large-scale environmental change scenarios.

[0054] In the initial path generation phase, the availability of accessible pathways and the spatial coordinates of hazardous areas are first converted into node attributes and edge weights in the passage network. Then, an appropriate path planning algorithm is selected based on the severity of the disaster. For example, when the fire spreads rapidly, the A* algorithm is prioritized to quickly generate paths to avoid the fire source; when the environment is relatively stable, the Dijkstra algorithm is used to obtain the globally optimal path. As the disaster situation dynamically changes, the system continuously monitors the expansion of hazardous areas and changes in facility status, performing spatial verification of existing paths by overlaying updated data in real time. If a path segment is detected to enter a hazardous area or its corresponding facility is unavailable, a local replanning module is triggered, recalculating alternative routes only for the affected path segment while retaining unaffected path segments to maintain evacuation continuity. When the hazard spread exceeds a preset threshold, the global path planning module is activated to regenerate complete evacuation routes and synchronously update path guidance information through a digital twin model.

[0055] Traditional methods often require manual reassessment of path feasibility when disaster situations change, leading to response delays. This solution, however, constructs a dynamically updated access network model and combines it with a multi-algorithm adaptive selection mechanism to automatically identify path failure areas and perform tiered replanning. Existing technologies typically use a single algorithm for path search, which is insufficient to handle complex and ever-changing disaster scenarios. This solution, through the flexible use of an algorithm library, can automatically switch to the optimal algorithm based on factors such as the rate of hazard spread and environmental complexity, improving the adaptability and real-time performance of path planning.

[0056] This application effectively solves the problem of inaccurate path planning caused by data fragmentation in existing technologies. Through dynamic network modeling and multi-algorithm collaboration, it ensures that the generated evacuation paths always meet the requirements for barrier-free passage and avoid dangerous areas in real time. As the disaster situation continues to evolve, the system maintains path effectiveness through a hierarchical replanning mechanism, avoiding global path failure caused by changes in the local environment, and significantly improving the reliability and response efficiency of emergency evacuation.

[0057] Example 7: Please refer to Figure 1 The specific method for step S6 is as follows: S6.1 Collect personnel characteristic information based on positioning devices, identity recognition systems or sensor networks. The characteristic information includes the personnel's real-time location, movement speed, auditory or visual sensitivity and dependence on assistive devices. Classify and label the collected data, group personnel according to their mobility and sensory characteristics, and analyze the evacuation needs of various personnel in combination with the evacuation paths in the digital twin model to form differentiated evacuation instruction generation conditions. S6.2 Based on the generation conditions formed in step S6.1, the instruction mapping module is invoked to match different groups of people with corresponding prompting methods. Vibration prompts are generated for people with mobility impairments, voice broadcasts are generated for visually impaired people, light strips or image prompts are generated for hearing impaired people, and LED ground indicators are generated for ordinary people. The generated evacuation instructions are sent to the corresponding prompting devices through the network communication interface, and the evacuation instructions are updated in real time according to changes in the location of people or changes in the environmental conditions during the evacuation process.

[0058] In this embodiment, this application further proposes a digital twin evacuation early warning method for barrier-free access in a substation, comprising the following steps: collecting personnel characteristic information based on positioning devices, identity recognition systems, or sensor networks, including the personnel's real-time location, movement speed, auditory or visual sensitivity, and dependence on assistive devices; classifying and labeling the collected data, grouping personnel according to their mobility and sensory characteristics, and analyzing the evacuation needs of various personnel in conjunction with the evacuation paths in the digital twin model to form differentiated evacuation instruction generation conditions; based on the formed generation conditions, calling the instruction mapping module to match different personnel groups with corresponding prompting methods, generating vibration prompts for personnel with mobility impairments, voice broadcasts for visually impaired personnel, light strips or image prompts for hearing impaired personnel, and LED ground indicators for ordinary personnel; and sending the generated evacuation instructions to the corresponding prompting devices through a network communication interface, and updating the evacuation instructions in real time according to changes in personnel location or environmental conditions during the evacuation process.

[0059] Positioning devices are used to acquire real-time location information of personnel. These can be implemented using Bluetooth beacons, UWB positioning modules, or RFID tags, determining personnel coordinates through triangulation or signal strength analysis. Identification systems are used to identify personnel and their physiological characteristics. These can be implemented using facial recognition terminals, smart bracelets, or IC card readers, acquiring information about personnel's mobility or sensory impairments. Sensor networks are collections of sensing devices deployed in passageways, using infrared sensor arrays, pressure-sensitive mats, or cameras, capturing personnel movement speed and behavioral patterns. Classification and labeling involve structuring the collected personnel data. This can be achieved using machine learning classification algorithms or rule engines, classifying personnel into groups such as those with mobility impairments, visual impairments, and hearing impairments based on preset thresholds. Differentiated evacuation instruction generation conditions define instruction triggering rules based on the needs of different groups. This can be implemented using decision tree models or state machines, for example, assigning vibration cues to wheelchair users to avoid reliance on visual information. Instruction mapping modules are software components that associate personnel groups with cues, using hash tables or relational databases, for example, establishing a mapping relationship between "visually impaired persons" and "voice broadcasts." The notification device refers to the terminal equipment that executes evacuation instructions. Specifically, it can be implemented using smart bracelets, LED floor tiles, voice speakers, or electronic displays, and ensures effective information transmission through multimodal interaction.

[0060] This method tracks personnel location in real time using positioning devices and combines this with an identity recognition system or sensor network to obtain information on personnel's movement speed, sensory sensitivity, and assistive device usage. For example, when a pressure sensor detects a wheelchair passing by, the system automatically identifies the person as having mobility impairments. The collected data is processed by a classification algorithm, categorized into different groups based on mobility capabilities, and the system analyzes the passage requirements of each group using pre-set evacuation routes in a digital twin model. The instruction mapping module calls preset prompt matching rules based on the grouping results; for example, it assigns light strip guidance to people wearing hearing aids to avoid reliance on auditory cues. Evacuation instructions are sent to corresponding devices via IoT communication protocols, such as controlling LED floor tiles to display arrow directions via a ZigBee network. During evacuation, if a temperature and humidity sensor detects environmental deterioration in a certain area, the system immediately updates the instruction content, for example, switching the original voice broadcast to a vibration prompt to avoid dangerous areas.

[0061] Existing solutions typically employ uniform evacuation instructions, such as simply broadcasting evacuation notices to all personnel, failing to differentiate between the specific needs of visually impaired, hearing impaired, or mobility-impaired individuals. This solution, however, accurately identifies personnel characteristics through multi-source data fusion and establishes a dynamic mapping between grouping and prompting methods. For example, existing LED signs may not effectively guide visually impaired individuals, while this solution ensures effective guidance for all groups by simultaneously triggering voice broadcasts and vibration prompts. Furthermore, existing systems often require manual reconfiguration of instructions when the environment changes, while this solution automatically updates instructions based on real-time data, reducing delays caused by human intervention.

[0062] This application addresses the low evacuation efficiency caused by the uniformity of instructions in existing technologies. It differentiates instructions to suit the sensory characteristics of different individuals, such as providing visual cues for the hearing impaired to avoid missing information. Simultaneously, a dynamic update mechanism ensures timely adjustments to guidance methods as disaster spreads or pathways are blocked; for example, the volume of voice broadcasts is immediately increased when smoke obscures LED floor tiles. This solution significantly improves the safety of special groups during emergency evacuations, such as helping wheelchair users avoid stairwells, and reduces the impact of single-device failures through multimodal redundancy design.

[0063] Example 8: Please refer to Figure 1 The specific method for step S7 is as follows: S7.1. The real-time location of personnel is collected through the positioning device, and the movement speed, movement direction and behavior pattern of personnel are obtained by combining the inertial measurement unit, video monitoring and behavior recognition algorithm. The trajectory analysis and group density calculation are used to determine whether congestion occurs, the path deviation detection algorithm is used to determine whether personnel deviate from the predetermined evacuation route, and the dwell time threshold is used to determine whether personnel are stranded, so as to realize real-time monitoring and anomaly detection of evacuation status. S7.2 When an abnormal situation is detected, different processing strategies are triggered based on the type of abnormality. When local congestion or path blockage occurs, the path planning module is called to dynamically replan the local path. When large-scale congestion or danger spread occurs, global path planning is executed. Simultaneously, the evacuation instruction generation module is called to issue updated evacuation instructions to the affected personnel and to provide prompts through a multimodal prompting device. After the personnel reach the safe area, the arrival time and location coordinates are recorded using a positioning device and the data is stored in the evacuation database to form an evacuation process record.

[0064] In this embodiment: This application further proposes to collect the real-time location of personnel during evacuation through a positioning device, and to obtain the personnel's movement speed, movement direction and behavior pattern by combining inertial measurement unit, video monitoring and behavior recognition algorithm. The system uses trajectory analysis and crowd density calculation to determine whether congestion occurs, uses path deviation detection algorithm to determine whether personnel deviate from the predetermined evacuation path, and uses a dwell time threshold to determine whether personnel are stranded, thereby realizing real-time monitoring and anomaly detection of the evacuation status. When an anomaly is detected, different processing strategies are triggered based on the anomaly type. When local congestion or path blockage occurs, the path planning module is invoked to dynamically replan the local path. When large-scale congestion or danger spread occurs, global path planning is executed. The evacuation command generation module is invoked simultaneously to issue updated evacuation commands to the affected personnel, and a multimodal prompting device is used to provide prompts. After personnel reach the safe area, the positioning device records the arrival time and location coordinates, and stores the data in the evacuation database to form an evacuation process record.

[0065] The formula for calculating the probability of personnel risk is as follows:

[0066] Where b1, b2, and b3 are weight values, and the values ​​of b1, b2, and b3 are adjusted and set by the user; AQ is the path safety coefficient; O is the path congestion degree, including 0 and 1, where 0 indicates smooth flow and 1 indicates congestion; V is the speed of personnel movement, and V < 1 indicates obstruction. Positioning devices are used to acquire real-time personnel location information. They can be implemented using GPS modules, UWB positioning modules, or Bluetooth beacons, transmitting location coordinate data wirelessly. Inertial measurement units (IMUs) are sensor modules integrating accelerometers and gyroscopes, typically implemented using MEMS devices, used to capture personnel's movement posture and gait characteristics. Behavior recognition algorithms are deep learning models based on computer vision, typically implemented using convolutional neural networks or temporal action detection models, used to analyze personnel movement patterns in surveillance videos. Trajectory analysis is a calculation method for clustering and trend prediction of continuous location points, typically implemented using Kalman filtering or hidden Markov models, used to identify abnormal movement trajectories. Crowd density calculation is a statistical method based on spatial distribution, typically implemented using kernel density estimation or grid counting algorithms, used to quantify the degree of personnel concentration within an area. Path deviation detection algorithms are methods for calculating the deviation between the actual path and the planned path, typically implemented using Euclidean distance thresholding or dynamic time warping algorithms, used to determine whether personnel have deviated from the safe route. The dwell time threshold refers to a pre-set time threshold, such as a range of 30 seconds to 2 minutes, used to determine whether a person has stayed in a fixed area for too long.

[0067] During evacuation, positioning devices continuously collect real-time coordinates of personnel, while inertial measurement units simultaneously measure acceleration and angular velocity data. Combined with video surveillance input and behavior recognition algorithms, this data extracts personnel's gait frequency, turning frequency, and limb movement characteristics. A trajectory analysis module filters the location data, generating a smooth movement trajectory and calculating its matching degree with the planned path. A path deviation alarm is triggered when the deviation exceeds a preset threshold. A crowd density calculation module generates a heat map based on real-time location distribution; congestion is identified when the density in a specific area exceeds a critical value. A dwell time threshold module tracks the duration of personnel's stay within a fixed coordinate range; a dwell time alarm is generated after a set time is exceeded. When an anomaly is detected, the system selects a local or global path replanning strategy based on the anomaly type. For example, in cases of localized congestion, only the evacuation routes in the affected area are adjusted; in cases of hazard spread, the entire station's path is recalculated. Updated evacuation instructions are issued to the corresponding personnel via vibration devices, voice broadcasts, or LED indicators, while the time and location information of the final arrival at the safe area are recorded in the database.

[0068] Existing solutions typically rely on single sensor data for anomaly detection, such as identifying congestion solely through video surveillance, but failing to distinguish between people remaining in a state of stagnation and those making brief stops, or detecting path deviations based solely on location data while ignoring behavioral pattern differences. This solution, by fusing location data, inertial measurement data, and behavioral recognition results, combined with trajectory analysis and crowd density calculation, can accurately distinguish between normal evacuation behavior and abnormal states. For example, it can differentiate slowly moving elderly groups from genuinely stagnant individuals, avoiding false alarms. Furthermore, while existing technologies often use fixed thresholds for anomaly detection, this solution dynamically adjusts the dwell time threshold and path deviation threshold to adapt to evacuation needs in different scenarios. For instance, a shorter dwell time threshold can be used in the early stages of a fire to improve response speed, while the threshold can be appropriately extended in the later stages of evacuation to reduce false alarms.

[0069] This application effectively solves the problem of misjudgment of abnormal states caused by data fragmentation in existing technologies, improving the accuracy and timeliness of dynamic adjustment of evacuation routes. Through multi-source data fusion and intelligent algorithm collaboration, the system can accurately identify abnormal situations such as congestion, deviation, and delays, and trigger targeted handling strategies according to different anomaly types, avoiding evacuation command conflicts caused by information redundancy or misjudgment. Simultaneously, the complete recording of evacuation process data provides reliable data support for subsequent emergency strategy optimization, helping to reduce the probability of secondary risks.

[0070] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0071] The specific embodiments of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this application. Equivalent transformations, modifications, and improvements made without departing from the spirit and principles of this application should be covered within the scope of this application.

Claims

1. A digital twin evacuation early warning method for barrier-free access in a substation, characterized in that: The specific steps are as follows: S1. Establish a digital twin model of the substation, collect building information, construct a digital twin model including barrier-free access and barrier-free facilities, and set geometric attributes, spatial location relationships and accessibility parameters for the facilities; S2. Collect sensor data deployed in and around the barrier-free facilities. The sensor data includes video surveillance, infrared detection, pressure detection and door magnetic detection data. Integrate the collected data into the digital twin model. S3. Perform time synchronization, spatial coordinate transformation and data fusion on multi-source sensor data, determine the availability status of barrier-free facilities based on the fusion results, and update the availability status to the digital twin model; S4. Receive disaster data, including fire alarm, smoke diffusion and abnormal temperature and humidity data. Overlay the disaster data onto the digital twin model to generate a dangerous area map and identify the risk range. S5. Based on the availability of accessibility facilities and the mapping of dangerous areas, use path planning algorithms to generate evacuation routes that meet accessibility constraints and avoid dangerous areas, and update the evacuation routes when the disaster situation changes. S6. Output evacuation instructions based on personnel characteristic information. Evacuation instructions include vibration prompts, light strips or image prompts, voice broadcasts and LED ground indicators to guide different personnel to evacuate. S7. During the evacuation process, use positioning devices and behavior recognition devices to monitor the evacuation status of personnel, determine whether there is congestion, deviation or stagnation. If there is an abnormality, repeat steps S5 and S6 to update the evacuation route and evacuation instructions, and record the information of personnel arriving at the safe area after the evacuation is completed.

2. The digital twin evacuation early warning method for barrier-free access in a booster station according to claim 1, characterized in that, The specific method for step S1 is as follows: S1.1 Collect data on the building structure, passage layout and accessibility facilities of the substation. The data includes geometric dimensions, spatial coordinates, hierarchical relationships and material properties, which are used to generate the basic dataset required for digital twin modeling. S1.

2. Based on the data collected step by step, construct a three-dimensional digital twin model containing barrier-free access and barrier-free facilities, and configure geometric attributes, spatial location relationships and accessibility parameters for each barrier-free facility to form a virtual model corresponding to the actual substation environment.

3. The digital twin evacuation early warning method for barrier-free access in a booster station according to claim 2, characterized in that, The specific method for step S2 is as follows: S2.

1. Deploy video surveillance devices, infrared detectors, pressure sensors and door magnetic sensors in and around the accessible facilities to collect real-time data on the occupancy status, structural status and surrounding environmental parameters of the accessible facilities and generate sensor data. S2.2 The collected sensor data is connected to the digital twin platform via the data transmission interface, the sensor data is formatted, and the processed data is mapped to the corresponding facility nodes in the digital twin model.

4. The digital twin evacuation early warning method for barrier-free access in a booster station according to claim 3, characterized in that, The specific method for step S3 is as follows: S3.1 Perform time synchronization processing and spatial coordinate transformation processing on the multi-source sensor data accessed in step S2, unify different types of data to the same time reference and the same spatial coordinate system, and generate a standardized dataset; S3.

2. Perform fusion processing on the standardized dataset obtained in step S3.

1. The fusion processing includes target recognition and contour matching of video surveillance data and infrared detection data to determine whether there are moving obstacles in the barrier-free passage area. Pressure detection data and door magnetic detection data are compared with threshold values ​​to determine whether the passage is occupied and whether the relevant facilities are in the open state. Based on this, the judgment results of different types of sensor data are cross-validated to generate the occupancy status, integrity status and accessibility status of the barrier-free facilities, and the status results are updated to the corresponding facility nodes in the digital twin model.

5. The digital twin evacuation early warning method for barrier-free access in a booster station according to claim 4, characterized in that, The specific method for step S4 is as follows: S4.1 Receive fire alarm data, smoke diffusion data, and abnormal temperature and humidity data; perform format conversion and parameter correction on data from different sources; and generate a disaster dataset in a unified format. S4.

2. Overlay the disaster dataset obtained in step S4.1 onto the digital twin model, perform disaster correlation calculations on the model's spatial regions. The disaster correlation calculations include determining the fire source location based on fire alarm data, simulating the diffusion trend based on smoke diffusion data, determining environmental thresholds based on abnormal temperature and humidity data, and spatially overlaying the calculation results to generate a hazardous area mapping. In the hazardous area mapping, the risk range is divided and marked according to the hazard level, and the division results are updated to the digital twin model.

6. The digital twin evacuation early warning method for barrier-free access in a booster station according to claim 5, characterized in that, The specific method for step S5 is as follows: S5.1 Based on the accessibility facility availability status obtained in step S3 and the hazardous area mapping obtained in step S4, the access network of the digital twin model is modeled, and the accessibility path nodes and hazardous area boundaries are set as constraints. Path planning algorithms are used to search for paths, including weighted graph shortest path search, heuristic search and Dijkstra's algorithm. The accessibility, path length and obstacle avoidance conditions of candidate paths are comprehensively calculated to generate an initial evacuation path that meets the accessibility constraints and avoids the hazardous area. S5.2 When disaster data changes, based on the updated availability of accessibility facilities and the mapping of dangerous areas, the effectiveness of existing evacuation routes is checked. The check includes determining whether path nodes fall into dangerous areas and whether path edges meet accessibility conditions. When a part of the path fails, the path planning algorithm is called to locally replan the failed path segment. When the overall path is not feasible, the global path is recalculated and the updated evacuation route is written into the digital twin model.

7. The digital twin evacuation early warning method for barrier-free access in a booster station according to claim 6, characterized in that, The specific method for step S6 is as follows: S6.

1. Collect personnel characteristic information based on positioning devices, identity recognition systems or sensor networks. The characteristic information includes the personnel's real-time location, movement speed, auditory or visual sensitivity and dependence on auxiliary equipment. Classify and label the collected data, group personnel according to their mobility and sensory characteristics, and analyze the evacuation needs of various personnel in combination with the evacuation paths in the digital twin model to form differentiated evacuation instruction generation conditions. S6.2 Based on the generation conditions formed in step S6.1, the instruction mapping module is invoked to match different groups of people with corresponding prompting methods. Vibration prompts are generated for people with mobility impairments, voice broadcasts are generated for visually impaired people, light strips or image prompts are generated for hearing impaired people, and LED ground indicators are generated for ordinary people. The generated evacuation instructions are sent to the corresponding prompting devices through the network communication interface, and the evacuation instructions are updated in real time according to changes in the location of people or changes in the environmental conditions during the evacuation process.

8. The digital twin evacuation early warning method for barrier-free access in a booster station according to claim 7, characterized in that, The specific method for step S7 is as follows: S7.

1. The real-time location of personnel is collected through the positioning device, and the movement speed, movement direction and behavior pattern of personnel are obtained by combining the inertial measurement unit, video monitoring and behavior recognition algorithm. The trajectory analysis and group density calculation are used to determine whether congestion occurs, the path deviation detection algorithm is used to determine whether personnel deviate from the predetermined evacuation route, and the dwell time threshold is used to determine whether personnel are stranded, so as to realize real-time monitoring and anomaly detection of evacuation status. S7.2 When an abnormal situation is detected, different processing strategies are triggered based on the type of abnormality. When local congestion or path blockage occurs, the path planning module is called to dynamically replan the local path. When large-scale congestion or danger spread occurs, global path planning is executed. Simultaneously, the evacuation instruction generation module is called to issue updated evacuation instructions to the affected personnel and to provide prompts through a multimodal prompting device. After the personnel reach the safe area, the arrival time and location coordinates are recorded using a positioning device and the data is stored in the evacuation database to form an evacuation process record.

Images