Mountainous rail transit station emergency command rescue system and method

Abstract

This invention discloses an emergency command and rescue system and method for mountain rail transit stations, including the system and method. The system includes a command center, n mobile rescue subsystems carried by rescue personnel during rescue operations, and s emergency navigation and positioning subsystems, image recognition-based emergency monitoring subsystems, and emergency voice recognition and dialogue subsystems, each installed in one of the mountain rail transit stations and communicating in real-time with the command center. The method includes steps such as direct positioning, image positioning and analysis, sound positioning and analysis, indicating trapped personnel in a scene model, indicating rescue routes, and real-time monitoring of the rescue. This invention allows rescue personnel to more conveniently and intuitively understand the situation on-site during emergencies, improving rescue efficiency.

Description

Technical Field

[0001] This invention relates to the technical field of transportation station rescue, and in particular to an emergency command and rescue system and method for mountain rail transit stations. Background Technology

[0002] Mountain rail transit is a new type of rail transit suitable for mountainous terrain, with advantages such as small footprint, large capacity, and comfort and convenience. However, because mountain rail transit stations are relatively enclosed and narrow spaces, a large number of passengers gather there in a short period of time, especially during peak hours. In the event of an accident, the enclosed and narrow space can cause casualties, property damage, and negative social impacts.

[0003] However, existing rescue methods require personnel to manually collect information on the location and situation of all trapped individuals on-site. This results in a long time lag for the department directing the rescue operation to formulate a rescue plan, leading to low rescue efficiency. In addition, when encountering situations where rescue routes are blocked, such as partial building collapses, it is impossible to notify all rescue personnel in a timely manner, resulting in the inability to conduct rescues through suitable routes in a timely manner. Summary of the Invention

[0004] To address the aforementioned shortcomings, this invention provides an emergency command and rescue system and method for mountain rail transit stations, which enables rescue personnel to more conveniently and intuitively understand the on-site situation and improve rescue efficiency in the event of an emergency.

[0005] In order to achieve the objectives of this invention, the following technologies are proposed:

[0006] An emergency command and rescue system for mountain rail transit stations includes:

[0007] The command center includes a control host and a command display module. The control host stores scene models of s mountain rail transit stations.

[0008] n mobile rescue subsystems carried by each rescuer during the rescue operation, each mobile rescue subsystem including a mobile camera module and a mobile display module;

[0009] The emergency command and rescue system also includes s mountain rail transit stations that communicate in real time with the command center, each equipped with:

[0010] The emergency navigation and positioning subsystem includes multiple indoor direct positioning modules. Each mountain rail transit station includes multiple areas, and each area is equipped with at least one indoor direct positioning module.

[0011] The emergency monitoring subsystem based on image recognition includes multiple fixed camera modules. Each area of ​​the mountain rail transit station is equipped with an image recognition array including at least two fixed camera modules.

[0012] An emergency voice recognition and dialogue subsystem includes multiple microphone modules, and microphone arrays containing at least two microphone modules are installed in some or all areas of mountain rail transit stations.

[0013] Furthermore, the command center also includes an editing interface, a command communication module, and a scene model that includes maps of the corresponding mountain rail transit stations;

[0014] The mobile rescue subsystem also includes a mobile communication module;

[0015] Each emergency navigation and positioning subsystem also includes a first transceiver module;

[0016] Each image recognition-based emergency monitoring subsystem also includes a second transceiver module;

[0017] Each emergency voice recognition and dialogue subsystem also includes a third transceiver module.

[0018] Furthermore, the emergency navigation and positioning subsystem also includes several emergency passage control modules, with each emergency passage control module in the mountain rail transit station being used to control one emergency passage.

[0019] An emergency command and rescue method for mountain rail transit stations, employing a mountain rail transit station emergency command and rescue system, includes the following steps:

[0020] S100: When an emergency occurs at a mountain rail transit station, the mountain rail transit station collects the location of trapped personnel through the indoor direct positioning modules of the emergency navigation and positioning subsystem, and sends the collected direct positioning coordinates of each trapped person in each area to the command center. Then, the command center generates each trapped person instance in the scene model of the mountain rail transit station. The trapped person instance includes the following variables: trapped person ID, location, and positioning coordinates.

[0021] S200: In the image recognition-based emergency monitoring subsystem of a mountain rail transit station in case of an emergency, the image recognition array in each area identifies the trapped personnel in the area, adds the trapped personnel instances that were not found through step S100 to the scene model in each area, and judges the urgency level of each trapped person.

[0022] S300: In the emergency voice recognition and dialogue subsystem of a mountain rail transit station in an emergency, each receiver array collects the voices of trapped personnel in its area and separates the voices into individual sound sources of several trapped personnel. The sound sources are then located separately. For trapped personnel in each area who were not found through steps S100 and S200, instances of trapped personnel are added to the scene model, and the urgency level of each trapped person is determined.

[0023] S400: The command center displays the scene model of the mountain rail transit station where the emergency has occurred on the command display module, and displays the instances of each trapped person at their predetermined positions on the command display module using the location coordinates of each trapped person;

[0024] S500: The command center receives rescue route instructions in group a. Each group of rescue route instructions includes multiple rescue routes that instruct rescuers to rescue trapped personnel in one of the a areas of the mountain rail transit station described in S400.

[0025] S600: The command center displays all the rescue routes described in S500 on the command display module. The command center communicates with all mobile rescue subsystems. When each rescuer arrives at the mountain rail transit station described in S400, the mobile rescue subsystem sends entry feedback to the command center. The command center adds rescuer instances to the scene model. The rescuer instances include the rescuer ID, the area, and the location coordinates. At the same time, the mobile camera module is activated to record video in real time and transmit it to the command center. When the rescuers enter the scene, the command center synchronously communicates the display content of the command display module to the mobile display module of the mobile rescue subsystem.

[0026] S700: After the rescuers select an area for rescue, if obstacles are found on every rescue path in the area during the rescue, the command center adds obstacle removal markers at the locations of the obstacles on the rescue paths selected by the rescuers after the rescuers report to the command center. The obstacle removal markers are displayed on the command display module and the mobile display modules of all mobile rescue subsystems. After the obstacles are removed, the command center and the mobile rescue subsystem delete the obstacle removal markers, and then S800 is executed. If at least one rescue path in the area is found to be unobstructed during the rescue, S800 is executed directly.

[0027] S800: In the event of an emergency at a mountain rail transit station, S700 is executed multiple times. After all rescueable trapped personnel at the mountain rail transit station have been rescued and left the station, the command center determines that the rescue is complete.

[0028] Furthermore, the S200 includes:

[0029] S210: Obtain the number a of mountain rail transit stations where an emergency has occurred;

[0030] S220: Set i=1;

[0031] S230: The image recognition array in the i-th region acquires images within the region and sends them to the command center;

[0032] S240: The command center separates multiple images corresponding to each trapped person from the acquired images, and calculates the image positioning coordinates of b trapped persons by identifying the position of each trapped person in the images captured by multiple fixed camera modules.

[0033] S250: Compare the image location coordinates of b trapped persons with the location coordinates of p trapped persons in the i-th region of the scene model of the mountain rail transit station described in S210. For the trapped persons identified in S240 whose location coordinates do not match the image location coordinates, generate several new trapped persons instances.

[0034] S260: Determine if i = a. If yes, execute S270. If no, i = i + 1, and then execute S230 again.

[0035] S270: After supplementation, the scene model includes q instances of trapped personnel. The command center calculates the urgency level based on the image of each trapped person corresponding to the instance and adds the urgency level to the instance.

[0036] Furthermore, the S250 includes:

[0037] S251: Obtain the number p of existing trapped personnel instances in the i-th region of the scene model of the mountain rail transit station described in S210;

[0038] S252: Set j=1;

[0039] S253: Determine whether there is a trapped person instance among the existing p trapped person instances whose distance from the image positioning coordinates of the j-th trapped person identified in S240 is less than the preset error w. If yes, execute S255 directly; otherwise, execute S254 first and then S255.

[0040] S254: Generate a new instance of a trapped person in the scene model, wherein the location coordinates of the trapped person instance are image location coordinates;

[0041] S255: Determine if j=b. If yes, S250 ends. If no, j=j+1, and S253 is executed again.

[0042] Furthermore, the S300 includes:

[0043] S310: Obtain the number of areas c with radio arrays in a mountain rail transit station where an emergency has occurred;

[0044] S320: Set x=1;

[0045] S330: The receiver array in the xth region acquires sound segments within the region and sends them to the command center;

[0046] S340: The command center separates the acquired sound segments to obtain individual sound sources corresponding to several trapped personnel. By identifying each individual sound source, the sound source is located for each trapped person, and the sound location coordinates of d trapped personnel are calculated.

[0047] S350: Compare the sound location coordinates of the d trapped persons with the location coordinates of the q trapped persons in the xth region of the scene model of the mountain rail transit station described in S310. For the trapped persons identified in S340 whose location coordinates do not match the sound location coordinates, generate several new trapped persons instances.

[0048] S360: Determine if x = c. If yes, execute S370. If no, x = x + 1, and then execute S330 again.

[0049] S370: After supplementation, the scene model includes k instances of trapped personnel.

[0050] Furthermore, the S350 includes:

[0051] S351: Obtain the number q of existing trapped personnel instances in the x-th region of the scene model of the mountain rail transit station described in S310;

[0052] S352: Set y=1;

[0053] S353: Determine whether there is an instance of a trapped person among the existing q instances whose distance from the sound location coordinates of the yth trapped person identified in S340 is less than the preset error w. If yes, execute S355 directly; otherwise, execute S354 first and then S355.

[0054] S354: Generate a new instance of a trapped person in the scene model, wherein the location coordinates of the trapped person instance are sound location coordinates;

[0055] S355: Determine if y = d. If yes, S350 ends; otherwise, y = y + 1, and S353 is executed again.

[0056] Furthermore, when executing S600~S700, the following steps are also executed simultaneously:

[0057] The command center sends status feedback instructions to the mobile rescue subsystems of all rescue personnel at preset time intervals t, requiring each rescue personnel to send a normal status feedback through the mobile rescue subsystem.

[0058] The rescuers are deemed to be trapped when at least one of the following two conditions occurs: the command center receives a trapped information from the mobile rescue subsystem of the trapped rescuers, or the command center does not receive a status feedback from the mobile rescue subsystem of the rescuers within a predetermined time interval t of m.

[0059] In the scenario model of a mountain rail transit station in an emergency, the control host of the command center converts the rescuer instances corresponding to each trapped rescuer into trapped personnel instances, deletes the rescuer IDs, generates trapped personnel IDs, and at the same time, the command center stops sending status feedback commands to the mobile rescue subsystem of the trapped rescuers.

[0060] For each trapped person instance converted from a rescuer instance, the location of each trapped rescuer is obtained through the location variable contained in the trapped person instance. The image is then acquired through the image recognition array of the area. The image is then sent to the command center by the emergency monitoring subsystem based on image recognition. The control host of the command center locates the trapped rescuer through the location coordinates of the trapped rescuer, separates the image of the trapped rescuer, calculates the urgency level, and adds the urgency level to the trapped person instance.

[0061] The beneficial effects of this technical solution are as follows:

[0062] 1. By using scenario models and instances of each trapped person detected within them, combined with rescue routes indicated by the command center, rescuers can make more accurate and intuitive on-site rescue decisions, thereby improving rescue efficiency and success rate.

[0063] 2. The system detects trapped individuals in mountain rail transit stations from multiple dimensions. Firstly, direct location is used to locate individuals carrying mobile phones or other communication devices. For elderly people, children, and others who cannot be directly located due to lack of such equipment, image recognition arrays are used for location and supplementary detection. Simultaneously, the severity level of the trapped individuals detected in these two steps is assessed, which can be used as a reference for rescuers, for example, in accordance with the principle of rescuing the easiest cases first. In some situations where visibility is obstructed or conditions such as smoke limit image recognition, a radio array is used for location and supplementary detection, further improving the accuracy of identifying those in need of rescue. Attached Figure Description

[0064] Figure 1 The diagram shows the overall architecture of the emergency command and rescue system for mountain rail transit stations according to an embodiment of this application.

[0065] Figure 2 The overall flowchart of the emergency command and rescue method for mountain rail transit stations according to an embodiment of this application is shown.

[0066] Figure 3 The flowchart of step S200 of the emergency command and rescue method for mountain rail transit stations according to an embodiment of this application is shown.

[0067] Figure 4 The flowchart of step S250 of the emergency command and rescue method for mountain rail transit stations according to an embodiment of this application is shown.

[0068] Figure 5 The flowchart of step S300 of the emergency command and rescue method for mountain rail transit stations according to an embodiment of this application is shown.

[0069] Figure 6 The flowchart of step S350 of the emergency command and rescue method for mountain rail transit stations according to an embodiment of this application is shown.

[0070] Figure 7 A partial flowchart of a preferred embodiment of the emergency command and rescue method for mountain rail transit stations according to this application is shown. Detailed Implementation

[0071] The present application will be further described below with reference to the accompanying drawings and embodiments.

[0072] like Figure 1The emergency command and rescue system for mountain rail transit stations shown includes a command center, n mobile rescue subsystems carried by rescue personnel during rescue operations, and an emergency navigation and positioning subsystem, an image recognition-based emergency monitoring subsystem, and an emergency voice recognition and dialogue subsystem, each of the s mountain rail transit stations communicating in real time with the command center.

[0073] The command center includes a control host, an editing interface, a command display module, and a command communication module. The control host stores scene models of s mountain rail transit stations. The scene models include maps of the corresponding mountain rail transit stations. Specifically, the maps are two-dimensional or three-dimensional maps.

[0074] Each of the n mobile rescue subsystems is carried by a rescuer during a rescue operation. The mobile rescue subsystem includes a mobile camera module, a mobile display module, and a mobile communication module.

[0075] The emergency navigation and positioning subsystem includes multiple indoor direct positioning modules, several emergency passage control modules, and a first transceiver module. Each mountain rail transit station includes multiple areas. Specifically, each area corresponds to the scope of a room or an independent space. Each area is equipped with at least one indoor direct positioning module. Each emergency passage control module of the mountain rail transit station is used to control one emergency passage.

[0076] Specifically, the indoor direct positioning module adopts a positioning method that primarily uses WIFI positioning and Bluetooth positioning, supplemented by RFID (Radio Frequency Identification) positioning. WIFI and Bluetooth positioning are common and relatively mature positioning methods with high accuracy indoors. If the person does not turn on Bluetooth or WIFI, positioning can also be achieved by embedding an RFID chip in the ticket or work ID.

[0077] The image recognition-based emergency monitoring subsystem includes multiple fixed camera modules. Specifically, both the fixed and mobile camera modules have thermal imaging capabilities to assist in low-light conditions. The image recognition-based emergency monitoring subsystem also includes a second transceiver module. Each area of ​​the mountain rail transit station is equipped with an image recognition array that includes at least two fixed camera modules.

[0078] The emergency voice recognition and dialogue subsystem includes multiple audio receiving modules, multiple broadcast alarm modules, and a third transceiver module. Some or all areas of the mountain rail transit station are equipped with audio receiving arrays including at least two audio receiving modules. Preferably, some or all areas of the mountain rail transit station are equipped with a broadcast alarm module, which is used by the command center to communicate with trapped personnel to calm their emotions and teach them self-rescue measures, and to guide trapped personnel who are still able to move to leave the mountain rail transit station on their own through the nearest emergency passage.

[0079] like Figures 2-7 As shown, an emergency command and rescue method for mountain rail transit stations using a mountain rail transit station emergency command and rescue system is operated according to the following steps:

[0080] S100: When an emergency occurs at a mountain rail transit station, the mountain rail transit station collects the location of trapped personnel through the indoor direct positioning modules of the emergency navigation and positioning subsystem, and sends the collected direct positioning coordinates of each trapped person in each area to the command center. Specifically, the data is sent through the first transceiver module and received through the command communication module. Then, the control host parses the direct positioning coordinates of each trapped person and generates each trapped person instance in the scene model of the mountain rail transit station. The trapped person instance includes the following variables: trapped person ID, location, and positioning coordinates. The trapped person ID is an automatically generated positive integer. In this embodiment, the trapped person ID is counted sequentially starting from 1.

[0081] S200: In the image recognition-based emergency monitoring subsystem of a mountain rail transit station in the event of an emergency, the image recognition arrays in each area identify trapped personnel within that area. For trapped personnel not identified in step S100, instances of trapped personnel are added to the scene model in each area, and the severity level of each trapped person is assessed, including:

[0082] S210: Obtain the number of mountain rail transit stations, a, where an emergency has occurred. Specifically, the number of stations, a, is pre-stored in the control host of the command center.

[0083] S220: Set i=1;

[0084] S230: The image recognition array in the i-th region acquires an image within the region and sends it to the command center. Specifically, it sends the image through the second transceiver module and receives it through the command communication module.

[0085] S240: The command center separates multiple images corresponding to each trapped person from the acquired images. By identifying the position of each trapped person in the images captured by multiple fixed camera modules, the image positioning coordinates of b trapped persons are calculated. Specifically, multiple fixed camera modules of the image recognition array simultaneously capture images of objects in space from different angles. By comparing and analyzing the common features in these images, algorithms such as triangulation are used to calculate the spatial position of the trapped persons. This is a common existing method of positioning through multiple images.

[0086] S250: Compare the image location coordinates of the b trapped persons with the location coordinates of the p trapped persons already existing in the i-th region of the scene model of the mountain rail transit station described in S210. For the trapped persons identified in S240 whose location coordinates do not match the image location coordinates, generate several new trapped person instances. S250 includes:

[0087] S251: Obtain the number p of existing trapped personnel instances in the i-th region of the scene model of the mountain rail transit station described in S210;

[0088] S252: Set j=1;

[0089] S253: Determine whether there is a trapped person instance among the existing p trapped person instances whose distance from the image positioning coordinates of the j-th trapped person identified in S240 is less than the preset error w. If yes, execute S255 directly; otherwise, execute S254 first and then S255.

[0090] S254: Generate a new instance of a trapped person in the scene model, wherein the location coordinates of the trapped person instance are image location coordinates;

[0091] S255: Determine if j = b. If yes, S250 ends. If no, j = j + 1, and S253 is executed again.

[0092] S260: Determine if i = a. If yes, execute S270. If no, i = i + 1, and then execute S230 again.

[0093] S270: After supplementation, the scene model includes q instances of trapped personnel. The command center calculates the severity level for each trapped person based on their image and adds the severity level to the trapped personnel instance. Specifically, a visual judgment model for severity level is first trained on the control host. In this embodiment, the visual judgment model for severity level is trained using the YOLO framework with millions of personnel feature images as the training set. This model categorizes the severity level of trapped personnel into three levels: unable to move, requiring assistance, etc. Trapped individuals are categorized into three levels: those who are mobile and capable of moving on their own, those deemed unable to move (indicating a critical level), and those requiring assistance (indicating a critical level). Other trapped individuals are categorized as capable of moving on their own. For example, in a rescue operation where several individuals were trapped in each of these three critical levels, those capable of moving on their own left after receiving prompts from the broadcast alarm module. For the other two categories, rescuers followed a principle of prioritizing easier cases, first assisting those requiring assistance, and then those unable to move.

[0094] S300: In the emergency voice recognition and dialogue subsystem of a mountain rail transit station in an emergency, each receiver array collects the voices of trapped personnel within its area, separates the voices into individual sound sources for each trapped person, and then locates each sound source. For trapped personnel not identified through steps S100 and S200 in each area, instances of trapped personnel are added to the scene model, and a severity level assessment is performed for each trapped person, including:

[0095] S310: Obtain the number c of areas with radio arrays in the mountain rail transit station where an emergency occurs. Specifically, the number c of areas with radio arrays is pre-stored in the control host of the command center.

[0096] S320: Set x=1;

[0097] S330: The receiver array in the xth region acquires sound segments within the region and sends them to the command center. Specifically, it sends the signals through the third transceiver module and receives them through the command communication module.

[0098] S340: The command center separates the acquired sound segments to obtain individual sound sources corresponding to several trapped personnel. By identifying each individual sound source, the sound source is located for each trapped person, and the sound location coordinates of d trapped persons are calculated. Specifically, through multiple receiving modules of the receiving array, existing speech models, such as deep learning-based models (e.g., Wave-U-Net, Conv-TasNet), are used for sound source separation. These models can be trained using a large number of labeled datasets, thus possessing the ability to separate different sound sources from mixed signals. These models can identify and separate multiple superimposed sounds, providing a foundation for subsequent sound source localization. For sound source localization, mature technologies such as multi-signal classification and signal parameter estimation based on rotation invariant techniques are used. By decomposing the covariance matrix of the array-received signals and constructing a spatial spectrum, the direction of the sound source is determined. These methods are applicable to multiple sound sources and have high direction-finding accuracy and resolution.

[0099] S350: Compare the sound location coordinates of the d trapped persons with the location coordinates of the q trapped persons already existing in the x-th region of the scene model of the mountain rail transit station described in S310. For the trapped persons identified in S340 whose location coordinates do not match the sound location coordinates, generate several new trapped person instances. S350 includes:

[0100] S351: Obtain the number q of existing trapped personnel instances in the x-th region of the scene model of the mountain rail transit station described in S310;

[0101] S352: Set y=1;

[0102] S353: Determine whether there is an instance of a trapped person among the existing q instances whose distance from the sound location coordinates of the yth trapped person identified in S340 is less than the preset error w. If yes, execute S355 directly; otherwise, execute S354 first and then S355.

[0103] S354: Generate a new instance of a trapped person in the scene model, wherein the location coordinates of the trapped person instance are sound location coordinates;

[0104] S355: Determine if y = d. If yes, S350 ends; otherwise, y = y + 1, and S353 is executed again.

[0105] S360: Determine if x = c. If yes, execute S370. If no, x = x + 1, and then execute S330 again.

[0106] S370: After supplementation, the scene model includes k instances of trapped personnel. Specifically, the urgency level variables of the (kq) instances of trapped personnel supplemented in S350 are temporarily empty. After the rescuers arrive, the urgency level of these (kq) trapped personnel is determined by the images through the mobile camera module of the mobile rescue subsystem they carry. The image recognition method is the same as the specific method described in S270.

[0107] S400: The control host of the command center converts the scene model of the mountain rail transit station in the emergency into visual data and displays it on the command display module. It also converts each trapped person instance into visual data through the location coordinates of each trapped person and displays it at each predetermined position on the command display module.

[0108] S500: The command center receives a group of rescue route instructions. Each group of rescue route instructions includes multiple rescue routes that instruct rescuers to rescue trapped personnel in one of the a areas of the mountain rail transit station described in S400. In this embodiment, three rescue routes are provided for each area. Specifically, receiving the rescue route instructions in this step means that the command personnel edit the rescue routes through the editing interface.

[0109] S600: The control host of the command center converts all the rescue paths described in S500 into visual data and displays it on the command display module. The command center communicates with all mobile rescue subsystems. When each rescuer arrives at the mountain rail transit station described in S400, the mobile rescue subsystem sends entry feedback to the command center. The command center adds rescuer instances to the scene model. The rescuer instance includes the rescuer ID, the area, and the location coordinates. The rescuer ID is an automatically generated positive integer. The rescuer carries a locator, so the area and location coordinates can be obtained directly and updated in real time and synchronized to the command center. At the same time, the mobile camera module is activated to record video in real time and transmit the video to the command center. When the rescuer enters the scene, the command center synchronizes the display content of the command display module to the mobile display module of the mobile rescue subsystem. Specifically, communication is carried out through the command communication module and the mobile communication module.

[0110] S700: After the rescuers select an area for rescue, if obstacles are found on every rescue path in the area during the rescue, such as a collapsed object making it impossible to pass, the command center's control host adds a clearing mark at the obstacle location on the rescue path selected by the rescuers after the rescuers report to the command center. The clearing mark is also displayed on the command display module and the mobile display modules of all mobile rescue subsystems. After the obstacle is removed, the command center and the mobile rescue subsystem delete the clearing mark, and then S800 is executed. If at least one rescue path in the area is found to be unobstructed during the rescue, S800 is executed directly.

[0111] S800: In the event of an emergency at a mountain rail transit station, S700 is executed multiple times by multiple rescue personnel until all rescueable trapped personnel at the mountain rail transit station have been rescued and left the station. The command center then determines that the rescue is complete. Specifically, the rescueable trapped personnel in this step exclude those who have been confirmed dead by rescue personnel. The remains of these individuals are not within the scope of the rescue and generally need to be handled by a specialized remains disposal team.

[0112] Preferably, in steps S100 to S300, there may still be trapped personnel who have not been detected through these steps. In this case, during step S700, the rescue personnel will report to the command center on their own.

[0113] More preferably, since rescuers themselves may also be trapped during the rescue operation, the following steps are also performed simultaneously when executing S600~S700:

[0114] The command center sends status feedback instructions to the mobile rescue subsystems of all rescue personnel at preset time intervals t, requiring each rescue personnel to send a normal status feedback through the mobile rescue subsystem.

[0115] The rescuers are deemed to be trapped when at least one of the following two conditions occurs: the command center receives a trapped information from the mobile rescue subsystem of the trapped rescuers, or the command center does not receive a status feedback from the mobile rescue subsystem of the rescuers within a predetermined time interval t of m.

[0116] In the scenario model of a mountain rail transit station in an emergency, the control host of the command center converts the rescuer instances corresponding to each trapped rescuer into trapped personnel instances, deletes the rescuer IDs, generates trapped personnel IDs, and at the same time, the command center stops sending status feedback commands to the mobile rescue subsystem of the trapped rescuers.

[0117] For each trapped person instance converted from a rescuer instance, the location of each trapped rescuer is obtained through the location variable contained in the trapped person instance. The image is then acquired through the image recognition array of the area. The image is then sent to the command center by the emergency monitoring subsystem based on image recognition. The control host of the command center locates the trapped rescuer through the location coordinates of the trapped rescuer, separates the image of the trapped rescuer, calculates the urgency level, and adds the urgency level to the trapped person instance.

[0118] The above are only some of the embodiments listed in this application and are not intended to limit this application.

Claims

1. A method for emergency command and rescue at a mountain rail transit station, characterized in that, An emergency command and rescue system for mountain rail transit stations is adopted, the system comprising: The command center includes a control host and a command display module. The control host stores scene models of s mountain rail transit stations. The command center also includes an editing interface and a command communication module. The scene models include maps of the corresponding mountain rail transit stations. A visual judgment model of the hazard level is trained on the control host. The model is trained using the YOLO framework with personnel feature images as the training set. There are n mobile rescue subsystems carried by each rescuer during the rescue operation. Each mobile rescue subsystem includes a mobile camera module, a mobile display module, and a mobile communication module. The emergency command and rescue system also includes s mountain rail transit stations that communicate in real time with the command center, each equipped with: The emergency navigation and positioning subsystem includes multiple indoor direct positioning modules and several emergency channel control modules. Each mountain rail transit station includes multiple areas, and each area is equipped with at least one indoor direct positioning module. Each emergency navigation and positioning subsystem also includes a first transceiver module. Each emergency channel control module of the mountain rail transit station is used to control one emergency channel. The emergency monitoring subsystem based on image recognition includes multiple fixed camera modules. Each area of ​​the mountain rail transit station is equipped with an image recognition array including at least two fixed camera modules. Each emergency monitoring subsystem based on image recognition also includes a second transceiver module. An emergency voice recognition and dialogue subsystem includes multiple sound receiving modules. Each part or all of the mountain rail transit station is equipped with a sound receiving array including at least two sound receiving modules and a broadcast alarm module. Each emergency voice recognition and dialogue subsystem also includes a third transceiver module. The sound receiving modules adopt a speech model based on deep learning. By decomposing the covariance matrix of the received signal of the array, a spatial spectrum is constructed to determine the direction of the sound source. The method includes the following steps: S100: When an emergency occurs at a mountain rail transit station, the mountain rail transit station collects the location of trapped personnel through the indoor direct positioning modules of the emergency navigation and positioning subsystem, and sends the collected direct positioning coordinates of each trapped person in each area to the command center. Then, the command center generates each trapped person instance in the scene model of the mountain rail transit station. The trapped person instance includes the following variables: trapped person ID, location, and positioning coordinates. S200: In the image recognition-based emergency monitoring subsystem of a mountain rail transit station in case of an emergency, the image recognition array in each area identifies the trapped personnel in the area, adds the trapped personnel instances that were not found through step S100 to the scene model in each area, and judges the urgency level of each trapped person. S300: In the emergency voice recognition and dialogue subsystem of a mountain rail transit station in an emergency, each receiver array collects the voices of trapped personnel in its area and separates the voices into individual sound sources of several trapped personnel. The sound sources are then located separately. For trapped personnel in each area who were not found through steps S100 and S200, instances of trapped personnel are added to the scene model, and the urgency level of each trapped person is determined. S400: The command center displays the scene model of the mountain rail transit station where the emergency has occurred on the command display module, and displays the instances of each trapped person at their predetermined positions on the command display module using the location coordinates of each trapped person; S500: The command center receives rescue route instructions in group a. Each group of rescue route instructions includes multiple rescue routes that instruct rescuers to rescue trapped personnel in one of the a areas of the mountain rail transit station described in S400. S600: The command center displays all the rescue routes described in S500 on the command display module. The command center communicates with all mobile rescue subsystems. When each rescuer arrives at the mountain rail transit station described in S400, the mobile rescue subsystem sends entry feedback to the command center. The command center adds rescuer instances to the scene model. The rescuer instances include the rescuer ID, the area, and the location coordinates. At the same time, the mobile camera module is activated to record video in real time and transmit it to the command center. When the rescuers enter the scene, the command center synchronously communicates the display content of the command display module to the mobile display module of the mobile rescue subsystem. S700: After the rescuers select an area for rescue, if obstacles are found on every rescue path in the area during the rescue, the command center adds obstacle removal markers at the locations of the obstacles on the rescue paths selected by the rescuers after the rescuers report to the command center. The obstacle removal markers are displayed on the command display module and the mobile display modules of all mobile rescue subsystems. After the obstacles are removed, the command center and the mobile rescue subsystem delete the obstacle removal markers, and then S800 is executed. If at least one rescue path in the area is found to be unobstructed during the rescue, S800 is executed directly. S800: In the event of an emergency at a mountain rail transit station, S700 is executed multiple times. After all rescueable trapped personnel at the mountain rail transit station have been rescued and left the station, the command center determines that the rescue is complete.

2. The emergency command and rescue method for mountain rail transit stations according to claim 1, characterized in that, S200 includes: S210: Obtain the number a of mountain rail transit stations where an emergency has occurred; S220: Set i=1; S230: The image recognition array in the i-th region acquires images within the region and sends them to the command center; S240: The command center separates multiple images corresponding to each trapped person from the acquired images, and calculates the image positioning coordinates of b trapped persons by identifying the position of each trapped person in the images captured by multiple fixed camera modules. S250: Compare the image location coordinates of the b trapped persons with the location coordinates of the p trapped persons already existing in the i-th region of the scene model of the mountain rail transit station described in S210. For the trapped persons identified in S240 whose location coordinates do not match the image location coordinates, generate several new trapped person instances. S250 includes: S251: Obtain the number p of existing trapped personnel instances in the i-th region of the scene model of the mountain rail transit station described in S210; S252: Set j=1; S253: Determine whether there is a trapped person instance among the existing p trapped person instances whose distance from the image positioning coordinates of the j-th trapped person identified in S240 is less than the preset error w. If yes, execute S255 directly; otherwise, execute S254 first and then S255. S254: Generate a new instance of a trapped person in the scene model, wherein the location coordinates of the trapped person instance are image location coordinates; S255: Determine if j = b. If yes, S250 ends. If no, j = j + 1, and S253 is executed again. S260: Determine if i = a. If yes, execute S270. If no, i = i + 1, and then execute S230 again. S270: After supplementation, the scene model includes q instances of trapped personnel. The command center calculates the severity level of each trapped person based on their image and adds the severity level to the instance. The severity level visual judgment model categorizes the severity levels of trapped personnel into three levels: unable to move, requiring assistance, and able to move independently. Trapped personnel classified as unable to move are considered injured persons who have fallen to the ground. Trapped personnel classified as requiring assistance are considered elderly, children, and pregnant women. Other trapped personnel are classified as able to move independently. Trapped personnel who can move independently leave on their own after being prompted by the broadcast alarm module. For the other two categories of trapped personnel, rescuers first assist those who require assistance, and then assist those who are unable to move.

3. The emergency command and rescue method for mountain rail transit stations according to claim 1, characterized in that, The S300 includes: S310: Obtain the number of areas c with radio arrays in a mountain rail transit station where an emergency has occurred; S320: Set x=1; S330: The receiver array in the xth region acquires sound segments within the region and sends them to the command center; S340: The command center separates the acquired sound segments to obtain individual sound sources corresponding to several trapped personnel. By identifying each individual sound source, the sound source is located for each trapped person, and the sound location coordinates of d trapped personnel are calculated. S350: Compare the sound location coordinates of the d trapped persons with the location coordinates of the q trapped persons already existing in the x-th region of the scene model of the mountain rail transit station described in S310. For the trapped persons identified in S340 whose location coordinates do not match the sound location coordinates, generate several new trapped person instances. S350 includes: S351: Obtain the number q of existing trapped personnel instances in the x-th region of the scene model of the mountain rail transit station described in S310; S352: Set y=1; S353: Determine whether there is an instance of a trapped person among the existing q instances whose distance from the sound location coordinates of the yth trapped person identified in S340 is less than the preset error w. If yes, execute S355 directly; otherwise, execute S354 first and then S355. S354: Generate a new instance of a trapped person in the scene model, wherein the location coordinates of the trapped person instance are sound location coordinates; S355: Determine if y = d. If yes, S350 ends; otherwise, y = y + 1, and S353 is executed again. S360: Determine if x = c. If yes, execute S370. If no, x = x + 1, and then execute S330 again. S370: After supplementation, the scenario model includes k instances of trapped personnel. The command center calculates the urgency level for each individual sound source corresponding to the (kq) instances of trapped personnel supplemented in S350, and adds the urgency level to the (kq) instances of trapped personnel.

4. The emergency command and rescue method for mountain rail transit stations according to claim 1, characterized in that, When executing S600~S700, the following steps are also executed simultaneously: The command center sends status feedback instructions to the mobile rescue subsystems of all rescue personnel at preset time intervals t, requiring each rescue personnel to send a normal status feedback through the mobile rescue subsystem. One or more rescuers are considered trapped when at least one of the following two situations occurs: the command center receives a trapped information from the mobile rescue subsystem of one or more trapped rescuers, or the command center does not receive a status feedback from the mobile rescue subsystem of one or more rescuers within a predetermined number of time intervals t. In the scenario model of a mountain rail transit station in an emergency, the control host of the command center converts the rescuer instances corresponding to each trapped rescuer into trapped personnel instances, deletes the rescuer IDs, generates trapped personnel IDs, and at the same time, the command center stops sending status feedback commands to the mobile rescue subsystem of the trapped rescuers. For each trapped person instance converted from a rescuer instance, the location of each trapped rescuer is obtained through the location variable contained in the trapped person instance. The image is then acquired through the image recognition array of the area. The image is then sent to the command center by the emergency monitoring subsystem based on image recognition. The control host of the command center locates the trapped rescuer through the location coordinates of the trapped rescuer, separates the image of the trapped rescuer, calculates the urgency level, and adds the urgency level to the trapped person instance.

Images