Urban rail transit service system and method based on cloud-edge-multi-agent

Through the cloud-edge-multi-integrated intelligent structure, combined with AI service robots and small AI computing power platforms, the problem of unreasonable information sharing and personnel scheduling in urban rail transit systems is solved, and unmanned intelligent services and cost savings are achieved. Improve the intelligence level of the system.

CN118898534APending Publication Date: 2024-11-05GUANGZHOU METRO DESIGN & RES INST CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411085786.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-08
Publication Date
2024-11-05

AI Technical Summary

Technical Problem

In the existing urban rail transit system, the integrated monitoring system is separated from passenger services, resulting in idle monitoring management personnel, and the cloud platform's computing power is insufficient, making it impossible to achieve information sharing and intelligent coordination, resulting in insufficient data sharing and unreasonable personnel scheduling. Increased labor and operation and maintenance costs.

Method used

Adopting a cloud-edge-multi-integrated structure, connecting it with the station-level edge cloud through the central cloud platform, configuring AI service robots and small AI computing power platforms to realize information sharing and unmanned intelligent services, and promote customer service and station service personnel. release.

Benefits of technology

The information sharing of all stations in the entire rail transit line has been realized, unmanned intelligent services have been promoted, labor costs have been reduced, and the AI ​​computing power platform has been mobilized to overcome the problem of insufficient computing power, and the system's intelligence level has been improved. Functional diversity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118898534A_ABST
    Figure CN118898534A_ABST
Patent Text Reader

Abstract

The invention provides an urban rail transit service system and method based on cloud-edge-multi-agent. The system comprises a central cloud platform, a station-level edge cloud and an AI service monitoring system. Each station-level edge cloud is provided with a small AI computing power platform, and the small AI computing power platform is used for inputting user demands and behavior information into the server intelligent model for analysis and transmitting an edge analysis result to the corresponding AI service robot; the AI service-oriented robot downloads a server-side intelligent model, if the computing power of the AI service-oriented robot is sufficient, user demands and behavior information are put into the server-side intelligent model for analysis, an intelligent-side analysis result is obtained, and a corresponding action is executed on a target passenger; and if the computing power is insufficient, executing a corresponding action on the target passenger according to an edge analysis result. Compared with the prior art, the system provided by the invention realizes the intelligent service of the urban rail transit, and solves the problems of unreasonable personnel allocation and insufficient computing power of related systems in the prior art.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of information processing technology, and specifically relates to a cloud-edge-multi-agent based urban rail transit service system and method. Background Art

[0002] The Integrated Monitoring and Control System (ISCS) is a highly integrated, comprehensive automated monitoring system that integrates multiple major weak-current systems of urban rail to form a unified monitoring platform and software platform, thereby realizing centralized monitoring and management of major weak-current system equipment of urban rail, such as train operation, power equipment, and electromechanical equipment, and realizing information sharing and coordinated interaction between various weak-current systems of urban rail. The passenger service system generally adopts a manual customer service model. When passengers have any service needs, they can seek services from station staff on site, mainly through verbal dialogue and interaction. The urban cloud platform, referred to as the cloud platform, is a basic cloud platform for comprehensive production and business of the subway, which is built using virtualization technology, distributed storage technology, and cloud resource management technology. It is used to realize the unified allocation, unified management, and centralized monitoring of information computing, storage, and network resources.

[0003] The above systems and platforms have the following defects:

[0004] 1. The integrated monitoring system is centered on electrical and environmental control, which are separated from passenger services and lack the function of taking passenger services into consideration. When some trains stop, monitoring and management personnel may be idle, so it is impossible to organically link them together to achieve the goal of reducing costs and increasing efficiency of urban rail transit.

[0005] 2. The existing customer service system of urban rail transit is based on the front-end self-service terminal of the station + the centralized customer service model in the background. Although it has released the customer service staff of the customer service center to a good extent, manual intervention is still required during the morning and evening rush hours of the station, and it cannot effectively save the customer service staff of the customer service center.

[0006] 3. Most of the existing urban rail city cloud platforms adopt the cloud-edge-end technology architecture, that is, the central cloud platform-edge computing cloud-on-site terminal architecture. The computing power centers are all located in the central cloud platform, and the on-site terminals (fixed terminals) have only a small amount of computing power, insufficient computing power, low degree of intelligence, and relatively single functions, which make it impossible to achieve coordination and information sharing between the comprehensive monitoring system and the customer service system.

[0007] Therefore, the above-mentioned systems and cloud platforms in the prior art each perform their own functions, and multiple systems run in parallel. For urban rail transit, data sharing is insufficient, and the operation, maintenance and station staff scheduling of each independent system are unreasonable. The joint operation of multiple systems will consume more manpower costs and maintenance costs. Summary of the invention

[0008] The present invention provides a cloud-edge-multi-agent based urban rail transit service system and method, which are mainly used to solve the problems of insufficient information sharing and unreasonable personnel scheduling in the prior art.

[0009] In a first aspect, to achieve the above-mentioned purpose, the present invention provides an urban rail transit service system based on cloud-edge-multi-agent, which includes a central cloud platform, a station-level edge cloud and an AI service monitoring system; the central cloud platform is connected to the station-level edge cloud deployed in multiple stations through backbone network communication, and each station-level edge cloud is connected to the AI ​​service monitoring system of the corresponding station through communication;

[0010] in,

[0011] The AI ​​service monitoring system includes at least a plurality of AI service robots; the central cloud platform trains the built-in initial intelligent model based on internal information, monitoring data collected by each station-level edge cloud, external information of the corresponding station of each station-level edge cloud, and user demand and behavior information collected by the AI ​​service robot to obtain a server-side intelligent model, and share it with each station-level edge cloud; each station-level edge cloud is equipped with a small AI computing power platform, which is used to extract user demand and behavior information monitored in real time by the AI ​​service robot; if the computing power of the AI ​​service robot is insufficient, the small AI computing power platform places the user demand and behavior information into the server-side intelligent model for analysis, and transmits the edge analysis results to the corresponding AI service robot; the AI ​​service robot downloads the server-side intelligent model, and if its own computing power is sufficient, it places the user demand and behavior information into the downloaded server-side intelligent model for analysis to obtain the intelligent end analysis results; the AI ​​service robot performs corresponding actions on the target passenger according to the edge analysis results or the intelligent end analysis results;

[0012] The central cloud platform includes the internal management network cloud sub-platform, the production network cloud sub-platform and the external service network cloud sub-platform; the internal network cloud sub-platform is connected to the production network cloud sub-platform by two-way communication, and the internal network cloud sub-platform is used to read and write internal information in real time and share and transmit it to the production network cloud sub-platform; the external service network cloud sub-platform is configured on the Internet terminal or mobile terminal and is connected to the production network cloud sub-platform by two-way communication, and the external service network cloud sub-platform is used to obtain the external information of the corresponding station of each station-level edge cloud (meteorology, fire, public security, flood control, external major emergencies (holidays, emergencies outside the station); the production network cloud sub-platform trains the initial intelligent model based on internal information, monitoring data, external information, user needs and behavior information.

[0013] It needs to be explained that the "cloud" in cloud-edge-multi-agent refers to the central cloud platform, while "edge" refers to the station-level edge cloud, and "multi-agent" refers to AI service robots.

[0014] Through the above technical solution, the system of the present invention adopts the cloud-edge-multi-agent structure to realize information sharing among all stations in the entire rail transit line. On the basis of realizing information sharing, the system of the present invention also configures a small AI computing power platform for intelligent control of AI service robots at each platform to promote unmanned intelligent services, freeing up customer service personnel in the customer service center and station staff patrolling the station hall and platform, thereby overcoming the phenomenon of unreasonable personnel scheduling.

[0015] Preferably, the AI ​​service monitoring system also includes multiple fixed monitoring terminals, which are communicatively connected to the small AI computing power platform of the station to which they belong; the fixed monitoring terminals are used to monitor the patrol blind spots of the AI ​​service robots and forward the blind spot signals to the small AI computing power platform of the station to which they belong; if the small AI computing power platform identifies abnormal passenger behavior from the blind spot signal, it notifies the corresponding AI service robot to go to the corresponding patrol blind spot.

[0016] Preferably, the AI ​​service robot includes a power supply module, a video acquisition module, a sound pickup module, a speaker module, an obstacle avoidance module, an electric drive motion module, an AI computing module, and a humanoid frame equipped with an electric drive multi-axis dual arm; the power supply module and the AI ​​computing module are arranged in the humanoid frame and are electrically connected to the video acquisition module, the sound pickup module, the speaker module, the obstacle avoidance module, the electric drive motion module and the electric drive multi-axis dual arm; the electric drive motion module is arranged on the chassis of the humanoid frame; multiple video acquisition modules are installed on the eyes and neck of the humanoid frame; multiple sound pickup modules are respectively arranged on the neck, torso and bottom of the humanoid frame; multiple speaker modules are respectively arranged on the neck, torso and bottom of the humanoid frame; the obstacle avoidance module is arranged at the bottom of the humanoid frame; the video acquisition module, the sound pickup module, the speaker module, the obstacle avoidance module, the electric drive motion module and the electric drive multi-axis dual arm are connected to the small AI computing power platform of the station to which they belong through AI computing module communication.

[0017] Preferably, the AI ​​service robot also includes a touch screen, a card reader and a banknote entry and exit identification port; the touch screen is installed in the middle of the torso of the human body frame, and the card reader and the banknote entry and exit identification port are installed at the bottom of the torso of the human body frame; the touch screen, the card reader and the banknote entry and exit identification port are communicatively connected to a small AI computing power platform of the station to which they belong.

[0018] Preferably, the user demand and behavior information includes passenger flow information, face recognition information, user abnormal behavior information, text demand information, language demand information, gesture demand information and ticket information; passenger flow information, user abnormal behavior information, gesture demand information and face recognition information are obtained by a video acquisition module; text demand information is obtained by a touch screen; language demand information is obtained by a sound pickup module; ticket information is obtained by a card reader and a banknote entry and exit identification port.

[0019] Preferably, the production network cloud sub-platform includes at least a big data platform and an AI computing power platform configured with a language big model and a visual big model; the AI ​​computing power platform trains the language big model and the visual big model based on internal information, monitoring data, external information, and user needs and behavior information to obtain a server-side intelligent model; when the computing power of the small AI computing power platform at any station is insufficient, the AI ​​computing power platform extracts the user needs and behavior information monitored in real time by the AI ​​service robot in the corresponding station with insufficient computing power, and uses the server-side intelligent model to analyze the user needs and behavior information, and finally forwards the analysis results of the total platform to the corresponding AI service robot through the corresponding small AI computing power platform; the big data platform is used to classify internal information, external information, monitoring data, user needs and behavior information, and the analysis results of the total platform, and place them in the template theme library, and push the template theme library to the Internet terminal or mobile terminal through the API interface.

[0020] It needs to be explained that the small AI computing power platform is also used to conduct in-depth training of the server-side intelligent models downloaded by the AI ​​service robot, and adjust the model parameters within the AI ​​service robot to enable it to better complete scene recognition such as voice recognition, video recognition, and security prevention.

[0021] Preferably, the internal information includes at least the personnel scheduling plan, operation plan arrangement, operation timetable and resource development plan of each station.

[0022] Preferably, the external information includes at least weather information, fire alarm information, public security information, flood prevention information, holiday information and off-site emergency information.

[0023] Preferably, the monitoring data includes at least environmental data of each station, operating data of all electromechanical equipment, energy consumption data, power data, fire data, video analysis data, passenger information data, ticket sales and inspection data, passenger flow data, broadcast data and security inspection data.

[0024] In a second aspect, in order to achieve the above-mentioned purpose, the present invention further provides a cloud-edge-multi-agent-based urban rail transit service method, which is built into the above-mentioned cloud-edge-multi-agent-based urban rail transit service system. The method of the present invention includes:

[0025] Parallel acquisition of internal information of the internal management network cloud sub-platform, monitoring data collected by each station-level edge cloud, and external information of the corresponding station of each station-level edge cloud;

[0026] Incorporate internal information, monitoring data, and external information into the big data platform;

[0027] Based on internal information, external information, monitoring data, and user needs and behavior information, the initial intelligent model is trained to obtain a server-side intelligent model;

[0028] When the computing power of the small AI computing power platform at any station is insufficient, the AI ​​computing power master platform is used to extract the user demand and behavior information monitored in real time by the AI ​​service robot in the corresponding station with insufficient computing power, and the server-side intelligent model is used to analyze the user demand and behavior information. Finally, the analysis results of the master platform are forwarded to the corresponding AI service robot through the corresponding small AI computing power platform;

[0029] A big data platform is used to classify internal information, external information, monitoring data, user needs and behavior information, and overall platform analysis results, and then placed in a template theme library, and the template theme library is pushed to Internet terminals or mobile terminals through an API interface.

[0030] The present invention has at least the following advantages:

[0031] 1) Compared with the existing technology, the system of the present invention adopts the cloud-edge-multi-agent structure to realize information sharing among all stations in the entire rail transit line. On the basis of realizing information sharing, the system of the present invention also configures a small AI computing platform for intelligent control of AI service robots at each platform to promote unmanned intelligent services, freeing up customer service personnel in the customer service center and station staff patrolling the station hall and platform, thereby overcoming the phenomenon of unreasonable personnel scheduling.

[0032] 2) Compared with the existing technology, the system of the present invention is equipped with an AI service robot, which is suitable for the complex service environment in urban rail transit. It has functions such as video analysis, behavior recognition and voice recognition, and can execute instructions obtained by the small AI computing platform through data training and conversion, so as to transform the traditional manual mode into an unmanned intelligent service mode. In addition, by sinking the station-level edge cloud to the intelligent AI terminal, the burden on the total AI computing platform can be reduced.

[0033] 3) Compared with the prior art, the method matching the system of the present invention can timely mobilize the AI ​​computing power platform to perform information analysis when the computing power of the small AI computing power platform is insufficient, so as to overcome the problem of insufficient computing power in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 This is a schematic diagram of the structure of the urban rail transit service system based on cloud-edge-multi-agent in Embodiment 1 of the present invention;

[0036] Figure 2 This is a schematic diagram of the structure of the AI ​​service robot in the second embodiment of the present invention;

[0037] Figure 3 This is a flow chart of the urban rail transit service method based on cloud-edge-multi-agent in Example 3 of the present invention.

[0038] Figure numerals: 1. Eye camera; 2. Neck camera; 3. Rotating bearing; 4. Neck sound pickup port; 5. Neck speaker; 6. Electric drive multi-axis dual arm; 7. Body speaker; 8. Body sound pickup port; 9. Touch screen; 10. Bottom speaker; 11. Obstacle avoidance module; 12. Electric drive motion module; 13. Card reader; 14. Banknote entry and exit identification port. DETAILED DESCRIPTION

[0039] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] In the description of the present invention, it should be noted that the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside", etc. indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they cannot be understood as limitations on the present invention.

[0041] This embodiment discloses a cloud-edge-multi-agent based urban rail transit service system and method, which are mainly used to solve the problems of insufficient information sharing and unreasonable personnel scheduling in the prior art.

[0042] Embodiment 1

[0043] like Figure 1 As shown, this embodiment discloses a cloud-edge-multi-agent based urban rail transit service system, which includes a central cloud platform, a station-level edge cloud and an AI service monitoring system.

[0044] Among them, the central cloud platform is connected to the station-level edge cloud deployed in multiple stations through backbone network communication, and each station-level edge cloud is connected to the AI ​​service monitoring system of the corresponding station. The AI ​​service monitoring system includes multiple AI service robots. Based on internal information, monitoring data collected by each station-level edge cloud, external information of the corresponding station of each station-level edge cloud, and user needs and behavior information collected by the AI ​​service robot, the central cloud platform trains the built-in initial intelligent model to obtain the server-side intelligent model and share it with each station-level edge cloud. Each station-level edge cloud is equipped with a small AI computing power platform, which is used to extract user needs and behavior information monitored in real time by the AI ​​service robot, and put the user needs and behavior information into the server-side intelligent model for analysis, and finally transmit the edge analysis results to the corresponding AI service robot.

[0045] It should be noted that the small AI computing power platform is also used to conduct in-depth training and model updates on the server-side intelligent models downloaded by the AI ​​service robots, and to adjust the model parameters within the AI ​​service robots so that they can better complete scene recognition such as voice recognition, video recognition, and security prevention.

[0046] The AI ​​service robot has downloaded a server-side intelligent model. If its own computing power is sufficient, it will place user needs and behavior information into the downloaded server-side intelligent model for analysis, obtain the intelligent-side analysis results, and then perform corresponding actions on the target passengers based on the edge analysis results or the intelligent-side analysis results.

[0047] The above-mentioned AI service robots have sufficient computing power and execute corresponding actions based on the analysis results of the intelligent end. If the computing power is insufficient, they execute corresponding actions based on the edge analysis results.

[0048] Compared with the existing technology, the system of the present invention adopts a cloud-edge-multi-agent structure to realize information sharing among all stations in the entire rail transit line. The edge clouds at each station level can be managed uniformly by the central cloud platform according to actual conditions or can be managed in a flat manner by themselves, so as to facilitate cost reduction, efficiency improvement and reasonable allocation of rail operation manpower. On the basis of the above, the system of the present invention also has a small AI computing power platform at each platform and an AI service robot controlled by the small AI computing power platform. The AI ​​service robot can serve passengers according to user needs and behavior information, thereby promoting unmanned intelligent service methods, effectively releasing the customer service personnel of the customer service center and the station staff patrolling the station hall and platform, and overcoming the phenomenon of unreasonable personnel scheduling.

[0049] It should be explained that during peak passenger flow periods, the AI ​​service monitoring system also includes a small number of mobile tablet terminals, which are held by station staff to assist station staff in serving passengers in the station hall or platform. During off-peak passenger flow periods, AI service robots alone are sufficient to meet passenger needs.

[0050] In this embodiment, the user demand and behavior information includes passenger flow information, face recognition information, user abnormal behavior information, text demand information, language demand information, gesture demand information and ticket information. The internal information at least includes the personnel scheduling plan, operation plan arrangement, operation timetable and resource development plan of each platform. The external information at least includes meteorological information, fire alarm information, public security information, flood prevention information, holiday information and emergency information outside the station. The monitoring data at least includes the environmental data of each station, all electromechanical equipment operation data, energy consumption data, power data, fire data, video analysis data, passenger information data, ticket sales and ticket inspection data, passenger flow data, broadcast data and security inspection data.

[0051] Furthermore, the central cloud platform includes an internal management network cloud sub-platform, a production network cloud sub-platform and an external service network cloud sub-platform.

[0052] Specifically, the internal network cloud sub-platform is bidirectionally connected to the production network cloud sub-platform for real-time reading and writing of internal information and sharing and transmitting to the production network cloud sub-platform. The external service network cloud sub-platform is configured on an Internet terminal or a mobile terminal and is bidirectionally connected to the production network cloud sub-platform for obtaining external information of the corresponding station of each station-level edge cloud. The production network cloud sub-platform includes at least a big data platform and an AI computing power platform. The AI ​​computing power platform and all small AI computing power platforms are deployed with language large models and visual large models. The above-mentioned language large model can adopt the existing Chat-GPT4, Wenxin Yiyan or BERT and other existing language large models, while the visual large model can adopt the existing AlexNet, VGGNet or GoogLeNet visual large model. The above-mentioned language large model and visual large model are both artificial intelligence models with multi-end input and multi-end output. The above-mentioned visual large model can be combined with the existing action recognition system to identify whether the passenger has abnormal behavior (special channel intrusion, passenger fall, barrier handover, passenger clearance detection, long queue, fight or left behind, etc.). The AI ​​computing power platform trains the language model and the visual model based on internal information, monitoring data, external information, and user needs and behavior information to obtain the server-side intelligent model. In order to overcome the problem of insufficient computing power in the prior art, when the computing power of the small AI computing power platform at any station is insufficient, the AI ​​computing power platform extracts the user needs and behavior information monitored in real time by the AI ​​service robot in the corresponding station with insufficient computing power, and uses the server-side intelligent model to analyze the user needs and behavior information, and finally forwards the analysis results of the total platform to the corresponding AI service robot through the corresponding small AI computing power platform. The above internal information, external information, monitoring data, user needs and behavior information, and the analysis results of the total platform will be classified by the big data platform and placed in the template theme library, and the template theme library will be pushed to the Internet terminal (webpage) or mobile terminal (APP) through the API wireless interface.

[0053] It should be noted that the production network cloud sub-platform is equipped with a basic database for storing all the above information and data. Similarly, the station-level edge cloud should also be equipped with a station-level database for storing the data and information obtained by monitoring the corresponding station. The following is an economic benefit analysis of the system of the present invention:

[0054] The staffing of a single station is generally: 1 station duty officer, 1 line duty officer, 1 passenger duty officer, 1 customer service officer, 1 platform officer, and 1 hall patrol officer, a total of 6 people, of which 5 shifts and 3 operations for station duty and line duty, 4 shifts and 2 operations for passenger duty and customer service, 3 shifts and 2 operations for platform, and 2 shifts and 2 operations for hall patrol. That is, a single station can generally have about 23 people.

[0055] In the single-station mode, AI service robots are responsible for passenger service and inspections. One service robot is deployed in the station hall and platform respectively. The original hall patrol, platform patrol, customer service, and customer service can be combined into one comprehensive service patrol post. This post uses a handheld mobile terminal to patrol the station hall and platform and provide passenger service, and uses the handheld terminal to work with the AI ​​service robot to achieve remote passenger assistance and station management and disposal. This post adopts a 4-shift 2-operation or 3-shift 2-operation mode. This method can save (4 people for customer service, 3 people for the platform, and 2 people for hall patrol), and a single station can save about 9-10 people. Calculated based on the saving of 9 people, the labor cost is 100,000 yuan / person, and the annual salary increase is 5%:

[0056] 9*10+9*10(1+5%)+……+9*10(1+5%)^9≈11.3196 million yuan.

[0057] That is to say, the average annual savings is about 1.13 million yuan per station.

[0058] The comprehensive benefits are as follows:

[0059] The labor cost of each station can be saved by about 1.13 million yuan per year. Based on 20 standard stations per line, the labor cost of each line can be saved by about 22.6 million yuan per year. As the scale of the line network increases, the labor cost will be saved more. If the regional centralized station management mode is adopted, the positions can be further optimized and the labor cost can be saved.

[0060] Embodiment 2

[0061] like Figure 2 As described above, this embodiment discloses an AI service robot, which is used in conjunction with the cloud-edge-multi-agent based urban rail transit service system in Embodiment 1.

[0062] The AI ​​service robot includes a power supply module, a video acquisition module, a sound pickup module, a speaker module, an obstacle avoidance module 11, an electric drive motion module 12, an AI computing module, a touch screen 9, a card reader 13, a banknote entry and exit identification port 14 and a humanoid frame equipped with an electric drive multi-axis dual arm 6.

[0063] Specifically, the power supply module and the AI ​​computing module are arranged in the human-like frame and are electrically connected to the video acquisition module, the sound pickup module, the speaker module, the obstacle avoidance module 11, the electric drive motion module 12 and the electric drive multi-axis dual arm 6. The power supply module adopts a fast-charging new energy lithium battery, which is equipped with a fast charging pile to enhance the endurance of the AI ​​service robot. The AI ​​computing module can be a single-chip microcomputer, which is used to receive the monitoring information or data of each module and serve as an information transfer element and a control element. The AI ​​computing module is connected to the small AI computing power platform of the station to which it belongs by wireless communication to forward the received and collected user needs and behavior information to the small AI computing power platform of the station to which it belongs. In order to further improve the real-time performance of data processing and reduce the data processing burden of the central cloud platform, the AI ​​computing module in the AI ​​service robot can also deploy a large language model and a large visual model, and the information or data that exceeds the computing power of the AI ​​computing module is processed by the small AI computing power platform of the station to which it belongs.

[0064] In this embodiment, the electric drive motion module 12 can use an electric drive universal wheel or an electric drive crawler. The electric drive motion module 12 is set on the chassis of the human body frame, and the electric drive motion module 12 is controlled by the AI ​​computing module to drive the AI ​​service robot to move. In this embodiment, the video acquisition module includes an eye camera 1 and a neck camera 2. The eye camera 1 is used to identify abnormal behavior information of passengers for urban rail transit safety prevention, including: special channel intrusion, passenger fall, barrier delivery, illegal gate, passenger clearance detection, long queue, fight, left behind items and passenger flow density detection. The neck camera 2 is used for passenger identification and video calls between passengers and backstage centralized customer service and station station handheld terminals. It can perform remote rescue services, and can also perform QR code identification and scanning. The sound pickup module includes a neck sound pickup port 4, a body sound pickup port 8 and a bottom sound pickup port. The neck sound pickup port 4 is used to receive voice for taller passengers, the body sound pickup port 8 is used to serve shorter passengers, and the bottom sound pickup port is used to receive voice information of passengers who fall. The speaker module includes a neck speaker 5, a body speaker 7 and a bottom speaker 10. The neck speaker 5 outputs voice for tall passengers and has a built-in power amplifier. The body speaker 7 is used to serve shorter passengers, and the bottom speaker 10 is used to serve passengers who fall. The obstacle avoidance module 11 is installed at the bottom of the humanoid frame. In this embodiment, the obstacle avoidance module 11 is a radar, which is used for the AI ​​service robot to detect the distance from the obstacle when moving to achieve the effect of obstacle avoidance and collision avoidance. The electric drive multi-axis double arm 6 can swing up and down and left and right for passenger guidance. Fixed gesture programming can be performed as a specific passenger guidance purpose. The touch screen 9 is used to provide passengers with self-service services such as route inquiries, surrounding service inquiries, operation information inquiries, ticket card inquiries, ticket card supplements, etc. The card reader 13 is used to read the passenger's boarding card, and is used for self-service operations such as querying the balance in the card, entry and exit time, and ticket supplements. The banknote entry and exit identification port 14 can be used to insert banknotes to achieve recharge, change, and ticket supplements.

[0065] Preferably, the humanoid frame neck of the AI ​​service robot is also equipped with an electric rotating bearing 3 to improve the monitoring range of its eye camera 1. The electric rotating bearing 3 is used as a camera ball machine and can be controlled by the station staff's handheld mobile terminal, the train control room and the fixed terminal in the dispatching hall of the control center. The rotation angle of the electric rotating bearing 3 is 360 degrees horizontally and 90 degrees in pitch.

[0066] The following is the working principle of the AI ​​service robot combined with the system of the present invention in different environments:

[0067] 1) In the peak passenger flow mode, the AI ​​service robot stands at a fixed point next to the customer service center in the station hall and in the middle of the platform as a self-service terminal. At this time, the AI ​​service robot turns on the voice recognition and self-service customer service functions. The touch screen 9 of the AI ​​service robot displays passenger service-related content and information such as the station environment, operation services, and entrances and exits. For large passenger flow stations, when the AI ​​service robot is fixed or moved in a small range within the entrances and exits of the large passenger flow that needs to be guided to guide passengers and take into account passenger self-service, if the passenger only needs to buy a ticket or make up a ticket, the AI ​​service robot will interact with the passenger through voice and video, reply to the customer service, query information and recognize the face, and finally process the ticket. If it is other needs other than buying or making up a ticket, and it cannot be solved through replies and action guidance, it will be reported to the station staff for processing. For example, if a passenger encounters difficulties in service or dialects and other events that cannot be recognized, he can click the rescue button of the AI ​​service robot to conduct a video connection with the train control room, the centralized customer service background or the mobile handheld tablet for rescue services. The station staff of the handheld mobile terminal can provide on-site services according to the location indicated by the tablet through the rescue needs on the tablet. When the station staff is handling other matters, they can call on the operating personnel in the vehicle control room to arrive at the scene, thereby achieving reasonable scheduling of personnel and avoiding idle personnel.

[0068] 2) In the off-peak passenger flow mode, the AI ​​service robot patrols in a circular manner in the station hall or platform, activates security patrol and passenger service functions, identifies abnormal situations in the station (fire, dangerous behavior, passenger injuries, etc.), and alerts the background and mobile platform after identifying the abnormality, and the on-site mobile station service or train control room operator handles the situation on the spot. When the voice and behavior recognize that a passenger is approaching and needs service, the AI ​​service robot stops and guides the passenger to solve the problem by voice. If it cannot be handled, the passenger can be prompted to connect to the background for help. If some passengers inquire about the station toilet, carriage location and other information, the AI ​​service robot can present the location information in real time on the touch screen 9.

[0069] 3) After the station operation is over, the AI ​​service robot will patrol the station hall and platform in a circular manner and start the security patrol. It will identify abnormal situations in the station (fire, flood control, intrusion, etc.), and after identifying the abnormality, it will alert the background and mobile platform, and the on-site mobile station staff or the operator in the train control room will deal with it on the spot. Return to the charging pile for charging according to the scheduled time.

[0070] 4) When the station staff does not hold the mobile terminal, they can switch the mode to the administrator mode on the touch screen 9 of the AI ​​service robot. In the administrator mode, all station management functions in the mobile tablet terminal can be realized.

[0071] 5) When the AI ​​service robot patrols and finds a passenger falling, it will drive to the fallen passenger, comfort the passenger through the bottom speaker 10, and synchronize information to the vehicle control room and the mobile tablet terminal for voice warning, urging relevant personnel to deal with passenger injuries and other incidents. At the same time, the nearby passenger service screen and broadcast are switched to guide the passengers, so that other passengers stay away from the scene, which is convenient for efficient safety handling.

[0072] 6) Flood emergency response mode: obtain rainfall information through the meteorological department or the Internet, and the system of the present invention notifies the station in advance to deploy flood prevention materials. The entrance and exit cameras analyze the height of the water level outside the station in real time through the AI ​​platform deployed at the station level. If it is too high, the broadcast is controlled in advance to guide passengers and deploy flood prevention at the entrance and exit. At this time, the AI ​​service robot can identify water leakage, seepage and rainwater backflow in the station through video analysis during the inspection, and promptly notify relevant personnel to check.

[0073] 7) Disaster prevention and emergency response mode: When the AI ​​service robot finds a fire source or a small amount of smoke in the early stage of smoldering, and the automatic fire alarm system in the station does not sound an alarm, it will send an alarm message to the mobile tablet terminal and the fixed terminal in the vehicle control room in advance, and project the camera image to the mobile tablet terminal and the fixed terminal in the vehicle control room so that relevant personnel can confirm the fire information in time.

[0074] Embodiment 3

[0075] like Figure 3 As shown, based on the first and second embodiments, this embodiment discloses a cloud-edge-multi-agent-based urban rail transit service method, which includes:

[0076] S1: Parallel acquisition of internal information of the internal management network cloud sub-platform, monitoring data collected by each station-level edge cloud, and external information of the corresponding station of each station-level edge cloud;

[0077] S2: Incorporate internal information, monitoring data, and external information into the big data platform;

[0078] S3: Based on internal information, external information, monitoring data, and user needs and behavior information, the initial intelligent model is trained to obtain a server-side intelligent model;

[0079] S4: When the computing power of the small AI computing power platform at any station is insufficient, the AI ​​computing power master platform is used to extract the user demand and behavior information monitored in real time by the AI ​​service robot in the corresponding station with insufficient computing power, and the server-side intelligent model is used to analyze the user demand and behavior information. Finally, the analysis results of the master platform are forwarded to the corresponding AI service robot through the corresponding small AI computing power platform;

[0080] S5: Use the big data platform to classify internal information, external information, monitoring data, user needs and behavior information, and overall platform analysis results, and place them in the template theme library, and push the template theme library to the Internet terminal or mobile terminal through the API interface.

[0081] It should be explained that the computing power mentioned in the embodiment refers to the CPU file processing rate. In the prior art, whether the computing power of the corresponding computing power platform is sufficient is often judged by setting a lower limit of the processing rate.

[0082] Compared with the existing technology, the above method can mobilize the AI ​​computing power platform in time to perform data training when the computing power of the small AI computing power platform is insufficient, so as to overcome the problem of insufficient computing power in the existing technology.

[0083] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above-mentioned method embodiment; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, etc. Various media that can store program codes.

[0084] For those skilled in the art, various other corresponding changes and deformations can be made according to the technical solutions and concepts described above, and all of these changes and deformations should fall within the protection scope of the claims of the present invention.

Claims

1. An urban rail transit service system based on cloud-edge-multi-agent, It is characterized in that It includes a central cloud platform, a station-level edge cloud and an AI service monitoring system; the central cloud platform is connected to the station-level edge cloud deployed at multiple stations through backbone network communication, and each station-level edge cloud is connected to the AI ​​service monitoring system of the corresponding station through communication; in, The AI ​​service monitoring system includes at least a plurality of AI service robots; the central cloud platform trains the built-in initial intelligent model based on internal information, monitoring data collected by the station-level edge clouds, external information of the stations corresponding to the station-level edge clouds, and user needs and behavior information collected by the AI ​​service robots to obtain a server-side intelligent model, and shares it with each station-level edge cloud; each station-level edge cloud is configured with a small AI computing power platform, which is used to extract user needs and behavior information monitored in real time by the AI ​​service robot; if the computing power of the AI ​​service robot is insufficient, the small AI computing power platform places the user needs and behavior information into the server-side intelligent model for analysis, and transmits the edge analysis results to the corresponding AI service robot; the AI ​​service robot downloads the server-side intelligent model, and if its own computing power is sufficient, places the user needs and behavior information into the downloaded server-side intelligent model for analysis to obtain the intelligent end analysis results; the AI ​​service robot performs corresponding actions on the target passenger according to the edge analysis results or the intelligent end analysis results; The central cloud platform includes an internal management network cloud sub-platform, a production network cloud sub-platform and an external service network cloud sub-platform; the internal network cloud sub-platform is connected to the production network cloud sub-platform by two-way communication, and the internal network cloud sub-platform is used to read and write the internal information in real time and share and transmit it to the production network cloud sub-platform; the external service network cloud sub-platform is configured on an Internet terminal or a mobile terminal and is connected to the production network cloud sub-platform by two-way communication, and the external service network cloud sub-platform is used to obtain external information of the station corresponding to each station-level edge cloud; the production network cloud sub-platform trains the initial intelligent model based on the internal information, the monitoring data, the external information and the user needs and behavior information.

2. The urban rail transit service system based on cloud-edge-multi-agent according to claim 1, It is characterized in that The AI ​​service monitoring system also includes multiple fixed monitoring terminals, which are communicatively connected to the small AI computing power platform of the station to which they belong; the fixed monitoring terminals are used to monitor the patrol blind spots of the AI ​​service robots and forward the blind spot signals to the small AI computing power platform of the station to which they belong; if the small AI computing power platform identifies abnormal passenger behavior from the blind spot signals, it notifies the corresponding AI service robot to go to the corresponding patrol blind spot.

3. The urban rail transit service system based on cloud-edge-multi-agent according to claim 1, It is characterized in that The AI ​​service robot comprises a power supply module, a video acquisition module, a sound pickup module, a speaker module, an obstacle avoidance module (11), an electric drive motion module (12), an AI computing module, and a humanoid frame equipped with an electric drive multi-axis dual arm (6); the power supply module and the AI ​​computing module are arranged in the humanoid frame and are electrically connected to the video acquisition module, the sound pickup module, the speaker module, the obstacle avoidance module (11), the electric drive motion module (12), and the electric drive multi-axis dual arm (6); the electric drive motion module (12) is arranged on the chassis of the humanoid frame; A plurality of video acquisition modules are installed on the eyes and neck of the humanoid frame; A plurality of sound pickup modules are respectively arranged on the neck, torso and bottom of the human-like frame; A plurality of speaker modules are respectively arranged at the neck, torso and bottom of the humanoid frame; the obstacle avoidance module (11) is arranged at the bottom of the humanoid frame; the video acquisition module, the sound pickup module, the speaker module, the obstacle avoidance module (11), the electric drive motion module (12) and the electric drive multi-axis dual arm (6) are connected to a small AI computing platform of the station to which they belong via the AI ​​computing module communication.

4. The urban rail transit service system based on cloud-edge-multi-agent according to claim 3, It is characterized in that The AI ​​service robot further comprises a touch screen (9), a card reader (13) and a banknote entry and exit identification port (14); the touch screen (9) is mounted on the middle part of the trunk of the human-like frame, and the card reader (13) and the banknote entry and exit identification port (14) are mounted on the bottom of the trunk of the human-like frame; the touch screen (9), the card reader (13) and the banknote entry and exit identification port (14) are communicatively connected to a small AI computing platform of the station to which they belong.

5. The urban rail transit service system based on cloud-edge-multi-agent according to claim 4, It is characterized in that The user demand and behavior information includes passenger flow information, face recognition information, user abnormal behavior information, text demand information, language demand information, gesture demand information and ticket information; the passenger flow information, user abnormal behavior information, gesture demand information and face recognition information are acquired by the video acquisition module; the text demand information is acquired by the touch screen (9); the language demand information is acquired by the sound pickup module; and the ticket information is acquired by the card reader (13) and the banknote entry and exit identification port (14).

6. The urban rail transit service system based on cloud-edge-multi-agent according to claim 1, It is characterized in that The production network cloud sub-platform at least includes a big data platform and an AI computing power platform configured with a language big model and a visual big model; the AI ​​computing power platform trains the language big model and the visual big model based on the internal information, the monitoring data, the external information and the user needs and behavior information to obtain a server-side intelligent model; when the computing power of the small AI computing power platform at any station is insufficient, the AI ​​computing power platform extracts the user needs and behavior information monitored in real time by the AI ​​service robot in the corresponding station with insufficient computing power, and uses the server-side intelligent model to analyze the user needs and behavior information, and finally forwards the analysis results of the total platform to the corresponding AI service robot through the corresponding small AI computing power platform; the big data platform is used to classify the internal information, the external information, the monitoring data, the user needs and behavior information, and the analysis results of the total platform, and place them in a template theme library, and push the template theme library to an Internet terminal or a mobile terminal through an API interface.

7. The urban rail transit service system based on cloud-edge-multi-agent according to claim 1, It is characterized in that The internal information at least includes the personnel scheduling plan, operation plan arrangement, operation timetable and resource development plan of each platform.

8. The urban rail transit service system based on cloud-edge-multi-agent according to claim 1, It is characterized in that The external information includes at least weather information, fire alarm information, public security information, flood prevention information, holiday information and off-site emergency information.

9. The urban rail transit service system based on cloud-edge-multi-agent according to claim 1, It is characterized in that The monitoring data includes at least environmental data of each station, operation data of all electromechanical equipment, energy consumption data, power data, fire data, video analysis data, passenger information data, ticket sales and inspection data, passenger flow data, broadcast data and security inspection data.

10. A cloud-edge-multi-agent-based urban rail transit service method. It is characterized in that Built into the cloud-edge-multi-agent based urban rail transit service system according to any one of claims 1 to 9, the method comprises: Parallel acquisition of internal information of the internal management network cloud sub-platform, monitoring data collected by each station-level edge cloud, and external information of the corresponding station of each station-level edge cloud; Incorporating the internal information, the monitoring data, and the external information into a big data platform; Based on the internal information, the external information, the monitoring data and the user needs and behavior information, the initial intelligent model is trained to obtain a server-side intelligent model; When the computing power of the small AI computing power platform at any station is insufficient, the AI ​​computing power master platform is used to extract the user demand and behavior information monitored in real time by the AI ​​service robot in the corresponding station with insufficient computing power, and the server-side intelligent model is used to analyze the user demand and behavior information, and finally the analysis results of the master platform are forwarded to the corresponding AI service robot through the corresponding small AI computing power platform; The big data platform is used to classify the internal information, the external information, the monitoring data, the user needs and behavior information, and the overall platform analysis results and place them in a template theme library, and the template theme library is pushed to an Internet terminal or a mobile terminal through an API interface.