CN-122027831-A - Rail transit visualization system integration system based on multi-agent cooperation and integration method thereof

CN122027831ACN 122027831 ACN122027831 ACN 122027831ACN-122027831-A

Abstract

The invention belongs to the technical field of information system integration and intelligent transportation, and relates to a rail transit visual system integration system based on multi-agent cooperation and an integration method thereof, wherein the rail transit visual system integration system comprises an agent community, an agent outer shell layer and a subsystem layer; the subsystem layer comprises a plurality of subsystems, an intelligent body shell is built for each subsystem to form a subsystem intelligent body, the intelligent body shell layer is formed, the subsystem intelligent bodies are communicated through a message middleware, and the intelligent body community is formed. By constructing the intelligent body shell for each heterogeneous subsystem and establishing the intelligent body community, the invention opens up a data circulation barrier, realizes automatic association analysis of multi-source data, and operation and maintenance personnel do not need to repeatedly switch and manually compare the data among a plurality of independent systems, so that the fault positioning time is greatly shortened, and the operation and maintenance response speed and the processing accuracy are fundamentally improved.

Inventors

ZOU FANG
LI XUEQI
YU DONGXU
ZHANG WEIHONG
TANG XIAO
LI LANXIN

Assignees

通号通信信息集团有限公司
北京国铁华晨通信科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260114

Claims (10)

1. A rail transit visualization system based on multi-agent cooperation is characterized by comprising an agent community, an agent outer shell layer and a subsystem layer; The subsystem layer comprises a plurality of subsystems, an intelligent body shell is built for each subsystem to form a subsystem intelligent body, the intelligent body shell layer is formed, the subsystem intelligent bodies are communicated through a message middleware, and the intelligent body community is formed.
2. The multi-agent collaboration-based rail transit visualization system of claim 1, wherein the subsystem comprises a video cloud platform system, a video cloud storage system, a video device network management system, and a video quality diagnostic system.
3. The multi-agent collaboration-based rail transit visualization system of claim 1, wherein the agent shell receives the task request message from the agent community and converts the task request message into a call operation password of an application program interface of a corresponding subsystem, wherein the agent shell encapsulates a response result of the subsystem, produces a standardized response result, and returns the standardized response result to the agent community, and wherein the agent shell simultaneously detects internal events in the subsystem and issues the internal events as notification messages to the agent community.
4. The rail transit visualization system based on multi-agent cooperation according to claim 1, wherein the message middleware needs to provide a deterministic delay or a bounded maximum delay communication guarantee and supports a priority or deadline-based message scheduling policy so as to meet the real-time transmission requirement of key operation and maintenance instructions and alarm information, and adopts a unified message format, and all subsystem agents perform asynchronous and decoupled communication through a subscription/release mode.
5. The multi-agent collaboration-based rail transit visualization system of claim 4, wherein the message format of the message middleware comprises at least a sender, a receiver, a communication primitive, a content load, and a session ID.
6. The multi-agent collaboration-based rail transit visualization system of claim 1, wherein the agent community comprises an interface agent, a decision recommendation agent, and a predictive maintenance agent; the interface agent is used for receiving user instructions and presenting response results in the unified monitoring view; The decision recommending agent is used for carrying out relevance analysis and reasoning on various events and state messages in the agent community through a predefined rule and a machine learning model; The predictive maintenance agent is used for actively predicting the service life decline trend and potential fault risk of the equipment and issuing a predictive maintenance notice in advance.
7. The multi-agent collaborative rail transit visualization system of claim 6, wherein the interface agents provide a customized visual monitoring interface that supports a user to freely configure their monitored content and layout of interest, receive operational instructions from the user via the visual monitoring interface, decompose the operational instructions into one or more subtasks, receive and fuse analysis results and content recommendations from the decision recommendation agent and predictive maintenance agent, and present the recommended content to the user, and set different real-time rendering priorities for all received agent messages, and immediately present the emergency alert messages sent by the decision recommendation agent or predictive maintenance agent to the user using a highest priority approach.
8. The multi-agent collaborative based rail transit visualization system of claim 6, wherein the decision recommendation agent subscribes to topics related to device status, video quality and storage capacity in message middleware, continuously monitors events and status change messages in an agent community, fuses and associates reasoning received multi-source information based on a predefined reasoning rule set and/or machine learning model, evaluates the urgency of an event corresponding to a failure or optimization opportunity when the potential failure or optimization opportunity is identified, sets decision deadlines for the event, generates an alarm message containing decision suggestions, and sends the alarm message to an interface agent for presentation to a user, the decision recommendation agent collects and analyzes historical operation records of the user, focused monitoring data and decision preference of a treatment alarm, establishes a user behavior portrait model by utilizing a machine learning algorithm, and actively calculates and recommends the information combination and content priority of the current best priority presentation to the interface agent based on the user behavior portrait model and a real-time system status.
9. The multi-agent collaboration-based rail transit visualization system of claim 6, wherein the predictive maintenance agent continuously collects multi-source historical and real-time data from message middleware, cleans, aligns and formats the collected data, stores and manages the processed data, trains an equipment health model through a predictive algorithm based on the stored data, periodically updates the equipment health model, provides a clear time window for prediction on life decay trend, residual usable life and potential failure occurrence probability of equipment through the equipment health model, automatically generates protection specific maintenance advice according to the health status obtained by the equipment health model, distributes the protection specific maintenance advice to an interface agent, presents the interface agent to a user or recommends the agent to take into consideration by decision, collects the subsequent processing result of maintenance advice and actual equipment status data, and returns the subsequent processing result and actual equipment status data as feedback signals to a training data set for continuously optimizing and updating the equipment health model.
10. An integration method of a rail transit visualization system based on multi-agent cooperation, which is used for the rail transit visualization system based on multi-agent cooperation as claimed in any one of claims 1 to 9, comprising the following steps: The method comprises the steps of decomposing an operation instruction into one or more subtasks through the operation instruction sent by an agent community user, generating a standard task request message according to the subtasks, issuing the task request message to a corresponding subsystem agent through a message middleware, sending a response message processed by the subsystem agent to the agent community, integrating through the agent community, and visually displaying an integrated result.

Description

Rail transit visualization system integration system based on multi-agent cooperation and integration method thereof Technical Field The invention relates to a rail transit visual system integration system based on multi-agent cooperation and an integration method thereof, belonging to the technical field of information system integration and intelligent transportation. Background The rail transit system is used as an aorta for urban traffic, and the safe, stable and efficient operation of the rail transit system is important. Under the background, the video monitoring and analyzing system forms a core infrastructure for operation, maintenance and safety management, and mainly comprises four key subsystems, wherein the video cloud platform is responsible for accessing and distributing massive video streams, the video cloud storage system ensures high-reliability storage and quick retrieval of historical video data, the video equipment network management system realizes full life cycle monitoring and state maintenance of a front-end camera, a rear-end server and network equipment, and the video quality diagnosis system automatically carries out intelligent detection and alarm on definition, consistency and the like of video pictures. The four systems have their own functions, and provide an indispensable technical support for real-time monitoring, post-hoc traceability, equipment operation and maintenance and safety decision-making of rail transit. However, in the actual deployment and application at present, the above system is usually constructed and operated as an independent entity, and there is a significant technical short board, which has become a bottleneck for improving the operation and maintenance efficiency: The independent operation of each system forms an information island, namely, the independent operation of each system and the data non-intercommunication cause the data to be unable to be associated and analyzed, and the same fault in the actual operation and maintenance often involves a plurality of systems, so that operation and maintenance personnel are required to search and analyze among the systems by themselves. For example, when video recording is found to be lost in a video monitoring interface, an operation and maintenance person cannot check the complete health state information and the video recording storage state of the camera, and the operation and maintenance person needs to log in a network management system and a video cloud storage system respectively to perform independent query, so that the operation is complex. The existing four systems cannot provide a unified monitoring interface to show the global view of the running condition, so that operation and maintenance personnel need to log in a plurality of independent system interfaces respectively, and global cognition of the overall running condition of the system is difficult to be quickly established. For example, the association relationship between "video stream-storage resource-device health-picture quality" cannot be intuitively grasped in real time, and the optimal treatment opportunity is easily musied in a scene requiring a quick decision such as emergency command. The intelligent degree is low, and the system is highly dependent on artificial operation and maintenance, and the existing system generally only has basic data presentation and alarm functions and lacks active decision making and prediction capabilities. The whole process of fault discovery, analysis, localization and processing is seriously dependent on personal experience and subjective judgment of operation and maintenance personnel. For example, the system cannot automatically correlate video quality degradation with equipment performance metrics, nor predict equipment life based on historical data and provide preventative maintenance recommendations. This responsive, manually driven mode of operation is inefficient and difficult to cope with increasingly complex system scales. Disclosure of Invention Aiming at the problems, the invention aims to provide a rail transit visualization system integration system based on multi-agent cooperation and an integration method thereof, which are used for integrating a plurality of heterogeneous subsystems in the field of rail transit monitoring so as to solve the problem of information island. The rail transit visualization system based on multi-agent cooperation comprises an agent community, an agent outer shell layer and a subsystem layer, wherein the subsystem layer comprises a plurality of subsystems, an agent shell is built for each subsystem to form a subsystem agent, the agent outer shell layer is formed, and the subsystem agents are communicated through message middleware to form the agent community. Further, the subsystem comprises a video cloud platform system, a video cloud storage system, a video equipment network management system and a video quality diagnosis system. Further, the intelligent s