Rail transit station-level intelligent agent implementation method and system

Abstract

The invention relates to the technical field of data communication, and discloses a rail transit station-level agent implementation method and system. The method comprises the following steps: constructing a professional body base through an ARM processor, designing a LiteIP security communication bus based on an FPGA, constructing a multi-core CPU cluster to construct an intelligent body center, establishing a real-time and entity database to form a data basis, fusing federated learning and a neural network to generate a decision, and executing edge calculation to realize station-level autonomous control. According to the invention, millisecond-level data exchange and intelligent cooperative control in the station-level system are realized, and the response speed and the processing efficiency of the system in an emergency are remarkably improved.

Description

Technical Field

[0001] The present application relates to the field of data communication technology, and in particular to a method and system for realizing a rail transit station-level intelligent entity. Background Art

[0002] As the scale and complexity of urban rail transit systems continue to increase, traditional station-level control systems face many challenges. Existing rail transit station-level systems usually adopt a decentralized architecture, independently deploying various professional subsystems in the station (such as building automation system BAS, fire alarm system FAS, passenger information system PIS, public broadcasting system PA, shielded door system PSD, etc.), and achieving limited data exchange between systems through standard protocols such as Modbus, BACnet or proprietary protocols. These systems are generally built on traditional programmable logic controllers (PLCs) or industrial computers, rely on preset logical rules for control, and usually require coordination and supervision through central-level systems. In terms of data communication, existing systems mostly use traditional TCP / IP protocol stacks to transmit data via Ethernet or industrial buses. The data exchange process involves multiple connection establishment, data transmission and disconnection, and the communication efficiency is low.

[0003] However, this traditional architecture has obvious shortcomings: first, each professional subsystem forms an "information island", data cannot be shared efficiently, and the coordination between systems is poor; second, the communication delay is high (usually at the level of hundreds of milliseconds), which makes it difficult to meet the needs of rapid response in emergency situations; third, the lack of a unified data processing platform makes it difficult to achieve cross-system intelligent analysis and decision-making; fourth, the system's scalability is limited, and it is difficult to access new functions or equipment; fifth, the security mechanism is simple and it is difficult to cope with increasingly complex network security threats; finally, the system's adaptive ability is weak and it is impossible to dynamically adjust the control strategy according to the actual operating conditions. These shortcomings lead to slow system response in emergency situations such as fire, large passenger flow, and equipment failure, making it difficult to achieve efficient and intelligent collaborative disposal, affecting rail transit operation safety and service quality. Summary of the invention

[0004] One purpose of the present application is to provide a method and system for realizing a rail transit station-level intelligent entity, which is used to realize millisecond-level data exchange and intelligent collaborative control within the station-level system, and significantly improve the system's response speed and processing efficiency in emergency situations.

[0005] In the first aspect, the present application provides a method for implementing a rail transit station-level intelligent body, including: building a general professional body base, configuring a real-time embedded operating system, and forming a modular professional body architecture; designing a secure communication bus according to the professional body architecture, performing differentiated transmission processing on data, and establishing a high-speed communication network; combining multiple ARM processors into a station-level multi-core CPU cluster based on the high-speed communication network, dynamically allocating workloads, and building a station-level intelligent body central system; relying on the station-level intelligent body central system to establish a unified real-time database URTDB and a unified entity database UDB, and hierarchically managing system data to form an intelligent body data foundation; performing multi-dimensional analysis and prediction of real-time station-level operation data to generate intelligent decision-making plans; executing edge computing and distributed linkage control based on the intelligent decision-making plan, and collaboratively scheduling various professional subsystems to achieve autonomous control of station-level intelligent bodies.

[0006] In a second aspect, the present application provides a rail transit station-level intelligent agent implementation system, including: Building modules are used to build a universal professional body base, configure a real-time embedded operating system, and form a modular professional body architecture; The transmission module is used to design a secure communication bus according to the professional body architecture, perform differentiated transmission processing on data, and establish a high-speed communication network; The allocation module is used to combine multiple ARM processors into a station-level multi-core CPU cluster based on a high-speed communication network, dynamically allocate workloads, and build a station-level intelligent agent central system; The hierarchical module is used to establish a unified real-time database URTDB and a unified entity database UDB based on the station-level intelligent body central system, and to manage the system data in a hierarchical manner to form the intelligent body data foundation; Prediction module, used to perform multi-dimensional analysis and prediction of real-time station-level operation data and generate intelligent decision-making solutions; The control module is used to perform edge computing and distributed linkage control based on intelligent decision-making solutions, coordinate the scheduling of various professional subsystems, and realize autonomous control of station-level intelligent bodies.

[0007] In a third aspect, a computer device is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the computer device executes the above-mentioned rail transit station-level intelligent agent implementation method.

[0008] In a fourth aspect, a computer-readable storage medium is provided, wherein instructions are stored in the computer-readable storage medium, and when the computer-readable storage medium is run on a computer, the computer executes the above-mentioned rail transit station-level intelligent agent implementation method.

[0009] In the technical solution provided by the present application, a universal professional body base is constructed by a high-performance 64-bit ARM processor, and a real-time embedded operating system is configured to form a modular professional body architecture, which effectively solves the problem of independent deployment and difficulty in collaborative work of various professional subsystems in traditional rail transit station-level systems, and realizes efficient integration and flexible configuration of hardware resources; the LiteIP secure communication bus is implemented by FPGA based on the professional body architecture design, and the data is transmitted and processed differently using a dynamic key encryption mechanism, which not only reduces the communication delay from the hundreds of milliseconds of traditional TCP / IP to the sub-millisecond level, but also greatly improves the data transmission efficiency and security by distinguishing and processing global shared data streams, group shared data streams, and point-to-point data streams; multiple ARM processors are combined into a station-level multi-core CPU cluster, and the workload is dynamically allocated through a task priority allocation algorithm. The constructed station-level intelligent body central system makes full use of computing resources, and the response time of key tasks is shortened to less than 20ms, which is much better than the 300ms response time of the traditional system; a unified real-time database URTDB and a unified entity database UDB are established, and the system is managed through a multi-level cache mechanism and a DSI data service interface. The data is managed in layers, which completely solves the "data island" problem in traditional systems, enables various professional entities to easily access global data, and provides a complete data foundation for intelligent analysis. In particular, in terms of artificial intelligence applications, this solution integrates federated learning and deep neural network technology, so that various professional entities can share model knowledge under the premise of protecting local data privacy, significantly improving the comprehensive performance of AI models. For example, the situational awareness engine successfully integrates and analyzes multimodal data (visual, time series, sensors, etc.) through the combination of convolutional neural networks and recurrent neural networks, accurately identifies complex scene states, predicts trend changes, and provides a reliable basis for intelligent decision-making. Finally, through edge computing and distributed linkage control, comprehensive autonomous control capabilities from millisecond-level response emergency control to long-term optimization are achieved. Especially in emergency situations such as fires and large passenger flows, the entire processing flow from data collection, analysis and processing to command execution only takes 80ms, which is nearly 4 times faster than the traditional system, greatly improving the safety level and operational efficiency of rail transit stations. At the same time, through adaptive parameter adjustment and continuous updating of the professional control strategy library, the system can continuously evolve and optimize to adapt to changing operating environments and needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] One or more embodiments are exemplarily described by pictures in the corresponding drawings, and these exemplified descriptions do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0011] Figure 1 This is a schematic diagram of an embodiment of a method for implementing a rail transit station-level intelligent agent in an embodiment of the present application; Figure 2 This is a schematic diagram of a station-level intelligent agent in an embodiment of the present application; Figure 3 A schematic diagram of the database system in this application; Figure 4 This is a schematic diagram of the distributed linkage in this application; Figure 5 This is a schematic diagram of an embodiment of a rail transit station-level intelligent agent implementation system in an embodiment of the present application; Figure 6 It is a schematic block diagram of the structure of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0012] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0013] See also Figure 1 , an embodiment of the rail transit station-level intelligent agent implementation method in the embodiment of the present application includes: Step S101, construct a universal professional body base, configure a real-time embedded operating system, and form a modular professional body architecture; Step S102: design a secure communication bus according to the professional architecture, perform differentiated transmission processing on data, and establish a high-speed communication network; Step S103: combining multiple ARM processors into a station-level multi-core CPU cluster based on a high-speed communication network, dynamically allocating workloads, and building a station-level intelligent agent central system; Step S104: Establish a unified real-time database URTDB and a unified entity database UDB based on the station-level intelligent agent central system, and manage the system data in layers to form an intelligent agent data foundation; Step S105: Perform multi-dimensional analysis and prediction on real-time station-level operation data to generate intelligent decision-making solutions; Step S106: Execute edge computing and distributed linkage control based on the intelligent decision-making solution, coordinate the scheduling of various professional subsystems, and realize autonomous control of station-level intelligent bodies.

[0014] It is understandable that the execution subject of the present application can be a rail transit station-level intelligent body implementation system, or a terminal or a server, which is not limited here. The present application embodiment is described by taking the server as the execution subject as an example.

[0015] Specifically, a general professional body base is constructed through a high-performance 64-bit ARM processor, and a real-time embedded operating system is configured to form a modular professional body architecture. In this process, the selection of ARM processors is based on their high performance and low power consumption characteristics. The processing capacity reaches 1.5GHz main frequency, and the multi-core structure supports parallel task processing. The real-time embedded operating system is configured with a time resolution of 1ms to ensure the accuracy of system response. The modular professional body architecture includes seven professional bodies: BAS, FAS, PIS, PA, PSD, ATS, and SCADA. Each professional body is customized for specific functions, but shares a unified base architecture. For example, the FAS professional body is specifically responsible for fire alarm signal processing. Its base is equipped with a 4-core ARM processor, 512MB memory and 1GB flash memory. Through the protocol adapter conversion module, different communication protocols such as Modbus and BACnet are converted into a unified standard data format to achieve seamless communication with fire detectors, manual alarm buttons and other equipment.

[0016] The station-level intelligent body is an intelligent cabinet, which includes a core processing cluster, professional bodies of various professional subsystems, and a secure communication bus. The core processing cluster is a multi-level multi-task cluster composed of 64-bit ARM processors based on domestic CPUs. It can realize the unified and efficient storage of real-time data urtdb and historical data udb. Each professional body and the intelligent body cluster are connected through efficient private FPGA communication, such as Figure 2 Shown is a schematic diagram of a station-level intelligent agent in an embodiment of the present application.

[0017] According to the professional body architecture, the LiteIP secure communication bus based on FPGA is designed, and the dynamic key encryption mechanism is used to perform differentiated transmission processing on the data to establish a high-speed communication network. The FPGA uses a hardware circuit with 100,000 logic gates to achieve efficient processing of the LiteIP protocol stack. The LiteIP communication bus divides the data stream into three types: global shared data stream, group shared data stream and point-to-point data stream, and assigns different transmission priorities to each. The dynamic key encryption mechanism is based on the principle of elliptic curve cryptography, automatically updates the key pair every 30 seconds, and encrypts the transmitted data. The data packet is marked by hardware timestamp and transmission frame sequence number, combined with the dynamic bandwidth allocation technology, to achieve sub-millisecond communication delay, which is 10 times more efficient than traditional TCP / IP. Based on the high-speed communication network, multiple ARM processors are combined into a station-level multi-core CPU cluster, and the workload is dynamically allocated through the task priority allocation algorithm to build a station-level intelligent body central system. Multiple ARM processors are networked and connected according to the master-slave architecture to form a processor array, which is divided into three roles: main control processor, computing processor and communication processor. The task priority allocation algorithm assigns weight coefficients to different tasks according to the urgency of the task, resource requirements and the current load status of the system, and allocates urgent tasks to low-load processors for execution. The station-level intelligent agent central system maintains real-time updates of the resource status diagram through the inter-processor communication channel to ensure balanced workload distribution and control the system response time within 20ms.

[0018] Relying on the station-level intelligent body central system, a unified real-time database URTDB and a unified entity database UDB are established. The system data is hierarchically managed through a multi-level cache mechanism and a DSI data service interface to form the intelligent body data foundation. URTDB allocates high-speed memory space to store real-time control data, and UDB manages parameter configuration and historical data through data persistence mapping technology. The multi-level cache mechanism stores high-frequency access data in the L1 cache, medium-frequency access data in the L2 cache, and low-frequency access data in the main storage area to accelerate data access speed. The DSI data service interface divides the data structure into a global concept layer, a local concept layer, a memory physical layer, an object layer, and an application physical layer, providing a unified access standard for each professional body to solve the problem of data islands. Based on the intelligent body data foundation, federated learning and deep neural network technology are integrated to perform multi-dimensional analysis and prediction of real-time station-level operation data to generate intelligent decision-making solutions. The data acquisition preprocessing unit extracts multi-source heterogeneous data from URTDB and UDB, and forms a standardized data set through denoising, normalization, and feature extraction. The hierarchical federated learning architecture enables each professional body to only transmit model gradient information after local calculation, protecting data privacy while improving learning efficiency. The situational awareness engine processes visual data through convolutional neural networks, analyzes time series data using recurrent neural networks, integrates multimodal information to identify station-level operating status, generates a multivariate time series prediction matrix, and outputs intelligent decision-making solutions in combination with reinforcement learning algorithms. Based on the intelligent decision-making solution, edge computing and distributed linkage control are executed to coordinate the scheduling of various professional subsystems to achieve autonomous control of station-level intelligent bodies. The intelligent decision mapper decomposes the decision solution into a professional body execution instruction set, and the edge computing task distribution mechanism divides the instruction set into real-time control instructions and non-real-time control instructions to build a dual-channel scheduling framework. The scheduling priority evaluator sorts the instruction execution order to form a linkage control timing table, and issues control commands to each professional subsystem through a distributed transaction mechanism. The state feedback collector obtains execution status data, generates a control execution report, and performs adaptive parameter adjustment in combination with non-real-time control instructions to complete the closed loop of autonomous control of station-level intelligent bodies.

[0019] Taking an emergency fire scenario as an example, when the FAS professional body detects a smoke signal, the data is transmitted to the station-level multi-core CPU cluster through the LiteIP secure communication bus accelerated by FPGA, and the data is updated in real time in the URTDB. The intelligent body uses a deep neural network to analyze the smoke data and compare it with historical data, confirm the fire situation and determine the spread trend, and generate an emergency decision-making plan. The intelligent decision mapper converts the plan into a professional body execution instruction set. The instructions trigger the BAS professional body to control the exhaust system, the PA professional body to issue an evacuation broadcast, the PSD professional body to control the screen door, and the ATS professional body to adjust the train operation plan through a dual-channel scheduling framework. The entire process from data collection to instruction execution takes only 80ms, which is much lower than the 300ms response time of the traditional system, greatly improving the handling efficiency in emergency situations, reflecting the significant advantages of high-speed communication of station-level intelligent bodies.

[0020] It should be noted that the station-level intelligent body is a station-level intelligent hub that can unify the management and control of all relevant sensors and controllers, and even data, models, and plans related to intelligence such as people and objects in the station. For each professional subsystem, it can be directly connected by PLC, or processed and executed by the corresponding professional body. The professional body is responsible for the data collection and control of a single profession. The intelligent body communicates with each other through a special network adapter based on FPGA, which greatly improves the transmission efficiency between the professional body base and the intelligent body.

[0021] The station-level intelligent body architecture introduces the multi-CPU cluster idea framework into the station intelligent body construction, and regards each professional subsystem as a CPU or human organ, which is efficiently connected to the brain center through vascular nerve links to become a human individual. The station-level intelligent body is an intelligent cabinet, which contains the core processing cluster, the professional body of each professional subsystem, and the security communication bus. The core processing cluster is a multi-level multi-tasking cluster composed of 64-bit ARM processors based on domestic CPUs. It can realize the unified and efficient storage of real-time data URTDB and historical data UDB. Each professional body and the intelligent body cluster are connected through efficient private FPGA communication.

[0022] The database system is based on an efficient memory library: Figure 3 , which is a schematic diagram of the database system in this application; The professional body has become a computer similar to multiple CPU cores. The time resolution of the real-time operating system is 1ms. The core cluster is the most powerful brain, working closely with the professional body. Each professional body is fully capable of controlling the response within 5-10ms. Combined with the actual computing response of the core cluster of 5-10ms, it is fully possible to achieve an autonomous driving level response efficiency of 20ms.

[0023] Working mechanism: UDB is a unified entity database that manages the parameters and historical data in the intelligent body. URTDB is a unified real-time database, which is a memory database for real-time control. The core cluster includes: UDB, URTDB, professional body mirroring, and cluster monitoring system. UDB and URTDB are used for high-speed computing and real-time response. Related operations are performed through the data bus related professional bodies.

[0024] Station-level intelligent bodies consist of a universal professional body base, similar to human genetic tissue --- unified domestic hardware, equipped with a real-time embedded operating system, an independent real-time database, and real-time message middleware and communication middleware. Each professional body directly forms a set of specialized systems with UDB and URTDB through the bus to achieve specialized operation control, and interacts with the core cluster of intelligent bodies for control. It is not only a professional control brain, but also a key element of the entire station-level intelligent body. It is closely united around the core cluster and responds to control commands at the fastest speed.

[0025] Each professional entity can realize flexible configuration of single-machine or multi-machine operation at the station level, and the device itself is plug-and-play. The communication protocols with each PLC are rich, and each protocol can achieve plug-and-play communication in the form of plug-ins. Internal private cloud deployment, the hardware of each professional entity is automatically configured and installed on demand.

[0026] Traditional TCP-based data standards often require multiple processes of link establishment, data exchange, and link disconnection, with average efficiency. In order to adapt to large-scale and efficient transmission during the design of this intelligent body, each professional subsystem uses a secure LiteIP communication protocol based on FPGA with private dynamic key management for communication. It effectively distinguishes between global shared data, group shared data, and point-to-point interconnected data to form an optimal communication strategy.

[0027] Security layer, dynamic code simple encryption, data is automatically encrypted during transmission serialization Customized transport layer, efficient transmission of the independent private protocol of LiteIP technology used by professional bodies. As mentioned above, even if the efficiency improvement of the transport layer is ignored, this method greatly improves the timeliness and reliability of station-level intelligence.

[0028] The AI ​​intelligent decision-making and self-learning subsystem introduces federated learning, computer vision and deep learning to achieve real-time emergency decision-making, predictive maintenance and customized information services. At the station system level, through vision and deep learning, combined with collected data, multi-dimensional data combing is used to confirm the reliability of data and the status of equipment and personnel. There is no need to rely on the computing power of the central system, and accurate and rapid judgments can be made. For example, if the shielding door signal alarms or there are fireworks and water-related alarms in the tunnel, through visual and infrared other spectrum analysis, it can be accurately confirmed whether it is a real alarm, reducing misjudgment. Improve the ability of operation and maintenance and operation.

[0029] Implement local data processing and intelligent scheduling at the station level to reduce the burden on the central system and improve the overall response time. As a station-level intelligent entity, the greatest value is the ability to realize station-level edge computing and distributed linkage, maximizing the safety level and response speed of the subway. A typical scenario is that when a FAS fire alarm occurs, an emergency plan or a temporary plan of AI decision-making will be triggered immediately, and the relevant parts of the BAS will be linked to carry out fire-related emergencies such as fan exhaust air intake, elevator two-way to one-way departure, and other triggered escape devices; and the station will broadcast and deploy personnel arrangements to avoid passengers outside the station from entering the station; and the emergency locking mechanism of the ATS system will be triggered to quickly drive away vehicles that have left the station, and trains that are about to enter the station will be urgently locked to avoid entering the station... Such a system engineering turns the original decisions made at the central level into independent completions by station-level intelligent entities, greatly improving immediacy and reliability. Figure 4 As shown, it is a schematic diagram of distributed linkage in this application; In the embodiment of the present application, a general professional body base is constructed by a high-performance 64-bit ARM processor, and a real-time embedded operating system is configured to form a modular professional body architecture, which effectively solves the problem of independent deployment and difficulty in collaborative work of various professional subsystems in traditional rail transit station-level systems, and realizes efficient integration and flexible configuration of hardware resources; the LiteIP secure communication bus is implemented by FPGA based on the professional body architecture design, and the data is transmitted and processed differently using a dynamic key encryption mechanism, which not only reduces the communication delay from the hundreds of milliseconds of traditional TCP / IP to the sub-millisecond level, but also greatly improves the data transmission efficiency and security by distinguishing and processing global shared data streams, group shared data streams, and point-to-point data streams; multiple ARM processors are combined into a station-level multi-core CPU cluster, and the workload is dynamically allocated through a task priority allocation algorithm. The constructed station-level intelligent body central system makes full use of computing resources, and the response time of key tasks is shortened to less than 20ms, which is much better than the response time of 300ms of the traditional system; a unified real-time database URTDB and a unified entity database UDB are established, and system data is accessed through a multi-level cache mechanism and a DSI data service interface The hierarchical management completely solves the "data island" problem in the traditional system, enables various professional entities to easily access global data, and provides a complete data foundation for intelligent analysis. In particular, in terms of artificial intelligence applications, this solution integrates federated learning and deep neural network technology, so that various professional entities can share model knowledge under the premise of protecting local data privacy, significantly improving the comprehensive performance of AI models. For example, the situational awareness engine successfully integrates and analyzes multimodal data (vision, time series, sensors, etc.) through the combination of convolutional neural networks and recurrent neural networks, accurately identifies complex scene states, predicts trend changes, and provides a reliable basis for intelligent decision-making. Finally, through edge computing and distributed linkage control, comprehensive autonomous control capabilities from millisecond-level response emergency control to long-term optimization are achieved. Especially in emergency situations such as fires and large passenger flows, the entire processing flow from data collection, analysis and processing to command execution only takes 80ms, which is nearly 4 times faster than the traditional system, greatly improving the safety level and operational efficiency of rail transit stations. At the same time, through adaptive parameter adjustment and continuous updating of the professional control strategy library, the system can continuously evolve and optimize to adapt to changing operating environments and needs.

[0030] In a specific embodiment, the process of executing step S101 may specifically include the following steps: By setting up multiple timing control units to accurately control the clock signal of the ARM processor, real-time processing capability with a time resolution of 1ms is obtained; The real-time embedded operating system is divided into a real-time task layer and an application layer, and the real-time task layer manages the allocation of hardware resources; A protocol adapter conversion module is built into the universal professional body base to convert different communication protocols into a unified standard data format; According to the functions, there are seven types of professional bodies, namely BAS professional body, FAS professional body, PIS professional body, PA professional body, PSD professional body, ATS professional body and SCADA professional body, which constitute professional body groups; The computing resources and storage space of the professional base are dynamically configured through the private cloud resource allocation algorithm to form an elastic computing architecture; Select the corresponding professional body from the professional body group to establish a communication connection with the corresponding PLC device to form a complete modular professional body architecture.

[0031] Specifically, the multi-channel timing control unit is a hardware module specially designed for high-precision clock signal management. It divides the basic clock signal of the ARM processor into multiple independent signals and performs precise control. Each signal is maintained stable by an independent crystal oscillator and a phase-locked loop. The multi-channel timing control unit performs frequency division processing on the clock signal and converts the main frequency into a multi-level clock frequency to ensure that tasks of different priorities obtain corresponding time accuracy, with the highest accuracy reaching 1ms. The control unit also has a built-in time compensation algorithm, which monitors clock drift in real time and automatically adjusts parameters to maintain clock synchronization, thereby obtaining real-time processing capabilities with a time resolution of 1ms. The real-time embedded operating system adopts a layered architecture, which is divided into a real-time task layer and an application layer. The real-time task layer is located at the bottom layer, directly controls hardware resources, and is responsible for core functions such as interrupt processing, task scheduling, and memory management. This layer maintains the task priority queue, arranges CPU time slice allocation according to priority, and implements priority inversion protection for key resources. The real-time task layer abstracts hardware resources into resource objects. Each resource object contains three attributes: status identification, occupancy information, and access rights. The resource allocation table records the current resource usage. Application tasks obtain the required resources through the resource request interface, and the real-time task layer makes allocation decisions based on priority and current resource status. The application layer is located in the upper layer and is responsible for the implementation of specific business logic. It interacts with the real-time task layer through standard APIs.

[0032] The universal professional body base has a built-in protocol adapter conversion module, which adopts a plug-in architecture and supports multiple industrial communication protocols such as Modbus, BACnet, CIP, Profibus, etc. The protocol adapter conversion module contains three core parts: protocol parsing engine, data mapper and standard data encapsulator. The protocol parsing engine is responsible for identifying and parsing data frames of different protocol formats and extracting valid data content; the data mapper maps the parsed data with the preset data model to establish a unified data structure; the standard data encapsulator encapsulates the mapped data into a unified standard data format for subsequent processing. After the data transmitted by various devices is converted through this module, no matter how different the original protocols are, they are converted into standard data packets with the same structure. According to the different functional requirements of rail transit stations, professional body bases are divided into seven types of professional bodies. The BAS professional body is responsible for station environment control and management of air conditioners, lighting, elevators and other equipment; the FAS professional body is responsible for fire alarm and processing data from smoke detectors, temperature sensors and other equipment; the PIS professional body is responsible for passenger information display and management of information screens, LED displays, etc.; the PA professional body is responsible for public broadcasting and controls loudspeakers, amplifiers and other equipment; the PSD professional body is responsible for platform screen door control; the ATS professional body is responsible for automatic train dispatching; and the SCADA professional body is responsible for monitoring and data collection. These seven professional bodies together constitute a professional body group, and each professional body optimizes the hardware configuration and software structure for its functional field.

[0033] The private cloud resource allocation algorithm dynamically configures the computing resources and storage space of the professional body base. Based on the resource pool management principle, the algorithm integrates all available hardware resources into the resource pool, including CPU cores, memory space, storage capacity, etc. The algorithm monitors the resource utilization rate and task queue length of each professional body through the load prediction model, calculates the resource demand trend, and then automatically adjusts the resource allocation plan according to the priority strategy and resource availability. In the resource allocation process, the resources required for key tasks are first reserved, and the remaining resources are dynamically allocated on demand. When resource competition occurs, it is processed in the preset priority order. The resource allocation algorithm also supports elastic scaling strategies, allowing automatic expansion or contraction of resource configuration when the load changes to form an elastic computing architecture. When selecting the corresponding professional body from the professional body family to establish a communication connection with the corresponding PLC device, the device is first scanned and identified to obtain the list of PLC devices in the connected network and their communication parameters. According to the preset function association table, different PLC devices are matched with the corresponding professional body to establish a device-professional body mapping relationship. Then, through the aforementioned protocol adaptation conversion module, the communication connection is initialized, the handshake authentication is completed, and the data channel is established. The professional body obtains control point information from the PLC device, builds a control point database, and forms a device abstraction layer. Multiple professional bodies work together, each responsible for device management and data processing within its own field, and together form a complete modular professional body architecture.

[0034] Take the fire emergency response of rail transit stations as an example. When a fire occurs in the station, the smoke detector detects the smoke signal and transmits the data to the PLC controller through the Modbus protocol. The FAS professional body parses the Modbus data frame into a standard data format through the protocol adaptation conversion module to extract information such as smoke concentration and temperature. The multi-channel timing control unit ensures that the FAS professional body processes real-time data with a time accuracy of 1ms, and the real-time task layer immediately allocates high-priority resources to process the fire alarm information. The FAS professional body transmits the fire alarm information to other professional bodies through a high-speed communication network, triggering a linkage response: the BAS professional body controls the air conditioning system to switch to emergency smoke exhaust mode, the PA professional body issues an evacuation broadcast, the PSD professional body controls the shield door to open, and the ATS professional body adjusts the train operation plan. The entire process is dynamically adjusted by the private cloud resource allocation algorithm to allocate more resources to the professional body that handles emergency tasks to ensure the efficient execution of the emergency response. From the detection of the smoke signal to the completion of the linkage response of the entire station, the entire process takes no more than 100ms, reflecting the efficient communication and collaboration capabilities of the modular professional body architecture in emergency response.

[0035] In a specific embodiment, the process of executing step S102 may specifically include the following steps: Customize the LiteIP protocol stack through FPGA hardware circuits to build a high-performance protocol processing unit; The communication data flow is divided into three types: global shared data flow, group shared data flow and point-to-point data flow, and different transmission priorities are assigned to each type; Generate dynamic key pairs based on the principles of elliptic curve cryptography and apply encryption strategies of different strengths to different types of data streams; The data packets are marked by hardware timestamp mechanism and transmission frame sequence number to form a reliable data packet transmission confirmation mechanism; Automatically calculate the optimal transmission path based on topology information and perform intelligent routing forwarding on data packets; The communication bandwidth dynamic allocation technology is used to coordinate and manage the data transmission between multiple professional entities and establish a high-speed communication network.

[0036] Specifically, the FPGA hardware circuit customizes the LiteIP protocol stack. The process of building a high-performance protocol processing unit first involves converting the software processing logic of the traditional network protocol stack into a hardware circuit design. The LiteIP protocol stack is a lightweight communication protocol. Compared with the traditional TCP / IP protocol, it removes the redundant handshake confirmation mechanism and directly implements the data transmission function on the data link layer. The FPGA hardware circuit implements data fragmentation, header encapsulation, checksum calculation, data encryption and other functions through dedicated logic units. Each functional module uses a pipeline architecture to achieve parallel processing. Inside the FPGA, a multi-channel processing structure is established. Each channel contains an independent transceiver buffer and state control logic, and the encryption engine and check unit are shared among multiple channels. When the data packet enters the FPGA, it first passes through the message classifier to identify the data type, and then is assigned to the corresponding processing channel. After pipeline processing, it is output as a standard LiteIP data packet. The entire processing process is completed at the hardware level without the participation of the CPU. The communication data flow is divided into three types: global shared data flow, group shared data flow and point-to-point data flow, which is a functional classification of data communication modes. Global shared data streams refer to public information that needs to be accessible to all professional entities, such as system clocks, global alarm information, etc. The broadcast flag is set in the data packet header to ensure that the data is received by all nodes. Group shared data streams refer to data shared only within a specific functional group, such as temperature and humidity information within the environmental control group. The data packet header contains a group identifier, and only the professional entities belonging to the group process the data. Point-to-point data streams are private communication data between two specific professional entities, such as control command transmission. The data packet header contains the source address and the destination address. The three data streams are assigned different transmission priorities. The point-to-point data stream is set with the highest priority (value 1), the group shared data stream has a middle priority (value 2), and the global shared data stream has the lowest priority (value 3). The priority value is transmitted through the priority field of the data packet header. When processing data exchange, the LiteIP protocol stack gives priority to data packets with higher priorities to ensure that key control instructions can be transmitted in a timely manner.

[0037] Generating dynamic key pairs based on the principle of elliptic curve cryptography and applying different encryption strategies to different types of data streams are the core security mechanisms of LiteIP secure communication bus. Elliptic curve cryptography (ECC) is a public key cryptography system based on elliptic curve mathematical problems. Compared with the traditional RSA algorithm, it can provide the same security strength with a shorter key length, which is suitable for resource-constrained environments. The dynamic key generator generates a basic key pair based on the preset elliptic curve parameters, and then adds timestamps and random number factors to calculate the dynamic key. During the generation process, an elliptic curve and a base point on the curve are first selected, a private key integer is randomly generated, and the public key point is calculated by elliptic curve point multiplication. Different types of data streams apply encryption strategies of different strengths: point-to-point data streams are fully encrypted with 256-bit ECC keys, including data headers and payloads; group shared data streams are encrypted with 192-bit ECC keys, only encrypting the data payload part; global shared data streams are simply encrypted with 128-bit ECC keys, mainly to protect data integrity. The key update mechanism automatically updates the key pair at a preset time interval (usually 30 seconds) and distributes the new public key to the communicating parties through a secure channel.

[0038] The hardware timestamp mechanism and transmission frame sequence number mark the data packet to form a reliable data packet transmission confirmation mechanism, which solves the reliability problem of data transmission. Each FPGA module has a built-in high-precision clock source to provide microsecond timestamps. When the data packet enters the transmission process, the FPGA automatically adds the current timestamp and the counter-based frame sequence number to the data packet header. After receiving the data packet, the receiver extracts the timestamp and sequence number information, compares it with the local clock to calculate the transmission delay, and detects whether there is packet loss through the sequence number. If packet loss is found, different measures are taken according to the severity of the packet loss: send a retransmission request when a single packet is lost; notify the sender to reduce the transmission rate when multiple packets are lost continuously; trigger link quality assessment when intermittent packet loss exceeds the threshold. The cumulative confirmation mechanism is used for data packet reception confirmation. The receiver periodically sends confirmation frames containing the highest sequence number that has been correctly received, effectively reducing confirmation traffic.

[0039] The process of automatically calculating the optimal transmission path based on the topological structure information and intelligently routing and forwarding data packets involves dynamic routing calculation and forwarding decisions. The topology discovery module collects network connection status information by regularly broadcasting detection packets and constructs a global topology map, which includes the connection relationship between nodes, link delay and bandwidth information. The path calculator calculates the shortest path from the source node to the target node based on the Dijkstra algorithm, using link delay as the basic metric while considering bandwidth and load factors. The path calculation results form a routing table, which includes the target address, next hop address and path delay information. The packet forwarding decision is based on the routing table. When the packet arrives at the forwarding node, the target address is extracted and the routing table is queried to obtain the next hop information. If there are multiple optional paths, the optimal path is selected based on the current load situation. The routing table is updated regularly to adapt to changes in the network topology and ensure that the data packet is always transmitted along the optimal path.

[0040] The communication bandwidth dynamic allocation technology coordinates and manages the data transmission between multiple professional entities, and the process of establishing a high-speed communication network realizes the rational use of network resources. The bandwidth monitoring module continuously collects the traffic data of each link, calculates the current bandwidth utilization and the remaining available bandwidth. The QoS (Quality of Service) manager allocates network resources according to the data flow priority and bandwidth demand, establishes a multi-level queue structure, and high-priority data is given priority in bandwidth allocation. The traffic shaper controls the data flow sent by each professional entity, limits the sending rate according to the preset strategy, and avoids a single professional entity from occupying too much bandwidth resources. The congestion control module monitors the network congestion status. When it detects that the link load is too high, it triggers a congestion notification and requires the relevant professional entity to reduce the data sending rate. The bandwidth reservation mechanism allows key businesses to apply for bandwidth resources in advance to ensure that sufficient transmission resources can still be obtained during the peak period of network load.

[0041] In the rail transit station-level application scenario, when an emergency occurs, such as a platform screen door failure, the PSD professional body first detects the abnormal status data of the platform screen door through sensors, encapsulates the data into a point-to-point data stream (priority 1), and processes it through the LiteIP protocol stack implemented by FPGA. After the data packet is encrypted with a 256-bit ECC key, the current timestamp and frame sequence number are attached. The path calculator determines the optimal path from the PSD professional body to the station-level multi-core CPU cluster based on the current network topology, and the data packet is forwarded via the shortest delay path. The bandwidth manager identifies the high-priority data stream and immediately allocates sufficient bandwidth resources to it to ensure that the transmission delay is minimized. After receiving the data, the station-level multi-core CPU cluster verifies the data integrity by decryption, extracts the timestamp to calculate the end-to-end delay (usually less than 1ms), and returns a confirmation frame to the PSD professional body. At the same time, the station-level intelligent body generates linkage instructions based on the platform screen door failure information, creates a set of point-to-point data streams and sends them to the PA professional body (broadcasting evacuation information), the ATS professional body (adjusting train operation) and other related professional bodies. The entire information transmission process is completed from the detection of shield door anomalies to the coordinated response of multiple professional bodies, and the total delay is controlled within 5ms, which reflects the high-speed communication capability of the FPGA-based LiteIP safety communication bus in rail transit station-level intelligent bodies.

[0042] In a specific embodiment, the process of executing step S103 may specifically include the following steps: Multiple ARM processors are physically networked and connected according to the master-slave architecture to form a processor array; Define the roles of processors in the processor array and divide them into three types: main control processor, computing processor and communication processor; Establish inter-processor communication channels based on high-speed communication networks and create a processor resource pool; The processor load in the processor resource pool is monitored in real time through the status monitoring mechanism to generate a resource status diagram; Identify the task type and urgency based on the resource status diagram, and assign priority weight coefficients to different tasks; The workload is dynamically allocated according to the priority weight coefficient, and the tasks are distributed to the corresponding processors for execution to build a station-level intelligent agent central system.

[0043] Specifically, the process of physically networking and connecting multiple ARM processors according to the active-standby architecture to form a processor array first involves hardware deployment at the physical level. The ARM processor adopts a 64-bit architecture, each processor contains a multi-core design, the main frequency is 1.5-2.0GHz, and has strong parallel computing capabilities. The physical network adopts a dual-trunk redundant topology, and each processor is connected to two independent high-speed communication backbone networks through dual network cards to ensure that communication is not interrupted when a single network fails. In the active-standby architecture design, processors are deployed in pairs, one of a pair of processors is in the active state, and the other processor is in the standby state, and the two keep data synchronized in real time. The active processor is responsible for actual business processing, and the standby processor receives processing status information in real time. When the active processor fails, the standby processor can take over the business within milliseconds to ensure continuous operation of the system. The processor array maintains state synchronization through the heartbeat mechanism. The active processor sends a heartbeat packet every 100ms, and the standby processor monitors the heartbeat packet interval. When the heartbeat packet is not received for three consecutive times, the active-standby switching process is triggered. The roles of the processors in the processor array are defined and divided into three types: main control processor, computing processor and communication processor, and a clear functional division of labor is established. The main control processor is the coordination center of the entire system, responsible for management functions such as task distribution, resource scheduling and system monitoring. It is usually equipped with a high-performance ARM processor and a built-in large-capacity cache to optimize the execution efficiency of management instructions. The main control processor in a station-level intelligent body central system is configured as dual-machine redundancy, with only one in active state and the other in standby state. The computing processor focuses on data-intensive computing tasks such as data analysis, pattern recognition and predictive computing. It is equipped with a multi-core ARM processor, optimized floating-point arithmetic units, and suitable for parallel data processing. The communication processor is specifically responsible for external data interaction, handling network communication, protocol conversion and data routing, etc. It is equipped with a high-speed network interface and a dedicated buffer to optimize data throughput performance. The number of each type of processor is configured according to the site scale and business needs. Generally, a medium-sized site is configured with 2 main control processors, 4-8 computing processors, and 2-4 communication processors.

[0044] Based on the high-speed communication network, the communication channel between processors is established. The process of creating the processor resource pool realizes the unified management and flexible call of resources. The high-speed communication network is implemented by the LiteIP protocol stack based on FPGA, providing sub-millisecond communication delay. The communication channel between processors includes two types: control channel and data channel. The control channel is used to transmit control data such as scheduling instructions and status information, and adopts a point-to-point communication mode; the data channel is used to transmit business data, and supports both point-to-point and broadcast modes. When each processor joins the resource pool, it first registers with the master processor and submits a resource description file containing information such as processor type, number of cores, memory capacity, and current load. The master processor collects all resource description files, builds a global resource pool view, and records all available computing resources and their status. The processor resource pool adopts a dynamic member mechanism, allowing processors to join or exit the resource pool at any time to adapt to system expansion and maintenance needs. The processor load in the processor resource pool is monitored in real time through the status monitoring mechanism, and the process of generating the resource status diagram establishes a global resource view. The status monitoring mechanism deploys a monitoring agent on each processor to regularly collect processor load data, including indicators such as CPU usage, memory occupancy, network traffic, and the number of active tasks. These data are sent to the master processor through the control channel and summarized and analyzed by the status aggregator. The resource status graph is a multidimensional data structure that records the current resource usage of each processor and the trend over time. The resource status graph uses color coding to intuitively display the resource load. Red indicates high load (over 85%), yellow indicates medium load (50%-85%), and green indicates low load (below 50%). The resource status graph also contains historical load data for load trend analysis and prediction. The master processor calculates the load change trend based on historical data and predicts resource demand in the short term in the future.

[0045] The process of identifying the task type and urgency based on the resource status diagram and assigning priority weight coefficients to different tasks realizes the differentiated processing of tasks. Task type identification is based on task feature analysis, including factors such as source, resource requirements, and execution time, and tasks are classified into three categories: control, computing, and communication. The urgency assessment is determined according to the time sensitivity of the task and the importance of the business, and is divided into four levels: emergency (needs to be processed immediately, such as equipment alarms), high (needs priority processing, such as personnel safety related), medium (routine business processing), and low (background analysis and processing). The priority weight coefficient calculation comprehensively considers the task type and urgency, and uses a weighted method to determine the final priority. The weight of control tasks is higher than that of computing and communication tasks, and the weight of urgent tasks is higher than that of non-urgent tasks. Tasks with higher weight coefficients receive more system resource allocation and are scheduled for execution first. The workload is dynamically allocated according to the priority weight coefficient, and tasks are distributed to the corresponding processors for execution. The process of building a station-level intelligent agent central system realizes efficient task scheduling. The task scheduler first arranges the queue of pending tasks in descending order according to the priority weight coefficient to ensure that high-priority tasks are processed first. Then, according to the task type, the appropriate processor category is selected. Control tasks are assigned to the master processor, computing tasks are assigned to the computing processor, and communication tasks are assigned to the communication processor. After determining the processor category, the task scheduler queries the processor with the lowest current load from the resource status diagram and assigns the task to the processor for execution. During the task distribution process, the scheduler generates a task description package containing information such as task ID, priority, and resource requirements, and sends it to the target processor through the control channel. After receiving the task, the target processor arranges the execution order according to the task priority. After the task is executed, it returns the execution result and resource usage to the master processor.

[0046] For example, in the rail transit station-level intelligent body, when a large passenger flow occurs at the station, the passenger flow monitoring sensor data is received and preliminarily processed through the communication processor to generate passenger flow density distribution data. The main control processor obtains this data from the communication processor, identifies that the passenger flow density in multiple areas of the current platform exceeds the safety threshold, and determines it as a high-urgency control task, assigning a priority weight coefficient of 0.9 (full score is 1). The main control processor queries the resource status diagram and finds that the current load of the main control processor MCU-02 is only 30%, which is much lower than the 75% load of MCU-01, so the passenger flow control task is assigned to MCU-02 for processing. After receiving the task, MCU-02 analyzes the passenger flow data and historical patterns, calculates the optimal evacuation path, and generates a linkage control plan. The plan includes multiple measures such as adjusting the station broadcast, display screen information, and shield door opening and closing time, which need to be sent to the PA professional body, PIS professional body, and PSD professional body for execution. Each control task is packaged and assigned a corresponding priority, and then distributed to the corresponding communication processor for execution through the task scheduler. The entire process takes only 65ms from receiving passenger flow data to issuing control instructions, achieving rapid response to sudden large passenger flow situations and demonstrating the efficient collaborative processing capabilities of the station-level intelligent multi-core CPU cluster in rail transit scenarios.

[0047] In a specific embodiment, the process of executing step S104 may specifically include the following steps: Divide the memory area in the station-level intelligent agent central system and allocate high-speed access memory space for the unified real-time database URTDB; A unified entity database (UDB) is built through data persistence mapping technology to store and manage parameter configuration information and historical data; Classify system data according to access frequency, store high-frequency access data in L1 cache, medium-frequency access data in L2 cache, and low-frequency access data in main storage area; Use the data mapping table to establish the data association relationship between URTDB and UDB, and build an automatic switching mechanism between hot data and cold data; Create DSI data service interfaces for each professional entity, and divide the data structure into global concept layer, local concept layer, memory physical layer, object layer, and application physical layer; Data reading and writing operations and transaction processing are performed based on the DSI data service interface to form the data foundation of the intelligent entity.

[0048] Specifically, the process of dividing memory areas in the station-level intelligent agent central system and allocating high-speed access memory space for the unified real-time database URTDB first involves the precise planning and allocation of memory resources. URTDB is a memory-based database structure that is specifically used to store and process real-time control data that requires millisecond response. The memory area division adopts the segment page memory management method to divide the physical memory space into multiple functional areas, among which the URTDB area is located in the high-speed access area closest to the CPU. When dividing, the physical memory space is first mapped to the logical address space through the memory page table, and then a dedicated segment is allocated for URTDB in the logical address space, which usually accounts for 30%-40% of the total memory. URTDB is further divided into three functional areas: data area, index area and buffer area. The data area stores actual data records, the index area stores fast query indexes, and the buffer area is used for temporary data exchange. The high-speed access mechanism prevents URTDB data from being exchanged to disk by the operating system through memory locking technology, ensuring that the data always resides in physical memory and achieving sub-millisecond data access performance. The unified entity database UDB is built through data persistence mapping technology. The process of storing and managing parameter configuration information and historical data solves the data persistence storage requirements. UDB adopts a disk-based relational database architecture and is designed as a multi-layer storage structure. Data persistence mapping technology is a mechanism that establishes a mapping relationship between memory data objects and disk storage space. Through this technology, UDB can efficiently manage a large amount of non-real-time data. In the specific implementation, UDB divides the data file into fixed-size data pages. Each data page contains two parts: a data header and a data body. The data header records page attributes and index information, and the data body stores actual data records. When a data record needs to be accessed, the data page position is first located through the index, and then the entire data page is loaded into the memory to complete the data mapping between the memory and the disk. UDB specifically stores parameter configuration information and historical data. Parameter configuration information includes static data such as device parameters, system configuration, and business rules; historical data includes operation records, device status changes, event logs, and other data that need to be stored for a long time. UDB adopts an incremental storage strategy to regularly write new data in batches to the disk to reduce the frequency of I / O operations and improve storage efficiency.

[0049] The system data is classified according to the access frequency, and the high-frequency access data is stored in the L1 cache, the medium-frequency access data is stored in the L2 cache, and the low-frequency access data is stored in the main storage area. The process realizes multi-level acceleration of data access. The access frequency classification is based on the statistical analysis of data access. By recording the number of accesses and time distribution of each data, the access heat value is calculated. The access heat value calculation comprehensively considers the recent access frequency and access time distribution, and gives higher weight to the access in the recent period. According to the heat value, the data is divided into three categories: the data with a heat value higher than the preset threshold T1 is the high-frequency access data, the data with a heat value between T1 and T2 is the medium-frequency access data, and the data with a heat value lower than T2 is the low-frequency access data. The L1 cache is located inside the CPU, with a small capacity but extremely fast speed, usually hundreds of KB; the L2 cache is located between the CPU and the main memory, with a moderate capacity and the second fastest access speed, usually a few MB; the main storage area is the main memory, with a large capacity but relatively slow access speed, usually a few GB. The data cache manager is responsible for maintaining the scheduling and updating of data between different cache levels, and adopts an improved LRU (least recently used) algorithm for cache replacement. When the cache space is insufficient, it prioritizes the elimination of data with low access frequency to ensure that high-frequency access data always maintains the best access performance.

[0050] The data mapping table is used to establish the data association relationship between URTDB and UDB, and the process of constructing the automatic switching mechanism for hot data and cold data realizes the intelligent flow of data between different storage media. The data mapping table is a centralized metadata management structure that records the position correspondence and data attributes of data items in URTDB and UDB. Each record in the mapping table contains fields such as data ID, URTDB address, UDB address, data type, timestamp, access counter, etc. When the data changes, it is first updated in URTDB, and the data item is marked as "dirty data", indicating that it is out of sync with the version in UDB; the data synchronization thread regularly scans the "dirty data" in URTDB and writes it to UDB in batches to complete data persistence. The automatic switching mechanism for hot data and cold data dynamically adjusts the distribution of data between URTDB and UDB according to the change of data access frequency. When the access frequency of low-frequency access data increases, the data loader reads the data from UDB and loads it into URTDB; conversely, when the access frequency of high-frequency access data decreases, the data aging device removes the data from URTDB and only retains it in UDB, freeing up valuable memory resources. During the switching process, the data mapping table is updated in real time to record the current location of the data and ensure that the data access path is correct.

[0051] The DSI data service interface is created for each professional entity, and the process of dividing the data structure into the global concept layer, local concept layer, memory physical layer, object layer and application physical layer establishes a unified data access standard. DSI (Data Service Interface) is a set of standardized APIs that provide a unified data access method for different professional entities. The global concept layer is at the highest level of abstraction, defining the data model and concepts common to the entire site, such as core concepts such as equipment, events, and alarms; the local concept layer defines specialized data models for specific professional fields, such as the smoke sensor and temperature sensor models of the FAS professional entity; the memory physical layer defines the physical storage structure of data in URTDB, including memory address mapping, index structure, etc.; the object layer abstracts data into object-oriented data entities and defines the properties and methods of data objects; the application physical layer defines the physical storage structure of data in UDB, including table structure, field type, etc. When each professional entity accesses data through the DSI interface, it does not need to care about the physical storage location and storage method of the data. It only needs to operate through a unified interface. The DSI service layer is responsible for converting logical operations into corresponding physical operations, which simplifies the complexity of data access and solves the problem of data islands.

[0052] Based on the DSI data service interface, data reading and writing operations and transaction processing are carried out to form the process of the intelligent agent data foundation, which realizes the standardization and consistency of data operations. The DSI interface provides four basic operations: read, write, query and subscribe. The read operation is used to obtain the current value of a single or multiple data items, supporting both synchronous and asynchronous modes; the write operation is used to update the data value, including single-item write and batch write; the query operation is used to retrieve data by condition, supporting complex condition combinations and aggregate functions; the subscription operation allows professional entities to register for data change notifications and automatically receive notifications when data changes. The transaction processing mechanism ensures the atomicity, consistency, isolation and persistence of complex operations, and implements distributed transactions through a two-phase commit protocol, solving the problem of data consistency when multiple professional entities collaborate. Data security control is based on a role-based access control model, which assigns different data access rights to different professional entities to prevent unauthorized access. Through this series of mechanisms, a complete, unified and efficient intelligent agent data foundation is established to provide support for data processing and intelligent decision-making of station-level intelligent entities.

[0053] Taking the platform door system monitoring in rail transit stations as an example, when the PSD professional body needs to monitor the status of all platform doors in real time, it first initiates a data subscription request to URTDB through the DSI interface to subscribe to the status data of all platform doors. The DSI interface converts this request into an internal data subscription operation and searches for the location of all platform door status data in the data mapping table. Since the platform door status is a high-frequency access data, this data has been cached in the L1 cache. The DSI interface directly obtains the latest data from the L1 cache and returns it to the PSD professional body, and establishes a data change notification mechanism. When the status of a platform door changes, the controller sends the status change data to the station-level intelligent body central system through the communication processor. The data update processor first writes the new status to the data area of ​​the URTDB, updates the relevant indexes in the index area, and marks the data as "dirty data". The DSI interface detects the data change and immediately sends a change notification to the PSD professional body that has subscribed to the data. At the same time, the data synchronization thread regularly writes the "dirty data" in the URTDB to the UDB to complete data persistence. If the PSD professional needs to query the historical fault records of a certain platform door, it will initiate a query request through the DSI interface. The DSI interface will identify it as a historical data query and directly retrieve and return the relevant records from the UDB. During the entire process, the PSD professional only needs to call a unified DSI interface, without having to worry about whether the data is stored in URTDB or UDB, and without having to deal with complex issues such as data synchronization and cache management, which greatly simplifies the data access process and improves data processing efficiency.

[0054] In a specific embodiment, the process of executing step S105 may specifically include the following steps: The data acquisition preprocessing unit extracts multi-source heterogeneous data from the unified real-time database URTDB and the unified entity database UDB, performs denoising, normalization and feature extraction on the multi-source heterogeneous data, and forms a standardized data set; Use feature selection algorithms to evaluate and screen standardized data sets, construct multidimensional feature vectors, and generate training sample libraries; A hierarchical federated learning architecture is designed based on the training sample library. After local computation of each professional volume data, only the model gradient information is transmitted to complete model parameter aggregation and build a global deep neural network model. Create a situational awareness engine based on the global deep neural network model, process visual data through convolutional neural networks, use recurrent neural networks to analyze time series data, fuse multimodal information, and identify the current station-level operating status; The situational awareness engine generates a multivariate time series prediction matrix, performs forward predictions on station-level key indicators, and builds a dynamic adjustment mechanism for early warning thresholds; Based on the dynamic adjustment mechanism of the early warning threshold and the preset rule base, decision tree analysis is performed to generate a multi-level set of alternative plans. The multi-level set of alternative plans is optimized and evaluated through the reinforcement learning algorithm to output an intelligent decision-making plan.

[0055] Specifically, the process of extracting multi-source heterogeneous data from the unified real-time database URTDB and the unified entity database UDB by the data acquisition preprocessing unit first involves data source identification and access. Multi-source heterogeneous data refers to a collection of data with different sources and formats. In the rail transit station-level environment, it includes real-time sensor data, video surveillance data, equipment status data, historical operation records and other types. The data acquisition preprocessing unit initiates data requests to URTDB and UDB through the DSI data service interface, specifying query conditions such as data type, time range and data source. For real-time data, it is mainly obtained from URTDB; for historical data and parameter configuration, it is extracted from UDB. After data acquisition, data quality detection is first performed to mark missing values, outliers and duplicate values, and then denoising is performed, using algorithms such as median filtering and wavelet transform to eliminate noise interference in the data. Data normalization processing unifies data of different dimensions to the same scale range. Commonly used normalization methods include minimum-maximum normalization and Z-score normalization. Feature extraction is the process of extracting representative features from raw data, including statistical feature extraction (mean, variance, kurtosis, etc.), time domain feature extraction (trend, periodicity, etc.) and frequency domain feature extraction (spectral analysis, etc.). After this series of processing, a standardized data set with consistent structure and reliable quality is formed, laying the foundation for subsequent analysis. The feature selection algorithm evaluates and screens the standardized data set, constructs a multidimensional feature vector, and generates a training sample library to achieve data dimension reduction and quality improvement. The feature selection algorithm evaluates the importance of each feature and selects the most representative feature subset to reduce data redundancy and noise. In rail transit station-level intelligent agents, the commonly used feature selection methods include filtering, packaging and embedding. The filtering method sorts the features according to statistical indicators (such as information gain, Pearson correlation coefficient, etc.) and selects the top-ranked features; the packaging method uses the performance of the target model as the evaluation criterion and finds the optimal feature subset through strategies such as forward selection or backward elimination; the embedding method integrates feature selection into the model training process, such as the LASSO algorithm based on L1 regularization. The feature evaluation process calculates the importance score for each feature, sorts them in descending order, and selects the top N features to form a feature subset. The multidimensional feature vector combines the selected features into a vector form, and each vector represents the data feature of a time point or a sample. The training sample library consists of a multidimensional feature vector and the corresponding label data. The label data indicates the category (such as the equipment status category) or target value (such as the predicted passenger flow) to which the sample belongs. The training sample library is divided into three parts: training set, validation set, and test set for subsequent model training and evaluation.

[0056] A hierarchical federated learning architecture is designed based on the training sample library. After the data of each professional body is locally calculated and processed, only the model gradient information is transmitted to complete the model parameter aggregation. The process of building a global deep neural network model solves the problems of data privacy protection and collaborative learning. Federated learning is a distributed machine learning method that allows multiple parties to jointly train models without sharing the original data. The hierarchical federated learning architecture is divided into two levels: intra-station federated learning and inter-station federated learning in rail transit station-level agents. In intra-station federated learning, each professional body (such as BAS, FAS, PIS, etc.) acts as a client, and the station-level agent central system acts as a server; in inter-station federated learning, the agents of each station act as clients, and the line network center server acts as an aggregation server. The federated learning process includes five steps: model initialization, local training, gradient upload, parameter aggregation, and model update. First, the server initializes the global model parameters and distributes them to each client; then, each client uses local data to train the model and calculates the parameter gradient; then, the client uploads the gradient information (not the original data) to the server; the server aggregates the gradients uploaded by all clients and updates the global model parameters; finally, the server distributes the updated global model parameters to each client and starts a new round of training. After multiple rounds of iterations, a global deep neural network model that integrates multi-party knowledge is finally obtained. This model has the ability to handle complex pattern recognition and predictive analysis. Based on the global deep neural network model, a situational awareness engine is created. The convolutional neural network is used to process visual data, and the recurrent neural network is used to analyze time series data. The process of integrating multimodal information and identifying the current station-level operating status realizes intelligent perception of complex scenes. The situational awareness engine is an intelligent system that can understand and interpret the current operating environment of rail transit stations. The convolutional neural network (CNN) mainly processes visual data from surveillance cameras, extracts spatial features through multi-layer convolution and pooling, and identifies visual information such as passenger flow density, abnormal behavior, and equipment status. Recurrent neural networks (RNNs), especially long short-term memory (LSTM) variants, mainly process time series data, such as passenger flow changes, equipment operating parameters and other time-dependent data, and capture long-term time series dependencies through memory units. Multimodal information fusion comprehensively analyzes different types of data (vision, sound, sensors, etc.), and uses attention mechanisms to assign different weights to different modal information to form a more comprehensive scene understanding. The output of the situational awareness engine is a comprehensive description of the current station-level operating status, including passenger flow conditions, equipment operating status, safety risk assessment and other dimensions, providing a basis for subsequent decision-making.

[0057] The situational awareness engine generates a multivariate time series prediction matrix, performs forward predictions on station-level key indicators, and constructs a dynamic adjustment mechanism for early warning thresholds to achieve deduction from the current state to future trends. The multivariate time series prediction matrix is ​​a three-dimensional data structure that contains the predicted values ​​of multiple key indicators at multiple time points in the future. Forward prediction uses a recurrent neural network or a time convolutional network to predict the changing trends of key indicators in the future based on historical data and current status. The mutual influence between indicators is considered during the prediction process, and the correlation between indicators is modeled through an attention mechanism or a graph neural network. The dynamic adjustment mechanism for early warning thresholds adaptively adjusts the early warning thresholds of each indicator based on the prediction results and historical data distribution. The traditional fixed threshold method cannot adapt to changes in different scenarios, while the dynamic threshold calculates the statistical distribution and trend changes of historical data, combines expert rules, and updates the early warning threshold in real time. The dynamic threshold calculation takes into account multi-dimensional influencing factors such as time factors (working days / holidays), environmental factors (weather, temperature), and operational factors (passenger flow, train schedules), forming a more accurate early warning judgment standard.

[0058] Based on the dynamic adjustment mechanism of the warning threshold and the preset rule base, a decision tree analysis is performed to generate a multi-level alternative solution set. The multi-level alternative solution set is optimized and evaluated through the reinforcement learning algorithm. The process of outputting the intelligent decision solution realizes the intelligent transformation from warning to decision. The preset rule base is a knowledge base defined by domain experts, which contains processing rules and disposal processes for various scenarios. The decision tree analysis builds a decision path based on the warning information and the rule base to form a hierarchical decision structure. The decision process first identifies the type and severity of the event, then matches the applicable disposal plan according to the rules, considers various constraints and resource limitations, and generates multiple feasible alternative solutions. The multi-level alternative solution set includes three levels: emergency disposal, short-term adjustment, and long-term optimization, which correspond to decision-making needs at different time scales. The reinforcement learning algorithm learns the optimal decision strategy by interacting with the environment. In the rail transit station-level intelligent agent, reinforcement learning mainly uses algorithms such as deep Q network (DQN) or policy gradient to guide the model to learn the optimal decision path through the reward function. The reward function design comprehensively considers multiple factors such as safety, efficiency, resource consumption, and passenger experience, and obtains a comprehensive evaluation score through weighted summation. The reinforcement learning model gradually optimizes the decision-making strategy through multiple simulations and training, and finally selects the best solution from the set of alternative solutions as the output of the intelligent decision-making solution.

[0059] Taking the fire emergency response in rail transit stations as an example, when the smoke detector in the station triggers an alarm, the data acquisition preprocessing unit first obtains multi-source data such as real-time data of smoke sensors, temperature sensor data, wind direction and speed data from URTDB. After these data are denoised (eliminating signal jitter) and normalized (unifying data from different units to the range of 0-1), key features such as smoke concentration change rate, temperature rise rate, etc. are extracted to form a standardized data set. The feature selection algorithm evaluates the importance of each feature, selects the three most representative features of smoke concentration, temperature change rate and wind speed, and constructs a multi-dimensional feature vector. In the hierarchical federated learning architecture, the FAS professional body, BAS professional body and SCADA professional body each train the initial model based on local data, and only uploads the model gradient to the station-level intelligent body central system. The central system aggregates the gradient information to update the global model. The situational awareness engine processes the surveillance camera images through a convolutional neural network and identifies the fire location as the northwest corner of the station hall; analyzes the smoke concentration time series data through a recurrent neural network to determine the fire development trend; and after fusing multimodal information, confirms that the current state is "initial fire, not affecting the main channel". The multivariate time series prediction matrix predicts the smoke spread path and speed in the next 30 minutes and dynamically adjusts the safety risk threshold of each area. Based on the early warning information and the preset rule base, the decision tree analysis generates a set of alternative solutions including "evacuation route planning", "ventilation system adjustment", "firefighting equipment activation" and other aspects. The reinforcement learning algorithm evaluates the comprehensive effect of each solution and finally outputs the optimal intelligent decision-making solution: activate the firefighting equipment in area B, adjust fans No. 1 and No. 3 to smoke exhaust mode, close the escalator, evacuate passengers through exits No. 2 and No. 4, and issue customized evacuation broadcasts through the PA system. The solution is sent to each professional body for execution via the station-level intelligent agent. The entire process from data collection to solution execution takes only 80ms, which reflects the efficiency and intelligence of the rail transit station-level intelligent agent implementation method in emergency handling.

[0060] In a specific embodiment, the process of executing step S106 may specifically include the following steps: The intelligent decision mapper is used to decompose the instructions of the intelligent decision plan and generate a professional execution instruction set; The edge computing task distribution mechanism is used to divide the professional execution instruction set into real-time control instructions and non-real-time control instructions according to functional attributes, and a dual-channel scheduling framework is constructed; The emergency response process is triggered according to the real-time control instructions in the dual-channel scheduling framework, and the scheduling priority evaluator sorts the instruction execution order to form a linkage control timing table; According to the linkage control timing table, control commands are issued to each professional subsystem through the distributed transaction mechanism, and the execution programs in the base of each professional body are started synchronously to produce a coordinated control effect; The status feedback collector obtains the execution status data from each professional subsystem, performs correlation analysis on the execution status data, and generates a control execution report; Adaptive parameter adjustment is performed based on control execution reports and non-real-time control instructions, and the professional body control strategy library is updated to complete the station-level intelligent body autonomous control closed loop.

[0061] Specifically, the intelligent decision mapper decomposes the intelligent decision scheme and generates the professional execution instruction set. The process first involves the structured processing of the decision content. The intelligent decision mapper is a functional module that converts high-level decisions into specific operation instructions. Its core function is to convert abstract decision intentions into specific control instructions. The instruction decomposition process adopts a top-down recursive decomposition strategy. First, the main action goals in the decision scheme are identified, then each goal is subdivided into specific control actions, and finally the control actions are matched to the corresponding professional functions. During the instruction decomposition process, the mapper queries the instruction mapping table, which maintains the correspondence between decision actions and professional functions to ensure that each decomposed instruction can find the execution subject. The professional execution instruction set is an instruction set organized according to the professional type. Each instruction contains attributes such as operation type, operation object, parameter value, execution time and priority. The instruction set is encapsulated in XML format for easy cross-system transmission and parsing. Each instruction is accompanied by a unique identification code for subsequent execution tracking and status feedback. The edge computing task distribution mechanism divides the professional execution instruction set into real-time control instructions and non-real-time control instructions according to functional attributes. The process of building a dual-channel scheduling framework realizes the differentiated processing of tasks. The edge computing task distribution mechanism is a technical architecture that delegates computing tasks to the vicinity of the data source for processing. In rail transit station-level intelligent entities, edge nodes are mainly the bases of various professional entities. When distributing tasks, the functional attributes of each instruction in the instruction set are first analyzed, and the instructions are divided into two categories according to time sensitivity: real-time control instructions are instructions with extremely high time response requirements (usually requiring millisecond-level response), such as emergency parking, fire alarm response, etc.; non-real-time control instructions have relatively loose time requirements (second-level or minute-level response), such as environmental parameter adjustment, equipment maintenance instructions, etc. The dual-channel scheduling framework is a parallel processing architecture that includes two independent processing pipelines, the real-time channel and the non-real-time channel. The real-time channel adopts interrupt-driven mode, has the highest processing priority, and specializes in processing real-time control instructions; the non-real-time channel adopts polling mode and processes non-real-time control instructions according to the preset scheduling strategy. Through this architectural design, it is ensured that time-critical tasks can get immediate response without blocking the execution of other ordinary tasks.

[0062] According to the real-time control instructions in the dual-channel scheduling framework, the emergency response processing flow is triggered, and the scheduling priority evaluator sorts the execution order of the instructions to form a linkage control timing table, which establishes accurate task execution timing control. The emergency response processing flow is a set of special processing mechanisms for real-time control instructions, which includes four main links: instruction parsing, resource preparation, execution scheduling and status monitoring. The scheduling priority evaluator is a functional module responsible for determining the execution order of instructions in the emergency response processing flow, and its core algorithm is based on a multi-factor weighted scoring mechanism. During the evaluation process, the internal priority tag, time sensitivity, resource dependency and other characteristic information of each instruction are first extracted; then, the comprehensive priority score is calculated according to the preset weight; finally, the execution queue is formed in descending order according to the priority score. For instructions with the same priority, they are sorted according to the arrival time; for instructions with dependencies, ensure that the preceding instruction is executed before the succeeding instruction. The linkage control timing table is a data structure that records the execution plan of the instruction in detail, including the execution time point, execution subject, expected execution time, preconditions and subsequent operations of each instruction, providing accurate execution references for each professional body to ensure the timing coordination of complex linkage operations.

[0063] According to the linkage control time table, control commands are issued to each professional subsystem through the distributed transaction mechanism, and the execution program in each professional base is started synchronously. The process of generating collaborative control effects realizes cross-system collaborative operations. The distributed transaction mechanism is a technical means to ensure that operations across multiple systems are either successfully completed or rolled back, and is suitable for scenarios that require multi-system collaboration. In the rail transit station-level intelligent body, distributed transactions adopt a two-phase commit protocol, including a preparation phase and a commit phase. In the preparation phase, the transaction coordinator sends a pre-execution request to each professional subsystem, and each subsystem verifies the validity of the instruction and locks the required resources, and then returns a ready or rejects the response; in the commit phase, if all subsystems return ready, the coordinator sends a formal execution command, otherwise it sends a cancel execution command. The control command is issued using the LiteIP secure communication bus based on FPGA to ensure the real-time and reliability of command transmission. After receiving the control command, each professional base calls the corresponding execution program according to the command content. Before the program is executed, parameter verification and resource checks will be performed to ensure that the execution conditions are met. The collaborative control effect is the result of multiple professional subsystems performing their respective tasks in a predetermined sequence and jointly completing complex control objectives. For example, the linkage control after a fire alarm involves the collaborative work of multiple systems such as fire protection, ventilation, broadcasting, and access control.

[0064] The state feedback collector obtains execution status data from each professional subsystem, performs correlation analysis on the execution status data, and generates a control execution report. The process establishes an execution monitoring and evaluation mechanism. The state feedback collector is a distributed data acquisition component that is responsible for collecting the execution status information of each professional subsystem in real time. The collection process adopts a combination of push and pull modes: for key state changes, the professional subsystem actively pushes the status data; for routine state monitoring, the collector regularly pulls the status data. The execution status data contains multiple dimensions such as instruction execution progress, execution results, resource consumption, and abnormal information, and is transmitted in a structured data format. Correlation analysis is the process of integrating and correlating state data from different professional subsystems. The state data in the same time period is combined through timestamp matching to construct a multidimensional state view. During the analysis process, the trigger relationship of state changes is identified through causal reasoning, the typical state sequence pattern is identified through pattern matching, and the deviation and problems in the execution process are discovered through anomaly detection. The control execution report is the output result of correlation analysis, which systematically records the entire process of instruction execution, including the overall execution status, the execution status of each stage, abnormal events and handling methods, resource usage, and execution effect evaluation.

[0065] Based on the control execution report and non-real-time control instructions, adaptive parameter adjustment is performed, the control strategy library of the professional body is updated, and the process of completing the autonomous control closed loop of the station-level intelligent body is realized to achieve continuous optimization of the system. Adaptive parameter adjustment is a technology that dynamically adjusts control parameters according to execution feedback. It is mainly used in non-real-time control scenarios in rail transit station-level intelligent bodies. The adjustment process first extracts key performance indicators from the control execution report, such as response time, resource utilization, control accuracy, etc.; then, it is compared with the expected target and the performance gap is calculated; then, according to the performance gap and the preset adjustment rules, the parameter adjustment direction and amplitude are determined; finally, the adjusted parameters are applied to the relevant control algorithm. The professional body control strategy library is a knowledge base that stores various control strategies and parameter configurations. It is organized in a hierarchical structure, including the basic strategy layer, the scenario strategy layer, and the optimization strategy layer. The strategy update mechanism marks inefficient or problematic strategies according to the evaluation results of the control execution report, introduces optimized new strategies, and updates the strategy scores. The autonomous control closed loop of the station-level intelligent body refers to a complete control cycle from perception, decision-making to execution, feedback, and optimization. Through continuous data collection, analysis, execution, and feedback, an adaptive control system is formed to continuously improve control performance and intelligence level.

[0066] Taking the emergency evacuation of large passenger flow at a rail transit station as an example, the intelligent decision mapper first receives the intelligent decision plan "Start large passenger flow evacuation plan A", and converts it into multiple specific instructions through instruction decomposition, including "PIS-display evacuation guidance information", "PA-play evacuation broadcast", "BAS-adjust air conditioning volume", "PSD-adjust door opening time", etc., forming a professional execution instruction set. The edge computing task distribution mechanism analyzes the instruction attributes, identifies "PIS-display evacuation guidance information" and "PA-play evacuation broadcast" as real-time control instructions and assigns them to the real-time channel, and identifies "BAS-adjust air conditioning volume" as a non-real-time control instruction and assigns it to the non-real-time channel. The scheduling priority evaluator calculates the priority of each real-time instruction. "PA-play evacuation broadcast" gets the highest priority of 90 points because it is directly related to personnel safety, and "PIS-display evacuation guidance information" gets 85 points. Based on this, a linkage control time table is formed: t=0 seconds, PA plays evacuation broadcast; t=0.5 seconds, PIS displays evacuation guidance information; t=2 seconds, PSD adjusts the door opening time. The distributed transaction mechanism starts a two-phase commit: first, a preparation request is sent to the three professional bodies of PA, PIS, and PSD. After the three professional bodies confirm that the resources are ready, they return a ready response; then a formal execution command is sent, and the three professional bodies start the execution program synchronously. The PA professional body activates the emergency broadcast, the PIS professional body switches the display screen content, and the PSD professional body modifies the control parameters to jointly complete the evacuation guidance. The status feedback collector collects the execution status in real time: the PA professional body returns "broadcast activation successful, volume 85 decibels", the PIS professional body returns "display content switching completed, 15 display screens are all updated", and the PSD professional body returns "parameter modification successful, door opening time adjusted to 12 seconds". The correlation analysis integrates these data to generate a control execution report, recording the execution status, time nodes and effect data of the entire evacuation process. Based on the execution report, it was found that the PSD door opening time was slightly insufficient. Combined with the passenger flow data, the adaptive parameter adjustment mechanism adjusted the door opening time from 12 seconds to 15 seconds, and updated the relevant parameters in the professional control strategy library to provide better configuration for similar scenarios in the future, thereby completing the station-level intelligent agent autonomous control closed loop. The whole process demonstrates the efficient collaborative control capabilities of the rail transit station-level intelligent agent implementation method in complex scenarios.

[0067] The above describes the rail transit station-level intelligent agent implementation method in the embodiment of the present application. The following describes the rail transit station-level intelligent agent implementation system in the embodiment of the present application. Figure 5 In the embodiment of the present application, one embodiment of the rail transit station-level intelligent agent implementation system includes: Building modules are used to build a universal professional body base, configure a real-time embedded operating system, and form a modular professional body architecture; The transmission module is used to design a secure communication bus according to the professional body architecture, perform differentiated transmission processing on data, and establish a high-speed communication network; The allocation module is used to combine multiple ARM processors into a station-level multi-core CPU cluster based on a high-speed communication network, dynamically allocate workloads, and build a station-level intelligent agent central system; The hierarchical module is used to establish a unified real-time database URTDB and a unified entity database UDB based on the station-level intelligent body central system, and to manage the system data in a hierarchical manner to form the intelligent body data foundation; Prediction module, used to perform multi-dimensional analysis and prediction of real-time station-level operation data and generate intelligent decision-making solutions; The control module is used to perform edge computing and distributed linkage control based on intelligent decision-making solutions, coordinate the scheduling of various professional subsystems, and realize autonomous control of station-level intelligent bodies.

[0068] Through the collaboration of the above components, a general professional body base is built through a high-performance 64-bit ARM processor, and a real-time embedded operating system is configured to form a modular professional body architecture, which effectively solves the problem of independent deployment and difficulty in collaborative work of various professional subsystems in traditional rail transit station-level systems, and realizes efficient integration and flexible configuration of hardware resources; the LiteIP secure communication bus is implemented based on the FPGA designed based on the professional body architecture, and the data is transmitted and processed differently using a dynamic key encryption mechanism, which not only reduces the communication delay from the hundreds of milliseconds of traditional TCP / IP to the sub-millisecond level, but also greatly improves the data transmission efficiency and security through the differentiated processing of global shared data streams, group shared data streams and point-to-point data streams; multiple ARM processors are combined into a station-level multi-core CPU cluster, and the workload is dynamically allocated through a task priority allocation algorithm. The constructed station-level intelligent body central system makes full use of computing resources, and the response time of key tasks is shortened to less than 20ms, which is much better than the 300ms response time of the traditional system; a unified real-time database URTDB and a unified entity database UDB are established, and multi-level cache mechanisms and DSI data service interfaces are used to The system data is managed in layers, which completely solves the "data island" problem in traditional systems, enables various professional entities to easily access global data, and provides a complete data foundation for intelligent analysis. In particular, in terms of artificial intelligence applications, this solution integrates federated learning and deep neural network technology, so that various professional entities can share model knowledge under the premise of protecting local data privacy, significantly improving the comprehensive performance of AI models. For example, the situational awareness engine successfully integrates and analyzes multimodal data (visual, time series, sensors, etc.) through the combination of convolutional neural networks and recurrent neural networks, accurately identifies complex scene states, predicts trend changes, and provides a reliable basis for intelligent decision-making. Finally, through edge computing and distributed linkage control, comprehensive autonomous control capabilities from millisecond-level response emergency control to long-term optimization are achieved. Especially in emergency situations such as fire and large passenger flow, the entire processing flow from data collection, analysis and processing to command execution only takes 80ms, which is nearly 4 times faster than the traditional system, greatly improving the safety level and operational efficiency of rail transit stations. At the same time, through adaptive parameter adjustment and continuous updating of the professional control strategy library, the system can continuously evolve and optimize to adapt to changing operating environments and needs.

[0069] Reference Figure 6 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 6As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0070] Those skilled in the art will understand that Figure 6 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0071] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0072] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (Double Data Rate SDRAM, referred to as DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0073] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0074] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0075] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A rail transit station-level agent implementation method, characterized in that: include: Build a universal professional body base, configure a real-time embedded operating system, and form a modular professional body architecture; Design a secure communication bus based on the professional architecture, perform differentiated transmission processing on data, and establish a high-speed communication network; Combine multiple ARM processors into a station-level multi-core CPU cluster based on a high-speed communication network, dynamically allocate workloads, and build a station-level intelligent agent central system; Relying on the station-level intelligent body central system, a unified real-time database URTDB and a unified entity database UDB are established, and system data is managed in layers to form the intelligent body data foundation; Conduct multi-dimensional analysis and prediction of real-time station-level operation data to generate intelligent decision-making solutions; Based on intelligent decision-making solutions, edge computing and distributed linkage control are executed to coordinate the scheduling of various professional subsystems and realize autonomous control of station-level intelligent bodies.

2. The rail transit station-level agent implementation method according to claim 1, characterized in that: The construction of a universal professional body base, configuration of a real-time embedded operating system, and formation of a modular professional body architecture include: By setting up multiple timing control units to accurately control the clock signal of the ARM processor, real-time processing capability with a time resolution of 1ms is obtained; The real-time embedded operating system is divided into a real-time task layer and an application layer, and the real-time task layer manages hardware resource allocation; A protocol adaptation conversion module is built into the universal professional body base to convert different communication protocols into a unified standard data format; According to the functions, there are seven types of professional bodies, namely BAS professional body, FAS professional body, PIS professional body, PA professional body, PSD professional body, ATS professional body and SCADA professional body, which constitute professional body groups; The computing resources and storage space of the professional base are dynamically configured through the private cloud resource allocation algorithm to form an elastic computing architecture; A corresponding professional body is selected from the professional body group to establish a communication connection with a corresponding PLC device to form a complete modular professional body architecture.

3. The rail transit station-level agent implementation method according to claim 1, characterized in that: The secure communication bus is designed according to the professional architecture, data is transmitted and processed in a differentiated manner, and a high-speed communication network is established, including: Customize the LiteIP protocol stack through FPGA hardware circuits to build a high-performance protocol processing unit; The communication data flow is divided into three types: global shared data flow, group shared data flow and point-to-point data flow, and different transmission priorities are assigned to each type; Generate dynamic key pairs based on the principles of elliptic curve cryptography and apply encryption strategies of different strengths to different types of data streams; The data packets are marked by hardware timestamp mechanism and transmission frame sequence number to form a reliable data packet transmission confirmation mechanism; Automatically calculate the optimal transmission path based on topology information and perform intelligent routing forwarding on data packets; The communication bandwidth dynamic allocation technology is used to coordinate and manage the data transmission between multiple professional entities and establish a high-speed communication network.

4. The rail transit station-level agent implementation method according to claim 1, characterized in that: The method combines multiple ARM processors into a station-level multi-core CPU cluster based on a high-speed communication network, dynamically allocates workloads, and builds a station-level intelligent agent central system, including: Multiple ARM processors are physically networked and connected according to the master-slave architecture to form a processor array; Defining roles of processors in the processor array and dividing them into three types: main control processor, computing processor and communication processor; Establishing an inter-processor communication channel based on the high-speed communication network and creating a processor resource pool; Monitor the processor load in the processor resource pool in real time through a status monitoring mechanism to generate a resource status diagram; Identify the task type and urgency according to the resource status diagram, and assign priority weight coefficients to different tasks; The workload is dynamically allocated according to the priority weight coefficient, and the tasks are distributed to the corresponding processors for execution, thereby building a station-level intelligent agent central system.

5. The rail transit station-level agent implementation method according to claim 1, characterized in that: The station-level intelligent agent central system is used to establish a unified real-time database URTDB and a unified entity database UDB, and hierarchical management of system data is performed to form an intelligent agent data foundation, including: Dividing memory areas in the station-level intelligent agent central system to allocate high-speed access memory space for the unified real-time database URTDB; A unified entity database (UDB) is built through data persistence mapping technology to store and manage parameter configuration information and historical data; Classify system data according to access frequency, store high-frequency access data in L1 cache, medium-frequency access data in L2 cache, and low-frequency access data in main storage area; Use the data mapping table to establish the data association relationship between URTDB and UDB, and build an automatic switching mechanism between hot data and cold data; Create DSI data service interfaces for each professional entity, and divide the data structure into global concept layer, local concept layer, memory physical layer, object layer, and application physical layer; Data read and write operations and transaction processing are performed based on the DSI data service interface to form the intelligent body data foundation.

6. The rail transit station-level agent implementation method according to claim 1, characterized in that: The multi-dimensional analysis and prediction of real-time station-level operation data to generate intelligent decision-making solutions includes: Extracting multi-source heterogeneous data from the unified real-time database URTDB and the unified entity database UDB through a data acquisition preprocessing unit, performing denoising, normalization and feature extraction processing on the multi-source heterogeneous data to form a standardized data set; Using a feature selection algorithm to evaluate and screen the standardized data set, construct a multidimensional feature vector, and generate a training sample library; A hierarchical federated learning architecture is designed based on the training sample library, and only the model gradient information is transmitted after the data of each professional volume is locally calculated and processed, so as to complete the aggregation of model parameters and build a global deep neural network model; A situational awareness engine is created based on the global deep neural network model, visual data is processed through a convolutional neural network, time series data is analyzed using a recurrent neural network, multimodal information is integrated, and the current station-level operating status is identified; The situational awareness engine generates a multivariate time series prediction matrix, performs forward predictions on station-level key indicators, and builds a dynamic adjustment mechanism for early warning thresholds; Based on the dynamic adjustment mechanism of the early warning threshold and the preset rule base, a decision tree analysis is performed to generate a multi-level set of alternative solutions. The multi-level set of alternative solutions is optimized and evaluated through a reinforcement learning algorithm to output an intelligent decision-making solution.

7. The rail transit station-level agent implementation method according to claim 1, characterized in that: The intelligent decision-making scheme is used to execute edge computing and distributed linkage control, coordinate the scheduling of various professional subsystems, and realize autonomous control of station-level intelligent bodies, including: Decomposing the intelligent decision-making scheme through an intelligent decision mapper to generate a professional execution instruction set; The edge computing task distribution mechanism is used to divide the professional body execution instruction set into real-time control instructions and non-real-time control instructions according to functional attributes, and a dual-channel scheduling framework is constructed; According to the real-time control instructions in the dual-channel scheduling framework, an emergency response process is triggered, and the scheduling priority evaluator sorts the execution order of the instructions to form a linkage control timing table; According to the linkage control timing table, control commands are issued to each professional subsystem through a distributed transaction mechanism, and the execution programs in the bases of each professional body are started synchronously to produce a coordinated control effect; The status feedback collector obtains execution status data from each professional subsystem, performs correlation analysis on the execution status data, and generates a control execution report; Based on the control execution report and non-real-time control instructions, adaptive parameter adjustment is performed, the professional body control strategy library is updated, and the station-level intelligent body autonomous control closed loop is completed.

8. A rail transit station-level intelligent agent implementation system, used to implement the rail transit station-level intelligent agent implementation method according to any one of claims 1 to 7, characterized in that: include: Building modules are used to build a universal professional body base, configure a real-time embedded operating system, and form a modular professional body architecture; The transmission module is used to design a secure communication bus according to the professional body architecture, perform differentiated transmission processing on data, and establish a high-speed communication network; The allocation module is used to combine multiple ARM processors into a station-level multi-core CPU cluster based on a high-speed communication network, dynamically allocate workloads, and build a station-level intelligent agent central system; The hierarchical module is used to establish a unified real-time database URTDB and a unified entity database UDB based on the station-level intelligent body central system, and to manage the system data in a hierarchical manner to form the intelligent body data foundation; Prediction module, used to perform multi-dimensional analysis and prediction of real-time station-level operation data and generate intelligent decision-making solutions; The control module is used to perform edge computing and distributed linkage control based on intelligent decision-making solutions, coordinate the scheduling of various professional subsystems, and realize autonomous control of station-level intelligent bodies.

9. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and is characterized in that when the processor executes the computer program, it implements the rail transit station-level intelligent agent implementation method described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the processor executes the rail transit station-level intelligent agent implementation method as described in any one of claims 1 to 7.

Images