An intelligent mapping method and system based on an AI large model

Abstract

The application provides an intelligent mapping method and system based on an AI large model, comprising: acquiring station dynamic data and preprocessing to generate fusion data; generating a running map based on the fusion data through a preset intelligent mapping model; sending the running map to a visual terminal and a vehicle-mounted controller, wherein the vehicle-mounted controller controls the vehicle running state according to the running map; and feeding back the vehicle running state to the intelligent mapping model to dynamically adjust the running map. The application solves the problems of low efficiency of existing rail transit running map compilation and difficulty in timely handling of emergency situations.

Description

Technical Field

[0001] This invention relates to the field of rail transit control technology, and in particular to an intelligent mapping method and system based on an AI large model. Background Technology

[0002] The city's rail transit train timetable (a graphic representation of a train timetable) specifies the arrival and departure times of trains at stations along the route. It serves as the premise and foundation for coordinating the work of various departments and ensuring the safety of train operations and the quality of passenger service.

[0003] Currently, emergency response plans for urban rail transit are primarily prepared manually, mostly in text form for emergency training and handling guidance. Alternatively, semi-automatic tools are used, employing fixed algorithms to generate basic operating lines, with manual adjustments made for conflicts (such as train delay compensation). However, actual emergency response relies more heavily on the experience and on-site judgment of emergency personnel, failing to fully utilize the capabilities of prepared emergency response plans. Furthermore, current emergency response plan systems for urban rail transit tend to be data management databases for text-based plans, offering only basic functions such as classification and retrieval, making them less practical in the event of emergencies. Problems include: delayed response (manual adjustments to operating schedules take 30+ minutes during sudden changes in passenger flow); inefficient conflict resolution (manual intervention is required in 85% of train delay propagation scenarios); resource waste (conservative scheduling leads to low vehicle utilization); and an inability to predict the impact of equipment failures on the operating schedule.

[0004] Therefore, in the field of urban rail transit emergency plan preparation, there is an urgent need to establish a digital and structured operation diagram preparation system that can integrate emergency organizational structure, emergency teams, departmental collaboration, disaster preparedness resource guarantee, and other contents, and can be applied to the actual emergency process of urban rail transit. Summary of the Invention

[0005] This invention provides an intelligent map compilation method and system based on an AI large model to solve the problems of low efficiency in the compilation of existing rail transit timetables and difficulty in handling emergencies in a timely manner.

[0006] This invention provides an intelligent graph compilation method based on a large AI model, comprising: Acquire dynamic station data and preprocess it to generate fused data; Based on the fused data, a running diagram is generated using a preset intelligent mapping model; The operation diagram is sent to the visualization terminal and the vehicle controller, and the vehicle controller controls the vehicle's operating status according to the operation diagram. The vehicle's operating status is fed back to the intelligent mapping model, and the operating map is dynamically adjusted.

[0007] According to the intelligent mapping method based on an AI large model provided by the present invention, the step of acquiring station dynamic data and preprocessing it to generate fused data includes: Real-time train location information is obtained through the train positioning system, signal equipment information is obtained through the trackside signal machine, and passenger flow information is obtained through the gate throughput and real-time collected pedestrian flow images. The real-time train location information, signal equipment information, and passenger flow information are spatiotemporally aligned, abnormal information is filtered, and compressed and fused to generate fused data.

[0008] According to the present invention, an intelligent graphing method based on an AI large model is provided, wherein the step of generating a running graph based on the fused data through a preset intelligent graphing model includes: Based on the fused data, a spatiotemporal dynamic matrix is ​​constructed using the intelligent mapping model to generate a conflict heatmap. A scheduling strategy is generated based on the conflict heatmap using a reinforcement learning algorithm. Based on the aforementioned scheduling strategy, the execution path is optimized using the ant colony algorithm. A running graph is generated according to the scheduling strategy corresponding to the optimal execution path, and the running graph is stored in a lightweight container.

[0009] According to the present invention, an intelligent mapping method based on an AI large model is provided, wherein the method generates a conflict heatmap by constructing a spatiotemporal dynamic matrix based on the fused data and the intelligent mapping model, comprising: The fused data is spatially discretized using the intelligent mapping model, divided into multiple grid units, and a spatiotemporal dynamic matrix is ​​constructed. A continuous time window is created at preset time intervals. Within each time window, the risk of rear-end collision, platform overcapacity, and route conflict in each grid cell of the spatiotemporal dynamic matrix are detected, and a conflict heat map is generated.

[0010] According to the present invention, an intelligent mapping method based on a large AI model is provided, wherein the generation of a scheduling strategy based on the conflict heatmap using a reinforcement learning algorithm includes: A dynamic decision-making environment is constructed based on the aforementioned conflict heatmap using a reinforcement learning algorithm. Action design is performed within the dynamic decision-making environment, and the value of the designed actions is calculated based on a preset reward mechanism. The scheduling strategy that resolves the most conflicts is determined based on the value of the actions.

[0011] According to the present invention, an intelligent mapping method based on an AI large model is provided, wherein the running map is sent to a visualization terminal and an on-board controller, and the on-board controller controls the vehicle's operating status according to the running map, comprising: The operation diagram is displayed through a visual terminal, and the operation diagram is broken down and identified through the vehicle controller to obtain control information; The vehicle's operating status is controlled based on the control information.

[0012] This invention also provides an intelligent graph editing system based on a large AI model, the system comprising: The data processing module is used to acquire station dynamic data and preprocess it to generate fused data; The runtime graph generation module is used to generate runtime graphs based on the fused data using a preset intelligent graph generation model; The operation control module is used to send the operation diagram to the visualization terminal and the vehicle controller, and the vehicle controller controls the vehicle operation status according to the operation diagram; The dynamic adjustment module is used to feed back the vehicle's operating status to the intelligent mapping model and dynamically adjust the operating map.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the intelligent graphing method based on the AI ​​large model as described above.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the intelligent graphing method based on the AI ​​large model as described above.

[0015] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the intelligent graphing method based on the AI ​​large model as described above.

[0016] This invention provides an intelligent timetable creation method and system based on an AI large model. After acquiring dynamic data of stations, the system designs operation plans through an intelligent timetable creation model. By automatically analyzing relevant data, it can quickly generate timetables that adapt to emergencies, improving the efficiency of timetable design. Furthermore, it can dynamically adjust the timetable in real time according to changes in station status, quickly resolve emergencies, and improve the throughput of rail transit. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1This is a flowchart illustrating the intelligent graph compilation method based on a large AI model provided by the present invention.

[0019] Figure 2 This is a schematic diagram of the module connections of the intelligent graphing system based on a large AI model provided by the present invention.

[0020] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention.

[0021] Figure label: 110: Data processing module; 120: Run chart compilation module; 130: Run control module; 140: Dynamic adjustment module; 310: Processor; 320: Communication interface; 330: Memory; 340: Communication bus. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0023] The following is combined Figure 1 The present invention describes an intelligent mapping method based on an AI large model, including: step 100, acquiring station dynamic data and preprocessing it to generate fused data.

[0024] Specifically, real-time train location information is obtained through the train positioning system, signal equipment information is obtained through the trackside signal machine, and passenger flow information is obtained through the gate throughput and real-time collected images of pedestrian flow. The real-time train location information, signal equipment information, and passenger flow information are spatiotemporally aligned, abnormal information is filtered, and compressed and fused to generate fused data.

[0025] In this invention, the real-time position information of the train is obtained through the ATS (Automatic Train Monitoring System); rail transit generally adopts a communication-based train control system. The train calculates its own position in real time by combining the onboard position detection unit with the speed and distance measurement unit and transponder / beacon information; the CI (Computer Interlocking System) uses track circuits to detect the status of the track section occupied by the train ("occupied" or "idle"), providing the ATS with coarse-grained train position information (i.e., which track section it is in).

[0026] During track circuit positioning, the track is divided into physically or electrically isolated sections (called "track circuits"). Each section is connected at both ends by a track relay. When there is no train in a section, the relay coil is energized, and the armature is pulled up, indicating that the section is "idle." When a train wheelset enters the section, the axle short-circuits the left and right rails, causing the track relay coil in that section to lose power (or a specific change in current or phase), and the relay armature falls, indicating that the section is "occupied." The interlocking system monitors the status changes of all track relays in real time. Once the status of a relay in a section changes from "idle" to "occupied," or from "occupied" to "idle," the CI system immediately sends this status change information to the ATS. By receiving the track circuit occupancy information sent by the CI system and combining it with the route map, the ATS system can determine which section the train is currently in, forming the basis for train positioning.

[0027] During speed measurement and distance determination for positioning, speed sensors, typically pulse-type speed sensors or Doppler radar, are installed on the train's axles or motors. The onboard computer continuously reads the pulse signals from the speed sensors (or radar speed data) and integrates them to convert them into travel distance (displacement). Starting from a known starting point (e.g., the transponder position), the train continuously calculates the travel distance relative to that point to obtain its relative position. The ATS (Automatic Train Protection) system, by receiving position information reported by the onboard system, can obtain a more accurate position than the track circuitry. However, this position information requires periodic calibration by external equipment; otherwise, errors will accumulate and affect the reliability of the ATS.

[0028] When locating a train via a transponder, the train's Base Query Antenna (BTM) emits electromagnetic energy to activate the transponder as it passes overhead. The transponder then sends its internally stored absolute position information (a unique geographical location within the railway network) to the onboard equipment. Upon receiving this information, the onboard equipment immediately resets its relative position, calculated based on speed and distance measurements, to the precise absolute position provided by the transponder. This eliminates accumulated errors. After calibrating its position, the onboard equipment sends the updated precise position information (typically including the location and timestamp of the transponder passage) to trackside equipment, ultimately reaching the ATS (Automatic Train Protection) system. The ATS system accurately knows when and where the train passed which transponder, and by combining this information with speed and distance measurement data, it can obtain the train's precise position at any given moment between transponders. Simultaneously, transponder information is also used to verify whether the train's reported consecutive positions conform to logic and the actual track conditions. This significantly enhances the accuracy and reliability of ATS monitoring.

[0029] In moving block signaling (MTS) positioning, precise train positioning is crucial for system operation. The onboard controller calculates a high-precision continuous absolute position by integrating speed and distance measurement data with transponder data. The train actively and periodically transmits its precise position, speed, and direction of travel to the upper-level area controller (ATS) via a vehicle-to-ground wireless communication network (such as WLAN, LTE-M, or CBTC dedicated communication). The ATS also receives track circuit occupancy status information from the interlocking system. The ATS then transmits its verified safe position information for each train to the ATS system. Simultaneously, the ATS can also directly receive position information from the train required for non-safety control purposes.

[0030] The ATS system simultaneously receives reports from interlocking (track circuit occupancy), train position, and transponder passage. This integrated information is used to display the train's symbol position on the route map at the dispatcher's workstation in real time.

[0031] Information on signaling equipment is obtained through trackside signals, and the number of people entering and exiting the station is counted by the number of turnstiles. Furthermore, the real-time images of pedestrian flow are analyzed to obtain overall passenger flow information.

[0032] After acquiring multiple station dynamic data, spatiotemporal alignment is performed to unify the timestamps and spatial coordinate systems; outliers are filtered out and abrupt data such as instantaneous position offsets >100 meters are automatically removed; finally, Delta encoding is used to reduce transmission load and obtain fused data.

[0033] Step 200: Generate a running diagram based on the fused data using a preset intelligent diagramming model.

[0034] Specifically, a spatiotemporal dynamic matrix is ​​constructed based on the fused data using the intelligent mapping model to generate a conflict heatmap; A scheduling strategy is generated based on the conflict heatmap using a reinforcement learning algorithm. Based on the aforementioned scheduling strategy, the execution path is optimized using the ant colony algorithm. A running graph is generated according to the scheduling strategy corresponding to the optimal execution path, and the running graph is stored in a lightweight container.

[0035] In this invention, generating a conflict heatmap for various emergencies within the station helps to quickly and intelligently generate subsequent scheduling plans.

[0036] The fused data is spatially discretized using the intelligent mapping model, divided into multiple grid units, and a spatiotemporal dynamic matrix is ​​constructed. A continuous time window is created at preset time intervals. Within each time window, the risk of rear-end collision, platform overcapacity, and route conflict in each grid cell of the spatiotemporal dynamic matrix are detected, and a conflict heat map is generated.

[0037] Specifically, during the spatial discretization process, the track network is divided into 100m×100m grid units, with the station area further refined to 10m precision; time slices create continuous time windows at 5-second intervals, with each window containing multi-dimensional data such as train position, signal status, and passenger flow load.

[0038] Rear-end collision risks include: triggering an early warning when the estimated time of the following train entering a grid and the time of the preceding train leaving the grid are less than 120 seconds (including braking distance buffer); platform overcapacity risks include: generating a red alarm when real-time passenger flow + predicted number of passengers entering the station within the next 2 minutes > 130% of the platform's design capacity; route conflicts include: detecting overlapping occupancy times of different trains on the same turnout / track section, and determining a conflict when the overlap duration exceeds the signal system switching time (45 seconds). The final result is a conflict heatmap including conflict location, estimated occurrence time, type (A / B / C level), and impact range.

[0039] Based on the conflict heatmap, a scheduling strategy is generated using a reinforcement learning algorithm, including: A dynamic decision-making environment is constructed based on the aforementioned conflict heatmap using a reinforcement learning algorithm. Action design is performed within the dynamic decision-making environment, and the value of the designed actions is calculated based on a preset reward mechanism. The scheduling strategy that resolves the most conflicts is determined based on the value of the actions.

[0040] In this invention, a dynamic decision-making environment is constructed by using a reinforcement learning framework within an intelligent mapping model to receive conflict heatmaps in real time and combining them with multi-dimensional state parameters such as the train's current position deviation, the platform's accumulated passenger flow, and the current energy consumption ratio.

[0041] When designing the action space, the basic strategy set includes four core operations: stop time adjustment (±5~30 seconds), operation mode degradation (full speed → economy → energy saving), temporary skipping, and route change. Action combination rules stipulate that stop compression and operation degradation can be implemented in parallel, but skipping operations must be used at least 3 stops apart. Based on a preset reward mechanism, the Deep Deterministic Policy Gradient Algorithm (DDPG) continuously explores action value, prioritizing scheduling strategies that can resolve multiple conflicts simultaneously.

[0042] After generating scheduling strategies, each strategy (e.g., "compress station A for 20 seconds of downtime + downgrade segment B operation") is treated as a solution path node, forming a multi-dimensional decision topology. A pheromone guidance mechanism is established, where the initial pheromone concentration is positively correlated with the value assessment of the RL strategy, and the path selection probability is calculated using a preset formula.

[0043] Through collaborative ant colony search, the optimal execution path is selected, including: Group exploration: Divide 50 "ants" into three groups (fast group / balanced group / conservative group), and each group adopts a different exploration strategy; Path construction involves each ant starting from its current state and selecting action nodes based on pheromone concentration and heuristic rules. Dynamic feedback: when a path reduces the red area on the conflict heatmap, the pheromone concentration of that path is immediately increased. Elite retention: In each iteration, the top 3 paths that resolve the most conflicts are retained to accelerate the diffusion of high-quality solutions.

[0044] An execution graph is generated based on the scheduling strategy corresponding to the optimal execution path, and stored in a lightweight container for easy real-time retrieval. Furthermore, seamless transfer of train status across interlocking zones is achieved through lightweight container management (JSON format data packets) (e.g., a two-phase commit protocol ensures data consistency).

[0045] Traditionally, when adjusting train schedules at stations, manual adjustments require station-by-station verification of conflict points. In the event of sudden delays or equipment failures, dispatchers must recalculate parameters such as stop times and route matching, which takes an average of 25-40 minutes and leads to a wider spread of delays. This invention utilizes an intelligent scheduling model to integrate multiple data sources in real time, including Train Positioning System (ATS), signal status (interlocking), and passenger flow (AFC). It employs spatiotemporal matrix dynamic modeling (predicting conflict points) and combines reinforcement learning (RL) and ant colony optimization to generate a global optimization solution (such as reducing stop times and dynamically downgrading operating levels) within ≤90 seconds.

[0046] Step 300: The operation diagram is sent to the visualization terminal and the vehicle controller, and the vehicle controller controls the vehicle operation status according to the operation diagram.

[0047] Specifically, the operation diagram is displayed through a visual terminal, and the operation diagram is broken down and identified by the vehicle controller to obtain control information; The vehicle's operating status is controlled based on the control information.

[0048] In this invention, conflict points are marked with red warnings on the visualization terminal, while the optimized solutions provided by the operation diagram are highlighted in green. Simultaneously, the operation diagram is sent to the vehicle controller. Upon receiving the operation diagram information, the vehicle controller breaks it down, identifies the planned content of the operation diagram, and controls the vehicle's operating status to maximize conflict mitigation.

[0049] In one specific embodiment, under normal conditions, the optimization objective is to minimize turnaround time, and global optimization is performed using an ant colony cooperative algorithm; when train delays are greater than or equal to three minutes, the optimization objective is to limit the spread of delays, and local timetable rearrangement is performed using a reinforcement learning algorithm; when signal equipment malfunctions, the optimization objective is to maximize the capacity of the faulty section, and timetables are dynamically planned using a degradation mode; under conditions of high passenger flow, the optimization objective is to balance transport capacity and passenger waiting time, and timetables are flexibly generated.

[0050] Step 400: Feed back the vehicle's operating status to the intelligent mapping model and dynamically adjust the operating map.

[0051] In this invention, the onboard controller manages the vehicle's operation according to the timetable, and feeds back the vehicle's operation status to the intelligent timetable model in real time. Combining the current vehicle operation status with dynamic data within the station, the timetable is quickly and dynamically adjusted to adapt to different emergencies, rapidly alleviate conflicts, and improve station throughput efficiency. Real-time monitoring of train position deviations from the plan triggers ATO (Automatic Train Operation) command loops (such as initiating a deceleration curve 300 meters before entering the station and executing energy-saving cruise after departure). In a specific application scenario, taking a sudden surge in passenger flow at a station on Metro Line 2 as an example, the AFC system detected that the platform passenger flow reached 142% of the design capacity. The spatiotemporal matrix predicted that a Class C platform overcapacity conflict would be triggered in 6 minutes. Within 70 seconds, a scheduling strategy was generated, initially locking 5 trains at the two stations before and after the current station, and constructing a path network with "skip / compression / degradation" as nodes. The basic strategy was verified, with a single train compressing its stop time by 60 seconds → the conflict resolution rate was only 38%. Ten paths were explored through an ant colony collaborative algorithm, and a combination scheme of skipping and multi-train collaborative compression was discovered. The ant colony scheme was adopted, and the reward function weight was adjusted, with 20 ants focusing on the skipping and degradation combination path. The simulated scheme was implemented, with train GXXX skipping the current station, while each of trains GXXX compresses its stop time by 25 seconds, increasing the optimal path conflict resolution rate to 91%.

[0052] The system generates a running graph through dynamic programming, automatically emphasizing on-time performance during peak periods. The reinforcement learning algorithm of the intelligent graph model works together with the ant colony collaboration algorithm to improve the efficiency of the plan planning. The reinforcement learning pre-screening reduces the ant colony search space by more than 70%, and the ant colony feedback helps the reinforcement learning algorithm break through local optima, increasing the discovery rate by 40%.

[0053] Based on the intelligent map-making method based on AI large model provided by this invention, the operation plan is designed by acquiring the dynamic data of the station and then using the intelligent map-making model. By automatically analyzing relevant data, the operation plan can be quickly generated to adapt to emergencies, thereby improving the design efficiency of the operation plan. Furthermore, the operation plan can be dynamically adjusted in real time according to changes in the station status, so as to quickly resolve emergencies and improve the throughput of rail transit.

[0054] refer to Figure 2 The present invention also discloses an intelligent graph editing system based on a large AI model, the system comprising: Data processing module 110 is used to acquire station dynamic data and preprocess it to generate fused data; The operation diagram compilation module 120 is used to compile and generate an operation diagram based on the fused data using a preset intelligent compilation model; The operation control module 130 is used to send the operation diagram to the visualization terminal and the vehicle controller, and the vehicle controller controls the vehicle operation status according to the operation diagram. The dynamic adjustment module 140 is used to feed back the vehicle's operating status to the intelligent mapping model and dynamically adjust the operating map.

[0055] This includes acquiring and preprocessing station dynamic data to generate fused data, including: Real-time train location information is obtained through the train positioning system, signal equipment information is obtained through the trackside signal machine, and passenger flow information is obtained through the gate throughput and real-time collected pedestrian flow images. The real-time train location information, signal equipment information, and passenger flow information are spatiotemporally aligned, abnormal information is filtered, and compressed and fused to generate fused data.

[0056] Based on the fused data, a running graph is generated using a preset intelligent graphing model, including: Based on the fused data, a spatiotemporal dynamic matrix is ​​constructed using the intelligent mapping model to generate a conflict heatmap. A scheduling strategy is generated based on the conflict heatmap using a reinforcement learning algorithm. Based on the aforementioned scheduling strategy, the execution path is optimized using the ant colony algorithm. A running graph is generated according to the scheduling strategy corresponding to the optimal execution path, and the running graph is stored in a lightweight container.

[0057] Based on the fused data, a spatiotemporal dynamic matrix is ​​constructed using the intelligent mapping model to generate a conflict heatmap, including: The fused data is spatially discretized using the intelligent mapping model, divided into multiple grid units, and a spatiotemporal dynamic matrix is ​​constructed. A continuous time window is created at preset time intervals. Within each time window, the risk of rear-end collision, platform overcapacity, and route conflict in each grid cell of the spatiotemporal dynamic matrix are detected, and a conflict heat map is generated.

[0058] Based on the conflict heatmap, a scheduling strategy is generated using a reinforcement learning algorithm, including: A dynamic decision-making environment is constructed based on the aforementioned conflict heatmap using a reinforcement learning algorithm. Action design is performed within the dynamic decision-making environment, and the value of the designed actions is calculated based on a preset reward mechanism. The scheduling strategy that resolves the most conflicts is determined based on the value of the actions.

[0059] The operation diagram is sent to the visualization terminal and the vehicle controller. The vehicle controller controls the vehicle's operating status according to the operation diagram, including: The operation diagram is displayed through a visual terminal, and the operation diagram is broken down and identified through the vehicle controller to obtain control information; The vehicle's operating status is controlled based on the control information.

[0060] Based on the intelligent map-making system based on AI large model provided by this invention, the system designs operation plans by acquiring dynamic data of stations and using intelligent map-making models. By automatically analyzing relevant data, it can quickly generate operation maps that adapt to emergencies, improving the efficiency of operation map design. Furthermore, it can dynamically adjust the operation map in real time according to changes in station status, quickly resolve emergencies, and improve the throughput of rail transit.

[0061] Figure 3 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 3 As shown, the electronic device may include a processor 310, a communications interface 320, a memory 330, and a communication bus 340. The processor 310, communications interface 320, and memory 330 communicate with each other via the communication bus 340. The processor 310 can call logical instructions in the memory 330 to execute an intelligent mapping method based on an AI large model. This method includes: acquiring and preprocessing station dynamic data to generate fused data; generating a running map based on the fused data using a preset intelligent mapping model; sending the running map to a visualization terminal and an onboard controller, whereby the onboard controller controls the vehicle's operating status according to the running map; and feeding back the vehicle's operating status to the intelligent mapping model to dynamically adjust the running map.

[0062] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0063] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the intelligent map-making method based on the AI ​​large model provided by the above methods. The method includes: acquiring station dynamic data and preprocessing it to generate fused data; generating a running map based on the fused data using a preset intelligent map-making model; sending the running map to a visualization terminal and an on-board controller, wherein the on-board controller controls the vehicle's operating status according to the running map; and feeding back the vehicle's operating status to the intelligent map-making model to dynamically adjust the running map.

[0064] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the intelligent map-making method based on the AI ​​large model provided by the above methods. The method includes: acquiring station dynamic data and preprocessing it to generate fused data; generating a running map based on the fused data using a preset intelligent map-making model; sending the running map to a visualization terminal and an on-board controller, wherein the on-board controller controls the vehicle's operating status according to the running map; and feeding back the vehicle's operating status to the intelligent map-making model to dynamically adjust the running map.

[0065] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0066] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent graph editing method based on a large AI model, characterized in that, include: Acquire dynamic station data and preprocess it to generate fused data; Based on the fused data, a running diagram is generated using a preset intelligent mapping model; The operation diagram is sent to the visualization terminal and the vehicle controller, and the vehicle controller controls the vehicle's operating status according to the operation diagram. The vehicle's operating status is fed back to the intelligent mapping model to dynamically adjust the operating map; The process of acquiring and preprocessing station dynamic data to generate fused data includes: Real-time train location information is obtained through the train positioning system, signal equipment information is obtained through the trackside signal machine, and passenger flow information is obtained through the gate throughput and real-time collected pedestrian flow images. The real-time train location information, signal equipment information, and passenger flow information are spatiotemporally aligned, abnormal information is filtered, and compressed and fused to generate fused data. The process of generating a runtime graph based on the fused data using a preset intelligent graphing model includes: Based on the fused data, a spatiotemporal dynamic matrix is ​​constructed using the intelligent mapping model to generate a conflict heatmap. A scheduling strategy is generated based on the conflict heatmap using a reinforcement learning algorithm. Based on the scheduling strategy, the execution path is optimized by the ant colony algorithm, and a running graph is generated according to the scheduling strategy corresponding to the optimal execution path. The running graph is stored in a lightweight container. The process of constructing a spatiotemporal dynamic matrix based on the fused data using the intelligent mapping model to generate a conflict heatmap includes: The fused data is spatially discretized using the intelligent mapping model, divided into multiple grid units, and a spatiotemporal dynamic matrix is ​​constructed. A continuous time window is created at a preset time interval. Within each time window, the risk of rear-end collision, platform overcapacity, and route conflict are detected in each grid cell of the spatiotemporal dynamic matrix, and a conflict heat map is generated. The generation of scheduling strategies based on the conflict heatmap using a reinforcement learning algorithm includes: A dynamic decision-making environment is constructed based on the aforementioned conflict heatmap using a reinforcement learning algorithm. Action design is performed within the dynamic decision-making environment, and the value of the designed actions is calculated based on a preset reward mechanism. The scheduling strategy that resolves the most conflicts is determined based on the value of the actions.

2. The intelligent graph editing method based on a large AI model according to claim 1, characterized in that, The step of sending the operation diagram to the visualization terminal and the vehicle controller, and the vehicle controller controlling the vehicle's operating status according to the operation diagram, includes: The operation diagram is displayed through a visual terminal, and the operation diagram is broken down and identified through the vehicle controller to obtain control information; The vehicle's operating status is controlled based on the control information.

3. An intelligent graphing system based on an AI large-scale model, based on the intelligent graphing method based on any one of claims 1 to 2, characterized in that, The system includes: The data processing module is used to acquire station dynamic data and preprocess it to generate fused data; The runtime graph generation module is used to generate runtime graphs based on the fused data using a preset intelligent graph generation model; The operation control module is used to send the operation diagram to the visualization terminal and the vehicle controller, and the vehicle controller controls the vehicle operation status according to the operation diagram; The dynamic adjustment module is used to feed back the vehicle's operating status to the intelligent mapping model and dynamically adjust the operating map.

4. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the intelligent graphing method based on a large AI model as described in any one of claims 1 to 2.

5. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the intelligent graphing method based on the AI ​​large model as described in any one of claims 1 to 2.

6. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the intelligent graphing method based on the AI ​​large model as described in any one of claims 1 to 2.

Images