Subway vehicle-mounted information intelligent display method and system

Abstract

The invention relates to the technical field of traffic transportation management, in particular to a subway vehicle-mounted information intelligent display method and system. The method comprises the following steps: acquiring train operation state data, compartment environment sensing data and train event flow data; passenger behavior analysis is carried out according to the compartment environment sensing data to obtain passenger behavior data; obtaining focusing situation data according to the passenger behavior data; performing state sensing scheduling according to the train operation state data, the train event flow data and the focusing situation data to obtain state sensing data; obtaining passenger attention screen data according to the state sensing data; and acquiring ground collaboration data, and carrying out collaboration linkage on the passenger attention screen data according to the ground collaboration data to obtain subway passenger collaboration data so as to carry out subway vehicle-mounted information intelligent display auxiliary operation. According to the invention, static passive to dynamic intelligent cooperative evolution of subway carriage information display is realized, and the overall rail transit service quality and passenger travel experience are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of transportation management, and in particular to a method and system for intelligently displaying subway vehicle-mounted information. Background Art

[0002] With the continuous expansion of rail transit networks and the increasing density of subway passenger traffic, traditional subway onboard information display systems are no longer able to meet the complex application requirements of multi-source situational awareness, event-driven response, and personalized passenger notifications. Currently, mainstream subway train onboard display systems mostly use timed carousels or station-triggered static information push. These systems lack the ability to perceive passenger behavior within the train, making it difficult to dynamically adapt displayed content to passenger attention areas, congestion levels, transfer intentions, and other information. In the face of emergencies (such as skip-station scheduling, temporary transfers, and service delays) or high-volume transfer scenarios, traditional display systems are unable to synchronize updates across multiple screens within the train, nor can they synchronize events with the ground dispatch center. This results in delayed and inconsistent information, or information that fails to reach key audiences, reducing passengers' efficiency in obtaining effective information and their ability to respond safely. Summary of the Invention

[0003] In order to solve the above technical problems, the present invention proposes a method and system for intelligent display of subway vehicle information to solve at least one of the above technical problems.

[0004] This application provides a method for intelligently displaying subway vehicle information, comprising the following steps: Step S1: Acquire train operation status data, carriage environment perception data, and train event flow data; perform passenger behavior analysis based on the carriage environment perception data to obtain passenger behavior data; perform focus situation analysis based on the passenger behavior data to obtain focus situation data; Step S2: Perform state-aware scheduling based on train operation status data, train event flow data, and focused situation data to obtain state-aware data; Step S3: driving the passenger attention screen according to the state perception data to obtain passenger attention screen data; Step S4: Acquire ground collaborative data, and coordinate and link the passenger attention screen data based on the ground collaborative data to obtain subway passenger collaborative data to perform intelligent display auxiliary operations for subway on-board information.

[0005] The present invention analyzes passenger behavior and identifies focus situations based on carriage environment perception data to dynamically perceive passenger focus areas and behavior trends. It then combines the train operation status and event information to perform state perception scheduling, and implements priority adjustment and strategy optimization of display content. At the same time, through the passenger attention-driven mechanism, it controls the display content, refresh frequency, and visual style of the screen to improve information communication efficiency and crowd adaptability. Combined with ground collaborative data, it realizes event synchronization and cross-carriage display linkage between the train and the dispatching center, providing consistent guidance and accurate information push in key scenarios such as emergencies and transfer pressure, thereby effectively improving the system's intelligent linkage capabilities and the global optimization effect of the display strategy.

[0006] Preferably, step S1 is specifically: Obtain train operation status data, carriage environment perception data, and train event stream data; Construct a spatial behavior graph based on the cabin environment perception data to obtain cabin behavior graph data; Extracting behavioral features from the carriage behavior graph data to obtain behavioral feature data; The behavioral feature data is generalized and enhanced to obtain passenger behavior data.

[0007] By acquiring train operating status, carriage environment perception, and event stream data, the system achieves spatiotemporal data fusion. Based on this environmental perception, a spatial behavior graph is then constructed, clarifying passenger movement paths, dwelling preferences, and behavior density in different areas while also providing a structured foundation for subsequent feature extraction. Through in-depth analysis of the behavior graph data, multi-dimensional behavioral features such as boarding and alighting patterns, dwell time distribution, and channel crossing frequency can be extracted. A behavioral generalization enhancement mechanism is then used to model and reinforce abnormal, marginal, or weak signal behaviors, effectively improving the generalization and robustness of passenger status recognition.

[0008] Preferably, the focus situation analysis specifically includes: Calculate regional attention based on passenger behavior data to obtain regional attention data; Classifying the regional attention data into focus states to obtain focus state data; Perform local congestion evolution on the focused state data to obtain local congestion data; Focusing situation data is generated according to the local congestion data and the focusing state data to obtain the focusing situation data.

[0009] The present invention calculates the passenger's attention in different areas and can quantify the passenger's eye contact, standing density and interaction frequency, thereby establishing a regional saliency model. The regional attention is classified into focus states, which can be subdivided into transfer focus, waiting focus, avoidance evacuation and other types, thereby improving the semantic recognition ability of passenger intentions. By modeling the evolution of local congestion based on the focus state data, dynamic changes such as density surges, transfer trends and chain reactions are captured, thereby achieving early perception of sudden gathering situations. Based on the focus state and local congestion data, focus situation data is generated, which not only has regional positioning accuracy, but also has time trend and event triggering, providing a quantifiable input basis for display scheduling, content priority adjustment and screen layout optimization, thereby improving the system's understanding of passenger behavior and response initiative.

[0010] Preferably, step S2 is specifically: Extract train status features based on train operation status data and train event stream data to obtain train status feature data; Perform train situation influence matrix mapping based on train state feature data and focus situation data to obtain train situation matrix data; Identify key response areas on the train situation matrix data to obtain key response area data; Generate a scheduling window based on key response area data to obtain scheduling window data; State-aware scheduling is generated based on the scheduling window data and the preset state-aware policy library to obtain state-aware data.

[0011] By extracting train status feature data, the present invention can comprehensively obtain core operating indicators such as the current train's operating stage, delay offset, station proximity, and event sensitivity. Subsequently, combined with focused situation data, a train situation influence matrix is ​​constructed to quantify the impact of different carriage areas on information scheduling strategies under specific operating conditions. This matrix can further identify the key response areas that require the most intervention in the current operating situation, providing a spatial range reference for display control. The scheduling window generated on this basis combines parameters such as operating rhythm, arrival countdown, and display duration to ensure that the display content is highly matched to changes in passenger behavior. The display strategy is generated by calling the state perception strategy library based on the scheduling window data, which can achieve intelligent control of screen content type, priority, and display rhythm, significantly improving the timeliness, pertinence, and scheduling response capabilities of the display, and meeting the information interaction needs under different operating scenarios.

[0012] Preferably, the train situation influence matrix mapping is specifically: Initialize the track network topology graph based on the train operation status data and train event stream data to obtain the track network graph data; Performing running track mapping on the track network diagram data according to the train status characteristic data to obtain running track diagram data; Perform tensor structure weighting on the running trajectory graph data to obtain weighted graph data; Performing spatiotemporal coupling on the weighted graph data to obtain coupled graph data; Calculate the propagation link impact on the coupling graph data to obtain the propagation graph data; Event-driven structural mutation is performed on the propagation graph data to obtain the train situation matrix data.

[0013] In this invention, the subway line structure is represented in the form of a graph through initialization of the track network topology graph, providing a basis for graph calculation. The running trajectory is mapped according to the train state characteristics, and the path distribution and state evolution path of the train in the network at the current moment can be constructed. Through the tensor structure weighting mechanism, multiple indicators such as speed, stop delay, and event intensity are embedded in the weight structure of nodes and edges in a multi-channel manner in the graph, forming high-dimensional weighted graph data. By performing spatiotemporal coupling processing on the weighted graph, the interactive relationship between state changes in different time slices and adjacent stations can be revealed. With the help of propagation link impact calculation, the diffusion process of scheduling events or abnormal passenger behavior in the track graph is simulated to form the event scope and response range. The graph structure is dynamically adjusted through the event-driven structural variation mechanism to generate train situation matrix data that can be used for scheduling decisions, thereby achieving precise control of the information display rhythm, response range, and priority strategy, significantly improving the global linkage and predictive scheduling capabilities of the display system.

[0014] Preferably, step S3 is specifically: Get the carriage screen data; Perform spatial modeling on the carriage screen data to obtain a carriage screen model; Calculate the attention probability field based on the car screen model and the focus situation data to obtain the attention probability field data; Multi-frequency drive generation is performed based on attention probability field data and state perception data to obtain passenger attention screen data.

[0015] The system of the present invention spatially models the carriage screen data to accurately obtain the geometric position, orientation, viewing angle, and coverage area of ​​each screen, constructing a carriage screen model to provide a physical basis for the display strategy. It then combines the focus situation data to calculate the attention probability field and quantify the distribution of attention of passengers in different areas on different screens, thereby capturing the passenger's line of sight concentration trend and regional attention dynamics. The integrated state perception data is used to differentiate the display rhythm, refresh rate, content type, and other parameters of each screen, forming a multi-frequency drive strategy for the crowd and outputting passenger attention screen data. This process can achieve regional adaptation and timing control of screen content based on the real-time behavior of passengers and the status of the train, significantly improving the efficiency and accuracy of passengers' information acquisition, while also increasing the display density and energy saving level of screen resources, and enhancing the system's intelligent adaptability and perception response capabilities in different scenarios such as peak congestion and transfer nodes.

[0016] Preferably, the spatial modeling is specifically: Carriage screen space modeling is performed based on carriage screen data to obtain a carriage screen space model; Construct a visual cone model for the cabin screen space model to obtain a screen visual cone model; Construct the crowd attention thermal field of the screen visual cone model to obtain the crowd attention thermal field data; Calculate the actual screen visibility probability based on the crowd attention thermal field data to obtain the actual screen visibility probability data; The actual screen visibility probability and the crowd attention thermal field data are used to annotate the cabin screen space model to obtain the cabin screen model.

[0017] In the present invention, through the spatial modeling of the car screen, the system can obtain the specific position and orientation relationship of each screen in the three-dimensional car structure; construct a visual cone model to clarify the theoretical visible area range of each screen, and spatially restrict the area in combination with the passenger distribution status. The system conducts crowd attention thermal field modeling, and dynamically generates a spatial heat map of the passenger's line of sight focus based on information such as passenger standing position, line of sight direction, and length of stay; calculates the actual visibility probability of the screen through the overlapping area of ​​the thermal field and the visual cone, and quantifies the effective information coverage and accessibility of each screen in the current car state. The attention thermal field and visibility probability results are annotated back into the car screen model to form a structured and parseable display visibility model.

[0018] Preferably, step S4 is specifically: Acquire ground collaborative data; Generate the full-line operation situation map based on the ground collaborative data to obtain the full-line operation situation map data; Process the cross-carriage passenger distribution trend based on the full-line operation situation diagram data to obtain cross-carriage passenger distribution trend data; Event synchronization is performed based on the passenger distribution trend data across carriages and the full-line operation status diagram data to obtain full-line event synchronization data; Based on the synchronized data of events along the entire line, the passenger attention screen data is linked across domains to obtain subway passenger collaborative data for assisting in the intelligent display of subway onboard information.

[0019] In the present invention, the system obtains ground collaborative data and generates a full-line operation situation map based on the line structure and train operation status, forming a spatiotemporal situation expression model for the entire network; then the situation map is processed to analyze the passenger distribution trend across carriages, identifying the flow trend and spatial load distribution of passengers between different trains or carriages, and accurately depicting the passenger flow dynamics of the entire line; combined with event information, spatiotemporal synchronization is performed to build a unified full-line event synchronization model to ensure that scheduling events are synchronized and responded to consistently across multiple vehicles. According to the event synchronization results, the passenger attention screen in the car is driven to perform cross-domain display linkage, and content simulcast, screen synchronization or partition adjustment are achieved in key transfer nodes, large passenger flow warnings or abnormal operation scenarios, and subway passenger collaborative data is output, which significantly improves the global coordination, dynamic adaptability and scheduling intervention efficiency of information display, and effectively ensures the consistency of information acquisition and guidance stability of passengers in different operation scenarios.

[0020] Preferably, the whole-line operation status diagram is generated as follows: Initialize the track line structure diagram based on the ground collaborative data to obtain the track line diagram data; Inject running status attributes into the track line map data to obtain attribute map data; Perform scheduling event impact domain subgraph mapping on the attribute graph data to obtain graph mapping data; Performing graph situation construction on the graph mapping data to obtain situation graph data; A time-series snapshot sequence is constructed based on the situation diagram data to obtain the full-line operation situation diagram data.

[0021] In the present invention, the relationship between line stations and intervals is modeled as a graph structure through track line structure graph initialization, providing a topological basis for multi-node concurrent state expression and propagation calculation; through the injection of running state attributes, the current train density, delay status, capacity load and other dynamic attributes of each line are mapped to the graph nodes and edges to form a high-dimensional attribute graph. The system performs scheduling event impact domain subgraph mapping, and can locally mark events such as construction skipping stations and transfer congestion in the form of propagable subgraphs in the track network, accurately modeling the impact boundary of events in the spatial dimension. Through graph situation construction, structural information and attribute changes can be integrated to generate node / edge embedding representations for situation identification and intervention judgment. The constructed time-series snapshot sequence retains the state evolution process of the graph at continuous moments, providing continuous contextual support for vehicle-ground collaborative scheduling and display linkage, and significantly enhancing the system's perception integrity and response foresight in the face of multiple vehicles, multiple events, and multiple regions.

[0022] Preferably, the present application further provides a subway vehicle information intelligent display system for executing the above-mentioned subway vehicle information intelligent display method, the subway vehicle information intelligent display system comprising: The passenger behavior perception and focus situation extraction module is used to obtain train operation status data, carriage environment perception data, and train event flow data; passenger behavior analysis is performed based on the carriage environment perception data to obtain passenger behavior data; and focus situation analysis is performed based on the passenger behavior data to obtain focus situation data; The state-aware intelligent scheduling decision module is used to perform state-aware scheduling based on train operation status data, train event flow data, and focus situation data to obtain state-aware data; A passenger attention sensing and driving screen control module is used to drive the passenger attention screen according to the state sensing data and obtain passenger attention screen data; The ground collaborative linkage display optimization module is used to obtain ground collaborative data, and to collaboratively link the passenger attention screen data based on the ground collaborative data to obtain subway passenger collaborative data for assisting in the intelligent display of subway on-board information.

[0023] The beneficial effects of the present invention are as follows: by acquiring and analyzing the train operation status, carriage environment perception and event flow data, it is possible to fully identify the behavior patterns and focus of passengers in the carriage, and realize dynamic modeling of passenger focus areas and behavior trends; by combining the train status and emergency information for state perception scheduling, it is possible to accurately control the information display content, priority and presentation method; by driving the carriage screen display strategy based on the state perception results, it is possible to realize adaptive refresh of screen content, rhythm control and optimization of visual parameters, and improve the efficiency and accuracy of passengers in obtaining key information; by acquiring ground collaborative data, constructing a full-line operation status map and realizing cross-carriage and cross-train synchronous linkage of display events, it is possible to ensure the consistency of information transmission and guidance stability in peak transfers, scheduling adjustments or abnormal event scenarios. The present invention enhances the timeliness, local adaptability and global coordination of the subway information display system, and provides passengers with smarter and more reliable travel information services. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Other features, objects and advantages of the present application will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings: Figure 1 A flowchart showing the steps of a method for intelligently displaying subway vehicle information according to an embodiment is shown; Figure 2 A flowchart showing the steps of a passenger behavior analysis method according to an embodiment is shown; Figure 3 A flowchart showing the steps of a state-aware intelligent scheduling decision-making method according to an embodiment is shown; Figure 4 A flowchart showing the steps of a passenger attention sensing driving screen control method according to an embodiment is shown; Figure 5 A flowchart of the steps of a ground collaborative linkage display optimization method according to an embodiment is shown. DETAILED DESCRIPTION

[0025] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0026] Furthermore, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0027] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0028] See also Figures 1 to 5 , this application provides a method for intelligently displaying subway vehicle information, comprising the following steps: Step S1: Acquire train operation status data, carriage environment perception data, and train event flow data; perform passenger behavior analysis based on the carriage environment perception data to obtain passenger behavior data; perform focus situation analysis based on the passenger behavior data to obtain focus situation data; In one embodiment, the system realizes the perception of the current train operation environment and passenger behavior by integrating train operation status data, carriage environment perception data and train event stream data. The train operation status data includes the current station number, train arrival timestamp, next station identifier, current operation speed (in The carriage environment perception data is collected based on the cameras, infrared sensors or thermal imaging equipment deployed inside the train, and mainly includes the following three indicators: spatial density distribution : used to represent the passenger density at the compartment spatial position (x, y) at time t; thermal trajectory of getting on and off the train : Thermal distribution of the frequency of passengers entering and exiting a specific area; statistics of passenger stay time :Indicates the number The cumulative time that passengers stay in each area; the train event flow data includes construction skipping station information, temporary transfer notices, and line-wide delay broadcasts and other emergency events. The recording format includes event type, triggering time, and the geographical or line range affected. The carriage space is divided into regular grid cells, each cell is denoted as , each unit is regarded as a node in a graph structure. According to the passenger's movement trajectory between grids and their residence transition, an edge weight function is constructed: the weight of the adjacent edge is calculated by the spatial overlap ratio of the movement frequency and the residence trajectory; the constructed result is a graph structure ,in Represents the set of all spatial unit nodes, Represents the set of mobile associated edges between nodes; each node is accompanied by the following attributes, including the passenger density value in the cell, average residence time, number of boarding and alighting times, and other behavioral indicators. The constructed carriage behavior graph is clustered, the node attributes are standardized, and clustering algorithms (such as K-Means or DBSCAN) are applied to cluster and classify the nodes; the nodes are labeled according to the clustering results (based on the cluster center and the preset label parameter library for mapping), for example: "dense residence area", "transfer concentration area" or "channel crossing area"; if the area where a node is located is a passenger passage path, and there is a high-attention area in the neighborhood (i.e., high density and high residence), the node is marked as a "secondary attention area"; the output is a structured passenger behavior feature data set, which records the behavior category of each node and its spatial attributes. Combining multiple passenger behavior characteristics, the regional attention is calculated and the focus situation data is generated, and the attention score of each node is calculated. It is defined by the following weighted formula: ,in is the weighted coefficient of the passenger density term, which is set to 0.5. is the weighted coefficient of the average residence time term, which is set to 0.3. is the weighted coefficient of the gaze direction concentration term, which is 0.2. The above weights are obtained based on empirical data fitting. For nodes The normalized value of passenger density is current passenger density / maximum passenger density; is the normalized value of the average residence time of the area, that is, the average residence time of the area / maximum residence time; It is an indicator that represents the concentration of passengers' gaze direction; the areas are classified according to the attention threshold. If the area has a high score and a long dwell time, it is marked as a "waiting focus area"; if the area is a door area with a high entry and exit frequency, it is marked as a "transfer focus area"; if the area is significantly avoided by people, it is marked as an "avoidance focus area"; the output is focus situation data, which includes focus type, corresponding area number, focus weight value and trend prediction direction.

[0029] Step S2: Perform state-aware scheduling based on train operation status data, train event flow data, and focused situation data to obtain state-aware data; In one embodiment, train status features are extracted from the running status: ,in is the train delay duration, is the actual arrival / departure time, Plan time for the timetable; extract the current operation phase identifier and use coding to represent different phase states: "0" indicates that the train is in the middle of operation; "1" indicates that the train is about to enter the station; "2" indicates that the train is stopped; "3" indicates that the train is about to depart; parse the event stream data, identify the event type (such as construction skipping, transfer anomalies, and full-line delays), and classify and manage them according to the event type code; map the scope of the event to the spatial structure of the train, for example: if the event affects the section from carriages C3 to C5, record its corresponding physical location interval. By integrating train status characteristics and focusing situation data, a two-dimensional train situation impact matrix is ​​constructed. , to evaluate the dispatch response priority of each carriage area at the current moment: ,in Score the dispatch response priority of the carriage area under the current dispatch, is the state fusion function, which represents the The first carriage area and the The response strength between the focus states, such as weighted sum, is the weighted coefficient of the carriage state factor, For the The state factors corresponding to each carriage area (such as delay, entry, stop, etc.) are first coded based on the preset state parameter table and then standardized. is the weighting coefficient of the focus situation weight, For the The attention intensity weight of the focus situation area. The top K area pairs with the highest weight values, such as 5-10, are selected as the target areas for priority display or broadcasting; based on the current train time status and the next station arrival time, a time window that can be used for display is generated: ,in, is the current timestamp, The estimated time of arrival of the train at the next station. A buffer time (e.g., 10 seconds) is reserved for display shutdown to prevent sudden changes in information from affecting passenger cognition. Based on the attributes of the key response area and the current operating status, the preset state perception display strategy library is called to generate a specific display plan. For example, if an area is identified as a "waiting focus area," the guidance route map will be displayed first, and the readability of the text content will be enhanced (e.g., increasing the font size); if an area is in a transfer focus area and there is currently an abnormal transfer event, a scrolling animation icon will be used in conjunction with a voice broadcast prompt; if there is evasive focusing behavior in the current carriage (e.g., people are clearly moving away from a certain area), a safety prompt display can be triggered; the output state perception data format includes fields such as the target screen identifier, display content type, priority level, animation style identifier, and whether voice broadcast is enabled.

[0030] Step S3: driving the passenger attention screen according to the state perception data to obtain passenger attention screen data; In one embodiment, the structural parameters of all information display screens in the vehicle cabin are obtained to construct a vehicle cabin screen space model. Specifically, the three-dimensional spatial coordinate position (x, y, z) of the screen in the vehicle cabin; the screen orientation vector, i.e., the normal direction perpendicular to the screen surface; the screen's viewing angle range, defined as the horizontal and vertical viewing angle boundaries of the viewing cone; and the above parameters are uniformly encoded into a spatial cone model, where the viewing angle of each screen is defined as a cone area with an angle range of , used to indicate whether the passenger's line of sight penetrates the visible area of ​​the screen. The output is the carriage screen model data with spatial cone structure attributes. Combined with the passenger focus status data, the degree of overlap of the projections of each focus area within the visible cone of each screen is calculated to quantify the degree of integration between screen visibility and passenger attention. Specifically, all focus areas are traversed to determine whether their center points or main activity trajectories fall within the visible cone of a certain screen; if overlap is established, the attention intensity of the focus area is used as the attention contribution of the screen; the passenger attention coverage probability of the screen is defined as: ,in For the The passenger attention coverage probability value of the screen, Indicates the screens; Indicates the Focus areas; Display screen The set of visible cone coverage areas; is the indicator function, indicating the Is the focus area on the screen? If it is within the visible range, it is 1, and if it is not, it is 0; Indicates the focus area The output is the passenger attention coverage probability value for each screen. Based on the attention probability value of each screen and the aforementioned state perception scheduling weight, all screens are sorted to determine the display update priority. For example, the display priority index is calculated as follows: ,in Indicates the policy weight of the screen in the state-aware scheduling stage (for example, transfer warning takes precedence over station guidance), where For the The passenger attention coverage probability value of the screen, For screen Policy scheduling weight in the state-aware scheduling stage; for all screens, press The values ​​are sorted from high to low; for screens with high priority, a high-frequency update display strategy is formulated, including setting the refresh rate to fast mode, such as updating every 3-5 seconds; giving priority to graphical full-screen display or scrolling broadcast mode for display; synchronously setting content category and voice control parameters (if enabled); outputting passenger attention screen data, including field information such as display content category, update frequency, display mode (full screen, scrolling, icon), priority label and voice control switch for each screen.

[0031] Step S4: Acquire ground collaborative data, and coordinate and link the passenger attention screen data based on the ground collaborative data to obtain subway passenger collaborative data to perform intelligent display auxiliary operations for subway on-board information.

[0032] In one embodiment, ground-based collaborative data is acquired, including operational status maps generated by the dispatch center, regional abnormal event synchronization plans, and transfer pressure prediction matrices. The format is {station identifier, event type, event severity (or priority), target action time range}. Based on the aforementioned system, a full-line operational status map is generated. Based on the rail network topology G = (V, E), where nodes V represent stations and edges E represent segments. The system injects state into each node, including attributes such as passenger density, delays, and construction impacts, and delay propagation probabilities into each edge. Distributed trend processing and event synchronization are performed, using predictive models (such as time-series LSTM) to generate cross-car trends, identifying which cars are likely to see transfer traffic. These trends are then mapped to the current train's focus area, generating a synchronization control table. Linked control is generated. If a line-wide transfer congestion trend is detected, a multi-screen simultaneous broadcast is triggered. If the current car is a dispatch hotspot, the screen highlights the warning information issued by the ground. Output: subway passenger collaborative data (including the screen IDs to be linked, display content, display strategy, and synchronization window).

[0033] Preferably, step S1 is specifically: Step S11: Acquire train operation status data, carriage environment perception data, and train event stream data; In one embodiment, the train operation status data is obtained by accessing the train control system interface (such as the train control and monitoring system TCMS). The system collects the following field information at fixed time intervals: the train's unique identification number, the current station number and platform number, the actual arrival time and departure time of the train (respectively ), current running speed v (unit: km / h), current acceleration and deceleration status, and running stage status code (for example: running, entering the station, stopping, etc.). The running stage status code can be used to determine the dynamic operation process of the train between stations. The carriage environment perception data is obtained through the multimodal edge perception equipment deployed inside the train, which includes edge computing cameras, thermal infrared imaging arrays, infrared counters and ultra-wideband (UWB) array nodes. The perception system collects the following spatial data indicators: real-time population density distribution , indicating that at time Next, the two-dimensional spatial position of the carriage Passenger density on the bus; hot zone trajectory point set for getting on and off the bus, used to describe the thermal trajectory area formed by passengers when getting on or off the bus, which can be further analyzed for the distribution of hot spot getting on / off areas; and the distribution of the concentration of the stay time. , represents the time the i-th passenger stays in a specific area, which is used to calculate regional congestion and activity frequency; the local channel flow vector field , described at the moment Next, various locations in the carriage The train event flow data is provided by the ground dispatching system or train event management middleware. The system receives event data through a message queue (such as the MQ protocol) or a REST-based asynchronous communication interface. Each event message includes at least the following fields: Event type: such as skip station scheduling, temporary transfer notification, construction closure notice, specific car disabled, etc.; Event effective period: indicates the time range between the occurrence and expiration of the event; Impact range identifier: includes the affected station number or car section number; Event priority tag: used for prioritizing event processing and response strategies.

[0034] Step S12: constructing a spatial behavior graph based on the cabin environment perception data to obtain cabin behavior graph data; In one embodiment, each carriage is divided into Equally divided grid units (e.g. 50cm per grid) 50cm); each grid Corresponding to a graph node , record the location index and attributes. If there is a person's movement trajectory connection between two adjacent grids, then establish an undirected edge ; Edge weight is defined as the passenger movement frequency per unit time , or the superposition value of thermal trajectory density. Graph structure definition: Each node contains the current population density; average length of stay; whether it belongs to the entry and exit area (marked by the door); each edge can contain direction probability, supporting the construction of directed graphs for flow analysis.

[0035] Step S13: extracting behavior features from the carriage behavior graph data to obtain behavior feature data; In one embodiment, the system extracts multi-dimensional behavioral features from each node and its adjacency relationship based on the aforementioned vehicle behavior graph. The extracted dimensions include the following four categories, including the residence intensity feature, for each graph node: (i.e. corresponding to a grid unit in the carriage) calculates the average passenger stay time, recorded as . The characteristics of boarding and alighting activity are to select grid cells within a certain spatial range from the door of the carriage, and count the number of passenger trajectories starting from or arriving at the node per unit time. The characteristics of migration kinetic energy are to extract the mobility intensity value associated with each node according to the weight of the edge in the carriage behavior graph (such as the frequency of passenger movement per unit time), which represents the degree of passenger mobility or behavior fluctuation in the area. The characteristics of path thermal sequence are to analyze the main path of passengers from the entrance area to the center or exit area of ​​the carriage, and count the cumulative thermal intensity along the way (which can be estimated by the movement frequency or trajectory density), which is used to identify the main traffic path and its heat distribution, and then infer the structured migration trend of the crowd.

[0036] Step S14: Perform behavior generalization enhancement on the behavior feature data to obtain passenger behavior data.

[0037] In one embodiment, for nodes located at the edge of the car or in areas with sparse passengers, whose original behavioral features are not significant, the system uses neighborhood information to enhance features. Specifically, if the neighboring nodes of a node (such as 4 or 8 neighborhoods above, below, left, and right) show high behavioral focus (such as high density), the features of the node are enhanced. Taking the passenger density as an example, the density value after enhancement is Expressed as ,in is the compensation factor (set by experience or obtained through training), Representation node The set of adjacent nodes of is the average density of adjacent nodes. The system analyzes the behavior graph sequence at several consecutive moments. Taking five frames as the time sequence, a time series graph set is constructed: The system performs time-weighted smoothing on the behavior states of each node in the behavior graph sequence, identifying regions with consistent behavior across multiple consecutive moments (e.g., continuous high density, stable stops, etc.). The system then classifies the characteristics of each behavior region based on rule templates or lightweight classification models (e.g., rule trees or shallow decision trees), converting multidimensional behavior features into unified behavioral semantic labels. The classification rules are as follows: waiting behavior (long stays and high density); transfer behavior (transfer focus): short stays and frequent boarding and alighting; and avoidance behavior (avoidance): rapid movement and low density. Each identified region is accompanied by structured attribute information, including a behavior type label (e.g., "transfer_focus"); a feature confidence score (e.g., the degree to which the behavior characteristics match the rules, ranging from 0 to 1); and a behavior evolution trend (e.g., "increasing," "stable," and "decreasing"), derived through time series analysis. The output format is a set of structured passenger behavior data items, each describing a semantically labeled passenger behavior region. For example, in the area numbered "Z07", the identified behavior type is "transfer concentration area", the behavior recognition confidence is 0.91, and it shows a trend of increasing activity.

[0038] Preferably, the focus situation analysis specifically includes: Calculate regional attention based on passenger behavior data to obtain regional attention data; In one embodiment, the entire compartment space is divided into several discrete area units based on the compartment space grid. , each region is of size Or it can be divided into equal parts according to the layout of the carriage. It can be calculated as follows: ,in For the The attention score of each region, is the weighted coefficient of the population density term, is the average population density per unit time, is the weighting coefficient of the residence time term, For the The average dwell time in each region, is the weighted coefficient of the action event frequency term, For the The frequency of action events in a unit time (such as looking up at the screen, walking, standing still); the system calculates the frequency of action events in each area unit based on the above calculation model. The attention level is evaluated, and the output results include the area number, attention score and related intermediate variables used for attention evaluation.

[0039] Classifying the regional attention data into focus states to obtain focus state data; In one embodiment, a fixed threshold is used to segment the attention value. The specific classification criteria are as follows: The attention value When , the area is marked as "strong focus" state; when When , the area is marked as "medium focus"; when , marks the area as "out of focus".

[0040] In one embodiment, the attention values ​​of all regional units are Constructing a feature vector input; executing a clustering algorithm to automatically divide the area into several clusters with similar attention levels; each cluster corresponds to a potential focus area, and its focus level can be inferred from the average attention level of the cluster center. For each area determined to be in focus, the system must also record the following: spatial coverage (i.e., a collection of several area unit numbers representing the spatial distribution of the focus area within the vehicle); historical focus duration (in seconds, used to measure the length of time the focus area has been continuously focused on); the associated in-vehicle display screen number (indicating the passenger information display device that the focus area is primarily facing); and the average gaze angle between the display screen and the crowd (which can be calculated by the eye tracking or camera pose estimation module to evaluate visibility and screen attractiveness).

[0041] Perform local congestion evolution on the focused state data to obtain local congestion data; In one embodiment, the system performs a Calculate its congestion value The formula is as follows: ,in For the region The congestion value, is the weighted coefficient of the population density term, which is set to 0.5. is the population density item, that is, the area at the current moment Average population density within the area / preset maximum population density, is the weighted coefficient of the inverse term of moving speed, and its value is 0.3. is the movement speed item, that is, the minimum movement speed / area The average moving speed of passengers in the car (unit: m / s), is the weighted coefficient of the in-out imbalance term, and its value is 0.2. It is an indicator of the imbalance of inflow and outflow in the area; the system’s regional congestion value Identify the evolution trend. Based on the historical continuous time window (length ) in the congestion value sequence, calculate the following two indicators, such as the volatility indicator, that is, the sliding window standard deviation, which is used to judge the magnitude of the change in congestion level; the trend indicator, that is, the sliding window derivative mean, is used to judge whether the congestion level continues to rise, tends to be stable, or decreases. When the congestion value shows an increasing trend for multiple consecutive moments and the change exceeds the preset threshold, the system marks the area as "evolving congestion". In order to identify the contiguous congested area composed of multiple adjacent areas, the system performs spatial diffusion analysis, which includes the following steps: calculating the spatial adjacency relationship and corresponding edge weight between all area units. The edge weight can be expressed as the travel probability or the coupling strength of pedestrian flow between areas; if a pair of adjacent areas Congestion value If both exceed the set congestion threshold and their edge weights are higher than the preset adjacency threshold, the system merges the two into a connected congested area. This fusion operation is repeated to form a cluster of areas with spatial connectivity and consistent congestion characteristics, which are output as local congestion areas. The local congestion data output by the system includes the following structured fields: congestion area number; congestion status level (e.g., increasing, stable, easing); average population density; regional inflow and outflow frequencies; and average movement speed.

[0042] Focusing situation data is generated according to the local congestion data and the focusing state data to obtain the focusing situation data.

[0043] In one embodiment, by focusing on the situation score function Perform the calculation: ,in For the attention score, is the weighted coefficient of the attention score, Score the congestion evolution trend, is the weighting coefficient of the congestion trend score, The historical duration factor (which gives higher credibility to long-term focus areas) is the historical maximum duration factor divided by the historical average duration factor. is the weighted coefficient of the historical duration factor. The concentration of the crowd is evaluated based on the entropy index (the more concentrated, the stronger the trend). Using the carriage map as the base map, superimpose the concentration of each focus area. , forming a two-dimensional heat map or area label map; supporting priority sorting of multiple areas in the same car for screen strategy matching.

[0044] Preferably, step S2 is specifically: Step S21: extracting train state features based on the train running state data and the train event stream data to obtain train state feature data; In one embodiment, it comes from the train control system (such as TCMS) or the train dispatching system, and the specific fields include the current station number (for positioning); the current running speed (unit: km / h); the train running direction (such as up / down); the relative distance between the trains in front (for judging the congestion situation); the planned arrival or entry time (for deviation assessment). The train event stream data is provided by the dispatching center or the event management system and is accessed in an asynchronous push manner, including temporary transfer plans (such as changes in transfer stations); skip station scheduling strategies (such as skipping some stations to save time); construction, temporary stops or station closure events (including duration and affected intervals); train fault records or delay status (such as power failure, system stagnation, etc.). Based on the above data, the system constructs a train state feature vector, which is recorded as: , where the meaning of each state dimension is as follows The time offset represents the difference between the actual arrival time and the planned arrival time (unit: seconds). The speed deviation is the difference between the current running speed and the historical average speed of the train throughout the journey, reflecting the acceleration or deceleration state; is the passability index of the preceding section, a Boolean variable (0 means blocked; 1 means passable), determined by the threshold of the distance between trains ahead; The event intensity code value is used to quantify the severity of the event currently affecting the train. For example, the construction impact is set to 0.8, the emergency is set to 1.0, and the temporary stop is set to 0.6. : Transfer pressure index, which is normalized and evaluated based on the expected number of transfer passengers at the transfer station that the current train passes through or is about to arrive at (the value range is 0~1).

[0045] Step S22: performing train situation influence matrix mapping based on the train state feature data and the focus situation data to obtain train situation matrix data; In one embodiment, the train status characteristic data includes the train delay time, speed deviation, forward traffic status, event level and transfer pressure and other operating indicators extracted in the above steps. The focus situation data includes the results of the carriage behavior diagram analysis, including passenger high density areas, dwell time concentrated areas, entry and exit active areas, etc. Each focus area contains the following parameters: area location index; real-time passenger density; dwell behavior focus; passenger attention or abnormal warning level. The system constructs a two-dimensional influence matrix , where each matrix element represents a state feature dimension of the train Focus situation area The impact intensity is expressed as: ,in Indicates the train status characteristic components (such as delay duration, speed deviation, etc.); Indicates the Behavioral characteristics of passenger focus areas (such as density or dwell time); The system uses an influence function, such as a linear function like weighted summation, to learn nonlinear relationships through deep learning training, and to map states through logical rules or conditional judgments. If the train's current stop is less than one station away from a station with a high passenger focus area, and the train delay exceeds 60 seconds, the train is marked as a "high interference source" for that area, and the corresponding matrix element is assigned a high value. If the train event type is "construction" or "skipped station," the system maps the event's impact to its impact interval and applies a disturbance factor to the relevant focus area, representing a situational disruption or behavioral disturbance. The system outputs a train situational impact matrix as a two-dimensional array, where the row dimension represents train state factors (such as delays, accidents, and transfer pressures); the column dimension represents the focus area within the carriage or platform; and the value of each element represents the corresponding impact intensity (which can range from 0 to 1).

[0046] Step S23: identifying key response areas of the train situation matrix data to obtain key response area data; In one embodiment, the system sets an influence threshold (e.g. 0.8) is used to filter out focus areas that are significantly disturbed or affected. Specifically, the system accumulates the influence score of each focus area j under its corresponding state factor dimension. If the accumulated value exceeds the set threshold , then the area is determined to be the key response area. It is expressed as: ,in Represents the train state factor For the region influence; Screening thresholds for key responses (empirically set or dynamically adjusted); is a set of focus area numbers identified as critical responses. For each candidate area, the higher the frequency of being triggered as a critical response in the past period and the lower the response delay, the more weight is applied to its impact score at the current moment to reflect its spatiotemporal inertia and response sensitivity. The system is limited to identifying critical areas within the following areas: the car where the current train is located; the previous or next car directly adjacent to the current car; and structural blind spots in the car (such as camera blind spots) and the platform end area (because their spatial structure is not conducive to intervention) are not considered. By setting the above spatial constraints, the system avoids incorporating difficult-to-intervene or inefficient sections into the response domain, thereby concentrating resources on controllable areas.

[0047] Step S24: Generate a scheduling window based on the key response area data to obtain scheduling window data; In one embodiment, for each passenger focus area identified as a critical response, the system generates a corresponding dispatch window, which is denoted as: ,in is the window start time, defined as 15 seconds before the train is expected to enter the critical response area; The window expiration time is defined as 5 seconds before the train is expected to depart the area. This time interval is used to optimize the timing of relevant information display or voice reminders, ensuring that passengers receive timely guidance or warnings before interacting with the area. If scheduling windows for multiple key response areas overlap, the system will implement the following merging or prioritization strategies: If the overlapping windows belong to adjacent areas, the system may merge the time ranges to form a larger display window. If window merging is not feasible, the window corresponding to the area with a higher impact score (i.e., one with greater focus and a higher risk of interference) will be prioritized. Each scheduling window is accompanied by a content type recommendation tag, indicating the primary type of information to be displayed within that time window. Recommended tags are automatically generated based on the dominant state factors triggering the key response. These include "Construction Avoidance Reminder" for areas with significant construction interference; "Transfer Guidance" for areas with high transfer pressure; "Train Head Congestion Warning" for areas with high density or delays at the front of the train; and "Station Entry Calm Reminder" for areas requiring guidance.

[0048] Step S25: Generate state-aware scheduling based on the scheduling window data and a preset state-aware policy library to obtain state-aware data.

[0049] In one embodiment, the system dynamically generates dispatch control instructions tailored to specific carriage situations based on generated dispatch window data and a pre-set state-awareness policy library. These instructions are used to control the switching, layout, and playback frequency of displayed content on the train's onboard screens or broadcasting equipment, thereby achieving passenger guidance, behavioral intervention, or risk warnings. The system's pre-set state-awareness policy library is a set of rule-based policies. Each policy includes trigger conditions, including event type (e.g., skipped station, construction, transfer), focus area characteristics (e.g., high density, concentration on a specific screen area), display templates specifying the graphic or animation display format to be switched when the trigger conditions are met, animation playback frequency (e.g., high frequency (every 5 seconds carousel), medium frequency, or low frequency (every 30 seconds refresh), visual cue color (e.g., red highlights indicate dangerous areas, blue guides indicate transfer directions), and screen layout and priority policies that determine display priority across multiple screens and the placement of content on a single screen (e.g., left corner, upper right corner, etc.) to avoid overlap with passenger gathering areas. The strategies include Strategy A: When the train event type is "skipping station", the content switches to "skipping station description diagram + voice warning template"; Strategy B: If the passenger focus area is in the upper right corner of the screen, the displayed content will be offset to the right side of the screen to avoid visual occlusion.

[0050] Preferably, the train situation influence matrix mapping is specifically: Initialize the track network topology graph based on the train operation status data and train event stream data to obtain the track network graph data; In one embodiment, the system abstractly models the subway line as a graph structure G=(V,E) to describe the spatial topology and operation characteristics of the rail network. Represents each subway station; edge set Indicates the track section that is accessible between stations, that is, stations and There is a direct running connection between them. The system further assigns the following attributes to the above nodes and edges, including node attributes, such as the line number: the line code to which the station belongs is identified, which is used to support multi-line network fusion modeling; station level label: used to distinguish the network role level of the station, such as "ordinary station", "transfer station", "transportation hub", etc.; transfer level: quantify the transfer capacity of the station, such as supporting transfers between several lines, the number of transfer channels, etc. Edge attributes, such as section length: record the physical distance of the track between stations; historical passing time statistics: represent statistical indicators such as the average running time and standard deviation of the train on the section; whether to support temporary skipping stations: a Boolean flag used to indicate whether the section has the ability to skip stations in response to scheduling needs. System marking abnormal mark field , used to represent edges Is it currently in invalid state? : Indicates that the track section is in a normal passable state; : Indicates that the section is temporarily closed or operation is restricted due to construction, emergencies, failures, etc. Through the above topology modeling process, the system completes the graph structure initialization of the rail network.

[0051] Performing running track mapping on the track network diagram data according to the train status characteristic data to obtain running track diagram data; In one embodiment, the system searches for a feasible path with a length of 3 to 5 stations in the constructed track network graph structure based on the node where the current train is located. This path represents the expected running range of the train in the next time window, which is recorded as the path sequence ,in is the current site node, is the expected target station node. For each track section in the above path (i.e., the edge in the diagram), the system injects the corresponding train scheduling prediction information, including the expected time of entry into the section; the path stability index of the section (such as historical passage success rate and event impact frequency); whether there is a passenger concentration area around the section and its spatial distance. The path with the above scheduling attributes injected is mapped as a "dynamic operation trajectory" to the track network diagram, and each node and edge is assigned a time label (such as entry time, expected stay time, etc.), thereby forming the operation trajectory diagram data containing time-series scheduling information.

[0052] Perform tensor structure weighting on the running trajectory graph data to obtain weighted graph data; In one embodiment, for each track segment edge in the running trajectory diagram (i.e., the path segment connecting any two station nodes in the diagram), the system constructs an edge weight structure in the form of a three-dimensional tensor. The tensor uses "starting node-ending node-weight dimension" as coordinates to form a set of multi-dimensional indicators to comprehensively represent the actual running characteristics and environmental influencing factors of the path segment. In this embodiment, the third dimension of the tensor (i.e., the weight dimension) includes but is not limited to the following four core weight factors, including the time delay factor, which represents the degree of running delay of the current path segment, and is calculated as the deviation value of the ratio of the actual running time to the planned time, ,in is the time delay factor weight, is the actual running time of the track section, is the planned running time of the track section; the passenger density interference factor characterizes the degree of passenger aggregation or congestion level in the area connected by the path segment, which can be obtained through the statistical value of the regional density sensor. , is the passenger density interference factor weight, is the average passenger density value in the area connected by the path segment; the transfer interaction intensity factor represents the transfer behavior frequency and transfer node activity level of the path segment within the time window. ,in is the transfer interaction intensity factor weight, The frequency of transfer interactions for route segment (i, j) within a time window is calculated based on historical or current data. The accident impact factor indicates whether the route segment is currently under construction, an accident, or a special event. This factor can be a binary value (affected / unaffected) or a continuous value modeled using historical accident frequencies. The tensorized weights of all edges are aggregated to form a weighted trajectory graph structure consisting of a node set, an edge set, and a weight tensor. The output graph structure is represented as a triplet consisting of a node set, an edge set, and a set of tensor weights corresponding to each edge.

[0053] Performing spatiotemporal coupling on the weighted graph data to obtain coupled graph data; In one embodiment, based on the track segment (i.e., the edge in the figure) as the basic unit, the operating status of each track edge in each time slice is gradually updated at set time intervals (for example, one minute is a time slice), forming a sequence structure of the path segment evolving over time. For the multi-dimensional tensor weights carried by each edge, the system presets a set of corresponding time-varying functions. This function is used to model the dynamic change law of different influencing factors on the time axis. For example, for accident impact factors, the event attenuation function is used to represent the decrease in its impact intensity over time; for passenger density factors, the peak response function is used to represent its rapid increase during the peak hours of getting on and off the bus; for transfer interaction factors, periodic functions or short-term perturbation functions are used to reflect its changing rhythm; the time delay factor can be combined with the historical regression trend to generate a smooth change curve. The state of each edge at each moment t is expressed as: ,in is a time-varying function of the kth feature (e.g., event decay function, peak function); this state characterizes the operational efficiency, stability, and potential risk level of the path segment at the current moment. The state sequence of each edge at different time slices together forms an evolutionary graph with a time dimension. The state score structures of all path segments at different time slices are unified and merged to construct a coupled graph structure consisting of a node set, an edge set, and a temporal state sequence. This coupled graph not only preserves the spatial connectivity of the rail network but also explicitly describes the dynamic evolution of the state of each path segment over time.

[0054] Calculate the propagation link impact on the coupling graph data to obtain the propagation graph data; In one embodiment, the event impact intensity of each site node at the initial moment is defined. This propagation intensity function is used to represent the degree of interference caused by a specific event to the site node at a certain point in time. The initial impact intensity can be derived from external input events (such as construction closures or accident notifications), and its value is set based on the event level, timeliness, and impact range. The system performs event impact modeling. ,in For nodes In time The intensity of the communication impact at each moment, is the currently affected target node, is the current propagation time step (discrete time), is the propagation factor from the weighted edges, For nodes The adjacent nodes of For nodes In time The impact strength at each moment. For example, graph convolutional propagation mechanisms or Bayesian network propagation methods are used to simulate the dynamic diffusion of influence between nodes. Within each iterative time step, the propagation strength of a node is determined by the influence of its neighboring nodes. For any target node, its propagation strength at the next moment is the weighted cumulative propagation strength of its neighboring nodes at the current moment. The weighting factor is derived from the propagation factor in the edge weights of the coupling graph, calculated based on the density influence of adjacent edges, spatiotemporal disturbance factors, and other factors. Adjacency relationships are defined based on the graph boundary within the rail network topology, limiting the propagation paths. Throughout the iterative process, the system progresses step by step, updating the propagation strength values ​​of all nodes with each iteration until the propagation strength converges or the propagation range reaches a set threshold. For example, in the initial stages of an accident, only the accident site has a high impact value. As the accident spreads, neighboring path segments and stations affected by associated transfers gradually experience varying degrees of diffusion interference. This process can identify the event propagation chain and its evolutionary rhythm. The system encodes the propagation impact results at each node in the rail graph, constructing the event propagation graph structure. The propagation graph retains the node and edge set structure of the original graph and adds the influence strength value at the current propagation moment to each node.

[0055] Event-driven structural mutation is performed on the propagation graph data to obtain the train situation matrix data.

[0056] In one embodiment, the system sets a propagation intensity threshold parameter, representing the critical condition for the event's impact to reach a significant level. For each node (i.e., a rail station) in the propagation graph, the system monitors its propagation intensity value in real time. If the propagation intensity does not fall within the range [0, 1], it is first normalized. If the propagation intensity of a node at a given moment exceeds the set threshold, the node is labeled as a high-situation node and considered a key area with significant event impact. The system sets a propagation intensity threshold θ (e.g., θ = 0.8∈[0, 1]), which represents the critical condition for the propagation impact to reach a significant level. This threshold can be set based on the 95th percentile of historical event propagation data or dynamically through manual tuning. After identifying multiple high-situation nodes, the system checks whether these nodes form continuous paths in the propagation graph. The system determines whether there are direct or two-hop or less edge connections between high-situation nodes. If several consecutive path segments are in a high-situation intensity state (e.g., the propagation intensity of three consecutive edges exceeds θ), they are aggregated into a high-risk propagation path. Path aggregation allows for the interruption of one node (e.g., a non-high-situation node), but the overall span does not exceed five nodes. The system classifies and analyzes the event types in the current propagation diagram and extracts key situational factors, including but not limited to construction impact, operation delays, transfer pressure, passenger detention density, etc. For each high-situation node or propagation path, the quantitative impact intensity of the above situation factors is calculated and coded by type. The system generates a two-dimensional train situation impact matrix. The row dimension of the matrix represents different categories of situation factors, such as construction factors, delay factors, transfer factors, etc.; the column dimension corresponds to the node or link index identified in the propagation diagram. Each element in the matrix represents the impact intensity of the corresponding factor at that position in the area or link.

[0057] Preferably, step S3 is specifically: Step S31: Acquire carriage screen data; In one embodiment, the system collects the following basic information for each car screen, including a screen identification code, which is used to uniquely identify each screen. The format generally includes the car number and the screen serial number; the installation location coordinates, which use a three-dimensional coordinate system (X, Y, Z) to represent the physical installation location of the screen in the car space, and the unit is meters; the installation orientation angle, which uses three rotation angles to represent the pitch angle (Pitch), yaw angle (Yaw) and roll angle (Roll) of the screen, respectively, and the unit is angle (°), which is used to describe the orientation of the screen; the screen size, which records the length and width of the screen display surface (unit: meter). To support content rendering and driver instruction compatibility for multiple types of screens, the system also needs to record the "type label" of each screen to distinguish the type of screen. For example, different types of devices such as LED strip screens, LCD advertising screens, dynamic interactive touch screens, etc. can be labeled with corresponding category labels.

[0058] Step S32: performing spatial modeling on the carriage screen data to obtain a carriage screen model; In one embodiment, a three-dimensional coordinate system is used to construct a vehicle compartment space model. , each screen is embedded into a directional viewing cone (visual cone): the center point of the screen is the cone top; the viewing angle range (set to 30° horizontally and 20° vertically) determines the cone angle; the visual radius is set according to the width of the car (such as 3 meters); based on the above parameters, the system constructs the viewing area of ​​each screen , which is defined as the spatial model The set of points that meet the following conditions: the point is within the coverage of the screen cone angle; the point is no more than the set visual radius from the screen center; the point orientation satisfies the screen installation direction constraint. It can be formally expressed as: The system must simultaneously mark the area each screen faces, specifically the crowd activity area or structural orientation (e.g., the entrance and exit areas, the center of the train compartment, etc.). This orientation information serves as a key reference dimension for matching focus with screen position. Through this spatial modeling process, the system completes the visual field spatial mapping of the screens within the train compartment.

[0059] Step S33: Calculating the attention probability field based on the car screen model and the focus state data to obtain attention probability field data; In one embodiment, the system performs Calculate its attention score , to measure the degree to which it can currently attract the passenger's attention. This calculation is based on the integral weighting of the focus situation within the visible area of ​​the screen, and the formula is as follows: ,in : No. Block screen attention score; : No. Block the viewing cone area of ​​the screen; : Passenger attention degree of the corresponding spatial point; : A viewing angle attenuation function, used to weight the degree of deviation from the center of the screen. This function is expressed as a cosine function. Points with a cosine value greater than 0 are retained, while those with a cosine value less than 0 are considered 0. This function exhibits a monotonically decreasing characteristic. For example, by modeling it as a cosine angle function or Gaussian attenuation, the weight of a point at the edge of the cone is lower than that of a point directly in the line of sight. : Spatial voxel integral element. The system performs the above calculations on all screens, and the output is the attention probability field dataset, defined as: ,in Indicates the The attention probability or priority index of the block screen at the current moment, It is the screen sequence item, and its value is 1, 2, 3…n.

[0060] Step S34: Generate multi-frequency drive according to the attention probability field data and the state perception data to obtain passenger attention screen data.

[0061] In one embodiment, the system scores The screen is graded and managed according to the set upper and lower thresholds. The driving strategy is as follows: If the attention score of a screen is , the system marks it as a high-attention screen. Such screens give priority to pushing important or urgent information, such as emergency transfer notices, accident handling progress or emergency broadcast information, to ensure that the information reaches the target audience in a timely manner; if , it is determined to be a low-attention screen. Such screens are only used to maintain the display of background information, such as the current station name, energy-saving prompts or static advertisements to avoid wasting resources; for screens between high and low thresholds, a periodic carousel strategy or prompt content switching can be implemented, which can be flexibly configured according to actual operational needs. Based on the above classification results, the system assigns a corresponding content refresh frequency to each screen For example, high-attention screens have a refresh cycle of 1 second, suitable for pushing high-frequency dynamic content; medium-attention screens have a refresh cycle of 5 seconds, suitable for light prompts or reminders; low-attention screens have a refresh cycle of more than 10 seconds, only maintaining basic information. The system has a multi-frequency content generation module, which generates content based on the refresh frequency. With the state event set E, call the backend content scheduling service to generate a specific information update package and generate a driving sequence to control each screen , whose structure includes display event information ( ): corresponds to the specific message content in the event set; animation style attributes ( ): Define dynamic color schemes for information presentation; layout parameters ( ): used to control the information display structure, such as horizontal scrolling, partitioned display, and parallel text and images. Through the above mechanism, the system can achieve dynamic screen driving for different attention levels.

[0062] Preferably, the spatial modeling is specifically: Carriage screen space modeling is performed based on carriage screen data to obtain a carriage screen space model; In one embodiment, a three-dimensional rectangular coordinate system is established with the vehicle cabin as the reference object. This coordinate system takes the floor at the lower left corner of the front end of the vehicle cabin as its origin, and defines three coordinate axes representing the horizontal (X-axis), vertical (Y-axis), and vehicle cabin depth (Z-axis). This coordinate system serves as a unified geometric reference for describing the position and orientation of all screens. For each display screen installed in the vehicle cabin, its core spatial attributes are extracted, including spatial position coordinates representing the absolute three-dimensional position of the screen center point in the vehicle cabin coordinate system; installation orientation information, defined in Euler angles, including pitch, yaw, and roll, representing the orientation vector of the screen surface; and display size parameters, including the screen's width and height. The system treats each screen as a rigid object and embeds it into the vehicle cabin coordinate space based on its spatial position and orientation parameters. All screen models in three-dimensional space are consistent with the actual vehicle cabin's physical structure, ensuring accurate representation of orientation, position, and scale. The spatial parameters of all screens are used to construct a cabin screen spatial model structure, identifying the precise three-dimensional layout of each screen in the cabin. This model serves as the basic structural data for constructing the visual cone, mapping passenger attention distribution, and driving screen content strategies.

[0063] Construct a visual cone model for the cabin screen space model to obtain a screen visual cone model; In one embodiment, a cone-shaped viewing area model is constructed for each in-car display screen. This cone's vertex is at the center of the screen, and its axis aligns with the screen's mounting orientation, defined by the screen's pitch, yaw, and roll angles. The cone's angle is determined by the viewing angle, with a horizontal range of ±30 degrees and a vertical range of ±20 degrees, simulating the natural viewing range of the human eye. A maximum viewing radius is set for the viewing cone. This radius is adjusted based on the vehicle's length and structure and the screen's resolution. For example, for a clear viewing distance of less than 3.5 meters, a maximum viewing radius of 3.5 meters is set. This parameter is used to prevent distant areas from focusing on the screen. The system uses spatial geometric ray tracing, emitting multiple line-of-sight rays from the cone's apex along the cone's direction to simulate the possible paths of a passenger's line of sight in three-dimensional space. The ray tracing is then intersected with the vehicle's spatial structure to generate a corresponding three-dimensional cone polygon, accurately representing the screen's visible coverage area. The cone can be approximated as a combination of multiple subdivided triangular pyramids, forming a closed viewing area volume. Each completed cone structure is encoded to generate a screen visibility cone model data structure, including the screen identifier, cone vertex coordinates, angle parameters, and the set of visible range boundary points. This model will serve as the spatial constraint foundation for constructing the passenger attention thermal field and calculating the actual visibility probability.

[0064] Construct the crowd attention thermal field of the screen visual cone model to obtain the crowd attention thermal field data; In one embodiment, the current passenger distribution data in the carriage is collected , each passenger contains position coordinates; facing direction (head orientation); stay time; construct three-dimensional thermal field function , the calorific value is calculated as follows: ,in A point in three-dimensional space The intensity of the crowd's attention thermal field, Passenger index number, is the total number of passengers in the current carriage, The passenger's attention coefficient (for example, those who stay for a long time and face the screen directly have a higher weight), is the natural exponential function, is any three-dimensional space coordinate point in the thermal field, For the The position coordinates of each passenger, To control the degree of diffusion, a value of 0.5 is taken. Passengers' attention is mainly concentrated within a radius of 0.5 meters, which is suitable for passengers in a stationary or dense state. The result is a three-dimensional attention distribution heat map of the entire car.

[0065] Calculate the actual screen visibility probability based on the crowd attention thermal field data to obtain the actual screen visibility probability data; In one embodiment, for each screen Visible cone Inner region, integrated thermal field intensity calculation: ,in For the The weighted thermal field integral value of the block screen (visible intensity score), is the three-dimensional coordinate of any voxel point in space, For the The visible cone space area of ​​the block screen, Heat up the crowd's attention at a specific point in space The intensity value of is the integral volume element, which represents the volume of the spatial element. is the viewing angle fitting weight function (the smaller the angle with the center line of the screen, the greater the weight); the actual visual probability vector is obtained by normalizing all screens: , output the visibility score of each screen.

[0066] The actual screen visibility probability and the crowd attention thermal field data are used to annotate the cabin screen space model to obtain the cabin screen model.

[0067] In one embodiment, the system annotates the actual visibility probability values ​​of each screen obtained in the previous step as a "visibility score" field in the train cabin screen model. This score reflects the relative probability of each screen being actually noticed given the current passenger distribution and gaze direction, and is a key indicator for display priority scheduling. The system extracts three-dimensional spatial slices from the attention thermal field that overlap with each screen's field of view and stores them in a structured form as "heat slice data." Each slice contains the spatial coordinates of that location and its attention thermal value, which is used for screen content projection simulation or heat distribution visualization analysis. The resulting train cabin screen model contains not only geometric information such as its original spatial position, orientation, and dimensions, but also its visibility score and corresponding thermal field slice. The system uses this structure as the basic unit of the enhanced train cabin screen model, enabling refined modeling and thermal-driven control of the multi-screen system within the train cabin. The model output is a structure that fuses spatial geometry, passenger attention probability, and thermal field information.

[0068] Preferably, step S4 is specifically: Step S41: Acquire ground collaborative data; In one embodiment, the system collects ground-based collaborative data from multiple external platforms or subsystems. For example, traffic management system data includes transfer route information, passenger density distribution, and transfer pressure warnings, which are used to infer passenger flow trends and transfer pressure at current stations or routes. Operation and maintenance system data includes daily operation and maintenance events, such as construction schedules, temporary scheduling changes, sudden station closures, and temporary stops, which affect train routes and passenger experience. The city-level big data platform interface accesses external environmental data, such as real-time weather conditions (rain, snow, high temperatures, etc.) and holiday peak passenger flow forecasts, through the city data platform to enhance the system's responsiveness to sudden passenger flow and environmental fluctuations. The system supports two real-time data access methods: a Kafka-based streaming data service for continuous push notifications of high-frequency events, such as traffic emergencies and sudden changes in passenger density, with millisecond-level transmission; and a RESTful interface-based polling acquisition method, suitable for operation and maintenance or forecasting events (such as holiday forecasts and construction plans), which are regularly pulled in accordance with scheduling strategies. The system is configured with a refresh window period of 3-5 seconds, meaning that data is updated or pushed every 3-5 seconds.

[0069] Step S42: Generate a full-line operation situation map based on the ground collaborative data to obtain full-line operation situation map data; In one embodiment, a graph structure is constructed ,in : All station nodes along the entire line; : accessible path (including transfers and parallel branches); : Node attributes, including the current passenger flow intensity, congestion index, transfer delay rate, etc. of the station; inject ground event impact factors into each station node, such as: , set threshold (Influence threshold) is marked as a "high-pressure point"; the output result is a graph structure with event association, transfer risk and node dynamic pressure.

[0070] Step S43: Processing the inter-carriage passenger distribution trend according to the full-line operation situation diagram data to obtain inter-carriage passenger distribution trend data; In one embodiment, a horizontal cross-carriage population heat map is performed on the currently running train to collect the number of passengers in each car, the stationary rate, and the boarding and alighting fluctuation rate; the system pre-sets a station stop prediction model (based on a linear function of historical data or a deep learning model trained) to predict future passenger flow changes at each station; and a time series trend function is constructed. : Represents the change in passenger density in the kth carriage in the next t seconds; uses sliding window regression prediction or LSTM neural network prediction; the output is the passenger cross-carriage flow trend matrix .

[0071] Step S44: performing event synchronization based on the cross-carriage passenger distribution trend data and the full-line operation status diagram data to obtain full-line event synchronization data; In one embodiment, an event synchronization table is constructed, such as the horizontal axis: carriage area ID; the vertical axis: ground collaborative event type (construction, delay, transfer fluctuation); the table value: whether there is a significant correlation (based on distance weight + synchronization window matching); the system determines whether the event is associated with the carriage based on the following criteria: calculate the current event affecting the station With train carriages The distance weight of the spatial position. The spatial weight can be calculated based on the station distance in the line map or the current running progress of the train. Trend linkage condition detection: If a certain car is predicted Passenger density change trend in the future time window (from the above trend function ) increased significantly and exceeded the set density threshold , then it is judged that the carriage is affected by The system checks whether the event's effective time overlaps with the time window of the train arriving at the target station or the target car entering the station interval (i.e., whether the synchronization condition is met). If these conditions are met, the system activates the event's linkage response for the corresponding car and marks it in the event synchronization table.

[0072] Step S45: Perform cross-domain display linkage on the passenger attention screen data based on the event synchronization data of the entire line to obtain subway passenger collaborative data to assist in the intelligent display of subway on-board information.

[0073] In one embodiment, when a synchronized event is marked as "urgent" or highly relevant to the train status (e.g., high-density overlays, construction closures, etc.), the system inserts or replaces the display content corresponding to that event into the current display plan, with higher-priority content overwriting the original content. Based on the attention score of each screen in the passenger attention data (e.g., the attention weight evaluation from step S3), the system selects the screen IDs with the highest scores for display. This process prioritizes displaying content within the passenger's primary field of view, improving information acquisition efficiency. Display frequency parameters are set based on the event level, including the number of loops, the duration of the display delay, and whether to highlight the original content in overlay mode. High-level events (e.g., emergencies) are pushed more frequently, while general transfer guidance information is pushed at a regular frequency. For example, on screens S1 and S2 in train C4, event reminder content numbered X2043 is displayed, with an overlay mode, a priority of 2, and a loop of three.

[0074] Preferably, the whole-line operation status diagram is generated as follows: Initialize the track line structure diagram based on the ground collaborative data to obtain the track line diagram data; In one embodiment, an undirected graph G = (V, E) is initialized, where V represents all subway stations (including main line and branch line nodes); E represents the track connection edges between stations (including transfer channels and transfer cost attributes); edge attributes include but are not limited to distance: the track distance between two stations; travel_time: the theoretical travel time; transfer_cost: the time cost of cross-line transfers; node attributes include: station_type: ordinary / hub / transfer node; capacity_max: the maximum passenger carrying capacity; structure sources include subway GIS layers, rail BIM model interfaces, official station structure diagrams, etc.

[0075] Inject running status attributes into the track line map data to obtain attribute map data; In one embodiment, state attributes are dynamically injected into each node (station) and edge (track segment) to construct a state attribute graph. : Node status field: crowding_index: congestion index; event_flag: whether there is a ground event (construction, closure, etc.); delay_time: estimated delay duration; edge status field: flow_velocity: passenger flow velocity; congestion_risk: current risk score; attribute data source: the dispatching system (SCADA) of the ground traffic control center; capacity estimation model output (based on flow meter or AI camera recognition); use timestamp t to mark the status.

[0076] Perform scheduling event impact domain subgraph mapping on the attribute graph data to obtain graph mapping data; In one embodiment, for each scheduling event, the system records its event identification information and triggering time. For example, an event example: 14:05, Station A is temporarily closed; event type: station closure event; triggering node: the starting station affected by the event, recorded as The system is based on the following two propagation control parameters to limit the identification boundary and propagation of the impact domain, including radius_threshold: propagation radius threshold, which defines the maximum distance that the event impact can propagate into the graph (usually measured by the number of adjacent nodes, such as 3 stations); transfer_dependency: transfer dependency flag, which indicates whether the transfer channel node or cross-line edge should be included in the impact domain expansion process to cover the affected transfer path. The system is based on a graph traversal algorithm from the event node Starting from the beginning, the impact propagation path search is performed, and the algorithms include breadth-first search (BFS) and depth-first search (DFS). During the traversal process, the system uses the propagation radius threshold as a constraint condition and constructs the event impact subgraph under the premise of meeting the propagation rules: , the subgraph contains all the nodes and edges affected by the current scheduling event. The system Set the Travel Time Decay Weight field , which is calculated as follows: ,in Indicates that the event source node to the edge starting point The path length (which can be the number of hops or weighted distance); is the initial impact strength parameter, with a value of 1.0; is the decay rate adjustment parameter, with a value of 0.5-2.0; is an exponential decay term, simulating the trend of event propagation having a strong impact on adjacent areas and a weak impact on remote areas. The graph mapping data output by the system includes the subgraph structure of the impact domain , indicating the area covered by the current event; the propagation attenuation weight field attached to each edge ; Updates to the operating status of the nodes involved (such as delay time, accessibility changes, transfer blockages, etc.).

[0077] Performing graph situation construction on the graph mapping data to obtain situation graph data; In one embodiment, in the graph structure, each node The operating status of the system is represented by a comprehensive indicator called "pressure value", which is recorded as , which is used to assess the operational complexity, load intensity, or scheduling risk currently faced by the site. It is calculated as follows: ,in For nodes Pressure value; For nodes The state function represents its operating status characteristics (such as congestion, delay, event flag, etc.); Representation node For Node The attention weight or propagation attenuation coefficient is calculated based on the propagation strength of the adjacent edge or the event influence coefficient; Represents the aggregation operation that affects the states of all adjacent nodes, which is one of weighted summation, maximum pooling, or attention-weighted averaging; The stress value is the state characteristic function value of the node itself, ensuring that local information is included in the stress assessment. The system performs the above stress value calculation operation on all nodes in the graph, and the stress values ​​of all nodes are aggregated into a situation heat map.

[0078] A time-series snapshot sequence is constructed based on the situation diagram data to obtain the full-line operation situation diagram data.

[0079] In one embodiment, a uniform time window interval is set , for example, sampling once every 1 minute, that is, at the time point , which means that from the first to the nth time point, the corresponding graph situation snapshots are saved respectively . Each snapshot graph contains the following core content, including a station situation score set, which records the pressure value or risk score of each station at that point in time; an edge flow estimation set, which records the passenger flow estimation value or traffic capacity score of the track side between adjacent stations, which is used to assist in track section load judgment or bottleneck identification. The graph snapshot structure at each moment can be organized into the following format, such as a timestamp field, which records the current sampling time point (the format is ISO 8601, for example: "2025-05-29T14:30:00"); a station situation score set, which uses the station number as the key and the pressure value as the value, indicating the situation strength of each station at that moment; a track connection flow estimation set, which uses the station pair number as the key and records the flow estimation value or traffic pressure index of the track section between the station pairs. Organize the graph snapshots at all time points in chronological order to form a time chain graph sequence: , represents the graph snapshot corresponding to each time point. This sequence constitutes the evolution trajectory of the operation status of the entire subway line network over a period of time.

[0080] Preferably, the present application further provides a subway vehicle information intelligent display system for executing the above-mentioned subway vehicle information intelligent display method, the subway vehicle information intelligent display system comprising: The passenger behavior perception and focus situation extraction module is used to obtain train operation status data, carriage environment perception data, and train event flow data; passenger behavior analysis is performed based on the carriage environment perception data to obtain passenger behavior data; and focus situation analysis is performed based on the passenger behavior data to obtain focus situation data; The state-aware intelligent scheduling decision module is used to perform state-aware scheduling based on train operation status data, train event flow data, and focus situation data to obtain state-aware data; A passenger attention sensing and driving screen control module is used to drive the passenger attention screen according to the state sensing data and obtain passenger attention screen data; The ground collaborative linkage display optimization module is used to obtain ground collaborative data, and to collaboratively link the passenger attention screen data based on the ground collaborative data to obtain subway passenger collaborative data for assisting in the intelligent display of subway on-board information.

[0081] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes that fall within the meaning and scope of equivalent elements of the application documents are included in the present invention.

[0082] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A method for intelligently displaying subway vehicle information, characterized in that: The following steps are involved: Step S1: Acquire train operation status data, carriage environment perception data, and train event stream data; Perform passenger behavior analysis based on the cabin environment perception data to obtain passenger behavior data; perform focus situation analysis based on the passenger behavior data to obtain focus situation data; Step S2: Perform state-aware scheduling based on train operation status data, train event flow data, and focused situation data to obtain state-aware data; Step S3: driving the passenger attention screen according to the state perception data to obtain passenger attention screen data; Step S4: Acquire ground collaborative data, and coordinate and link the passenger attention screen data based on the ground collaborative data to obtain subway passenger collaborative data to perform intelligent display auxiliary operations for subway on-board information.

2. The method according to claim 1, characterized in that Step S1 is specifically as follows: Obtain train operation status data, carriage environment perception data, and train event stream data; Construct a spatial behavior graph based on the cabin environment perception data to obtain cabin behavior graph data; Extracting behavioral features from the carriage behavior graph data to obtain behavioral feature data; The behavioral feature data is generalized and enhanced to obtain passenger behavior data.

3. The method according to claim 1, characterized in that The focus of situation analysis is as follows: Calculate regional attention based on passenger behavior data to obtain regional attention data; Classifying the regional attention data into focus states to obtain focus state data; Perform local congestion evolution on the focused state data to obtain local congestion data; Focusing situation data is generated according to the local congestion data and the focusing state data to obtain the focusing situation data.

4. The method according to claim 1, wherein Step S2 is specifically as follows: Extract train status features based on train operation status data and train event stream data to obtain train status feature data; Perform train situation influence matrix mapping based on train state feature data and focus situation data to obtain train situation matrix data; Identify key response areas on the train situation matrix data to obtain key response area data; Generate a scheduling window based on key response area data to obtain scheduling window data; State-aware scheduling is generated based on the scheduling window data and the preset state-aware policy library to obtain state-aware data.

5. The method according to claim 4, characterized in that The train situation influence matrix mapping is specifically as follows: Initialize the track network topology graph based on the train operation status data and train event stream data to obtain the track network graph data; Performing running track mapping on the track network diagram data according to the train status characteristic data to obtain running track diagram data; Perform tensor structure weighting on the running trajectory graph data to obtain weighted graph data; Performing spatiotemporal coupling on the weighted graph data to obtain coupled graph data; Calculate the propagation link impact on the coupling graph data to obtain the propagation graph data; Event-driven structural mutation is performed on the propagation graph data to obtain the train situation matrix data.

6. The method according to claim 1, wherein Step S3 is specifically as follows: Get the carriage screen data; Perform spatial modeling on the carriage screen data to obtain a carriage screen model; Calculate the attention probability field based on the car screen model and the focus situation data to obtain the attention probability field data; Multi-frequency drive generation is performed based on attention probability field data and state perception data to obtain passenger attention screen data.

7. The method according to claim 6, characterized in that The spatial modeling is specifically as follows: Carriage screen space modeling is performed based on carriage screen data to obtain a carriage screen space model; Construct a visual cone model for the cabin screen space model to obtain a screen visual cone model; Construct the crowd attention thermal field of the screen visual cone model to obtain the crowd attention thermal field data; Calculate the actual screen visibility probability based on the crowd attention thermal field data to obtain the actual screen visibility probability data; The actual screen visibility probability and the crowd attention thermal field data are used to annotate the cabin screen space model to obtain the cabin screen model.

8. The method according to claim 1, characterized in that Step S4 is specifically as follows: Acquire ground collaborative data; Generate the full-line operation situation map based on the ground collaborative data to obtain the full-line operation situation map data; Process the cross-carriage passenger distribution trend based on the full-line operation situation diagram data to obtain cross-carriage passenger distribution trend data; Event synchronization is performed based on the passenger distribution trend data across carriages and the full-line operation status diagram data to obtain full-line event synchronization data; Based on the synchronized data of events along the entire line, the passenger attention screen data is linked across domains to obtain subway passenger collaborative data for assisting in the intelligent display of subway onboard information.

9. The method according to claim 8, characterized in that The whole line operation status diagram is generated as follows: Initialize the track line structure diagram based on the ground collaborative data to obtain the track line diagram data; Inject running status attributes into the track line map data to obtain attribute map data; Perform scheduling event impact domain subgraph mapping on the attribute graph data to obtain graph mapping data; Performing graph situation construction on the graph mapping data to obtain situation graph data; A time-series snapshot sequence is constructed based on the situation diagram data to obtain the full-line operation situation diagram data.

10. A subway vehicle information intelligent display system, characterized in that: For executing the method for intelligently displaying subway vehicle information according to claim 1, the intelligent display system for subway vehicle information comprises: The passenger behavior perception and focus situation extraction module is used to obtain train operation status data, carriage environment perception data, and train event flow data; passenger behavior analysis is performed based on the carriage environment perception data to obtain passenger behavior data; and focus situation analysis is performed based on the passenger behavior data to obtain focus situation data; The state-aware intelligent scheduling decision module is used to perform state-aware scheduling based on train operation status data, train event flow data, and focus situation data to obtain state-aware data; A passenger attention sensing and driving screen control module is used to drive the passenger attention screen according to the state sensing data and obtain passenger attention screen data; The ground collaborative linkage display optimization module is used to obtain ground collaborative data, and to collaboratively link the passenger attention screen data based on the ground collaborative data to obtain subway passenger collaborative data for assisting in the intelligent display of subway on-board information.

Images