A subway train PIS intelligent interaction method and system
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUIYANG CRRC PUZHEN URBAN RAIL TRANSIT EQUIP SERVICE CO LTD
- Filing Date
- 2026-04-08
- Publication Date
- 2026-07-14
Smart Images
Figure CN122392300A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent interactive control technology, specifically to a PIS intelligent interactive method and system for subway trains. Background Technology
[0002] With the rapid development of urban rail transit systems, subway train passenger information systems (PIS) are gradually evolving from traditional single information dissemination systems towards intelligent, networked, and multimedia-integrated systems. Existing PIS systems typically integrate multiple services such as video playback, voice broadcasting, emergency alerts, and passenger guidance, and achieve real-time information updates and distribution through vehicle-to-ground wireless communication systems. Meanwhile, with the development of 5G communication technology, MIMO technology, and beamforming technology, vehicle-to-ground communication capabilities have been significantly improved, making concurrent transmission of multiple services possible in high-speed mobile environments. Furthermore, in line with the development trend of smart subways, PIS systems are gradually introducing data-driven and intelligent decision-making mechanisms to enhance passenger interaction experience and operational service levels.
[0003] However, existing PIS systems still have significant shortcomings in multi-service concurrency and complex communication environments. Existing technologies mostly employ static priority allocation based on service type, lacking the ability to dynamically model service semantic characteristics (such as urgency, timeliness, and scope of impact). This makes it difficult to accurately reflect the actual importance of different services in different scenarios, potentially leading to critical services not receiving priority. Existing methods typically allocate resources based on Quality of Service (QoS) indicators, failing to effectively establish a mapping relationship between Quality of Experience (QoE) and communication parameters, resulting in discrepancies between communication resource allocation results and the actual passenger experience. In situations where vehicle-to-ground wireless communication links dynamically change, existing technologies lack a collaborative scheduling mechanism oriented towards MIMO spatial resources and beam direction, making it difficult to fully utilize spatial resources, leading to insufficient spectrum utilization efficiency and link stability. Furthermore, existing PIS systems generally lack closed-loop feedback and adaptive optimization mechanisms based on service execution results, failing to dynamically adjust priority models and resource allocation strategies according to operational performance, thus limiting the system's continuous optimization capabilities in complex scenarios. Therefore, how to achieve semantic-driven priority modeling, effective mapping of QoE and QoS, dynamic collaborative allocation of communication resources, and closed-loop adaptive optimization of interaction strategies in multi-service concurrent scenarios has become a key technical problem that urgently needs to be solved in this field. Summary of the Invention
[0004] In view of the above-mentioned problems, the present invention is proposed.
[0005] Therefore, the technical problem solved by this invention is that existing subway train PIS information interaction and communication resource scheduling methods have static service priority division and lack semantic modeling capabilities, lack of effective mapping between QoE and QoS leading to a mismatch between resource allocation and actual experience, insufficient utilization efficiency of vehicle-to-ground wireless communication resources (including MIMO spatial resources and beam resources), and how to achieve semantic-driven intelligent interaction and dynamic collaborative optimization of communication resources in multi-service concurrent scenarios.
[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a smart interaction method for subway train PIS, comprising collecting multi-service data, train-to-ground communication status data, and passenger interaction context information from the subway train PIS system; performing semantic parsing on the multi-service data to extract the urgency, timeliness, and scope of impact of the services; constructing an interaction semantic priority model based on the semantic parsing results to generate corresponding interaction semantic priority values; establishing a service interaction experience QoE target based on the interaction semantic priority values, and mapping the QoE target to communication service quality QoS parameter constraints; dynamically allocating bandwidth resources, MIMO spatial stream resources, and beamforming resources in conjunction with the current train-to-ground wireless communication link status; adaptively adjusting the video playback, broadcast insertion, and emergency information display strategies of the PIS system based on the dynamic allocation results; and collecting service execution feedback data to perform closed-loop correction of the interaction semantic priority model and communication resource allocation strategy.
[0007] As a preferred embodiment of the intelligent interaction method for subway train PIS described in this invention, the following steps are included: The collection of multi-service data, vehicle-to-ground communication status data, and passenger interaction context information from the subway train PIS system includes: acquiring video service streams, broadcast service streams, and emergency information data streams through an onboard data acquisition unit, and classifying and labeling them according to service type; simultaneously acquiring the bandwidth utilization status, transmission delay, packet loss, and channel quality indicators of the vehicle-to-ground link in real time through the communication interface, and periodically updating them according to a preset time window; acquiring passenger interaction context information, including carriage location, current station status, passenger flow density, and indicators of any abnormal events, through onboard sensors and system logs; uniformly encapsulating the multi-source data to form a comprehensive status set including communication status, environmental status, and service status; triggering data updates when the data acquisition cycle reaches a set time interval, otherwise maintaining the data from the previous cycle.
[0008] As a preferred embodiment of the subway train PIS intelligent interaction method described in this invention, the generation of corresponding interaction semantic priority values includes: extracting multiple semantic attribute parameters for each business data item, including business urgency, remaining effective time of information, scope of passenger impact, and tolerance for business interruption; normalizing each semantic attribute and comprehensively evaluating it according to a pre-set weight relationship; when an abnormal event flag is detected and triggered, raising the urgency of the corresponding business to the highest level; when passenger flow density exceeds a set threshold, increasing the weight of the impact range of the business involving the corresponding area; generating interaction semantic priority values based on the comprehensive evaluation results, and forming a sorting queue according to priority; when there are businesses with the same priority, further sorting them according to a preset priority order of business types.
[0009] As a preferred embodiment of the subway train PIS intelligent interaction method described in this invention, the establishment of the business interaction experience (QoE) objective includes: establishing corresponding interaction experience evaluation index systems for different types of services; for video services, focusing on image clarity, playback continuity, and loading response time; for broadcast services, focusing on voice continuity and playback stability; and for emergency information, focusing on information arrival timeliness and coverage. Based on the interaction semantic priority value, services are divided into three levels: high priority, medium priority, and low priority, and different experience assurance strategies are matched for each level. For high-priority services, response time and stability requirements are set; for medium-priority services, a dynamic adjustment strategy is adopted; and for low-priority services, a certain degree of experience degradation is allowed. Based on the experience objective, the corresponding communication requirements are deduced, including the required minimum bandwidth range, the allowed maximum latency range, and the acceptable packet loss range, establishing a correspondence between interaction experience and communication performance.
[0010] As a preferred embodiment of the PIS intelligent interaction method for subway trains described in this invention, the dynamic allocation includes: dividing available communication resources into multiple resource units according to the communication requirements of each service, and allocating resource units hierarchically according to service priority; during resource allocation, prioritizing high-priority services to meet minimum communication requirements, and allocating remaining resources to medium and low-priority services when resources are sufficient; dynamically selecting transmission paths based on current channel state information, and prioritizing the selection of the spatial channel with the best channel quality for data transmission in a multi-input multi-output communication structure; simultaneously adjusting the signal transmission direction according to train position and channel direction information to concentrate signal energy in the effective reception area; and triggering a resource reallocation mechanism when link quality degradation or resource conflict is detected, compressing, delaying, or suspending low-priority services.
[0011] As a preferred embodiment of the PIS intelligent interaction method for subway trains described in this invention, the adaptive adjustment includes: determining the current service operation status based on the matching degree between the allocated communication resources and service requirements; executing normal playback or broadcasting strategies when resources meet service requirements; reducing the resolution or adjusting the playback frequency of video services and performing delay compensation or buffering playback for broadcast services when resources are lower than service requirements; pausing or delaying low-priority services when resources are severely insufficient; immediately triggering a preemption mechanism when emergency information has a higher priority than the currently executed service, interrupting the current service and prioritizing the display or broadcasting of emergency information; and simultaneously making differentiated adjustments to the display content in different areas based on the passenger flow status of the carriages.
[0012] As a preferred embodiment of the PIS intelligent interaction method for subway trains described in this invention, the closed-loop correction of the interaction semantic priority model and communication resource allocation strategy includes: collecting feedback data during business execution, including actual transmission delay, data loss, video playback smoothness, and broadcast continuity, and comparing and analyzing the feedback data with preset targets; when the actual operating state deviates from the target range, determining whether the deviation originates from inaccurate priority assessment or unreasonable resource allocation; if it is a priority assessment problem, adjusting the semantic weights to increase the weights of urgency or impact range; if it is a resource allocation problem, adjusting the resource allocation strategy to increase the resource proportion of key businesses; and maintaining the current parameter settings when the system operates stably for multiple consecutive cycles.
[0013] As a preferred embodiment of the subway train PIS intelligent interaction system described in this invention, it includes: a data processing module, a resource allocation module, and a strategy adjustment module; the data processing module is used to collect multi-service data, train-to-ground communication status data, and passenger interaction context information from the subway train PIS system, and performs semantic parsing on the multi-service data to extract the urgency, timeliness, and scope of impact of the services, and constructs an interaction semantic priority model based on the semantic parsing results to generate corresponding interaction semantic priority values; the resource allocation module is used to establish a service interaction experience QoE target based on the interaction semantic priority values, and maps the QoE target to communication service quality QoS parameter constraints, and dynamically allocates bandwidth resources, MIMO spatial stream resources, and beamforming resources in combination with the current train-to-ground wireless communication link status; the strategy adjustment module is used to adaptively adjust the video playback, broadcast insertion, and emergency information display strategies of the PIS system based on the dynamic allocation results, and collect service execution feedback data to perform closed-loop correction of the interaction semantic priority model and communication resource allocation strategy.
[0014] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program as a step in implementing a PIS intelligent interaction method for subway trains.
[0015] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of a PIS intelligent interaction method for subway trains.
[0016] The beneficial effects of this invention are as follows: The subway train PIS intelligent interaction method provided by this invention achieves a shift from traditional coarse-grained scheduling based on service type to refined decision-making based on semantics by structurally modeling semantic elements such as service urgency, timeliness, impact scope, and interruption tolerance. This allows priorities to be dynamically adjusted according to the operating scenario. Furthermore, by establishing a mapping relationship between QoE targets and QoS parameters, user experience requirements are transformed into executable communication constraints, realizing a shift in communication resource allocation from network-side driven to service experience-driven. On this basis, combined with a collaborative scheduling mechanism of bandwidth, MIMO spatial stream, and beamforming, the efficiency of wireless resource utilization and link stability are improved. Simultaneously, by constructing a resource matching and execution feedback mechanism, adaptive adjustment of interaction strategies and closed-loop optimization of model parameters are achieved, enabling the system to continuously approach the optimal operating state in a dynamic environment. Overall, this invention improves the critical service assurance capability, system real-time response, and long-term adaptive optimization capability. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0018] Figure 1 The above is an overall flowchart of a subway train PIS intelligent interaction method provided in Embodiment 1 of the present invention.
[0019] Figure 2 This is a QoE mapping diagram of a subway train PIS intelligent interaction method provided in Embodiment 1 of the present invention.
[0020] Figure 3 This is a resource allocation diagram for a subway train PIS intelligent interaction method provided in Embodiment 1 of the present invention.
[0021] Figure 4 This is a schematic diagram of a computer device for a subway train PIS intelligent interaction method provided in Embodiment 3 of the present invention. Detailed Implementation
[0022] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.
[0023] Example 1, referring to Figures 1-3 As an embodiment of the present invention, a PIS intelligent interaction method for subway trains is provided, comprising:
[0024] S1: Collect multi-service data, vehicle-to-ground communication status data, and passenger interaction context information from the PIS system of subway trains, and perform semantic parsing on the multi-service data to extract the urgency, timeliness, and scope of impact of the services. Based on the semantic parsing results, construct an interaction semantic priority model to generate corresponding interaction semantic priority values.
[0025] Furthermore, the collection of multi-service data, vehicle-to-ground communication status data, and passenger interaction context information from the subway train PIS system includes: acquiring video service streams, broadcast service streams, and emergency information data streams through the onboard data acquisition unit and classifying and labeling them according to service type; simultaneously acquiring the bandwidth utilization status, transmission latency, packet loss, and channel quality indicators of the vehicle-to-ground link in real time through the communication interface and updating them periodically according to a preset time window; acquiring passenger interaction context information, including carriage location, current station status, passenger flow density, and indicators of whether abnormal events exist, through onboard sensors and system logs; unifying the multi-source data into a comprehensive status set that includes communication status, environmental status, and service status; triggering data updates when the data collection cycle reaches a set time interval, otherwise maintaining the data from the previous cycle.
[0026] It should also be noted that a preferred scheme for collecting multi-service data, vehicle-to-ground communication status data, and passenger interaction context information from the subway train PIS system specifically includes: firstly, collecting multi-service data, including video service streams, broadcast service streams, and emergency information data streams. Video service streams may include arrival announcement videos, transfer guidance videos, promotional videos, and service announcement videos; broadcast service streams may include automatic arrival announcements, transfer announcements, dispatch announcements, and manual announcements; emergency information data streams may include fault alarms, evacuation instructions, temporary operation adjustment information, and safety reminders. When a service enters edge computing, a service type identifier is added to each service record. subscript Indicates the first Items of pending processing This indicates the category to which the service belongs. In this embodiment, The system can use one of three preset categories: video, broadcast, and emergency, or further subdivide it into subcategories such as arrival broadcast, manual broadcast, guidance video, promotional video, and emergency announcement. While collecting business data, link status information is also collected in real time via the vehicle-to-ground communication interface. Link status information includes current bandwidth utilization, end-to-end transmission latency, link packet loss, and channel quality indicators. To ensure that data from different sources can be included in decision-making calculations at the same time, this embodiment sets the link status collection to a fixed time window. Periodic updates. The refresh rate can be configured according to the refresh capability of the vehicle-to-ground communication system, for example, set to 100 milliseconds, 200 milliseconds, or 500 milliseconds. At the end of each cycle, a link status record is generated, including the effective bandwidth, average latency, average packet loss ratio, and smoothed channel quality value for the current cycle. A moving average or median filter is applied to the sampling results of the current cycle and the previous few cycles to prevent instantaneous link fluctuations from directly causing abnormal priority jumps in subsequent cycles. This embodiment does not limit the use of a specific filtering method; it only requires that the output link status value represents the actual link level within the current time window.
[0027] In addition to business data and link status, passenger interaction context information is also obtained through onboard sensors and system logs. Context information includes carriage location, current station status, passenger flow density, and abnormal event flags. Carriage location distinguishes the physical area where the service operates, such as the lead car, middle car, or last car; station status indicates the train's operating phase, such as running between sections, entering a station, stopping at a station, or leaving a station; passenger flow density reflects the current level of passenger concentration within the carriage; abnormal event flags indicate whether there are special circumstances such as malfunctions, temporary stops, emergency evacuations, abnormal doors, or chirp takeovers. Passenger flow density can be obtained from carriage counting sensors, visual passenger flow statistics modules, or door access records; abnormal event flags can come from the train control system, fault logs, or manually monitored terminals.
[0028] Because the data sources, refresh frequencies, and data structures differ, they cannot be directly used in unified computation. Therefore, this embodiment further incorporates a data alignment and encapsulation mechanism. Specifically, based on the link state update cycle... As the primary time reference, at the end of each primary time period, the latest business data, link status, and context data of that period are written into the same comprehensive status record, generating a record consistent with the previous one. The state object corresponding to the business item .in, Indicates the current update time. Indicates the first This business is at all times The overall status. This overall status includes the following fields: Business Type Identifier. The fields include the original business content identifier, current link status, current carriage or region, station status, current passenger flow density, abnormal event flag, and timeliness information required for subsequent semantic extraction.
[0029] To avoid inconsistencies in status caused by different data sources due to varying sampling frequencies, if a field has no new data within the current time window, it will preferentially inherit the valid value that has passed verification in the previous period; if the field is missing for multiple consecutive periods, it will be marked as invalid and the default safe value will be called. If the passenger flow density statistics module does not output new results in the current period, it is allowed to use the passenger flow density value from the previous period; if passenger flow information cannot be obtained for multiple consecutive periods, the passenger flow density can be marked as unknown, and no gain correction will be made for the affected range in subsequent priority calculations. This processing ensures that the system can still execute when some sensor information is missing for a short period of time, rather than the entire priority calculation being interrupted due to the absence of a few fields. In terms of the update triggering mechanism, this embodiment adopts a combination of periodic update and event-triggered update. Firstly, when the data collection period reaches the set time interval... First, the system automatically triggers a state set update. Second, when an abnormal event flag changes from invalid to valid, the link quality continuously degrades beyond a preset threshold, or a new emergency service arrives, it immediately triggers an instant update without waiting for the next cycle. This ensures that security-related or time-sensitive services can enter the priority recalculation process as soon as possible. The system does not mechanically refresh according to a fixed cycle but allows for advance updates in emergency scenarios, thereby ensuring that the priority model has sufficient responsiveness to changes in the scenario.
[0030] It should be noted that generating the corresponding interactive semantic priority value includes: extracting multiple semantic attribute parameters for each piece of business data, including business urgency, remaining effective time of information, scope of passenger impact, and business tolerance for interruption; normalizing each semantic attribute and comprehensively evaluating it according to a pre-set weight relationship; when an abnormal event flag is detected and triggered, raising the urgency level of the corresponding business to the highest level; when passenger flow density exceeds a set threshold, increasing the weight of the impact scope of the business in the corresponding area; generating interactive semantic priority values based on the comprehensive evaluation results, and forming a sorting queue according to priority; when there are businesses with the same priority, further sorting them according to the preset priority order of business types.
[0031] It should also be noted that a preferred scheme for generating the corresponding interaction semantic priority value specifically includes obtaining the comprehensive state set. Subsequently, semantic parsing is performed on each business to extract semantic attribute parameters that can directly participate in priority calculation. The semantic parsing in this embodiment does not require complex natural language understanding of video footage or full-text audio; instead, it prioritizes an engineering implementation combining business type, metadata fields, scene tags, and a rule base. It primarily generates semantic attribute values for various businesses based on their source, category, control fields, scheduling flags, and scene status, thus ensuring a simple implementation that can be deployed in an in-vehicle environment. Specifically, for the first... For each business item, four semantic attributes are extracted, which are denoted as the business urgency level. Remaining validity period of the information Scope of passenger impact And the business's tolerance for disruption. .in, Used to characterize the urgency of the business for immediate processing; Used to represent the remaining time that the service will remain valid even after the current time is further delayed; Used to characterize the number of passengers or the coverage area affected by the service; This is used to characterize the level of urgency that is acceptable to passengers and system operation in the event of an interruption during transmission or playback. This embodiment uses a combination of a predefined service level table and event flag escalation rules to determine the level. Specifically, the following basic levels can be pre-set: ordinary promotional videos are low-level, arrival notification videos and guidance broadcasts are medium-level, manual dispatch broadcasts and fault notification information are high-level, and emergency evacuation instructions and safety warning information are the highest level. When an abnormal event flag is triggered, the urgency level of the services directly related to the event is determined. For ordinary business processes not directly related to the abnormal event, the priority level is directly elevated to the highest level; for ordinary business processes, the original level is maintained or the relative priority is reduced as needed. This avoids processing business processes according to the half-time rule in emergency scenarios, which would prevent critical business processes from quickly seizing resources.
[0032] Regarding the remaining validity period of the information In this embodiment, the time of failure is determined based on the difference between the service failure time and the current time. For services with clear time limits, such as pre-arrival transfer reminders, upcoming door opening reminders, and temporary dispatch notifications, the target failure time can be directly recorded in the service metadata. The system calculates the remaining time until the service failure time during each update and determines the time accordingly. The size. For services lacking a clear expiration time, such as promotional videos, a default expiration period can be preset to make them exhibit higher latency tolerance in priority calculation. To maintain comparability between different attributes, this embodiment will... The data is uniformly converted to a standard value between 0 and 1, with the timeliness value being higher for services closer to the failure threshold and requiring immediate display. (This applies to services affecting a wider range of passengers.) This embodiment determines the scope of influence by considering both the service's effective area and passenger flow density. If a service is broadcast or announced to all carriages of the train, its basic influence range is large; if it only targets a single carriage or a specific area, its basic influence range is small. Furthermore, the influence range is adjusted based on the current passenger flow density of the area where the service operates. When guidance information only affects a single carriage, but that carriage is currently experiencing high passenger flow, its actual influence range can be considered expanded; conversely, when promotional information affects multiple carriages, but the current number of passengers in those carriages is low, its influence range gain is relatively limited. Therefore, in engineering practice… It's not solely determined by physical coverage area, but rather by both the size of the coverage area and the passenger flow density within that area. A base score is first assigned based on the coverage area, and then adjustments are made according to the passenger flow density level; for example, adjustments are made when the passenger flow density exceeds a threshold. At that time, the impact level of the corresponding region for this service will be increased. Regarding the level of interruption tolerance... This embodiment pre-sets parameters based on service type and interaction characteristics. Emergency services typically require continuous, immediate, and complete display, therefore their interruption tolerance is the lowest; while manual broadcasts and arrival broadcasts can tolerate very short buffer compensation, they are still generally low-tolerance services; guidance videos can usually accept slight frame rate reduction, reduced resolution, or short-term delays, therefore their interruption tolerance can be set to medium; promotional videos do not directly affect passengers' travel decisions, and can usually be delayed, re-uploaded in segments, or suspended by higher-priority services, therefore their interruption tolerance is relatively high. It should be noted that in this embodiment, The significance of this is the level of acceptability of the service to interruption; therefore, in the calculation of overall priority, its direction of action is the same as... , , different, The higher the value, the more tolerant the business is of interruptions, and its ultimate priority should be lowered accordingly.
[0033] After the extraction of the four types of semantic attributes, in order to ensure that attributes of different scales and from different sources can enter the same priority evaluation process, the following steps are taken: , , , For standardization processing, this embodiment preferably employs a tiered mapping or interval mapping method. Taking urgency level as an example, a preset level can be directly mapped to several discrete values within the range of 0 to 1; taking remaining effective time as an example, it can be converted into a corresponding standard value according to a preset upper limit of effective time; taking impact range and interruption tolerance as examples, they can also be converted into standard values according to predetermined tiers. In this way, priority calculation deals with a set of attributes with a uniform numerical range, avoiding the problem of a single attribute dominating the result due to excessively large or small dimensions.
[0034] After completing semantic attribute extraction and standardization, the system proceeds to generate interaction semantic priority values. In this embodiment, the interaction semantic priority value essentially quantifies the degree to which a service should be "prioritized for transmission, processing, and display" in the current scenario. This embodiment uses a weighted aggregation + rule correction + queue sorting approach to generate the final priority. Specifically, for the first... For each business, the system is based on its standardized version. , , and Generate a priority value In this embodiment, In calculations, urgency and timeliness positively increase priority, while the scope of impact has a scenario-dependent positive effect on priority. Conversely, interruption tolerance has a negative inhibitory effect on priority. In other words, the more urgent the service, the closer it is to failure, the wider its impact on passengers, and the less acceptable the interruption, the higher its priority. The larger the value.
[0035] In a preferred embodiment, a priority value is calculated. , represented as:
[0036]
[0037] in, Indicates the first The semantic priority value of the interaction of the business item. , , and These represent the weighting coefficients corresponding to urgency, timeliness, scope of impact, and tolerance for interruption, respectively. All weighting coefficients are non-negative and can be pre-written into the configuration table of the in-vehicle edge computing module, switching according to the operating scenario. In normal operation scenarios, it is preferable that the weight corresponding to urgency is no less than the weight corresponding to timeliness; in high-passenger-flow or locally congested scenarios, it is permissible to increase the weight corresponding to scope of impact; in emergency scenarios, it is preferable to further increase the weight corresponding to urgency.
[0038] Furthermore, to avoid overly static weight configuration, this embodiment also introduces a rule correction mechanism. Firstly, when an anomaly event flag is valid, the services related to that anomaly event directly execute the priority enhancement rule, that is, they are given priority. Upgrade to the highest level and allow the corresponding... An emergency scenario weighting table is used to ensure that the service ranks high in the queue during the current calculation period. Secondly, when the passenger flow density exceeds a threshold... At the same time, for businesses operating in high-traffic areas, increase the weighting of their influence scope or directly increase their... Standard values are used to reflect the actual importance of the same type of information in crowded carriages compared to the interaction logic in empty carriages. Thirdly, when a service itself is of a low-interruption-tolerance type, such as emergency alerts or critical guidance broadcasts, even if its impact is temporarily small, its queue priority should not be suppressed indefinitely by other high-interruption-tolerance services. This is done after obtaining the priority values corresponding to each service. Then, press Generate a sorted queue to be scheduled from largest to smallest. ,in Indicates time A priority queue. During the sorting process, if two or more services... If values are the same or within a preset approximate range, a secondary sorting rule with equal priority is triggered. This secondary sorting rule can be pre-configured as follows: emergency information takes precedence over manual broadcasts, manual broadcasts over automated arrival broadcasts, automated arrival broadcasts over guidance videos, and guidance videos over regular promotional videos. If the service categories are still the same, priority is given to those with shorter remaining valid time. If they still cannot be distinguished, they are sorted by arrival time. This ensures unique sorting results and avoids inconsistencies in system execution due to a lack of detailed rules when calculated values are similar.
[0039] This embodiment can also include a priority jitter suppression mechanism. Specifically, if a service experiences jitter suppression in two consecutive update cycles... If the change is less than a preset threshold, its relative position in the queue will remain unchanged to avoid frequent switching of service order due to slight fluctuations in passenger flow or short-term disturbances in the link. Only when... Significant adjustments to the sorting queue are only made when changes reach the reordering threshold or when new emergency services enter the system. This reduces the frequency of resource scheduling and PIS display strategy switching, improving the overall system stability.
[0040] It should also be noted that in a metro PIS system with multiple concurrent services, while it can identify service types, it struggles to depict the true importance of services in specific operational scenarios. This is especially true under conditions of rapid changes in emergency situations, passenger flow, and information timeliness. Traditional scheduling methods based on fixed priorities or simple rules cannot reflect the dynamic value of services, leading to a mismatch between critical and non-critical services in resource competition. The solution involves transforming the originally unquantifiable semantics of services into computable structured variables. By constructing a multi-dimensional semantic model that includes urgency, timeliness, scope of impact, and interruption tolerance, services are elevated from "category-driven" to "semantic-driven." Simultaneously, by introducing a comprehensive state set... By incorporating communication status and passenger environment information into the same computational framework, priority is no longer a static label isolated from communication and scenario, but rather a function result that dynamically changes with link quality and passenger flow distribution. Furthermore, by combining weight adjustment and rule correction mechanisms, the model possesses adaptive reinforcement capabilities under abnormal events or high passenger flow scenarios.
[0041] S2: Establish the business interaction experience QoE target based on the interaction semantic priority value, and map the QoE target to the communication service quality QoS parameter constraint. Combine the current vehicle-to-ground wireless communication link status to dynamically allocate bandwidth resources, MIMO spatial stream resources and beamforming resources.
[0042] Furthermore, establishing business interaction experience (QoE) goals includes: establishing corresponding interaction experience evaluation index systems for different types of services; for video services, focusing on picture clarity, playback continuity, and loading response time; for broadcast services, focusing on voice continuity and playback stability; and for emergency information, focusing on information arrival timeliness and coverage. Based on interaction semantic priority values, services are divided into three levels: high priority, medium priority, and low priority, and different experience assurance strategies are matched for each level. For high-priority services, response time and stability requirements are set; for medium-priority services, a dynamic adjustment strategy is adopted; and for low-priority services, a certain degree of experience degradation is allowed. Based on the experience goals, corresponding communication requirements are deduced, including the minimum required bandwidth range, the maximum allowable latency range, and the acceptable packet loss range, establishing a correspondence between interaction experience and communication performance.
[0043] It should also be noted that a preferred approach for establishing business interaction experience QoE goals based on interaction semantic priority values specifically includes, as referenced... Figure 2For video services, the focus is on three core metrics: image clarity, playback continuity, and loading response time. For broadcast services, the focus is on audio continuity and playback stability. For emergency information services, the focus is on the timeliness and coverage of information delivery. Video clarity can be represented by resolution levels or bitrate ranges; playback continuity can be represented by the number of pauses per unit time or the duration of pauses; broadcast continuity can be represented by the number of audio interruptions or the duration of interruptions; and the timeliness of emergency information can be represented by the time interval from triggering to display. Priority values are then obtained. Then, based on preset thresholds, all services are divided into three priority levels: high priority, medium priority, and low priority. Let the high priority threshold be... The low priority threshold is Then when At that time, this service was classified as a high-priority service; when When classified as medium priority service; The system categorizes services into low-priority and high-priority segments. Thresholds can be set based on line operation experience or system configuration and can be switched between different operating modes (peak, off-peak, emergency). Different QoE protection strategies are matched for different priority ranges. For high-priority services, the system sets strict experience targets, such as requiring video services to maintain high resolution and minimal buffering, broadcast services to maintain continuous output and no significant jitter, and emergency information to be displayed within a preset very short time. For medium-priority services, the system allows for adaptive adjustments when the link fluctuates, such as reducing video resolution, increasing buffers, and delaying non-critical broadcasts. For low-priority services, a significant degradation in experience is allowed when resources are insufficient, such as delayed playback, segmented loading, or temporary pausing.
[0044] After the QoE objective is determined, a process of working backward from user experience to deduce communication requirements is further executed. The core idea of this process is that different QoE objectives correspond to different communication performance requirements; therefore, corresponding communication constraint parameters need to be generated for each service. In this embodiment, each service... There are at least three corresponding communication constraint parameters, denoted as minimum bandwidth requirements. Maximum allowable delay and the maximum allowable packet loss ratio Specifically, for video services, when in the high-priority range, the base bitrate requirement is determined based on the target resolution and target frame rate, with additional redundant bandwidth added to mitigate link fluctuations. For medium-priority video services, a lower bitrate range is allowed to maintain basic continuous playback under lower bandwidth conditions. For low-priority video services, only the minimum playable bitrate is guaranteed, or complete pause is permitted. For broadcast services, continuity is prioritized; therefore, the maximum permissible latency and jitter range are constrained, and corresponding minimum continuous transmission bandwidth requirements are provided. For emergency information services, the maximum permissible latency is the primary constraint. And by reducing the allowable packet loss ratio To ensure reliable information delivery. Every business Each can obtain a set of communication requirement parameters corresponding to its QoE target. These parameters will serve as input constraints for subsequent resource allocation, thereby completing the process from semantic priority. Mapping to communication needs.
[0045] It should be noted that dynamic allocation includes: dividing available communication resources into multiple resource units according to the communication requirements of each service, and allocating resource units hierarchically according to service priority; during resource allocation, prioritizing high-priority services to meet minimum communication requirements, and allocating remaining resources to medium and low-priority services when resources are sufficient; dynamically selecting transmission paths based on current channel state information, and prioritizing the selection of the spatial channel with the best channel quality for data transmission in a multiple-input multiple-output communication structure; adjusting the signal transmission direction according to train position and channel direction information to concentrate signal energy in the effective reception area; and triggering a resource reallocation mechanism when link quality degradation or resource conflicts are detected, compressing, delaying, or suspending low-priority services.
[0046] It should also be noted that a preferred scheme for dynamic allocation specifically includes, as referenced Figure 3 After obtaining the communication requirement parameters corresponding to each service, the system enters the dynamic resource allocation stage. Under the constraints of the current link state, higher-priority services are given priority in meeting their communication requirements, while medium- and low-priority services are considered as much as possible within the resource allowance. Specifically, the currently available communication resources are first structurally divided. In this embodiment, communication resources include three categories: bandwidth resources, spatial domain resources, and directional domain resources. Bandwidth resources correspond to the available spectrum or transmission rate range; spatial domain resources correspond to the number of available spatial streams in the MIIMO system; and directional domain resources correspond to the set of selectable beam directions. Based on the sorting queue... Services are processed sequentially from highest to lowest priority. For the first service in the queue (i.e., the highest priority service), it is first determined whether the currently available bandwidth meets its minimum bandwidth requirement. Simultaneously determine whether the current link latency and packet loss level meet the requirements. and The constraints are as follows: If satisfied, corresponding resources are allocated to the service; if not satisfied, the resource structure is adjusted first, such as compressing resources occupied by low-priority services or reselecting spatial channels to satisfy the high-priority service as much as possible. Regarding spatial domain resource allocation, for the MIMO communication structure, the quality of all available spatial channels is evaluated. Specifically, the channel quality index of each spatial channel is recorded over several consecutive sampling periods, and channels that maintain high quality over multiple periods are prioritized as the primary transmission path, rather than being selected solely based on instantaneous optimal values, thus avoiding instability caused by frequent switching. Regarding directional domain resources (i.e., beamforming), based on the train's current position, direction of travel, and historical link quality records, the beam direction with the highest matching degree with the current environment is selected from a preset beam set. The matching degree indicates that the received signal strength and stability of the corresponding beam are at a relatively good level over the most recent several periods. This method concentrates signal energy as much as possible in the effective receiving area, thereby improving link quality.
[0047] After allocating high-priority services, the remaining resources are allocated to medium-priority services according to the queue order. When the remaining resources are insufficient to simultaneously meet the needs of all medium-priority services, scheduling can be carried out using round-robin or priority-weighted allocation. For low-priority services, allocation is only performed when resources are sufficient; otherwise, delays or suspensions are allowed. This embodiment also sets up a resource reallocation trigger mechanism. When any of the following conditions are detected, the reallocation process is immediately triggered: first, the link quality continuously declines and falls below a preset threshold; second, the actual latency or packet loss exceeds the constraint threshold corresponding to any service; third, a new high-priority or emergency service enters the queue. During the reallocation process, the system prioritizes reclaiming the resources occupied by low-priority services and re-executes the allocation process to ensure that the communication needs of critical services are met first. Through QoE target establishment and dynamic resource allocation processes, the system achieves a mechanism that prioritizes semantic priority. and sorted queue The continuous mapping between the actual allocation of communication resources and the actual results enables different services in the PIS system to obtain communication guarantees commensurate with their interaction importance in complex link environments.
[0048] It should also be noted that in the integration of existing vehicle-to-ground communication and PIS systems, the communication system typically allocates resources based on QoS indicators, while the PIS system focuses on the user-perceived interactive experience (QoE). The lack of a clear mapping relationship between the two leads to unsatisfactory user experience even when communication indicators meet requirements, and vice versa. By constructing an intermediate mapping link from semantic priority to QoE target to QoS constraints, the abstract importance of services is transformed into specific communication resource requirement parameters. Specifically, by dividing priority values into different level ranges, differentiated experience targets are matched, and minimum bandwidth, maximum latency, and dropout tolerance are further derived, making communication resource allocation directly guided by experience targets. Simultaneously, at the resource scheduling level, it is no longer limited to single bandwidth allocation but incorporates MIMO spatial streams and beam directions into a unified scheduling framework, combining link status for multi-dimensional resource collaborative optimization. This solves the problem of how to finely match communication resources under multiple services and constraints, ensuring that resource allocation not only meets technical indicators but also serves the service experience. This improves resource utilization efficiency and critical service assurance capabilities.
[0049] S3: Based on the dynamic allocation results, the video playback, broadcast insertion and emergency information display strategies of the PIS system are adaptively adjusted, and business execution feedback data is collected to perform closed-loop correction of the interaction semantic priority model and communication resource allocation strategy.
[0050] Furthermore, adaptive adjustments include determining the current service operation status based on the matching degree between allocated communication resources and service requirements; executing normal playback or broadcasting strategies when resources meet service requirements; reducing resolution or adjusting playback frequency for video services and providing delay compensation or buffering playback for broadcast services when resources are less than service requirements; pausing or delaying low-priority services when resources are severely insufficient; immediately triggering a preemption mechanism when emergency information has a higher priority than currently executed services, interrupting the current service and prioritizing the display or broadcast of emergency information; and simultaneously, making differentiated adjustments to the display content in different areas based on passenger flow in the carriage.
[0051] It should also be noted that a preferred approach for adaptive adjustments specifically includes, for each business function... Based on the actual resource allocation results obtained, and their corresponding communication requirement parameters Perform matching analysis to determine the current operational status of the service. Resource matching coefficient. This is used to represent the relationship between the actual allocation of resources and the minimum demand. It can be understood as the ratio between the actual available bandwidth and the minimum bandwidth requirement, and is combined with whether the latency and packet loss meet the constraints for a comprehensive judgment.
[0052] When the actual allocated bandwidth is not less than And the current link latency and packet loss both meet the requirements. and When required, determine When the bandwidth is satisfied, the service is operating normally; when the actual bandwidth is lower than the specified value... However, if the latency or packet loss is still higher than the preset degradation operation threshold, and the latency or packet loss only slightly exceeds the constraint range, then it is determined that... In a partially satisfied state; when the actual bandwidth is much lower than If the latency or packet loss significantly exceeds the constraints, a judgment will be made. It is in a state of dissatisfaction.
[0053] Differentiated interaction strategies are implemented for different types of services. When a service is in a satisfied state, the corresponding QoE target strategy is maintained, i.e., video services play normally at the target resolution and frame rate, broadcast services broadcast continuously, and emergency information is displayed in a predetermined manner. When a service is in a partially satisfied state, adaptive degradation processing is implemented. For video services, this involves reducing the resolution level, frame rate, or increasing buffer length to reduce bandwidth dependence; for broadcast services, this involves introducing a buffer compensation mechanism or appropriately delaying non-critical broadcasts to avoid voice interruptions caused by link jitter; for emergency information services, compression encoding or prioritizing the display of key information fields can be used to shorten transmission time, provided that the core content is not affected. When a service is in a dissatisfied state, a stricter control strategy is implemented. Specifically, for low-priority services, their transmission or display can be directly suspended to release resources to higher-priority services; for medium-priority services, the remaining effective time is used to determine the appropriate level of service. Determine whether to postpone execution or move to a waiting queue; for high-priority services, even if resources are insufficient, prioritize the transmission of their core content, such as retaining only key frames or key voice segments.
[0054] This embodiment also includes an emergency preemption mechanism. When the sorting queue... New emergency information services have emerged, and their priority values are... When a priority service is triggered, an immediate preemption operation is performed, interrupting the transmission and display of the current low-priority or medium-priority service. The emergency information is inserted at the head of the output queue, and resources are allocated preferentially to complete its display. After the emergency information is executed, the remaining valid time of the interrupted service is used to determine the next priority. The system determines whether to continue execution or re-queue the train. Furthermore, based on the obtained passenger flow density information, the display strategy for different carriages or areas is adjusted accordingly. For example, when the passenger flow density in a certain carriage exceeds a threshold... When passenger flow is high, guidance or safety information is prioritized for display in that area, while lower-priority video content can continue to be displayed in areas with lower passenger flow. This approach allows the interaction strategy to not only rely on communication resources but also to be optimized in conjunction with passenger distribution. It achieves a dynamic mapping from communication resource allocation results to PIS interaction behavior, enabling the service display strategy to adaptively adjust to changes in link status.
[0055] It should be noted that the closed-loop correction of the interaction semantic priority model and communication resource allocation strategy includes: collecting feedback data during business execution, including actual transmission latency, data loss, video playback smoothness, and broadcast continuity, and comparing and analyzing the feedback data with preset targets; when the actual operating status deviates from the target range, determining whether the deviation originates from inaccurate priority assessment or unreasonable resource allocation; if it is a priority assessment problem, adjusting the semantic weights to increase the weight of urgency or scope of impact; if it is a resource allocation problem, adjusting the resource allocation strategy to increase the resource proportion of critical businesses; and maintaining the current parameter settings when the system runs stably for multiple consecutive cycles.
[0056] It should also be noted that a preferred scheme for closed-loop correction of the interaction semantic priority model and communication resource allocation strategy specifically includes, after the interaction strategy is executed, further monitoring of the business execution effect, and performing closed-loop correction of the priority model and resource allocation strategy based on the feedback results, thereby improving the long-term stability and adaptability of the system. Specifically, business execution feedback data is collected at the end of each time period to form a feedback set. Feedback data includes actual transmission latency, actual packet loss, video playback smoothness metrics, and broadcast continuity metrics. Video playback smoothness can be represented by the number of stutters or stutter duration per unit time, while broadcast continuity can be represented by the number of audio interruptions or interruption duration. This feedback data is then compared with the target parameters set for each service. A comparative analysis is performed with the QoE target. When the actual operating indicators continuously deviate from the target range, a deviation is determined, and the source of the deviation is further identified. In a preferred embodiment, the sources of deviation are divided into two categories: semantic priority assessment deviation and resource allocation deviation. When a certain business has a high priority (i.e., Located in the high priority range), and its actual resource allocation has been satisfied. If the expected user experience is still not achieved, it is considered a semantic evaluation deviation. This indicates that the weighting of the service in the priority model is insufficient to reflect its actual importance. When a service has a high priority, but its resource allocation is consistently lower than demand, and there is competition for resources in the link, it is considered a resource allocation deviation. To address semantic evaluation deviation, the weight parameters in step S1 are adjusted. For example, the urgency level is increased. or scope of influence The corresponding weight coefficients are used to improve the ranking of this type of service in subsequent priority calculations. For resource allocation deviations, the resource scheduling strategy in step S2 is adjusted, for example, by increasing the resource reservation ratio for high-priority services or decreasing the maximum resource ratio that low-priority services can occupy. To avoid frequent adjustments leading to system instability, a stability determination mechanism is set up in this embodiment. Specifically, when the actual latency, packet loss, and playback continuity of all key services remain within the target range for multiple consecutive periods, the system is determined to have entered a stable state, at which point the current weight parameters and resource allocation strategy remain unchanged; only when the deviation continuously exceeds a preset period threshold is a new round of parameter adjustments triggered.
[0057] It should also be noted that in complex dynamic environments, even after priority modeling and resource allocation are completed, there is still a discrepancy between the decision-making results and the actual operational effects. This discrepancy accumulates over time, and without a feedback mechanism, system performance will gradually deteriorate. Existing PIS and communication scheduling systems generally employ static strategies or weak feedback mechanisms, making it difficult to effectively correct the model based on the operational results. This invention proposes a closed-loop mechanism that drives model optimization from the execution results. This is achieved by constructing resource matching coefficients... This involves directly comparing the communication resource allocation results with business requirements and dynamically adjusting the interaction strategy accordingly, enabling the business layer to respond instantly to the communication status; simultaneously, it collects feedback sets... By comparing actual latency, packet loss, and playback smoothness metrics with the target QoE, the source of deviation is identified as either priority model distortion or insufficient resource allocation. Based on this, semantic weights or resource strategies are adjusted accordingly, enabling the system to gradually approach the optimal state in subsequent cycles. This mechanism overcomes the limitations of traditional single-cycle scheduling, giving the system continuous learning and adaptive capabilities. This solves the problem of unstable long-term performance optimization in complex scenarios, ultimately achieving a closed-loop synergistic improvement between interactive experience, communication resources, and the decision-making model.
[0058] Example 2, an embodiment of the present invention, provides a subway train PIS intelligent interaction system, including a data processing module, a resource allocation module, and a strategy adjustment module.
[0059] The data processing module collects multi-service data, vehicle-to-ground communication status data, and passenger interaction context information from the subway train's PIS system. It performs semantic parsing on the multi-service data to extract the urgency, timeliness, and scope of impact of the services. Based on the semantic parsing results, it constructs an interaction semantic priority model to generate corresponding interaction semantic priority values. The resource allocation module establishes a service interaction experience QoE target based on the interaction semantic priority values and maps the QoE target to communication service quality (QoS) parameter constraints. It dynamically allocates bandwidth resources, MIMO spatial stream resources, and beamforming resources in conjunction with the current vehicle-to-ground wireless communication link status. The strategy adjustment module adaptively adjusts the video playback, broadcast insertion, and emergency information display strategies of the PIS system based on the dynamic allocation results. It also collects service execution feedback data to perform closed-loop correction of the interaction semantic priority model and communication resource allocation strategy.
[0060] Example 3, referring to Figure 4 This embodiment also provides a computer device applicable to the PIS intelligent interaction method for subway trains, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the PIS intelligent interaction method for subway trains as proposed in the above embodiment.
[0061] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.
[0062] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the PIS intelligent interaction method for subway trains as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.
Claims
1. A PIS intelligent interaction method for subway trains, characterized in that, include: Collect multi-service data, vehicle-to-ground communication status data, and passenger interaction context information from the PIS system of subway trains. Perform semantic parsing on the multi-service data to extract the urgency, timeliness, and scope of impact of the services. Based on the semantic parsing results, construct an interaction semantic priority model to generate corresponding interaction semantic priority values. The business interaction experience QoE target is established based on the interaction semantic priority value, and the QoE target is mapped to the communication service quality QoS parameter constraint. The bandwidth resources, MIMO spatial stream resources and beamforming resources are dynamically allocated in combination with the current vehicle-to-ground wireless communication link status. Based on the dynamic allocation results, the video playback, broadcast insertion, and emergency information display strategies of the PIS system are adaptively adjusted, and business execution feedback data is collected to perform closed-loop correction of the interaction semantic priority model and communication resource allocation strategy.
2. The subway train PIS intelligent interaction method as described in claim 1, characterized in that: The data collected from the subway train PIS system includes multi-service data, vehicle-to-ground communication status data, and passenger interaction context information. The vehicle-mounted data acquisition unit acquires video service streams, broadcast service streams, and emergency information data streams, and classifies and labels them according to service type. Meanwhile, the bandwidth utilization status, transmission latency, packet loss, and channel quality indicators of the vehicle-to-ground link are obtained in real time through the communication interface, and are periodically updated according to a preset time window. Passenger interaction context information is obtained through onboard sensors and system logs, including carriage location, current station status, passenger flow density, and indicators of whether there are abnormal events. Multi-source data is uniformly encapsulated to form a comprehensive status set that includes communication status, environmental status, and business status; Data updates are triggered when the data collection cycle reaches the set time interval; otherwise, the data from the previous cycle is maintained.
3. The subway train PIS intelligent interaction method as described in claim 2, characterized in that: The generation of the corresponding interactive semantic priority value includes, For each piece of business data, extract multiple semantic attribute parameters, including business urgency, remaining valid time of information, scope of passengers affected, and business tolerance for interruption; Each semantic attribute is normalized and then comprehensively evaluated based on a pre-defined weighting relationship. When an abnormal event flag is detected and triggered, the urgency level of the corresponding service will be raised to the highest level. When passenger flow density exceeds a set threshold, the weight of the impact range involving the corresponding area's business is increased. Based on the comprehensive evaluation results, interactive semantic priority values are generated, and sorted queues are formed according to priority. When there are services with the same priority, they are further sorted according to the preset priority order of service types.
4. The subway train PIS intelligent interaction method as described in claim 3, characterized in that: The goals for establishing a business interaction experience (QoE) include: Establish corresponding interactive experience evaluation index systems for different types of services. For video services, focus on picture clarity, playback continuity and loading response time; for broadcast services, focus on voice continuity and playback stability; and for emergency information, focus on the timeliness and coverage of information arrival. Based on the interaction semantic priority value, the business is divided into three levels: high priority, medium priority and low priority, and different experience guarantee strategies are matched for each level. For high-priority services, set response time and stability requirements; for medium-priority services, adopt a dynamic adjustment strategy; and for low-priority services, allow for a certain degree of performance degradation. Based on the experience goals, the corresponding communication requirements are deduced, including the minimum required bandwidth range, the maximum allowable latency range, and the acceptable packet loss range, thus establishing a correspondence between interactive experience and communication performance.
5. The subway train PIS intelligent interaction method as described in claim 4, characterized in that: The dynamic allocation includes, Based on the communication requirements of each service, the available communication resources are divided into multiple resource units, and the resource units are allocated hierarchically according to the service priority. During resource allocation, priority is given to ensuring that high-priority services meet the minimum communication requirements, and when resources are sufficient, the remaining resources are allocated to medium and low-priority services. Based on the current channel state information, the transmission path is dynamically selected, and in the multiple-input multiple-output communication structure, the spatial channel with the best channel quality is selected first for data transmission. At the same time, the signal transmission direction is adjusted according to the train's position and channel direction information so that the signal energy is concentrated in the effective receiving area; When a degraded link quality or resource conflict is detected, a resource reallocation mechanism is triggered to compress, delay, or suspend low-priority services.
6. The subway train PIS intelligent interaction method as described in claim 5, characterized in that: The adaptive adjustment includes, Based on the degree of matching between the allocated communication resources and service requirements, the current service operation status is determined, and when the resources meet the service requirements, the normal playback or broadcasting strategy is executed. When resources are lower than business needs, reduce the resolution or adjust the playback frequency of video services, and compensate for delays or cache the playback of broadcast services. When resources are severely insufficient, low-priority services will be suspended or delayed. When emergency information has a higher priority than currently executed business, the preemption mechanism is immediately triggered, interrupting the current business and prioritizing the display or broadcast of emergency information; At the same time, the displayed content in different areas is adjusted according to the passenger flow in the carriages.
7. The subway train PIS intelligent interaction method as described in claim 6, characterized in that: The closed-loop correction of the interaction semantic priority model and communication resource allocation strategy includes... Collect feedback data during business execution, including actual transmission latency, data loss, video playback smoothness, and broadcast continuity, and compare and analyze the feedback data with preset targets; When the actual operating status is detected to deviate from the target range, determine whether the deviation is due to inaccurate priority assessment or unreasonable resource allocation. If it is a priority assessment problem, the semantic weights are adjusted to increase the weights of urgency or scope of impact; If the issue is related to resource allocation, then adjust the resource allocation strategy and increase the resource allocation ratio for key business operations. When the system runs stably for several consecutive cycles, maintain the current parameter settings.
8. A subway train PIS intelligent interaction system, employing the subway train PIS intelligent interaction method as described in any one of claims 1 to 7, characterized in that: It includes a data processing module, a resource allocation module, and a strategy adjustment module; The data processing module is used to collect multi-service data, vehicle-to-ground communication status data and passenger interaction context information from the PIS system of the subway train, and to perform semantic parsing on the multi-service data to extract the urgency, timeliness and scope of impact of the services. Based on the semantic parsing results, an interaction semantic priority model is constructed to generate the corresponding interaction semantic priority value. The resource allocation module is used to establish the business interaction experience QoE target based on the interaction semantic priority value, and map the QoE target to the communication service quality QoS parameter constraint, and dynamically allocate bandwidth resources, MIMO spatial stream resources and beamforming resources in combination with the current vehicle-to-ground wireless communication link status; The strategy adjustment module is used to adaptively adjust the video playback, broadcast insertion, and emergency information display strategies of the PIS system based on the dynamic allocation results, and to collect service execution feedback data to perform closed-loop correction of the interactive semantic priority model and communication resource allocation strategy.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the subway train PIS intelligent interaction method as described in any one of claims 1 to 7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the subway train PIS intelligent interaction method as described in any one of claims 1 to 7.