Urban rail train multi-subsystem intelligent fusion management method and system

Abstract

The invention relates to the technical field of fusion management, in particular to an urban rail train multi-subsystem intelligent fusion management method and system, and the method comprises the steps: obtaining operation data through multiple types of sensing terminals inside and outside each carriage of a train; constructing a situation awareness model of a passenger and environment interaction relationship, and predicting passenger interaction data in a future time period; adopting an anomaly detection model to identify emergencies and classifying emergencies; the method comprises the following steps: constructing a hierarchical fusion management architecture comprising a compartment-level local management unit and a train-level global coordination unit, wherein the local management unit dynamically adjusts the configuration of an operation subsystem in each compartment; and the global coordination unit constructs a global coordination model, and sets a priority regulation and control weight for a compartment area corresponding to an emergency by taking the passenger comfort and the system energy consumption as targets. According to the invention, the operation state of each subsystem can be sensed in real time and intelligently adjusted according to passenger flow and demand changes, so that the passenger experience is optimized, the energy consumption is reduced, and the overall system efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of integrated management, and specifically provides an intelligent integrated management method and system for multiple subsystems of urban rail trains. Background Art

[0002] With the rapid development of smart city construction and urban rail transit, urban rail trains are gradually entering a new stage of high integration and intelligent collaboration. Traditional train control systems are mostly independently composed of multiple subsystems such as traction, braking, doors, air conditioning, passenger information, and security. There is a lack of efficient data linkage and intelligent collaboration mechanisms between these systems. This "each doing its own thing" structure gradually exposes problems such as slow response, uncoordinated scheduling, and low energy utilization rate during daily operation. Especially during peak passenger flows and special operating states, it is easy to cause subsystem control failures or feedback delays, affecting train safety, efficiency, and the passenger experience. Although existing technologies have introduced the concept of cloud platform integrated management or proposed integrated prediction mechanisms, they still remain at the passive monitoring and fault warning at the platform level or system level, lacking the ability to actively and intelligently optimize the micro-coupling between core operation control subsystems, and it is difficult to meet the trend of multi-system collaborative evolution.

[0003] Especially during actual operation, the passenger flow on urban rail trains shows typical instantaneous mutation characteristics. For example, during morning and evening rush hours or large station transfers, passengers gather or disperse extremely quickly in an instant. At this time, if the door system, air conditioning system, and braking system lack integrated perception and collaborative control, it will not only cause problems such as delayed opening / closing of doors and misjudgment, but also may lead to uncoordinated air conditioning cooling intensity and braking response deviation due to fluctuations in the carriage load, thereby triggering a series of chain reactions such as passenger congestion, decreased comfort, and increased energy consumption. Therefore, there is an urgent need to propose a multi-subsystem intelligent integration method that can real-time sense the behavior and context changes of passengers and intelligently adjust the operating states of each subsystem according to the changes in passenger flow and demand, so as to optimize the passenger experience, reduce energy consumption, and improve the overall system efficiency.

[0004] Therefore, an intelligent integrated management method and system for multiple subsystems of urban rail trains are proposed. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent integrated management method and system for multiple subsystems of urban rail trains, so as to realize intelligent adjustment of the operating states of each subsystem by real-time sensing of passenger flow and demand changes, thereby optimizing the passenger experience, reducing energy consumption, and improving the overall system efficiency.

[0006] To achieve the above purpose, the present invention provides the following technical solutions: An intelligent integrated management method for multiple subsystems of urban rail trains, including: Obtain operation data through various types of perception terminals inside and outside each carriage of the train; the various types of perception terminals include: thermal imaging devices, pressure detectors, video acquisition devices, gas detectors, and motion sensors; Based on the operation data, construct a situation awareness model for passenger-environment interaction, and predict passenger interaction data within a future time period, including first data, second data, and third data; Analyze the operation data and passenger interaction data, use an anomaly detection model to identify emergencies, and classify the emergency levels; Construct a hierarchical management architecture including a carriage-level local management unit and a train-level global coordination unit, where: The carriage-level local management unit dynamically adjusts the configuration parameters of the operation subsystems inside each carriage based on the passenger interaction data and the emergency level, in combination with the global scheduling strategy; The train-level global coordination unit summarizes the operation data, passenger interaction data, and emergency levels of the entire train, constructs a global collaboration model, and sets priority control weights for the emergency impact areas in combination with passenger comfort and system energy consumption reduction, generates a global scheduling strategy, and issues it to the carriage-level local management unit.

[0007] Preferably, the operation data includes: passenger behavior data, environmental status data, and on-vehicle equipment data; The passenger behavior data includes: the number of passengers, distribution density, staying positions, movement paths, body temperature, and boarding and alighting frequencies; The environmental status data includes: carriage temperature, humidity, CO 2 concentration, odor concentration, and noise level; The on-vehicle equipment data includes: real-time operation data of each subsystem, vehicle position, and speed.

[0008] Preferably, the situation awareness model includes: a physiological feature extraction unit, an emotion recognition unit, an interaction modeling unit, and a behavior prediction unit; The physiological feature extraction unit uses face recognition and speech recognition technologies to extract the physiological features of passengers, including: expression features, intonation features, and body posture features; The emotion recognition unit constructs a multi-modal emotion recognition model based on the physiological features to identify the current emotional state of passengers; The interaction modeling unit constructs a passenger-environment situation awareness model based on passenger behavior data, environmental status data, and the emotional state to identify the interaction behaviors between passengers and environmental facilities inside the carriage; the interaction behaviors include: the proximity, residence time, and usage frequency of passengers to doors, air-conditioning outlets, seats, and handrails; The behavior prediction unit predicts the passenger interaction data within a future time period based on the emotional state and interaction behaviors.

[0009] Preferably, the first data is the passenger flow trend, which is obtained by analyzing historical operation data and current passenger distribution and using time series analysis to predict the passenger entry and exit situations in each carriage during future time periods. The second data is the passenger behavior pattern, which is obtained by clustering analysis to identify common passenger behavior patterns. The third data is the change in passenger demand, which is obtained by predicting the change in passengers' demand for environmental facilities based on their emotional states and interaction behaviors.

[0010] Preferably, the anomaly detection model includes: a multi-dimensional feature extraction unit, an anomaly detection unit, and an emergency event level determination unit; The multi-dimensional feature extraction unit extracts statistical features, spatio-temporal features, and behavior pattern features based on operation data and passenger interaction data. The anomaly detection unit calculates the anomaly scores for each carriage area through an anomaly detection algorithm using the statistical features, spatio-temporal features, and behavior pattern features, and identifies emergency events. The emergency event level determination unit classifies the identified emergency events according to the anomaly scores, anomaly duration, and influence scope.

[0011] Preferably, the carriage-level local management unit includes: Evaluating the passenger comfort level and system operation efficiency of the current carriage according to the passenger interaction data and the emergency event level; combining with the global scheduling strategy to determine the operation subsystems that need to be adjusted and their configuration parameters; sending adjustment instructions to the corresponding operation subsystems to dynamically adjust the configuration parameters of the operation subsystems in each carriage; monitoring the adjusted operation data and passenger interaction data, evaluating the adjustment results, and feeding back the evaluation results to the train-level global coordination unit.

[0012] Preferably, the train-level global coordination unit includes: Collecting operation data, passenger interaction data, and emergency event levels from all carriages, and performing preprocessing and feature extraction to identify the passenger distribution, environmental status, and system operation status of the entire train; constructing a global cooperation model, comprehensively considering passenger comfort and system energy consumption, and using a multi-objective optimization algorithm to solve for the optimal regulation strategy; setting priority regulation weights according to the emergency event level and influence scope; generating a global scheduling strategy for each carriage based on the global cooperation model and priority regulation weights; and sending the global scheduling strategy to the carriage-level local management unit of each carriage.

[0013] Preferably, an urban rail train multi-subsystem intelligent fusion management system includes: A data acquisition module for acquiring operation data through multi-type sensing terminals inside and outside each carriage of the train. A situation awareness module, configured to construct a situation awareness model of the interaction relationship between passengers and the environment based on the operation data, and predict passenger interaction data in a future period, including first data, second data, and third data; An emergency detection module, configured to analyze the operation data and passenger interaction data, and use an anomaly detection model to identify emergencies and classify the emergency levels; A local-global fusion management module, configured to construct a hierarchical management architecture including a carriage-level local management unit and a train-level global coordination unit, where: The carriage-level local management unit dynamically adjusts the configuration parameters of the operation subsystems in each carriage based on the passenger interaction data and the emergency level, in combination with the global scheduling strategy; The train-level global coordination unit summarizes the operation data, passenger interaction data, and emergency levels of the entire train and constructs a global collaboration model. With the goal of passenger comfort and system energy consumption, it sets priority control weights for the carriage areas corresponding to emergencies, generates a global scheduling strategy, and issues it to the local management unit.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. The present invention proposes a situation awareness model that models the behavior, decision-making preferences of passengers, and their interaction behavior data with the carriage environment through multi-source information. It can not only capture data such as the emotional state and behavior tendency of passengers, but also predict their possible next actions based on the psychological state and physiological reactions of passengers. Through situation awareness and prediction, the system can real-time sense and adapt to different passenger behavior patterns, providing more forward-looking and proactive perception support for the intelligent linkage of each subsystem, helping to optimize the passenger experience and improve the active control effect at the same time.

[0015] 2. The anomaly detection model constructed by the present invention can effectively identify various types of emergencies, including passenger aggregation, emotional intensification, environmental anomalies, operation deviations, etc., and reasonably classify their influence scope and severity by introducing a multi-index fusion mechanism based on behavior patterns and spatial features. This model improves the detection accuracy and response efficiency of the system for abnormal states in a complex operation environment, providing a basis for subsequent priority control in combination with the emergency level.

[0016] 3. The present invention constructs a two-layer integrated management architecture consisting of a car-level local management unit and a train-level global coordination unit, taking into account the two major requirements of local rapid response and global collaborative optimization. In this architecture, when an emergency occurs, emergency strategies can be immediately executed based on the real-time status and event level of the car, such as forced ventilation or delayed door closing, to ensure local safety and comfort; at the same time, the train-level global coordination unit gathers data from each car, builds a multi-subsystem collaborative model, takes passenger comfort and system energy consumption as dual goals, and dynamically adjusts the resource allocation and operating parameters of the entire train through a multi-objective optimization algorithm to ensure a dynamic balance between overall energy consumption and passenger comfort. This architecture can not only respond quickly in emergency situations, but also achieve the optimal balance between energy consumption and comfort at the overall level, significantly improving train operating efficiency and passenger satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of a flow chart of a multi-subsystem intelligent integration management method for an urban rail train provided by an embodiment of the present invention; Figure 2 A schematic diagram of the structure of a multi-subsystem intelligent fusion management system for urban rail trains provided by an embodiment of the present invention; Figure 3 A working principle diagram of a context awareness model provided by an embodiment of the present invention; Figure 4 This is a working principle diagram of the local-global fusion management module proposed in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] In actual operation, the passenger flow on and off urban rail trains presents typical instantaneous mutation characteristics. For example, during rush hours in the morning and evening or when transferring at major stations, passengers gather or evacuate very quickly. At this time, if the door system, air-conditioning system and braking system lack integrated perception and coordinated control, it will not only cause problems such as door opening / closing lag and misjudgment, but may also cause air-conditioning cooling intensity imbalance and braking response deviation due to carriage load fluctuations, which in turn may cause a series of chain reactions such as passenger congestion, reduced comfort and increased energy consumption.

[0020] The present invention proposes a method and system for intelligent fusion management of multiple subsystems of urban rail trains to achieve real-time perception and intelligent adjustment of the operating states of each subsystem according to the changes in passenger flow and demand, thereby optimizing the passenger experience, reducing energy consumption, and improving the overall system efficiency. The core hardware support platform for this method and system is the novel intelligent fusion host device provided by the present invention. This device is a fusion analysis and processing host integrating the functions of multiple subsystems (such as doors, air conditioners, ventilation, lighting, braking, etc. involved in this embodiment). This host device provides high processing speed and powerful AI computing power, enabling each subsystem to simply send the original data collected at the front end to this device. Relying on its own powerful AI computing power, this device completes all operations and processing (including data acquisition, model construction, event recognition, policy generation, etc. detailed below), makes a quick response, and transmits relevant data or instructions to the TCMS network and each subsystem through its communication proxy program.

[0021] Based on the above intelligent fusion host device platform, the method for intelligent fusion management of multiple subsystems of urban rail trains proposed by the present invention can operate effectively. To illustrate that the method of the present invention can play a role in optimizing the passenger experience, reducing energy consumption, and improving the overall system efficiency, the effectiveness of the present invention will be described below with two embodiments.

[0022] Embodiment 1 In the embodiment of the present application, the method and system proposed by the present invention (running on the aforementioned intelligent fusion host device) are used to detail the process of intelligently adjusting the operating states of each subsystem by real-time perceiving the changes in passenger flow and demand, thereby optimizing the passenger experience, reducing energy consumption, and improving the overall system efficiency. The embodiment of the present application is directed to the intelligent fusion management of multiple subsystems in urban rail train A with high-density operation and frequent emergency states. The following will be based on Figure 1 the content to detail the intelligent fusion management process of multiple subsystems during the operation of this urban rail train; among them, Figure 1The specific flowchart of the method proposed by the present invention includes: obtaining operation data through various types of perception terminals inside and outside each carriage of the train; constructing a situation awareness model of passenger-environment interaction based on the operation data to predict passenger interaction data in future time periods, including first data, second data, and third data; analyzing the operation data and passenger interaction data, using an anomaly detection model to identify emergencies and classify the emergency levels; constructing a hierarchical management architecture including a carriage-level local management unit and a train-level global coordination unit, where: the carriage-level local management unit dynamically adjusts the configuration parameters of the operation subsystems inside each carriage based on the passenger interaction data and the emergency levels, in combination with the global scheduling strategy; the train-level global coordination unit summarizes the operation data, passenger interaction data, and emergency levels of the entire train, constructs a global collaboration model, and sets priority control weights for the emergency impact areas with the dual objectives of improving passenger comfort and reducing system energy consumption, generates a global scheduling strategy, and issues it to the local management unit. Combined with Figure 1 and Figure 2 the content in An intelligent fusion management method for multiple subsystems of an urban rail train includes: Obtaining operation data through various types of perception terminals inside and outside each carriage of the train; the various types of perception terminals include: thermal imaging devices, pressure detectors, video acquisition devices, gas detectors, and motion sensors; The operation data includes: passenger behavior data, environmental state data, and on-vehicle equipment data; The passenger behavior data includes: the number of passengers, distribution density, staying positions, moving paths, body temperature, and boarding and alighting frequencies; The environmental state data includes: carriage temperature, humidity, CO 2 concentration, odor concentration, and noise level; The on-vehicle equipment data includes: real-time operation data of each subsystem, vehicle position, and speed.

[0023] Specifically, in this embodiment, the operation data is collected in real time by various types of perception terminals inside and outside the carriage during the train operation, specifically including: the thermal imaging device is used to detect the temperature field distribution inside the carriage and identify local overheating areas; the pressure detector is deployed in the door area and the seat area to obtain the passenger distribution density and dynamic changes; the video acquisition device is used to record the boarding and alighting behaviors and staying trajectories of passengers; the gas detector is used to detect air quality parameters such as CO2 concentration and odor intensity; the motion sensor includes an accelerometer and a gyroscope, and is used to monitor the acceleration and deceleration states of the train and the shaking characteristics of the carriage.

[0024] These multi-source heterogeneous data collected are sent to the intelligent fusion host through the train network for unified processing.

[0025] In this embodiment, multi-type sensing terminals including a thermal imaging device, a pressure detector, an odor detector, a high-frequency video acquisition device, and a motion sensor are introduced in the operation data collection link to construct a multi-source heterogeneous and real-time high-frequency sensing network. The thermal imaging device can visualize the temperature distribution in the carriage and identify the aggregated hot areas; the pressure detector can sense the passenger distribution and the dynamics of getting on and off the train in real time; the odor detector is used to identify abnormal air quality and improve the environmental comfort management; the high-frequency video acquisition supports the structured recognition of passenger behaviors; and the motion sensor can accurately sense the coupling relationship between the train dynamics and the passenger responses. This solution significantly improves the system's refined sensing ability for complex passenger flows and environmental changes, provides a solid data foundation for subsequent situation awareness and control optimization by the intelligent fusion host, and significantly improves the system's real-time sensing ability and environmental response accuracy.

[0026] Preferably, according to the operation data, with the support of the AI computing power unit of the intelligent fusion host, a situation awareness model of passenger-environment interaction is constructed, and the passenger interaction data in the future time period is predicted, including the first data, the second data, and the third data; as Figure 3 shown; The situation awareness model includes: a physiological feature extraction unit, an emotion recognition unit, an interaction modeling unit, and a behavior prediction unit; The physiological feature extraction unit extracts the physiological features of passengers using face recognition and speech recognition technologies, including: expression features, intonation features, and body posture features; The emotion recognition unit constructs a multi-modal emotion recognition model based on the physiological features to identify the current emotion state of passengers; The interaction modeling unit constructs a situation awareness model of passenger-environment based on passenger behavior data, environmental state data, and the emotion state to identify the interaction behaviors between passengers and the environmental facilities in the carriage; the interaction behaviors include: the proximity, residence time, and usage frequency of passengers to the door, air-conditioning outlet, seat, and handrail; The behavior prediction unit predicts the passenger interaction data in the future time period based on the emotion state and the interaction behaviors.

[0027] Specifically, the feature extraction unit passes the video stream collected by the video acquisition device through the OpenCV and dlib face detection modules to detect the face area and extract 68 key face points; calculates the expression features (such as smiling, frowning) based on the key points; first pre-emphasizes the speech in the video stream through a filter, and then extracts 13-dimensional MFCC features, combined with short-time energy and fundamental frequency, to form intonation features; extracts the body posture features (such as standing, sitting, leaning forward, etc.) through key bone point detection according to the video stream; obtains the passenger body temperature through the thermal imaging device; The emotion recognition unit splices the facial expression features, intonation features, body posture features, and body temperature and inputs them into a multi-modal attention mechanism network (1-layer self-attention, 2-layer fully connected, 64 units in the hidden layer); it outputs the probability distributions of 5 types of emotional states (calm, anxious, tense, pleasant, fatigued), which are updated every 0.5 s; The interaction modeling unit inputs the passenger behavior data, environmental state data, and the current emotional state recognized by the emotion recognition unit into the behavior prediction unit together, and outputs the passenger-environment interaction behavior data within the next 30 s.

[0028] These complex model operation tasks are all efficiently executed on the AI computing power unit of the intelligent fusion host.

[0029] Table 1 is a comparison table of the accuracies between the present invention (multi-modal feature input: image + voice + thermal imaging + smell + dynamic trajectory, deep semantic fusion modeling, time series prediction) and traditional methods (single camera image + behavior rule threshold inference).

[0030] Table 1 Comparison Table of Passenger Interaction Behavior Prediction Accuracies

[0031] Through the situation awareness model, high-precision modeling of the passenger's mental state, physiological reactions, and environmental interaction behaviors can be achieved, which can capture the physiological and emotional changes of passengers in real time and accurately; at the same time, integrating the emotional state into behavior prediction can improve the system's understanding ability of passenger needs and behavior trends, providing high-precision and fine-grained dynamic information on passenger needs for subsequent local and global scheduling, and effectively supporting the coordination of multiple subsystems.

[0032] Preferably, the first data is the passenger flow trend, which is obtained by analyzing historical operation data and the current passenger distribution and using time series analysis to predict the passenger inflow and outflow in each carriage within the future time period; The second data is the passenger behavior pattern, which is obtained by identifying common passenger behavior patterns through cluster analysis; The third data is the change in passenger demand, which is obtained based on the emotional state and interaction behavior to predict the change in the passenger's demand for environmental facilities.

[0033] Specifically, for the first data: the emotional state output by the emotion recognition unit and the interaction behavior recognized by the interaction modeling unit are used as input factors, and the behavior prediction unit uses time series analysis methods (such as LSTM, GRU) combined with historical entry and exit data and the current passenger distribution characteristics to predict the passenger inflow and outflow in each carriage within the future time period, obtaining the first data; For the second data: the behavior prediction unit extracts typical behavior types based on historical interaction behaviors using cluster analysis (such as K-means, DBSCAN), obtaining the second data; Third data: By combining the current passenger emotional state identified by the emotion recognition unit and the interaction behavior of the passenger with the carriage environmental facilities analyzed by the interaction modeling unit, the future adjustment requirements of the passenger for the facilities, such as whether expecting to cool down, delaying door opening and closing, improving ventilation, etc., are predicted through an intention recognition model (such as Transformer) to obtain the third data.

[0034] Through the collaborative action of each functional unit in the context awareness model, the present invention outputs three types of passenger interaction data: passenger flow trend, behavior pattern, and demand change, realizing fine modeling and forward-looking prediction of the passenger state. Compared with the traditional static statistical method based only on a single sensor or rule threshold, this solution integrates passenger physiological characteristics, emotional state, and interaction behavior, and can dynamically identify the behavior intentions and service demands of passengers in different operation scenarios, not only improving the semantic richness and accuracy of the prediction, but also significantly enhancing the response forward-looking and humanization of the multi-subsystem control strategy. The above three types of data respectively provide an active optimization basis for subsystems such as door scheduling, air-conditioning ventilation adjustment, and lighting control, realizing the synchronous improvement of the system operation efficiency and passenger comfort. Table 2 is a comparison table of the prediction output of the present invention's solution and the conventional method.

[0035] Table 2 Comparison table of the prediction output of the present invention's solution and the conventional method

[0036] Preferably, the operation data and passenger interaction data are analyzed, and using the computing power of the intelligent fusion host, an anomaly detection model is adopted to identify emergencies and classify the emergency levels; The anomaly detection model includes: a multi-dimensional feature extraction unit, an anomaly detection unit, and an emergency level determination unit; The multi-dimensional feature extraction unit extracts statistical features, spatio-temporal features, and behavior pattern features based on the operation data and passenger interaction data; The anomaly detection unit calculates the anomaly scores of each carriage area through an anomaly detection algorithm for the statistical features, spatio-temporal features, and behavior pattern features to identify emergencies; The emergency level determination unit classifies the identified emergencies according to the anomaly scores, anomaly duration, and influence range.

[0037] Specifically, the multi-dimensional feature extraction unit extracts statistical features (mean, variance), spatio-temporal features (density gradient, flow vector), and behavior pattern features (behavior frequency, change rate) based on the operation data and passenger interaction data; The anomaly detection unit inputs multi-dimensional features into an autoencoder or an isolation forest model (running on the intelligent fusion host), determines emergencies (such as sudden increase in passenger flow, braking delay, emotional intensification, sudden environmental change, operation deviation, etc.) in the current carriage, and outputs an anomaly score. The emergency level determination unit classifies the emergencies as follows: Minor: Score 0.7 - 0.8, lasting less than 2 minutes, affecting a single carriage; Medium: Score 0.8 - 0.9, lasting 2 - 5 minutes, affecting 2 - 3 carriages; Severe: Score > 0.9, lasting more than 5 minutes, affecting the entire train.

[0038] Through a high-dimensional and model-based anomaly detection mechanism, it can detect instantaneous passenger aggregation and dispersion events in real time during peak travel hours, effectively identify states such as sudden passenger flow, emotional intensification, and environmental anomalies, and quantitatively grade their impact levels, providing a scientific basis for subsequent priority control in combination with the emergency level.

[0039] Preferably, a hierarchical management architecture including a carriage-level local management unit and a train-level global coordination unit is constructed, and the management logic of this architecture mainly runs on the intelligent fusion host. Among them, the function of the train-level global coordination unit is undertaken by the host core processing unit, and part of the logic of the carriage-level local management unit can also run on the host or be realized through the interaction between the host and carriage-level devices.

[0040] Based on the passenger interaction data and emergency level processed by the intelligent fusion host, the carriage-level local management unit dynamically adjusts the configuration parameters of the operation subsystems in each carriage in combination with the global scheduling strategy generated by the host. The carriage-level local management unit includes: Evaluates the passenger comfort and system operation efficiency of the current carriage according to the passenger interaction data and the emergency level; Combines the global scheduling strategy to determine the operation subsystems that need to be adjusted and their configuration parameters; Sends adjustment instructions to the corresponding operation subsystems to dynamically adjust the configuration parameters of the operation subsystems in each carriage; Monitors the adjusted operation data and passenger interaction data, evaluates the adjustment results, and feeds back the evaluation results to the train-level global coordination unit.

[0041] Specifically, the passenger comfort is calculated by non-linearly fusing the passenger interaction data and the emergency level through a non-linear fusion formula; the system operation efficiency is calculated from the subsystem power and the rated power; Based on the calculation results of passenger comfort and system operation efficiency, dynamically adjust the operation parameters of the air conditioning system, door system, ventilation system, lighting system, and braking system in this carriage; for example: when the passenger comfort level < 0.6 and the emergency level ≥ medium, adjust the air volume of the air conditioner by ±10%, and delay the door opening and closing by 20 ms; When receiving that the emergency level is higher than the set threshold (0.8 in this embodiment), the carriage-level local management unit preferentially executes the emergency response strategy, including but not limited to switching the air conditioner to the forced ventilation mode, delaying the closing of the door, or giving an in-car broadcast prompt.

[0042] After receiving the train-level global scheduling policy instruction, the carriage-level local management unit in this embodiment can autonomously issue adjustment instructions such as air volume of the air conditioner, door opening and closing sequence, ventilation mode, lighting brightness, and braking response within milliseconds by non-linearly integrating the passenger flow trend, behavior pattern, demand change, and emergency level in this carriage, making precise compensation for local congestion, uneven heating and cooling, and high-risk behaviors, thereby improving the overall operation efficiency and passenger satisfaction under high-density operation; at the same time, it real-time feeds back the adjustment effect and the latest passenger interaction data to the global coordination unit on the intelligent fusion host, enabling the global model to continuously fine-tune the weights and thresholds based on the feedback closest to the scene, ensuring the closed-loop collaborative optimization of "global - local - global".

[0043] Preferably, the train-level global coordination unit (its function is implemented on the intelligent fusion host) summarizes the operation data, passenger interaction data, and emergency level of the whole train, constructs a global collaborative model, and combines passenger comfort and system energy consumption to set priority regulation weights for the emergency impact area, generates a global scheduling policy, and issues it to the carriage-level local management unit of each carriage. The train-level global coordination unit includes: Collect the operation data, passenger interaction data, and emergency level from all carriages, and perform preprocessing and feature extraction to identify the passenger distribution, environmental status, and system operation status of the whole train; Construct a global collaborative model, comprehensively consider passenger comfort and system energy consumption, and use a multi-objective optimization algorithm to solve the optimal regulation strategy; Set priority regulation weights according to the emergency level and the affected range; Generate a global scheduling policy for each carriage based on the global collaborative model and the priority regulation weights; Issue the global scheduling policy to the carriage-level local management unit of each carriage.

[0044] Specifically, after formatting standardization, missing value filling, and noise filtering of the received operation data, passenger interaction data, and emergency event levels of all carriages, it enters the subsequent feature extraction stage to extract the passenger distribution status map of the whole train, the load distribution map of each system, and the risk event hot zone map; Construct a global collaborative model including the following elements: State variables: including passenger density, environmental quality score, mutation level, subsystem operation load, etc. of each carriage; Control variables: including adjustable parameters of each subsystem, such as air conditioning set temperature, lighting brightness, door delay time, braking sensitivity, etc.; Objective function: Set two main optimization objectives, including: maximizing passenger comfort (based on comprehensive environmental score and behavior matching degree) and minimizing system energy consumption (based on equipment energy consumption model and adjustment frequency); Constraint conditions: including the control parameter ranges of each subsystem, adjustment cycle limit, priority regulation rules, etc.

[0045] According to the emergency event levels and influence ranges uploaded by each carriage, construct a "sudden response priority map" to generate a global scheduling strategy, including: assigning higher regulation weights to the carriages within the emergency event influence area; limiting the adjustment response time in this area to not be higher than the set response time threshold; increasing the available adjustment resources of this carriage (such as ventilation power, lighting intensity, etc.); forming a "local priority, overall coordination" scheduling mechanism with local units.

[0046] Based on the setting of the global collaborative model and priority regulation weights, the intelligent fusion host calls multi-objective optimization algorithms (such as NSGA-II, particle swarm, multi-objective ant colony algorithm, etc.) to calculate the optimal scheduling strategy.

[0047] Table 3 is a performance comparison table between the system fusion management solution of the present invention and the traditional solution.

[0048] Table 3 Performance Comparison Table between the System Fusion Management Solution of the Present Invention and the Traditional Solution

[0049] The train-level global coordination unit in this embodiment aggregates the real-time operation data, interaction data, and event levels of all carriages, takes into account passenger comfort and system energy consumption with a multi-objective optimization model, and assigns priority control weights to high-level emergency areas to generate an optimal scheduling strategy for the whole train and send it to each local unit; when each carriage feeds back the latest evaluation results, the global coordination unit can dynamically update the optimization objectives and weights to achieve iterative upgrade of the strategy, and effectively improve the average passenger comfort and reduce the total energy consumption of the whole train under sudden emergencies and high-density operations with the high-performance computing support of the intelligent fusion host, so as to form seamless coordination at the macroscopic scheduling and microscopic execution levels.

[0050] The present invention constructs a data acquisition system based on multi-type sensing terminals, relies on a high-performance and highly reliable intelligent fusion host device platform, integrates context awareness modeling, abnormal event recognition, hierarchical control architecture and global optimization strategy generation mechanism, and proposes a multi-subsystem fusion management method for urban rail trains with real-time sensing, autonomous analysis and intelligent control capabilities. Compared with the traditional control mode of "independent operation and passive response" of each subsystem, this method realizes multi-source data fusion in the sensing dimension, realizes deep interaction modeling among passengers - environment - system in the decision-making dimension (supported by the powerful computing power of the intelligent fusion host), and realizes the unified coordination of local fast response and global optimal scheduling in the control dimension. Through the linkage adjustment of the carriage-level local management unit and the train-level global coordination unit (whose core logic runs on the intelligent fusion host), this method can actively predict potential risks and dynamically optimize the configuration parameters of each operating subsystem when facing complex scenarios such as sudden changes in passenger behavior and environmental fluctuations, significantly improving the control response speed, system energy consumption efficiency and overall passenger comfort experience under emergency conditions, and having good adaptability, scalability and intelligent level.

[0051] Embodiment 2 In Embodiment 1, the method proposed by the present invention successfully realizes intelligent adjustment of the operating states of each subsystem by real-time sensing of passenger flow and demand changes, thereby optimizing the passenger experience, reducing energy consumption, and improving the overall system efficiency. To further verify the effectiveness of the present invention, the intelligent fusion management of multiple subsystems during the operation of urban rail train B is also carried out in the embodiments of this application.

[0052] An intelligent fusion management system for multiple subsystems of an urban rail train, comprising: A data acquisition module, configured to acquire operation data through multi-type sensing terminals inside and outside each carriage of the train; the multi-type sensing terminals include: a thermal imaging device, a pressure detector, a video acquisition device, a gas detector, and a motion sensor; The operation data includes: passenger behavior data, environmental state data, and on-vehicle equipment data; The passenger behavior data includes: the number of passengers, distribution density, staying positions, moving paths, body temperature, and getting on and off frequencies; The environmental state data includes: carriage temperature, humidity, CO 2 concentration, odor concentration, and noise level; The on-vehicle equipment data includes: real-time operation data of each subsystem, vehicle position, and speed.

[0053] Preferably, a context awareness module, configured to construct a context awareness model of the interaction relationship between passengers and the environment according to the operation data, and predict passenger interaction data in future time periods, including first data, second data, and third data; The described situation awareness model includes: a physiological feature extraction unit, an emotion recognition unit, an interaction modeling unit, and a behavior prediction unit; The physiological feature extraction unit extracts the physiological features of passengers using face recognition and voice recognition technologies, including: expression features, intonation features, and body posture features; the emotion recognition unit constructs a multi-modal emotion recognition model based on the physiological features to identify the current emotional state of passengers; the interaction modeling unit constructs a passenger-environment situation awareness model based on passenger behavior data, environmental state data, and the emotional state to identify the interaction behaviors between passengers and in-car environmental facilities; the interaction behaviors include: the proximity, residence time, and usage frequency of passengers to doors, air-conditioning outlets, seats, and armrests; the behavior prediction unit predicts passenger interaction data in future time periods based on the emotional state and interaction behaviors.

[0054] Preferably, the first data is the passenger flow trend, obtained by analyzing historical operation data and current passenger distribution and predicting the passenger entry and exit situations of each carriage in future time periods using time series analysis; the second data is the passenger behavior pattern, obtained by identifying common passenger behavior patterns through clustering analysis; the third data is the change in passenger demand, obtained by predicting the change in passenger demand for environmental facilities based on the emotional state and interaction behaviors.

[0055] Preferably, an emergency detection module is used to analyze operation data and passenger interaction data, and an anomaly detection model is used to identify emergencies and classify the emergency levels; The anomaly detection model includes: a multi-dimensional feature extraction unit, an anomaly detection unit, and an emergency level determination unit; The multi-dimensional feature extraction unit extracts statistical features, spatio-temporal features, and behavior pattern features based on operation data and passenger interaction data; the anomaly detection unit calculates the anomaly scores of each carriage area through an anomaly detection algorithm using the statistical features, spatio-temporal features, and behavior pattern features to identify emergencies; the emergency level determination unit classifies the identified emergencies according to the anomaly scores, anomaly duration, and influence range.

[0056] Preferably, a local-global fusion management module is used to construct a hierarchical management architecture including a carriage-level local management unit and a train-level global coordination unit, refer to Figure 4 , where: The carriage-level local management unit dynamically adjusts the configuration parameters of the in-car operation subsystems based on passenger interaction data and emergency levels, combined with the global scheduling strategy; the carriage-level local management unit includes: Evaluate the passenger comfort level and system operation efficiency of the current carriage according to the passenger interaction data and the emergency level; combine the global scheduling strategy to determine the operating subsystems that need to be adjusted and their configuration parameters; send adjustment instructions to the corresponding operating subsystems to dynamically adjust the configuration parameters of the operating subsystems in each carriage; monitor the adjusted operation data and passenger interaction data, evaluate the adjustment results, and feedback the evaluation results to the train-level global coordination unit.

[0057] Preferably, the train-level global coordination unit summarizes the operation data, passenger interaction data, and emergency level of the entire train, constructs a global collaboration model, sets priority control weights for the areas affected by emergencies in combination with passenger comfort and reduced system energy consumption, generates a global scheduling strategy, and issues it to the carriage-level local management unit.

[0058] The train-level global coordination unit includes: Collect the operation data, passenger interaction data, and emergency level from all carriages, perform preprocessing and feature extraction, identify the passenger distribution, environmental status, and system operation status of the entire train; construct a global collaboration model, comprehensively consider passenger comfort and system energy consumption, and use a multi-objective optimization algorithm to solve the optimal control strategy; set priority control weights according to the emergency level and the affected range; generate a global scheduling strategy for each carriage based on the global collaboration model and the priority control weights; issue the global scheduling strategy to the carriage-level local management unit of each carriage.

[0059] Table 4 is the intelligent fusion management and control effect table based on the present invention.

[0060] Table 4 Intelligent Fusion Management and Control Effect Table

[0061] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent integration management of multiple subsystems of urban rail trains, characterized in that: include: Acquire operational data through multiple types of sensing terminals inside and outside each train compartment; Based on the operational data, a situational awareness model of the interaction between the passenger and the environment is constructed, and passenger interaction data in a future period is predicted, including first data, second data, and third data; Analyze operational data and passenger interaction data, use anomaly detection models to identify emergencies, and classify emergencies; Construct a hierarchical management architecture including a car-level local management unit and a train-level global coordination unit, where: The carriage-level local management unit dynamically adjusts the configuration parameters of the operating subsystems in each carriage based on the passenger interaction data and the emergency level in combination with the global scheduling strategy; The train-level global coordination unit summarizes the operational data, passenger interaction data and emergency levels of the entire train, builds a global coordination model, and combines passenger comfort and reducing system energy consumption to set priority control weights for emergency-affected areas, generates a global scheduling strategy and sends it to the carriage-level local management unit.

2. According to claim 1, a multi-subsystem intelligent integration management method for urban rail trains is characterized in that: The multi-type sensing terminal includes: a thermal imaging device, a pressure detector, a video acquisition device, a gas detector and a motion sensor; The operational data includes: passenger behavior data, environmental status data and vehicle-mounted equipment data; Passenger behavior data includes: number of passengers, distribution density, stop location, movement path, body temperature and frequency of boarding and alighting; Environmental status data include: cabin temperature, humidity, CO2 concentration, odor concentration and noise level; The on-board equipment data includes: real-time operation data of each subsystem, vehicle position and speed.

3. According to claim 1, a multi-subsystem intelligent integration management method for urban rail trains is characterized in that: The context perception model includes: a physiological feature extraction unit, an emotion recognition unit, an interaction modeling unit and a behavior prediction unit; The physiological feature extraction unit uses face recognition and voice recognition technology to extract the physiological features of the passenger, including: facial expression features, intonation features and body features; the emotion recognition unit constructs a multimodal emotion recognition model based on the physiological features to identify the passenger's current emotional state; the interaction modeling unit constructs a passenger-environment situational awareness model based on passenger behavior data, environmental state data and the emotional state to identify the interaction between the passenger and the environmental facilities in the car; the interaction behavior includes: the passenger's proximity to the door, air-conditioning outlet, seat and armrest, the residence time and the frequency of use; the behavior prediction unit predicts the passenger's interaction data in the future time period based on the emotional state and interaction behavior.

4. The method for intelligent integration management of multiple subsystems of urban rail trains according to claim 1 is characterized in that: The first data is the passenger flow trend, which is obtained by analyzing historical operation data and current passenger distribution and using time series analysis to predict the passenger entry and exit of each carriage in the future period; The second data is the passenger behavior pattern, which is obtained by identifying common passenger behavior patterns through cluster analysis; the third data is the change in passenger demand, which is obtained by predicting the change in passenger demand for environmental facilities based on emotional state and interactive behavior.

5. The method for intelligent integration management of multiple subsystems of urban rail trains according to claim 1 is characterized in that: The anomaly detection model includes: a multi-dimensional feature extraction unit, an anomaly detection unit and an emergency level determination unit; The multidimensional feature extraction unit extracts statistical features, spatiotemporal features and behavioral pattern features based on operational data and passenger interaction data; the anomaly detection unit calculates the anomaly score of each car area using the statistical features, spatiotemporal features and behavioral pattern features through an anomaly detection algorithm to identify emergencies; the emergency level determination unit classifies the identified emergencies according to the anomaly score, duration of the anomaly and scope of impact.

6. The method for intelligent integration management of multiple subsystems of urban rail trains according to claim 1 is characterized in that: The carriage-level local management unit comprises: According to the passenger interaction data and the emergency level, the passenger comfort and system operation efficiency of the current car are evaluated; in combination with the global scheduling strategy, the operating subsystems and their configuration parameters that need to be adjusted are determined; adjustment instructions are sent to the corresponding operating subsystems to dynamically adjust the configuration parameters of the operating subsystems in each car; the adjusted operating data and passenger interaction data are monitored, the adjustment results are evaluated, and the evaluation results are fed back to the train-level global coordination unit.

7. The method for intelligent integration management of multiple subsystems of urban rail trains according to claim 1 is characterized in that: The train-level global coordination unit includes: collecting operational data, passenger interaction data and emergency levels from all carriages, and performing preprocessing and feature extraction to identify the passenger distribution, environmental status and system operation status of the entire train; building a global coordination model, comprehensively considering passenger comfort and system energy consumption, and using a multi-objective optimization algorithm to solve and build the global coordination model; setting priority control weights according to the emergency level and impact range; generating a global scheduling strategy for each carriage based on the global coordination model and priority control weights; and sending the global scheduling strategy to the carriage-level local management unit of each carriage.

8. An intelligent fusion management system for multi-subsystems of urban rail trains, characterized in that: include: A data acquisition module, used to acquire operational data through multiple types of sensing terminals inside and outside each train compartment; A situational awareness module, used to construct a situational awareness model of the interaction relationship between passengers and the environment based on the operation data, and predict passenger interaction data in a future period, including first data, second data and third data; The emergency detection module is used to analyze operational data and passenger interaction data, identify emergencies using anomaly detection models, and classify emergency events; The local-global fusion management module is used to build a hierarchical management architecture including a car-level local management unit and a train-level global coordination unit, where: The carriage-level local management unit dynamically adjusts the configuration parameters of the operating subsystems in each carriage based on the passenger interaction data and the emergency level in combination with the global scheduling strategy; The train-level global coordination unit summarizes the operational data, passenger interaction data and emergency levels of the entire train and constructs a global coordination model. It also sets priority control weights for emergency-affected areas in combination with passenger comfort and reduction of system energy consumption, generates a global scheduling strategy and sends it to the carriage-level local management unit.

Images