Information dynamic management and control service system for traffic station operation

Abstract

The application discloses an information dynamic management and control service system for traffic station operation, which comprises information collection, data processing, dynamic scheduling, precise interaction, execution feedback modules and terminals, storage modules, and each module cooperatively realizes whole-process management and control; through collecting multiple types of dynamic information of the station and standardizing preprocessing, multi-dimensional fusion analysis is carried out to obtain operation evaluation, abnormal early warning and scheduling demand; scheduling instructions are generated according to operation rules and are sent to execution terminals, and personalized interaction information is pushed according to interaction objects; execution effect and interaction feedback data are collected, and after comparative analysis, algorithm parameters and scheduling strategies are continuously optimized. The application constructs a whole-process closed-loop intelligent management and control system, improves real-time performance and reliability of management and control, improves resource utilization, reduces labor cost, improves user travel experience, and is suitable for management and control of various traffic stations.

Description

Technical Field

[0001] This invention relates to the field of transportation hub operation and management technology, specifically to an information dynamic management and control service system and method for transportation hub operation. Background Technology

[0002] As core hubs for the gathering and dispersal of people and vehicles, transportation hubs directly impact the capacity of the transportation network and the user travel experience through their operational efficiency. During operation, real-time management and control of multi-dimensional information, including passenger flow, vehicle flow, facility status, and service resources, is essential. Currently, existing technologies for managing and controlling operational information at transportation hubs primarily suffer from the following problems:

[0003] 1) Fragmented information collection and poor data interoperability: In the existing management and control model, functions such as passenger flow statistics, vehicle dispatching, and facility monitoring rely on independent hardware equipment and software systems. Different systems use different data formats and communication protocols, lacking a unified information collection interface, which makes it impossible to achieve collaborative analysis of multi-dimensional information.

[0004] 2) Delayed control response and lack of dynamic adaptive capability: Existing technologies mostly adopt fixed threshold trigger control strategies, without dynamically adjusting to the real-time changes in station operations (such as surges in passenger flow during peak hours, sudden equipment failures, etc.). There is a significant delay in the generation and issuance of control instructions, making it impossible to respond to emergencies in the operation process in a timely manner, which can easily lead to operational chaos.

[0005] 3) Low level of intelligence in resource scheduling and reliance on manual intervention: The scheduling of existing service resources relies heavily on human experience and judgment, without combining real-time operational information for accurate matching, resulting in low resource utilization, increased manual management costs, and a high risk of scheduling errors.

[0006] 4) Inaccurate information interaction and poor user experience: The information pushed to users by the existing stations is mostly general content, which makes it impossible for users to obtain accurate guidance information and ticketing information in a timely manner, increasing the uncertainty of users' travel.

[0007] While some existing technologies attempt to address the aforementioned issues, none have established a complete management and control system encompassing "information collection, analysis and processing, dynamic scheduling, precise interaction, and closed-loop optimization." Furthermore, they still exhibit significant shortcomings in data processing efficiency, the adaptability of management and control strategies, and the accuracy of resource scheduling, failing to meet the high-density and highly dynamic operational management and control needs of large transportation hubs. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention discloses an information dynamic management and control service system for transportation hub operations, in order to solve the problems mentioned in the background section.

[0009] To achieve the above objectives, the present invention provides the following technical solution: an information dynamic management and control service system for transportation hub operations, comprising:

[0010] Information acquisition module: Composed of multiple types of acquisition devices, with built-in isolated forest algorithm and hash deduplication algorithm, used to collect various dynamic information in the operation of transportation stations in real time, and to perform standardized preprocessing on the collected dynamic information, including data cleaning, format conversion and data deduplication, and output the collected data in a unified format.

[0011] Data processing module: Communicates with the information acquisition module and has built-in multi-dimensional fusion analysis algorithms (including PCA algorithm and weighted fusion algorithm), BP neural network anomaly risk prediction model, LSTM operational load prediction model, and reinforcement learning model. It is used to receive the collected data output by the information acquisition module.

[0012] Dynamic scheduling module: It communicates with the data processing module and has built-in improved genetic algorithm and traffic station operation rules. It is used to receive the analysis results and prediction results output by the data processing module, optimize the resource scheduling scheme through the improved genetic algorithm, generate targeted dynamic control and scheduling instructions, and send the scheduling instructions to the corresponding execution terminals.

[0013] Precise Interaction Module: It communicates with the data processing module and the dynamic scheduling module respectively, and has a built-in collaborative filtering recommendation algorithm. It is used to receive the analysis results and prediction results output by the data processing module and the scheduling instructions output by the dynamic scheduling module.

[0014] Execution feedback module: It communicates with the dynamic scheduling module and the precise interaction module, and is used to collect the execution effect data of scheduling instructions and the feedback data of each interactive object in real time. It transmits the execution effect data and feedback data to the data processing module to provide data support for closed-loop optimization.

[0015] Storage module: Communicates with the data processing module and is used to store collected data, analysis results, prediction results, scheduling instructions, execution effect data, feedback data, historical data of station operation, and parameter data of various algorithms and models, providing data support for data processing, analysis and prediction, and closed-loop optimization;

[0016] Execution terminal: Communicatively connected to the dynamic scheduling module, including personnel scheduling terminal, vehicle scheduling terminal and equipment control terminal, which correspond to the personnel management, vehicle management and equipment management positions in the station operation, respectively. It is used to receive and execute dynamic control and scheduling instructions, and to feed back the instruction execution status to the execution feedback module.

[0017] Interactive Terminal: Communicates with the precision interaction module and includes passenger service terminal, operator handheld terminal, and management monitoring terminal, corresponding to three types of interaction objects: passengers, operators, and managers, respectively. It is used to receive interactive information pushed by the precision interaction module and transmit the feedback data of the interaction object to the execution feedback module.

[0018] This invention also provides a method for dynamic information management and control services for transportation hub operations, comprising the following steps:

[0019] S1. Information Collection: By deploying various types of collection devices in different areas of the transportation hub, various dynamic information during the operation of the hub is collected in real time. The dynamic information includes personnel information, vehicle information, equipment operation information, environmental information and operation service information. The collected dynamic information is standardized and preprocessed to obtain collected data in a unified format.

[0020] S2. Analysis and processing: The collected data obtained in step S1 is transmitted to the data processing module. A multi-dimensional fusion analysis algorithm is used to analyze the collected data in real time to obtain the site operation status assessment results, abnormal early warning information and resource scheduling requirements.

[0021] S3. Dynamic scheduling: Based on the analysis results obtained in step S2, and combined with the operation rules and resource allocation of transportation hubs, generate targeted dynamic control and scheduling instructions, and send the scheduling instructions to the corresponding execution terminals.

[0022] S4. Precise Interaction: The analysis results of step S2 and the scheduling instructions of step S3 are classified and processed according to the type of the interaction object, and targeted interaction information is generated and transmitted to each interaction object through the corresponding interaction terminal.

[0023] S5. Closed-loop optimization: Real-time collection of execution effect data of scheduling instructions in step S3 and feedback data of each interactive object in step S4, transmission of execution effect data and feedback data to data processing module, comparison and analysis with analysis results of step S2, and continuous optimization of parameters and dynamic scheduling strategy of multi-dimensional fusion analysis algorithm based on comparison results.

[0024] Preferably, in step S1, the standardized preprocessing is used to solve the problems of low data processing efficiency caused by messy collected data, large outlier interference, and inconsistent formats in the prior art. Specifically, it includes three sub-steps: data cleaning, data format conversion, and data deduplication. In addition, the isolated forest algorithm is introduced in the data cleaning process to achieve accurate identification and removal of outlier data, as detailed below:

[0025] S11. Data Cleaning: The Isolation Forest algorithm is used to construct a forest model consisting of K isolated trees. Anomaly detection is performed on the collected raw dynamic information to remove outliers, noisy data, and missing values. In the Isolation Forest algorithm, outlier data, due to its features deviating from the normal data distribution, is quickly isolated. The anomaly detection threshold is calculated using the following formula: Where score(x) is the anomaly score of the data to be detected x, when When the data is determined to be abnormal, E(h(x)) is the average path length of the data to be detected x in K isolated trees; c(n) is the average path length of normal data in isolated trees when the sample size is n, which is used as a normalization factor; n is the total number of samples in this data cleaning, and K is the number of isolated trees, with a value range of 80-120.

[0026] S12. Data Format Conversion: Convert heterogeneous data output from different acquisition devices, including image data, numerical data, and text data, into a preset unified JSON format to ensure data interoperability and provide unified data support for subsequent multi-dimensional fusion analysis.

[0027] S13. Data Deduplication: A hash deduplication algorithm is used to perform hash calculations on the converted uniform format data. By comparing the hash values, duplicate data is deleted, including monitoring data of the same passenger, the same vehicle, and the same equipment. This ensures the uniqueness and validity of the collected data and reduces the computing power consumption of subsequent data processing.

[0028] Preferably, in step S2, the multi-dimensional fusion analysis algorithm is used to solve the problems of low data processing efficiency, inaccurate operational status assessment, and delayed anomaly warning in the prior art. Specifically, it includes four sub-steps: feature extraction, data fusion, anomaly identification, and demand matching. It integrates principal component analysis (PCA) algorithm and weighted fusion algorithm, as detailed below:

[0029] S21. Feature Extraction: Principal Component Analysis (PCA) algorithm is used to perform feature dimensionality reduction on the preprocessed data from step S1, eliminating redundant features and extracting the core feature parameters of each dimension of the data, thereby reducing data processing pressure and improving data processing efficiency; the core feature parameters include peak traffic flow. Vehicle turnover efficiency Equipment failure rate Environmental parameter thresholds and service response time ;

[0030] The feature dimensionality reduction process of the PCA algorithm is described by the formula... The implementation is as follows: Y is the core feature matrix after dimensionality reduction, X is the preprocessed acquisition data matrix, and W is the feature vector matrix, which is composed of the feature vectors corresponding to the eigenvalues ​​of the covariance matrix of the acquisition data matrix X.

[0031] S22. Data Fusion: A weighted fusion algorithm is used to fuse the core feature parameters to obtain comprehensive feature parameters, which are used to comprehensively evaluate the station's operational status. The formula is as follows:

[0032] ;

[0033] Where F is the comprehensive feature parameter, reflecting the overall operational status of the station; m is the number of dimensions of the core feature parameter, and m=5 in this step; Let be the weight coefficient of the i-th core feature parameter. This coefficient is dynamically allocated based on the degree of influence of each core feature parameter on the operation and management of the facility. The higher the degree of influence, the larger the weight coefficient, and it satisfies the following condition: ; This is the normalized value of the i-th core feature parameter, eliminating the influence of different dimensions on the fusion result;

[0034] S23. Anomaly Detection: Compare the comprehensive feature parameter F with the preset operational threshold range. A comparison is made, and an anomaly risk prediction model based on a BP neural network is introduced to output the probability P of an anomaly risk occurring; when and When an operation is deemed abnormal, an abnormality warning is generated, which includes the abnormality type, abnormality location, abnormality level, and probability of occurrence.

[0035] The BP neural network anomaly risk prediction model includes an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is consistent with the dimension of the core feature parameters, i.e., m=5. The number of nodes in the hidden layer is determined by cross-validation. The output layer outputs the probability P of anomaly occurrence. The model's weights are updated using the following formula:

[0036] ;

[0037] in, The connection weight between the i-th node in the input layer and the j-th node in the hidden layer at the (k+1)-th iteration; The connection weights at the k-th iteration; The learning rate, with a value ranging from 0.01 to 0.05; This represents the error term of the j-th node in the hidden layer. The input value of the i-th node in the input layer is the normalized value of the core feature parameters. ;

[0038] S24. Demand Matching: Based on the comprehensive feature parameter F and anomaly warning information, combined with the operational status prediction results for the next 1-2 hours output by the Long Short-Term Memory (LSTM) network-based operational load prediction model, the corresponding resource scheduling requirements are matched. The LSTM network model is used to solve the problems of inability to predict operational status and scheduling lag in existing technologies, capturing the time-series characteristics of operational data. Its prediction output formula is as follows:

[0039] ;

[0040] in, This is the output value of the LSTM model at time t, which is the predicted value of the operating load at a future time. The cell state at time t is used to store long-term temporal features; t represents the output value of the output gate at time t; tanh is the activation function used to normalize the cell state to the interval -1 to 1. This is a dot product operation; This is the sigmoid activation function, used to control the output strength of the output gate.

[0041] Preferably, in step S22, the weight coefficients of each core feature parameter are... The Analytic Hierarchy Process (AHP) was used to determine the appropriate method to address the inaccurate operational status assessment caused by unreasonable weight allocation in existing technologies. The specific determination process is as follows:

[0042] S221. Constructing the Judgment Matrix: Based on the impact of each core feature parameter on the operation and management of the station, construct a 5×5 judgment matrix. ,in The value represents the importance of the i-th core feature parameter relative to the j-th core feature parameter. It is an integer from 1 to 9 and its reciprocal. 1 means that the two are equally important, and 9 means that the i-th parameter is much more important than the j-th parameter.

[0043] S222. Consistency test: Calculate the largest eigenvalue of the judgment matrix A. Through consistency indicators The random consistency index RI, where RI = 1.12 when m = 5, and the consistency ratio. Perform a consistency check; if CR < 0.1, the judgment matrix meets the consistency requirement; otherwise, adjust the judgment matrix.

[0044] in, 5 represents the largest eigenvalue of the judgment matrix A; 5 represents the number of dimensions of the core feature parameters; CI is the consistency index, reflecting the degree of deviation of the judgment matrix; RI is the random consistency index, which is a preset empirical value; CR is the consistency ratio, used to judge whether the consistency of the judgment matrix is ​​acceptable.

[0045] S223. Determine the weighting coefficients: Normalize the judgment matrix A that meets the consistency requirements to obtain the weighting coefficients of each core feature parameter. And satisfy .

[0046] Preferably, in step S3, the dynamic scheduling process introduces an improved genetic algorithm to optimize the resource scheduling scheme, which is used to solve the problems of low accuracy of resource scheduling and poor adaptability of control strategies in the existing technology. It can find the optimal scheduling scheme in a short time and adapt to the high-density and high-dynamic operation needs of large transportation hubs, as detailed below:

[0047] S31. Encoding: The resource scheduling scheme, including personnel allocation, vehicle scheduling, equipment management and resource allocation, is encoded in binary to form an initial population. Each individual corresponds to a scheduling scheme. The population size is N, where N ranges from 40 to 60.

[0048] S32. Fitness Function: A fitness function is constructed with the objectives of maximizing resource utilization, maximizing scheduling response speed, and minimizing management costs.

[0049] ;

[0050] Where fit(x) is the fitness value of individual x, i.e., the scheduling scheme. The higher the fitness value, the better the scheduling scheme. Let be the weighting coefficient, satisfying These correspond to the importance of resource utilization, scheduling response speed, and management and control costs, respectively; U(x) represents the resource utilization of scheduling scheme x. The maximum resource utilization rate is set to 100%; T(x) is the response time of scheduling scheme x. Let C(x) be the maximum allowable value for scheduling response time; C(x) is the management cost of scheduling scheme x. To minimize control costs;

[0051] S33. Selection: The roulette wheel selection method combined with the elite retention strategy is used to select individuals with higher fitness values ​​from the initial population to enter the next generation of the population. The elite retention strategy is used to retain the best individuals in each generation of the population to avoid losing the optimal scheduling scheme.

[0052] S34. Crossover and Mutation: The crossover operation uses a single-point crossover method, with a crossover probability... The value range is 0.6-0.8; the mutation operation uses a random mutation method, and the mutation probability is... The value ranges from 0.01 to 0.05; crossover and mutation are used to increase population diversity and prevent the algorithm from getting trapped in local optima.

[0053] S35. Termination: When the fitness value of the population tends to stabilize, that is, the change in fitness value is less than 0.01 over 10 consecutive generations; or when the preset number of iterations K is reached, where K ranges from 80 to 120, the iteration stops and the optimal scheduling scheme is output.

[0054] S36. Instruction Generation and Issuance: Based on the optimal scheduling scheme, generate targeted dynamic control and scheduling instructions. The scheduling instructions include instruction type, execution object, execution content and execution time limit, and are issued to the corresponding execution terminal through the wireless communication network to ensure that the scheduling instructions are delivered quickly and accurately.

[0055] Preferably, in step S4, the interaction process incorporates a collaborative filtering recommendation algorithm to address the problems of poor targeting of interaction information and inability to meet the needs of different interaction objects in the prior art, thereby achieving personalized and precise interaction, as detailed below:

[0056] S41. User Profile Construction: Based on the historical behavior data of the interaction objects, construct user profiles for different interaction objects; the interaction objects include passengers, operators and managers, and the historical behavior data includes passengers' query records and travel preferences, operators' job positions and work habits, and managers' focus and operation records.

[0057] S42. Similarity Calculation: The cosine similarity algorithm is used to calculate the similarity between different interactive objects, or the similarity between an interactive object and interactive information. The formula is:

[0058] ;

[0059] in, The similarity value ranges from 0 to 1. The closer the similarity value is to 1, the higher the similarity between the two. , These are the user profile vector of the interactive object and the feature vector of the interactive information, respectively. For vectors and The dot product; , They are vectors , The modulus length;

[0060] S43. Interactive Information Generation and Push: Based on user profiles and similarity calculation results, generate targeted interactive information for different interactive objects and transmit it through the corresponding interactive terminals; push transfer guidance, abnormality alerts, and rest area recommendations to passengers that match their travel preferences; push dispatch instructions, operational status of responsible areas, and work tasks to operations personnel that match their job positions; and push operational status summaries, abnormality warning details, and dispatch effect analysis of the areas they are interested in to managers.

[0061] Preferably, in step S5, the closed-loop optimization process introduces a reinforcement learning model to address the problems of lack of a closed-loop optimization mechanism, poor adaptability of control strategies, and inability to continuously improve control performance in existing technologies, thereby achieving dynamic optimization of various algorithm parameters and scheduling strategies, as detailed below:

[0062] S51. Feedback Data Collection: Real-time collection of execution effect data of scheduling instructions in step S3 and feedback data of each interactive object in step S4; the execution effect data includes the execution completion rate of scheduling instructions, execution time, resource utilization rate and operational efficiency improvement value; the feedback data includes the satisfaction rating and suggestions of each interactive object;

[0063] S52. Comparative Analysis: Compare the execution effect data and feedback data with the operational status evaluation results and resource scheduling requirements of step S2, analyze the rationality of scheduling instructions, the accuracy of interactive information and the accuracy of multi-dimensional fusion analysis algorithms, and determine the optimization direction.

[0064] S53, Parameter and Strategy Optimization: Introduce a reinforcement learning model to continuously optimize the parameters and dynamic scheduling strategy of the multi-dimensional fusion analysis algorithm;

[0065] The state update formula for the Q-learning algorithm is:

[0066] ;

[0067] Where Q(s, a) is the action value function for performing action a in state s; is the learning rate, ranging from 0.1 to 0.3; r is the reward value, determined based on the comparative analysis results, with a positive reward for improved control effectiveness and a negative reward for decreased control effectiveness; is the discount factor, ranging from 0.8 to 0.95, used to measure the importance of future rewards; s is the current system state, which consists of execution effect data, feedback data, and operational status; a is the currently executed optimization action, i.e., algorithm parameter adjustment and scheduling strategy adjustment; s' is the new system state after executing action a; a' is the optimal action under the new state s'. The maximum action value under the new state s';

[0068] S54. Optimization and Iteration: Synchronize the optimized algorithm parameters and scheduling strategies to the corresponding modules, repeat steps S1-S5, and realize a complete closed-loop iteration of information collection, analysis and processing, dynamic scheduling, precise interaction and closed-loop optimization to continuously improve control performance and adapt to the high-density and high-dynamic operation and control needs of large transportation stations.

[0069] Preferably, in step S2, the station operation status assessment results include the operation load level, including low load, normal load, high load, and overload; the operation status of each area includes orderly, basically orderly, congested, and abnormal; the abnormal warning information includes the abnormal type, namely, personnel congestion, vehicle delays, equipment failure, and environmental abnormality; the abnormal location and abnormal level, including general warning, severe warning, and emergency warning; the probability of occurrence and handling suggestions; the resource scheduling requirements include the resource type, namely, manpower, vehicles, equipment and site, resource quantity, scheduling priority, and scheduling time limit.

[0070] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0071] 1. This invention enables integrated management and control of transportation station operation information, breaking through the technical bottlenecks of fragmented information collection and poor data interoperability. It achieves unified format conversion of multi-type and heterogeneous collected data through standardized preprocessing, and completes collaborative analysis of multi-source information by combining multi-dimensional fusion analysis algorithms, making the information support for station operation management and control more comprehensive and efficient.

[0072] 2. This invention enhances the dynamic adaptive capability of transportation hub operation management and control, abandons the traditional fixed threshold management and control mode, and combines abnormal risk prediction and operational load prediction to achieve real-time perception and advance judgment of the hub's operational status. Through an improved genetic algorithm, it completes the intelligent optimization of resource scheduling schemes, which can quickly respond to sudden operational anomalies and dynamic load changes at the hub, effectively avoid operational chaos, and greatly improve the timeliness and accuracy of management and control.

[0073] 3. This invention constructs a full-process operation and control system encompassing information collection, analysis and processing, dynamic scheduling, precise interaction, and closed-loop optimization. It achieves continuous iterative optimization of algorithm parameters and scheduling strategies through reinforcement learning models, and realizes differentiated and precise information interaction based on user profiles and collaborative filtering algorithms. This not only improves the utilization rate of station resources and reduces the cost of manual management, but also meets the actual needs of different interaction objects, significantly improving the travel experience of passengers and enhancing the work efficiency of operation and management personnel. Attached Figure Description

[0074] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0075] In the attached diagram:

[0076] Figure 1 This is a simplified framework diagram of an information dynamic management and control service system for transportation station operation according to the present invention;

[0077] Figure 2This is a flowchart of an information dynamic management and control service method for transportation station operation according to the present invention. Detailed Implementation

[0078] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0079] Example: This example uses a large hub railway station as an application scenario. Existing management and control methods have problems such as incomplete management and control system, data disorder, scheduling lag, and inaccurate interaction. The technical solution of the present invention can effectively solve the above problems. The following describes in detail the information dynamic management and control service method and system for transportation station operation of the present invention.

[0080] like Figure 1 As shown in the figure, the information dynamic management and control service system for transportation station operation in this embodiment is as follows: Information collection module: composed of multiple types of collection devices, with built-in isolated forest algorithm and hash deduplication algorithm, used to collect various dynamic information in the transportation station operation process in real time, and to perform standardized preprocessing on the collected dynamic information, including data cleaning, format conversion and data deduplication, and output collection data in a unified format; wherein, the multiple types of collection devices include high-definition cameras, infrared sensors, traffic flow statistics instruments, vehicle recognition devices, equipment status sensors, environmental monitoring sensors and service terminals; high-definition cameras, infrared sensors and traffic flow statistics instruments are used to collect personnel information; vehicle recognition devices are used to collect vehicle information; equipment status sensors are used to collect equipment operation information; environmental monitoring sensors are used to collect environmental information; service terminals are used to collect operation service information; all collection devices are distributed in the passenger waiting area, ticket gate, transfer channel, parking lot, equipment room and surrounding roads of the transportation station to achieve information collection without blind spots throughout the station.

[0081] Data Processing Module: Communicates with the information acquisition module and incorporates multi-dimensional fusion analysis algorithms (including PCA algorithm and weighted fusion algorithm), a BP neural network anomaly risk prediction model, an LSTM operational load prediction model, and a reinforcement learning model. It receives the collected data from the information acquisition module, performs real-time analysis and prediction using the aforementioned algorithms and models, and outputs station operation status assessment results, anomaly warning information, resource scheduling requirements, and operational status prediction results. It also receives execution effect data and feedback data from the execution feedback module and continuously optimizes the parameters and dynamic scheduling strategies of the multi-dimensional fusion analysis algorithms using the reinforcement learning model.

[0082] Dynamic scheduling module: It communicates with the data processing module and has built-in improved genetic algorithm and traffic station operation rules. It is used to receive the analysis results and prediction results output by the data processing module, optimize the resource scheduling scheme through the improved genetic algorithm, generate targeted dynamic control and scheduling instructions, and send the scheduling instructions to the corresponding execution terminals.

[0083] Precise Interaction Module: It communicates with the data processing module and the dynamic scheduling module respectively. It has a built-in collaborative filtering recommendation algorithm and is used to receive the analysis results and prediction results output by the data processing module and the scheduling instructions output by the dynamic scheduling module. Based on the type of interaction object and user profile, it generates personalized and targeted interaction information through the collaborative filtering recommendation algorithm and transmits it to each interaction object through the corresponding interaction terminal.

[0084] Execution Feedback Module: Communicates with the Dynamic Scheduling Module and the Precise Interaction Module to collect real-time execution effect data of scheduling instructions and feedback data of each interactive object. It transmits the execution effect data and feedback data to the data processing module to provide data support for closed-loop optimization.

[0085] Storage module: Communicates with the data processing module and is used to store collected data, analysis results, prediction results, scheduling instructions, execution effect data, feedback data, historical data of station operation, and parameter data of various algorithms and models, providing data support for data processing, analysis and prediction, and closed-loop optimization;

[0086] Execution terminal: Communicatively connected to the dynamic scheduling module, including personnel scheduling terminal, vehicle scheduling terminal and equipment control terminal, which correspond to the personnel management, vehicle management and equipment management positions in the station operation, respectively. It is used to receive and execute dynamic control and scheduling instructions, and to feed back the instruction execution status to the execution feedback module.

[0087] Interactive Terminal: Communicates with the precision interaction module and includes passenger service terminal, operator handheld terminal, and management monitoring terminal, corresponding to three types of interaction objects: passengers, operators, and managers, respectively. It is used to receive interactive information pushed by the precision interaction module and transmit the feedback data of the interaction object to the execution feedback module.

[0088] like Figure 2 As shown in the figure, this embodiment also provides a method for dynamic information management and control services for transportation hub operations. The specific implementation process of each step is as follows:

[0089] Step S1: Information Collection. Each collection device collects various dynamic information during the station's operation in real time. The information collection module performs standardized preprocessing on the raw collected data. The specific sub-steps are as follows:

[0090] S11. Data Cleaning: Activate the Isolation Forest algorithm (K=100 isolated trees) to perform anomaly detection on the original collected data using the formula... Anomaly scores are calculated, where n=10000, representing the total sample size for this batch of data cleaning; c(n)=8.9 is the normalization factor when the sample size n=10000; for example, if the passenger flow data collected by the camera at the departure level ticket gate is 800 people / 10 minutes, which is much higher than the normal data for the same period (300-500 people / 10 minutes), the calculated score(x)=0.78≥0.7 is determined to be abnormal data and is removed; at the same time, the nearest neighbor interpolation method is used to supplement missing data, such as 5 minutes of missing data due to a malfunction of a certain flow meter, to ensure data integrity.

[0091] S12. Data format conversion: Convert heterogeneous data output from different acquisition devices into a unified JSON format.

[0092] S13. Data Deduplication: The hash deduplication algorithm is started to perform hash calculation on the converted JSON format data and compare the hash values ​​to delete duplicate data. When the personnel flow data collected by the high-definition camera and the infrared sensor at the same time in the same area have the same hash value, the duplicate entries are deleted. Finally, the collected data with a unified format and no errors is obtained and transmitted to the data processing module.

[0093] Step S2: Analysis and Processing. The data processing module receives the collected data after preprocessing in Step S1, and starts the multi-dimensional fusion analysis algorithm, BP neural network anomaly risk prediction model, and LSTM operational load prediction model to perform real-time analysis and prediction. The specific sub-steps are as follows:

[0094] S21. Feature Extraction: The PCA algorithm is started, and feature dimensionality reduction is performed using the formula Y=XW. The preprocessed collected data matrix X is 10000×20 (10000 samples, 20 original features). After dimensionality reduction by the PCA algorithm, the core feature matrix Y is 10000×5. The real-time values ​​of 5 core feature parameters are extracted, including f1 (peak traffic), f2 (vehicle turnover efficiency), f3 (equipment operation failure rate), f4 (environmental parameter threshold), and f5 (service response time), and normalization is performed.

[0095] S22. Data Fusion: A weighted fusion algorithm is used, through the formula... Calculate comprehensive characteristic parameters; preset operational threshold range Preliminary assessment indicates that the current station operation status is "basically orderly".

[0096] S23. Anomaly Detection: Activate the BP neural network anomaly risk prediction model, input the normalized values ​​of the five core feature parameters into the model, and update the model weights using the following formula:

[0097] ;

[0098] Output the probability of an anomaly occurring, P; when an equipment anomaly is detected, generate an anomaly warning message: anomaly type = equipment failure, anomaly location = elevator in the waiting hall, anomaly level = severe warning, high probability of occurrence, handling suggestion = immediately arrange maintenance personnel to handle the situation, and set up warning signs.

[0099] S24. Demand Matching: Start the LSTM operational load forecasting model, using historical operational data from the past 3 months plus real-time collected data as input;

[0100] Through formula Output the operational status prediction results for the next hour. The data processing module synchronously transmits the operational status assessment results, anomaly warning information, resource scheduling requirements and prediction results to the dynamic scheduling module and the precise interaction module.

[0101] Step S3: Dynamic scheduling. The dynamic scheduling module receives the analysis and prediction results from step S2, and, in conjunction with the station operation rules, initiates an improved genetic algorithm to optimize the resource scheduling scheme. The specific sub-steps are as follows:

[0102] S31. Encoding: The resource scheduling scheme, namely personnel allocation, equipment management and resource allocation, is encoded in binary to form an initial population, with each individual corresponding to a scheduling scheme;

[0103] S32. Fitness Function: The fitness value is calculated using the following formula:

[0104] ;

[0105] Among them α=0.4, β=0.3, γ=0.3, U max =100%, T max =10 minutes, C min =500 yuan; Suppose that the resource utilization rate of a certain scheduling scheme x is U(x)=90%, the response time is T(x)=6 minutes, and the control cost is C(x)=600 yuan, substitute into the calculation:

[0106] f it(x)=0.4×90% / 100%+0.3×(10-6) / 10+0.3×500 / 600

[0107] =0.36+0.12+0.25=0.73.

[0108] S33. Selection: Using a roulette wheel selection method combined with an elite retention strategy, individuals with a fitness value ≥ 0.7 are selected to enter the next generation of the population. The best individual in each generation, fit(x) = 0.85, is retained to avoid losing the optimal scheduling scheme.

[0109] S34. Crossover and Mutation: Using the single-point crossover method p c =0.7 and random mutation method p m =0.03, crossover and mutation operations are performed on the population to increase population diversity and avoid the algorithm getting trapped in local optima.

[0110] S35. Termination: After 100 iterations, the population fitness value tends to stabilize, and the optimal scheduling scheme fit(x) = 0.85 is output.

[0111] S36. Command Generation and Issuance: Based on the optimal scheduling scheme, generate targeted dynamic management and scheduling commands, specifying the command type, execution target, execution content, and execution time limit, and issue them to the corresponding execution terminals via the 5G network. Specific scheduling commands are as follows:

[0112] 1) Personnel dispatch instructions (issued to maintenance personnel A and B's handheld terminals): "Maintenance personnel A and B are requested to immediately go to the elevator equipment in the waiting hall to handle the fault. Execution deadline: arrive at the station before the specified time and report the execution result after the task is completed."

[0113] 2) Personnel dispatch instructions (issued to operator C's handheld terminal): "Operator C, please go to the area around the elevators in the waiting hall, guide passengers to use other escalators, maintain order on site, and set an execution time limit."

[0114] 3) Equipment control instructions (issued to the equipment control terminal): "Immediately lock the elevator in a certain location in the waiting hall, stop its operation, display a fault message, and unlock it after the fault is resolved."

[0115] 4) Advance dispatch instructions (issued to operators Ding and Wu's handheld terminals): "Operators Ding and Wu, please set the execution time limit and go to the corresponding area of ​​the waiting hall to maintain order and deal with the increase in passenger flow."

[0116] After the scheduling instruction is issued, the dynamic scheduling module receives the receipt confirmation information from the execution terminal in real time. All instructions are delivered within 0.5 seconds to ensure timely scheduling.

[0117] Step S4: Precise Interaction. The precise interaction module receives the analysis results and prediction results from step S2, as well as the scheduling instructions from step S3, and starts the collaborative filtering recommendation algorithm to achieve personalized and precise interaction. The specific sub-steps are as follows:

[0118] S41. User Profile Construction: Based on the historical behavior data of the interacting objects, construct three types of user profiles:

[0119] 1) Passenger profile: Passenger Zhang (historical query records: multiple queries for waiting hall ticket gates and transfer routes; travel preference: quick transfers).

[0120] 2) Operations personnel profile: Operations personnel C (Position: Order maintenance; Work habits: Focus on crowd congestion; Historical execution record: Timely response).

[0121] 3) Management personnel profile: Management personnel (focus on: equipment operating status, personnel management during peak hours, operation records: frequently check details of abnormal warnings).

[0122] S42. Similarity Calculation: The cosine similarity algorithm is used to calculate the similarity between the interactive object and the interactive information using a formula:

[0123] ;

[0124] Let a user profile vector of a certain passenger be given. =(0.8,0.2,0.1) (Fast transfer preference, rest area preference, dining preference), feature vector of the interaction information "First elevator is out of service, recommend the second escalator for transfer" =(0.9,0.1,0), the calculation yields:

[0125] cosθ=(0.8×0.9+0.2×0.1+0.1×0) / (√ ×√ The similarity is approximately 0.92, indicating a high degree of similarity.

[0126] S43. Interactive Information Generation and Push: Based on user profiles and similarity calculation results, targeted interactive information is pushed to different interaction objects.

[0127] 1) For passengers: Information will be pushed to passenger Zhang through the waiting hall display screen, service terminal, and the station's official APP;

[0128] 2) For operations personnel: A [dispatch instruction] is pushed to operations personnel C; an [emergency task] is pushed to maintenance personnel A and B. The current fault risk probability is 85%. Handling suggestion: First check the circuit fault, and provide timely feedback after the problem is resolved.

[0129] 3) For management personnel: Push the "Operations Summary" to manager Li through the control center's large screen and office computer, and simultaneously push the execution progress of dispatch instructions.

[0130] Step S5: Closed-loop optimization. The execution feedback module collects real-time data on the execution effect of scheduling instructions and feedback data from each interactive object, transmits it to the data processing module, and starts the reinforcement learning model for closed-loop optimization. The specific sub-steps are as follows:

[0131] S51. Feedback Data Collection: The collected execution effect data and feedback data are as follows:

[0132] 1) Execution Results Data: Maintenance personnel A and B responded promptly within the stipulated timeframe, completing the first elevator malfunction repair within the allotted time, achieving a 100% completion rate; Operations personnel C effectively guided passengers, with no congestion occurring around the first elevator; the peak passenger flow in the lobby is projected to be 640 people per 10 minutes in the next hour (prediction error 1.5%), with a resource utilization rate of 92% and an operational efficiency improvement of 18%;

[0133] 2) Feedback data: Feedback on overall passenger satisfaction; Operations staff member C reported that dispatch instructions were clear and interactive information was tailored to job requirements; Management staff reported that anomaly warnings were accurate, dispatch responses were timely, and prediction results were accurate.

[0134] S52. Comparative Analysis: By comparing the execution effect data and feedback data with the operational status assessment results and resource scheduling requirements of step S2, it was found that: 1) The BP neural network anomaly risk prediction model has a high accuracy in predicting equipment failures, but the prediction error for passenger flow can be further reduced; 2) The scheduling response time of the improved genetic algorithm can be further shortened; 3) The collaborative filtering recommendation algorithm has a high accuracy in personalized passenger push and does not require significant adjustments.

[0135] S53, Parameter and Policy Optimization: Launch the reinforcement learning model, i.e., the Q-learning algorithm, using the formula... Update the action value function, where η=0.2, γ=0.9, r=0.92 (positive reward due to improved control effectiveness), and optimize the parameters:

[0136] 1) Optimize the parameters of the LSTM operational load prediction model: adjust the weights of hidden layer nodes to improve the accuracy of personnel flow prediction; 2) Optimize the parameters of the improved genetic algorithm to shorten the scheduling response time; 3) Fine-tune the weight coefficients of the multi-dimensional fusion analysis algorithm: increase the weight coefficient w1 of the peak personnel flow to enhance the focus on personnel flow.

[0137] S54. Optimization and Iteration: Synchronize the optimized algorithm parameters and scheduling strategies to the corresponding modules, repeat steps S1-S5, and complete one closed-loop optimization. In subsequent operations, through continuous iteration, the control system will continuously adapt to the dynamic changes in station operations, thoroughly solving the problems of incomplete control system, low data processing efficiency, poor adaptability of control strategies, and insufficient accuracy of resource scheduling in existing technologies, and meeting the high-density and high-dynamic operation and control needs of large hub railway stations.

[0138] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An information dynamic management and control service system for transportation hub operations, characterized in that, include: The information collection module is used to collect various dynamic information during the operation of transportation hubs in real time, and to perform standardized preprocessing on the collected dynamic information. Data processing module: Communicates with the information acquisition module and is used to receive the acquired data output by the information acquisition module; Dynamic scheduling module: It communicates with the data processing module and is used to receive the analysis results and prediction results output by the data processing module, generate targeted dynamic control and scheduling instructions, and send the scheduling instructions to the corresponding execution terminals; Precision Interaction Module: It communicates with the data processing module and the dynamic scheduling module respectively, and is used to receive the analysis results and prediction results output by the data processing module and the scheduling instructions output by the dynamic scheduling module. Execution feedback module: It communicates with the dynamic scheduling module and the precise interaction module, and is used to collect the execution effect data of scheduling instructions and the feedback data of each interactive object in real time. It transmits the execution effect data and feedback data to the data processing module to provide data support for closed-loop optimization. Storage module: Communicates with the data processing module to store data for various parameters, providing data support for data processing, analysis and prediction, and closed-loop optimization; Execution terminal: Communicates with the dynamic scheduling module, used to receive and execute dynamic management and scheduling instructions, and to feed back the instruction execution status to the execution feedback module; Interactive terminal: Communicates with the precision interaction module, receives interactive information pushed by the precision interaction module, and transmits the feedback data of the interactive object to the execution feedback module.

2. A method for dynamic information management and control services for transportation hub operations, used to implement the dynamic information management and control service system for transportation hub operations as described in claim 1, characterized in that, Includes the following steps: S1. By deploying data collection devices in various areas of transportation hubs, various dynamic information during the operation of the hubs is collected in real time. The collected dynamic information is then preprocessed in a standardized manner to obtain data in a unified format. S2. Transmit the collected data obtained in step S1 to the data processing module, and use a multi-dimensional fusion analysis algorithm to analyze the collected data in real time to obtain the site operation status assessment results, abnormal early warning information and resource scheduling requirements. S3. Based on the analysis results obtained in step S2, and combined with the operation rules and resource allocation of transportation hubs, generate targeted dynamic control and dispatch instructions, and send the dispatch instructions to the corresponding execution terminals. S4. Classify the analysis results of step S2 and the scheduling instructions of step S3 according to the type of the interactive object, generate targeted interactive information, and transmit it to each interactive object through the corresponding interactive terminal. S5. Real-time acquisition of the execution effect data of the scheduling instructions in step S3 and the feedback data of each interactive object in step S4. The execution effect data and feedback data are transmitted to the data processing module and compared with the analysis results of step S2. Based on the comparison results, the parameters and dynamic scheduling strategy of the multi-dimensional fusion analysis algorithm are continuously optimized.

3. The information dynamic management and control service method for transportation station operation according to claim 2, characterized in that: In step S1, the standardization preprocessing specifically includes three sub-steps: data cleaning, data format conversion, and data deduplication. The isolated forest algorithm is introduced during the data cleaning process to accurately identify and remove abnormal data, as detailed below: S11. Data Cleaning: The Isolation Forest algorithm is used to construct a forest model consisting of K isolated trees. Anomaly detection is performed on the collected raw dynamic information to remove outliers, noisy data, and missing values. In the Isolation Forest algorithm, outlier data, due to its features deviating from the normal data distribution, is quickly isolated. The anomaly detection threshold is calculated using the following formula: ； Where score(x) is the anomaly score of the data to be detected x; E(h(x)) is the average path length of the data to be detected x in K isolated trees; c(n) is the average path length of normal data in isolated trees when the sample size is n, which is used as a normalization factor; n is the total number of samples in this data cleaning, and K is the number of isolated trees; S12. Data format conversion: Convert heterogeneous data output from different acquisition devices into a preset unified JSON format; S13. Data Deduplication: A hash deduplication algorithm is used to perform hash calculations on the converted uniform format data, and duplicate data is deleted by comparing the hash values.

4. The information dynamic management and control service method for transportation station operation according to claim 2, characterized in that: In step S2, the multi-dimensional fusion analysis algorithm, which integrates principal component analysis algorithm and weighted fusion algorithm, is as follows: S21. Feature Extraction: Principal component analysis (PCA) is used to perform feature dimensionality reduction on the preprocessed data from step S1, eliminating redundant features; core feature parameters of each dimension of data are extracted, including peak traffic flow. Vehicle turnover efficiency Equipment failure rate Environmental parameter thresholds and service response time ; The feature dimensionality reduction process of the PCA algorithm is described by the formula... The implementation is as follows: Y is the core feature matrix after dimensionality reduction, X is the preprocessed acquisition data matrix, and W is the feature vector matrix, which is composed of the feature vectors corresponding to the eigenvalues ​​of the covariance matrix of the acquisition data matrix X. S22. Data Fusion: A weighted fusion algorithm is used to fuse the core feature parameters to obtain comprehensive feature parameters, which are used to comprehensively evaluate the station's operational status. The formula is: ； Where F is the comprehensive feature parameter, reflecting the overall operational status of the station; m is the number of dimensions of the core feature parameter; The weight coefficient for the i-th core feature parameter is dynamically allocated based on the degree of influence of each core feature parameter on the operation and management of the station. This is the normalized value of the i-th core feature parameter; S23. Anomaly Detection: Compare the comprehensive feature parameter F with the preset operational threshold range. A comparison is made, and an anomaly risk prediction model based on a BP neural network is introduced to output the probability P of anomaly risk occurrence; when and When an operation is deemed abnormal, an abnormality warning is generated, which includes the abnormality type, abnormality location, abnormality level, and probability of occurrence. The BP neural network anomaly risk prediction model includes an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is consistent with the dimension of the core feature parameters. The number of nodes in the hidden layer is determined by cross-validation. The output layer outputs the probability P of anomaly occurrence. The model's weights are updated using the following formula: ； in, The connection weight between the i-th node in the input layer and the j-th node in the hidden layer at the (k+1)-th iteration; The connection weights at the k-th iteration; The learning rate; This represents the error term of the j-th node in the hidden layer. The input value of the i-th node in the input layer is the normalized value of the core feature parameters. ; S24. Demand Matching: Based on the comprehensive feature parameter F and anomaly warning information, combined with the future operational status prediction results output by the LSTM-based operational load prediction model, corresponding resource scheduling demands are matched. The LSTM model captures the time-series characteristics of operational data, and its prediction output formula is as follows: ； in, This is the output value of the LSTM model at time t, which is the predicted value of the operating load at a future time. Let be the cell state at time t; is the output value of the output gate at time t; tanh is the activation function used to normalize the cell state; This is a dot product operation; This is the sigmoid activation function, used to control the output strength of the output gate.

5. The information dynamic management and control service method for transportation station operation according to claim 4, characterized in that: In step S22, the weight coefficients of each core feature parameter The determination was made using the Analytic Hierarchy Process (AHP), and the specific process is as follows: S221. Constructing the Judgment Matrix: Based on the impact of each core feature parameter on the operation and management of the station, construct a 5×5 judgment matrix. ,in This indicates the importance of the i-th core feature parameter relative to the j-th core feature parameter; S222. Consistency test: Calculate the largest eigenvalue of the judgment matrix A. Through consistency indicators Random consistency index (RI) and consistency ratio Perform a consistency check; if CR < 0.1, the judgment matrix meets the consistency requirement; otherwise, adjust the judgment matrix. in, 5 represents the largest eigenvalue of the judgment matrix A; 5 represents the number of dimensions of the core feature parameters; CI is the consistency index, reflecting the degree of deviation of the judgment matrix; RI is the random consistency index, which is a preset empirical value; CR is the consistency ratio, used to judge whether the consistency of the judgment matrix is ​​acceptable. S223. Determine the weighting coefficients: Normalize the judgment matrix A that meets the consistency requirements to obtain the weighting coefficients of each core feature parameter. And satisfy .

6. The information dynamic management and control service method for transportation station operation according to claim 2, characterized in that: In step S3, the dynamic scheduling process introduces an improved genetic algorithm to optimize the resource scheduling scheme, as follows: S31. Encoding: The resource scheduling scheme, including personnel allocation, vehicle scheduling, equipment management, and resource allocation, is encoded in binary to form an initial population. Each individual corresponds to a scheduling scheme, and the population size is N. S32, Fitness Function: Constructing the fitness function: ； Where fit(x) is the fitness value of individual x; Let be the weighting coefficient, satisfying These correspond to the importance of resource utilization, scheduling response speed, and management and control costs, respectively; U(x) represents the resource utilization of scheduling scheme x. T(x) represents the maximum resource utilization rate; T(x) represents the response time of scheduling scheme x. Let C(x) be the maximum allowable value for scheduling response time; C(x) is the management cost of scheduling scheme x. To minimize control costs; S33. Selection: Using a combination of roulette wheel selection and elite retention strategy, individuals with higher fitness values ​​are selected from the initial population to enter the next generation of the population. S34. Crossover and Mutation: Crossover operation uses single-point crossover; mutation operation uses random mutation; crossover and mutation increase population diversity and prevent the algorithm from getting trapped in local optima. S35. Termination: When the fitness value of the population tends to stabilize or reaches the preset number of iterations K, stop the iteration and output the optimal scheduling scheme. S36. Instruction generation and issuance: Based on the optimal scheduling scheme, generate targeted dynamic control and scheduling instructions. The scheduling instructions include instruction type, execution object, execution content and execution time limit, and are issued to the corresponding execution terminal through the wireless communication network.

7. The information dynamic management and control service method for transportation station operation according to claim 2, characterized in that: In step S4, the interaction process introduces a collaborative filtering recommendation algorithm, as follows: S41. User Profile Construction: Based on the historical behavior data of the interaction objects, construct user profiles for different interaction objects; the interaction objects include passengers, operators and managers, and the historical behavior data includes passengers' query records and travel preferences, operators' job positions and work habits, and managers' focus and operation records. S42. Similarity Calculation: The cosine similarity algorithm is used to calculate the similarity between different interactive objects, or the similarity between an interactive object and interactive information. The formula is: ； in, This represents the similarity value. , These are the user profile vector of the interactive object and the feature vector of the interactive information, respectively. For vectors and The dot product; , They are vectors , The modulus length; S43. Interactive Information Generation and Push: Based on user profiles and similarity calculation results, generate targeted interactive information for different interactive objects and transmit it through the corresponding interactive terminals.

8. The information dynamic management and control service method for transportation station operation according to claim 2, characterized in that: In step S5, a reinforcement learning model is introduced into the closed-loop optimization process to achieve dynamic optimization of various algorithm parameters and scheduling strategies, as detailed below: S51. Feedback Data Collection: Real-time collection of execution effect data of scheduling instructions in step S3 and feedback data of each interactive object in step S4; the execution effect data includes the execution completion rate of scheduling instructions, execution time, resource utilization rate and operational efficiency improvement value; the feedback data includes the satisfaction rating and suggestions of each interactive object; S52. Comparative Analysis: Compare the execution effect data and feedback data with the operational status assessment results and resource scheduling requirements of step S2 to determine the optimization direction; S53. Parameter and Strategy Optimization: Introduce a reinforcement learning model to continuously optimize the parameters and dynamic scheduling strategy of the multi-dimensional fusion analysis algorithm; The state update formula for the Q-learning algorithm is as follows: ； Where Q(s, a) is the action value function for performing action a in state s; is the learning rate; r is the reward value, determined based on the comparative analysis results. s is the discount factor used to measure the importance of future rewards; s is the current system state, which consists of execution effect data, feedback data, and operational status; a is the currently executed optimization action, which is the adjustment of algorithm parameters and scheduling strategy; s' is the new system state after executing action a; a' is the optimal action under the new state s'. The maximum action value under the new state s'; S54. Optimization Iteration: Synchronize the optimized algorithm parameters and scheduling strategy to the corresponding modules, and repeat steps S1-S5 to achieve a complete closed-loop iteration.

9. A method for dynamic information management and control services for transportation hub operations according to claim 2, characterized in that: In step S2, the station operation status assessment results include the operation load level; the operation status of each area includes orderly, basically orderly, congested, and abnormal; the abnormal warning information includes the abnormal type, abnormal location, abnormal level, probability of occurrence, and handling suggestions; and the resource scheduling requirements include resource type, resource quantity, scheduling priority, and scheduling time limit.

Images