Comprehensive transportation junction passenger flow control optimization method for smart city

Abstract

The invention discloses a smart city comprehensive transportation junction passenger flow control optimization method comprising the following steps: collecting data of a sensor, a camera and a mobile device, cleaning and preprocessing the data, using a data mining technology to identify a passenger flow rule, and using the data mining technology to determine the passenger flow rule; and establishing a passenger flow prediction model based on time sequence analysis, a regression model and a neural network, and training and verifying the accuracy of the model through historical data. The invention provides a passenger flow management and control optimization method for a comprehensive transportation junction of a smart city, and aims to realize accurate prediction and real-time regulation and control of passenger flow by integrating multiple data sources and applying artificial intelligence and machine learning technologies. And according to the prediction result, dynamically adjusting the scheduling schemes of the entrance and exit channels of the transportation hub and the transportation means, providing transfer path suggestions and reducing the transfer time of passengers.

Description

Technical Field

[0001] The present invention relates to the fields of smart city technology, traffic engineering, data analysis, artificial intelligence, optimization theory and information systems, and in particular to a method for optimizing passenger flow control at a comprehensive transportation hub in a smart city. Background Art

[0002] The construction of smart cities has received extensive attention in recent years, aiming to improve urban management and service levels by using modern information technology and communication technology. In smart cities, passenger flow control in integrated transportation hubs is an important research field, involving the integration and application of multiple technologies, including sensor technology, data mining, machine learning, artificial intelligence, and big data analysis. Traditional methods of passenger flow control in transportation hubs mainly rely on manual experience and static scheduling schemes, which have problems such as low efficiency, slow response, and lack of intelligence.

[0003] In the existing technology, the collection of passenger flow data usually relies on a single data source, such as a sensor or camera. The data is less comprehensive and accurate, and it is difficult to reflect the real passenger flow dynamics. Traditional data processing and analysis methods cannot effectively identify the patterns and characteristics of passenger flow, resulting in low prediction accuracy and inability to provide effective support for passenger flow control. Existing machine learning algorithms and prediction models often face problems such as training difficulties, model overfitting, and inaccurate predictions when processing large-scale and complex passenger flow data.

[0004] In terms of passenger flow forecasting, existing technologies mostly use simple regression models or time series analysis, ignoring the nonlinearity and complex relationships in the data, resulting in poor forecasting results. In addition, existing scheduling schemes and optimization strategies are mostly static or experience-based scheduling, which cannot dynamically adjust the opening and closing status of the import and export channels of the transportation hub and the departure frequency of transportation vehicles, making it difficult to cope with real-time changes in passenger flow.

[0005] In response to the above problems, the present invention proposes a method for optimizing passenger flow control in integrated transportation hubs in smart cities, which aims to comprehensively improve the passenger flow control capabilities of transportation hubs by integrating multiple data sources, applying advanced machine learning and artificial intelligence technologies, performing real-time optimization and dynamic adjustment, and realizing system integration and information sharing.

[0006] Specifically, the innovative points and solutions of the present invention include:

[0007] 1. Application of comprehensive data sources: Through the integration of sensor data, camera data and mobile device data, more comprehensive and accurate passenger flow information is provided, solving the problem of single data source and incomplete data in existing technologies.

[0008] 2. Data cleaning and processing: Through the data cleaning and preprocessing module, noise and outliers are removed, overcoming the problem of poor data quality and affecting analysis results in the existing technology.

[0009] 3. Advanced machine learning algorithms: Apply time series analysis, regression models and neural network advanced machine learning algorithms to establish a high-precision passenger flow prediction model, which improves the accuracy and reliability of passenger flow prediction and makes up for the defects of low prediction accuracy and overly simple models of traditional methods.

[0010] 4. Real-time optimization and dynamic adjustment: According to the passenger flow forecast results, the opening and closing status of the import and export channels of the transportation hub and the departure frequency and scheduling plan of the transportation tools are dynamically adjusted to achieve real-time response and dynamic adjustment, solving the problem that the scheduling plan in the existing technology is static and cannot cope with real-time changes.

[0011] 5. System integration and information sharing: Establish a comprehensive information management system to realize data sharing and collaborative work among different departments and systems, improve the overall management level and decision-making support capabilities, and overcome the problems of information islands and lack of coordination in existing technologies.

[0012] Therefore, how to provide a smart city's integrated transportation hub passenger flow control optimization method is a problem that technical personnel in this field urgently need to solve. Summary of the invention

[0013] One purpose of the present invention is to propose a method for optimizing passenger flow control in a comprehensive transportation hub of a smart city. The present invention makes full use of multiple data sources, artificial intelligence and machine learning technologies, and describes in detail the algorithms for real-time optimization and dynamic adjustment. It has the advantages of comprehensive data, accurate prediction, rapid response and efficient management.

[0014] According to an embodiment of the present invention, a method for optimizing passenger flow control in a comprehensive transportation hub of a smart city comprises the following steps:

[0015] Optionally, the method is characterized by comprising the following steps:

[0016] S1. Collect passenger flow data inside and outside the transportation hub. Data sources include sensor data, camera data, and mobile device data.

[0017] S2. Clean and process the collected raw data to remove noise and erroneous data;

[0018] S3. Use data mining technology to analyze the cleaned data and identify the patterns and characteristics of passenger flow;

[0019] S4. Apply machine learning algorithms to establish a passenger flow prediction model;

[0020] S5. Use historical data to train the passenger flow prediction model and evaluate the accuracy of the model through validation data;

[0021] S6. Dynamically adjust the opening and closing status of the import and export channels of the transportation hub according to the prediction results;

[0022] S7. Provide the best transfer route suggestion;

[0023] S8. Dynamically adjust the departure frequency and dispatching plan of buses and subways;

[0024] S9. Establish a comprehensive information management system to enable data sharing and collaborative work between different departments and systems.

[0025] Optionally, it is characterized in that step S4 specifically includes:

[0026] S4.1. Collect historical passenger flow data and real-time passenger flow data to construct a time series data set;

[0027] S4.2. Use machine learning algorithms based on large amounts of crowd data to conduct preliminary analysis of time series data sets to analyze basic trends and patterns in passenger flow changes;

[0028] S4.3. Apply the neural network model to conduct deep learning training on passenger flow data and build a passenger flow prediction model, which specifically includes the following steps:

[0029] Construct the input feature matrix X and the target vector y, where X contains time series data and related variables, and the target vector y represents the actual passenger flow;

[0030] Design a neural network architecture, including an input layer, several hidden layers, and an output layer. The structure of the neural network is as follows:

[0031]

[0032] in, is the predicted passenger flow, W i is the weight matrix of the i-th layer, b i is the bias vector of the i-th layer, and f is the activation function;

[0033] The neural network is trained using the back-propagation algorithm and the gradient descent optimization algorithm. The update rules for updating the weight matrix W and the bias vector b are as follows:

[0034]

[0035] Among them, η is the learning rate, L is the loss function, and the loss function is as follows:

[0036]

[0037] Use cross-validation to verify the prediction performance of the neural network model and calculate the mean square error:

[0038]

[0039] According to the verification results, optimize the hyperparameters of the neural network model, such as learning rate, number of hidden layers, and number of neurons in each layer;

[0040] Finally, the passenger flow prediction results based on the neural network model are output;

[0041] S4.4. Combining the prediction results of time series analysis and neural network model, conduct multi-dimensional analysis on passenger flow change trend and generate a future passenger flow prediction data set;

[0042] S4.5. Input the forecast data into the passenger flow control system of the transportation hub to dynamically adjust the opening and closing status of the import and export channels and optimize the transportation vehicle scheduling plan, which specifically includes the following steps:

[0043] Adjust the opening and closing status of the import and export channels of the transportation hub in real time according to the predicted passenger flow;

[0044] Use the control system to adjust the opening and closing of automatic doors to guide passenger flow, dynamically adjust channel signs and the display content of electronic display screens, provide real-time transfer and passage information, and adjust the number of security inspection channels open to adapt to changes in passenger flow and reduce waiting time;

[0045] Dynamically adjust the departure frequency and dispatching plan of buses and subways according to the predicted peak passenger flow time period and flow rate. The specific steps include:

[0046] Increase bus and subway frequency during peak hours;

[0047] Optimize the dispatching plan of transportation tools to match the transportation capacity with the passenger flow demand;

[0048] Use real-time traffic data to adjust the routes of vehicles;

[0049] S4.5.3. Establish a real-time feedback mechanism to continuously adjust and optimize strategies based on real-time monitoring data within the transportation hub. Specific steps include:

[0050] Monitor passenger flow changes in real time and obtain real-time data through cameras and sensors;

[0051] Use data analytics systems to analyze real-time data and identify potential congestion and safety issues;

[0052] Adjust optimization strategies in time

[0053] Optionally, it is characterized in that the S4.2 specifically includes:

[0054] S4.2.1. Construct a feature vector X, which contains time series data and related variables such as weather, events, holidays, and transportation scheduling:

[0055] S4.2.2. Select the random forest algorithm to train the feature vector X and passenger flow data y. The training process of the random forest model is as follows:

[0056]

[0057] in, represents the predicted passenger flow, M is the number of decision trees, and f m is the mth decision tree, and the random forest integrates multiple decision trees;

[0058] S4.2.3. Calculate the importance of each feature to passenger flow prediction. Calculate the feature importance I(xi):

[0059]

[0060] Among them, Im(xi) represents the feature x in the mth decision tree i The importance of

[0061] S4.2.4. Use cross-validation to verify the prediction performance of the random forest model and calculate the mean square error:

[0062]

[0063] Where N is the number of samples, y i is the actual passenger flow, To predict passenger flow;

[0064] S4.2.5. Optimize the random forest model and adjust model parameters based on feature importance and cross-validation results;

[0065] S4.2.6. Finally, the passenger flow prediction results based on the random forest algorithm are output.

[0066] 4. Optionally, it is characterized in that the step S5 specifically includes:

[0067] S5.1. Collect historical passenger flow dataset D train and validation dataset D val , construct training set and validation set, the data set contains time series data and related variables;

[0068] S5.2. Input the historical passenger flow data into the machine learning model, including the random forest model and the neural network model, and conduct preliminary training on the model. The training process uses the following loss function:

[0069]

[0070] Among them, L is the loss function, N is the number of training samples, y is the actual passenger flow, To predict passenger flow;

[0071] S5.3. Apply the back propagation algorithm and gradient descent optimization algorithm to update the weight matrix W and bias vector b of the neural network model and the parameters of the random forest model:

[0072]

[0073] Where η is the learning rate, and are the gradients of the loss function with respect to the weight matrix and the bias vector respectively;

[0074] S5.4. Use the validation dataset D val Verify the trained model and calculate the verification error and the mean square error of the verification error:

[0075]

[0076] Among them, M is the number of validation samples, y j is the actual passenger flow, Forecast passenger flow during verification;

[0077] S5.5. According to the verification results, adjust the hyperparameters of the model, including the learning rate α, the number of hidden layers, the number of neurons in each layer, and the number of trees M of the random forest model:

[0078]

[0079]

[0080] Among them, γ and λ are weight parameters adjusted by learning rate, α and β are weight parameters adjusted by the number of hidden layers, representing the gradient influence on the validation set and training set respectively;

[0081] S5.6. Perform cross-validation, split the data set into multiple subsets, use each subset as a validation set in a loop, and use the remaining subsets as training sets, and calculate the average validation error:

[0082]

[0083] Among them, K is the number of cross-validation folds, D val,k and D train,k are the number of samples in the validation set and training set of the kth fold, μ is the adjustment weight parameter of the training set error to the cross-validation error, and y j,k and are the actual and predicted passenger flows in the k-th fold validation set, respectively, and yi,k and are the actual and predicted passenger flows in the k-th fold training set, respectively;

[0084] S5.7. Finally determine the optimized model parameters, train the final model, and output the passenger flow prediction results.

[0085] 5. Optionally, it is characterized in that step S6 specifically includes:

[0086] S6.1. Predicted passenger flow output by the passenger flow prediction model Determine the expected passenger flow at each entrance and exit channel in the future time period and

[0087]

[0088] in, and are the capacity parameters of the i-th inlet and outlet channels, n and m are the numbers of inlet and outlet channels, respectively;

[0089] S6.2. Using real-time monitoring data real , monitor the actual passenger flow of the import and export channels, compare it with the predicted value, and calculate the error ∈:

[0090]

[0091] S6.3, dynamically adjust the opening and closing status of the import and export channels according to the error ∈, the specific steps include:

[0092] S6.3.1, if ∈ entrance If the preset threshold value θ is exceeded, the number of open import and export channels will be increased, and the adjustment rule is:

[0093]

[0094] Among them, Δn entrance is the number of inlet channels to be added, α is the adjustment coefficient, represents the ceiling function;

[0095] S6.3.2, if ∈ exit If the preset threshold value θ is exceeded, the number of open exit channels will be increased, and the adjustment rule is:

[0096]

[0097] Among them, Δn exit is the number of outlet channels to be added, and β is the adjustment coefficient;

[0098] S6.4. Redistribute passenger flow paths and optimize passenger transfer and travel routes. The specific steps include:

[0099] S6.4.1. Use the Dijkstra algorithm to calculate the optimal transfer path for passengers based on the current passenger flow and channel status. The objective function of path optimization is:

[0100]

[0101] Among them, T i is the time consumption of the i-th path segment, C i is the congestion level of the ith path segment, S i is the safety factor of the i-th path segment;

[0102] S6.4.2. Update electronic display screens and intelligent guidance systems;

[0103] S6.5. Continuously monitor and record the adjusted passenger flow data and evaluate the effect of the adjustment. The specific steps include:

[0104] S6.5.1. Conduct real-time analysis of adjusted passenger flow data to evaluate the effects of channel adjustment and route optimization;

[0105] S6.5.2. Based on the evaluation results;

[0106] S6.5.3. Record the adjustment process and results.

[0107] 6. Optionally, it is characterized in that the step S7 specifically includes:

[0108] S7.1. Predicted passenger flow output by the passenger flow prediction model Determine the expected passenger flow at each transfer node in the future time period

[0109]

[0110] in, is the weight parameter of the i-th transfer node, n is the number of transfer nodes;

[0111] S7.2. Using real-time monitoring data real , monitor the actual passenger flow at the transfer node and compare it with the predicted value, and calculate the error ∈ node :

[0112]

[0113] S7.3, according to the error ∈ node Dynamically adjust transfer routes to optimize passengers' transfer time and distance. Specific steps include:

[0114] S7.3.1, if ∈ node If the preset threshold value θ is exceeded, the optimal transfer path is recalculated and the adjustment rule is:

[0115]

[0116] Among them, P new New transfer path, T k is the time consumption of the kth path segment, C k is the congestion level of the kth path segment, S k is the safety factor of the kth path segment, λ and μ are weight parameters;

[0117] S7.3.2 Use intelligent guidance systems and electronic display screens to provide passengers with real-time transfer route suggestions:

[0118]

[0119] Among them, I path is the path guidance information, T total , C total and S total They are total time consumption, total congestion level and total safety factor respectively;

[0120] S7.4. Continuously monitor and record the adjusted transfer data and evaluate the effect of the adjustment. The specific steps include:

[0121] Conduct real-time analysis on the adjusted transfer data to evaluate the effect of route optimization:

[0122]

[0123] Among them, E opt To optimize the effect evaluation index, T before , C before and S before are the time consumption, congestion level and safety factor before optimization, T after , C before and S before They are respectively the optimized time consumption, congestion level and safety factor;

[0124] According to the optimization effect evaluation results, adjust the parameters of the path planning algorithm to increase the path capacity weight during peak hours, reduce the path capacity weight during off-peak hours, and increase the path safety weight during peak hours;

[0125] Record the specific process of each path adjustment, including the parameters used, adjustment steps, problems encountered and solutions, and analyze the effectiveness of the adjustment results;

[0126] Regularly update the passenger flow prediction model and route optimization algorithm, retrain the model every week, so that the model and algorithm can reflect the latest passenger flow changes and traffic conditions, adjust the model's training data set, add the latest passenger flow data, and delete outdated data;

[0127] Work closely with traffic management departments and system development teams to hold regular coordination meetings to discuss the implementation effects of route adjustment strategies, solve problems encountered during implementation, and jointly develop implementation plans and timetables for optimization measures.

[0128] 7. Optionally, it is characterized in that step S8 specifically includes:

[0129] S8.1. Predicted passenger flow output by the passenger flow prediction model Determine the expected passenger flow of each mode of transportation in the future time period

[0130]

[0131] in, is the capacity parameter of the ith transportation vehicle, n is the number of transportation vehicles;

[0132] S8.2. Using real-time monitoring data real , monitor the actual passenger flow of each mode of transportation, compare it with the predicted value, and calculate the error ∈ veh icle :

[0133]

[0134] S8.3, according to the error ∈ veh icle Dynamically adjust the departure frequency and dispatching plan of transportation tools. The specific steps include:

[0135] S8.3.1, if ∈ veh icle If the preset threshold value θ is exceeded, the departure frequency of the vehicle will be increased, and the adjustment rule is:

[0136]

[0137] Among them, f new is the new departure frequency, f current is the current departure frequency, α is the adjustment coefficient, represents the ceiling function;

[0138] S8.3.2. Re-plan the transportation dispatch plan based on passenger flow changes and actual traffic conditions, and adjust the rules as follows:

[0139]

[0140] Among them, D newis the new scheduling scheme, T k is the time consumption of the kth path segment, C k is the congestion level of the kth path segment, S k is the safety factor of the kth path segment, λ and μ are weight parameters;

[0141] S8.4. Use intelligent traffic management systems and electronic display screens to update real-time departure frequency and dispatch information. The specific steps include:

[0142] S8.4.1. Update the departure frequency and dispatch information in the intelligent traffic management system;

[0143] S8.4.2. Provide real-time frequency and dispatch information to passengers through electronic display screens and mobile applications:

[0144]

[0145] Among them, I schedule is the scheduling information, T total , C total and S total They are total time consumption, total congestion level and total safety factor respectively;

[0146] S8.5. Continuously monitor and record the adjusted transportation operation data and evaluate the adjustment effect. The specific steps include:

[0147] S8.5.1. Conduct real-time analysis of adjusted transportation operation data to evaluate the effectiveness of departure frequency and dispatch plans:

[0148]

[0149] Among them, E opt To optimize the effect evaluation index, T before , C before and S before are the time consumption, congestion level and safety factor before optimization, T after , C after and S before They are respectively the optimized time consumption, congestion level and safety factor;

[0150] S8.5.2. Further optimize the departure frequency and dispatching plan based on the optimization effect evaluation results;

[0151] S8.5.3. Record the adjustment process and results;

[0152] S8.5.4. Regularly update the passenger flow prediction model and scheduling optimization algorithm to reflect the latest passenger flow changes and traffic conditions;

[0153] S8.5.5. Work closely with the transportation operations department and system development team to hold regular coordination meetings to discuss the implementation effects of the dispatch strategy, solve problems encountered during implementation, and jointly develop an implementation plan and timetable for optimization measures.

[0154] 8. Optionally, it is characterized in that step S9 specifically includes:

[0155] S9.1. Establish an integrated information management system to integrate and process data from different data sources, including sensor data, camera data, and mobile device data. The specific steps include:

[0156] S9.1.1. Design a data integration module to receive and integrate various types of data to form a unified data structure;

[0157] S9.1.2. Establish a data cleaning and preprocessing module to clean and preprocess the integrated data to remove noise and outliers;

[0158] S9.2. Design a data sharing platform to achieve data sharing between different departments and systems. The specific steps include:

[0159] S9.2.1. Define data sharing protocols and interfaces to ensure seamless data transfer between different systems:

[0160] S9.2.2. Data sharing security and privacy protection through secure encryption and permission control:

[0161] D encrypted =E(D shared ,K);

[0162] Where E is the encryption function and K is the encryption key;

[0163] S9.3. Build a collaborative work module to improve the collaborative work capabilities among departments. The specific steps include:

[0164] S9.3.1. Establish a real-time communication and collaboration platform to support information exchange and collaborative decision-making among different departments:

[0165]

[0166] Among them, C collab For collaborative work efficiency, M i is the information exchange volume of the i-th department, T i is the information processing time of the i-th department;

[0167] S9.3.2. Develop collaborative workflow and task allocation systems to automatically generate optimization strategies and task allocation plans based on real-time data and forecast results:

[0168]

[0169] Among them, W task is the optimized task allocation scheme, T j is the time consumption of the jth task, C j is the resource consumption of the jth task, R j is the priority of the jth task, λ and μ are weight parameters;

[0170] S9.4. Design an information display and decision support system to provide managers with real-time data and analysis results to support scientific decision-making. The specific steps include:

[0171] S9.4.1. Develop a data visualization module to dynamically display passenger flow data and prediction results:

[0172] V data =g(D clean ,P visual );

[0173] Among them, V data is the visualization data set, g is the visualization function, P visual is the visualization parameter;

[0174] S9.4.2. Establish a decision support model to provide optimization suggestions and decision support based on real-time data and forecast results:

[0175]

[0176] Among them, D decision is the decision support data set, h is the decision support function, S policy is a policy parameter;

[0177] S9.5. Continuously monitor and evaluate the operation of the information management system. Specific steps include:

[0178] S9.5.1. Design a system performance monitoring module to monitor the system operation status and performance indicators in real time:

[0179]

[0180] Among them, P system is the system performance index, U k is the utilization rate of the kth module, R k is the response time of the kth module;

[0181] S9.5.2. Optimize system architecture and performance based on monitoring results:

[0182]

[0183] Among them, A optimized For the optimized system architecture, S l is the benefit of the lth optimization measure, C l is the cost of the lth optimization measure;

[0184] S9.5.3. Record the system’s operation logs and performance data.

[0185] Beneficial effects:

[0186] The smart city integrated transportation hub passenger flow control optimization method proposed in the present invention has many beneficial effects:

[0187] 1. Improved data comprehensiveness and accuracy: By integrating sensor data, camera data, and mobile device data, more comprehensive and accurate passenger flow information is provided. This method overcomes the problem of single data source and incomplete data in traditional data, making passenger flow control more scientific and effective.

[0188] 2. Efficient data cleaning and processing: Data cleaning and preprocessing modules are introduced to remove noise and outliers.

[0189] 3. Accurate passenger flow prediction: We applied time series analysis, regression models and neural network advanced machine learning algorithms to establish a high-precision passenger flow prediction model. Compared with traditional methods, the prediction accuracy is significantly improved, providing a solid data foundation for passenger flow control.

[0190] 4. Real-time optimization and dynamic adjustment: According to the prediction results, the opening and closing status of the import and export channels of the transportation hub and the departure frequency and scheduling plan of the transportation tools are dynamically adjusted. This method realizes real-time response and dynamic adjustment, overcoming the problem that traditional static scheduling cannot cope with real-time changes.

[0191] 5. System integration and collaborative work: A comprehensive information management system has been established to achieve data sharing and collaborative work between different departments and systems. Through real-time communication and collaboration platforms, various departments can effectively cooperate and jointly respond to changes in passenger flow, thus improving the overall management level.

[0192] 6. Optimized passenger experience: By providing the best transfer route suggestions, travel efficiency and comfort are improved. The application of intelligent guidance system and electronic display screen further enhances the travel experience of passengers.

[0193] 7. Transportation dispatch optimization: Dynamically adjust the departure frequency and dispatch plan of buses and subways to effectively deal with passenger flow peaks and special situations. This method ensures the reasonable dispatch of transportation and reduces the waiting time of passengers.

[0194] 8. System performance monitoring and continuous optimization: Continuously monitor and record system operation data and evaluate optimization effects. Based on the monitoring results, further optimize the system architecture and performance, regularly update models and algorithms, and the system can reflect the latest passenger flow changes and traffic conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0195] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0196] Figure 1 This is a flow chart of a method for optimizing passenger flow control in a comprehensive transportation hub of a smart city proposed by the present invention;

[0197] Figure 2 A schematic diagram of a passenger flow control optimization method for a comprehensive transportation hub in a smart city proposed by the present invention, which mainly describes the process in claim 4;

[0198] Figure 3 The data flow diagram of the passenger flow control optimization method for the comprehensive transportation hub of a smart city proposed by the present invention mainly describes claim 2. DETAILED DESCRIPTION

[0199] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0200] refer to Figure 1-3 , a smart city comprehensive transportation hub passenger flow control optimization method, comprising the following steps:

[0201] In this embodiment, it is characterized by comprising the following steps:

[0202] S1. Collect passenger flow data inside and outside the transportation hub. Data sources include sensor data, camera data, and mobile device data.

[0203] S2. Clean and process the collected raw data to remove noise and erroneous data;

[0204] S3. Use data mining technology to analyze the cleaned data and identify the patterns and characteristics of passenger flow;

[0205] S4. Apply machine learning algorithms to establish a passenger flow prediction model;

[0206] S5. Use historical data to train the passenger flow prediction model and evaluate the accuracy of the model through validation data;

[0207] S6. Dynamically adjust the opening and closing status of the import and export channels of the transportation hub according to the prediction results;

[0208] S7. Provide the best transfer route suggestion;

[0209] S8. Dynamically adjust the departure frequency and dispatching plan of buses and subways;

[0210] S9. Establish a comprehensive information management system to enable data sharing and collaborative work between different departments and systems.

[0211] In this implementation, it is characterized in that step S4 specifically includes:

[0212] S4.1. Collect historical passenger flow data and real-time passenger flow data to construct a time series data set;

[0213] S4.2. Use machine learning algorithms based on large amounts of crowd data to conduct preliminary analysis of time series data sets to analyze basic trends and patterns in passenger flow changes;

[0214] S4.3. Apply the neural network model to conduct deep learning training on passenger flow data and build a passenger flow prediction model, which specifically includes the following steps:

[0215] Construct the input feature matrix X and the target vector y, where X contains time series data and related variables, and the target vector y represents the actual passenger flow;

[0216] Design a neural network architecture, including an input layer, several hidden layers, and an output layer. The structure of the neural network is as follows:

[0217]

[0218] in, is the predicted passenger flow, W i is the weight matrix of the i-th layer, b i is the bias vector of the i-th layer, and f is the activation function;

[0219] The neural network is trained using the back-propagation algorithm and the gradient descent optimization algorithm. The update rules for updating the weight matrix W and the bias vector b are as follows:

[0220]

[0221] Among them, η is the learning rate, L is the loss function, and the loss function is as follows:

[0222]

[0223] Use cross-validation to verify the prediction performance of the neural network model and calculate the mean square error:

[0224]

[0225] According to the verification results, optimize the hyperparameters of the neural network model, such as learning rate, number of hidden layers, and number of neurons in each layer;

[0226] Finally, the passenger flow prediction results based on the neural network model are output;

[0227] S4.4. Combining the prediction results of time series analysis and neural network model, conduct multi-dimensional analysis on passenger flow change trend and generate a future passenger flow prediction data set;

[0228] S4.5. Input the forecast data into the passenger flow control system of the transportation hub to dynamically adjust the opening and closing status of the import and export channels and optimize the transportation vehicle scheduling plan, which specifically includes the following steps:

[0229] Adjust the opening and closing status of the import and export channels of the transportation hub in real time according to the predicted passenger flow;

[0230] Use the control system to adjust the opening and closing of automatic doors, guide the flow of passengers, dynamically adjust channel signs and the display content of electronic display screens, provide real-time transfer and passage information, and adjust the number of open security channels;

[0231] Dynamically adjust the departure frequency and dispatching plan of buses and subways according to the predicted peak passenger flow time period and flow rate. The specific steps include:

[0232] Increase bus and subway frequency during peak hours;

[0233] Optimize the dispatching plan of transportation tools to match the transportation capacity with the passenger flow demand;

[0234] Use real-time traffic data to adjust the routes of vehicles;

[0235] S4.5.3. Establish a real-time feedback mechanism to continuously adjust and optimize strategies based on real-time monitoring data within the transportation hub. Specific steps include:

[0236] Monitor passenger flow changes in real time and obtain real-time data through cameras and sensors;

[0237] Use data analytics systems to analyze real-time data and identify potential congestion and safety issues;

[0238] Adjust optimization strategies in a timely manner.

[0239] In this implementation, it is characterized in that the S42 specifically includes:

[0240] S4.2.1. Construct a feature vector X, which contains time series data and related variables such as weather, events, holidays, and transportation scheduling:

[0241] S4.2.2. Select the random forest algorithm to train the feature vector X and passenger flow data y. The training process of the random forest model is as follows:

[0242]

[0243] in, represents the predicted passenger flow, M is the number of decision trees, and f m is the mth decision tree, and the random forest integrates multiple decision trees;

[0244] S42.3. Calculate the importance of each feature to passenger flow prediction, feature importance I(xi):

[0245]

[0246] Among them, im(xi) represents the feature x in the mth decision tree i The importance of

[0247] S4.2.4. Use cross-validation to verify the prediction performance of the random forest model and calculate the mean square error:

[0248]

[0249] Where N is the number of samples, y i is the actual passenger flow, To predict passenger flow;

[0250] S4.2.5. Optimize the random forest model and adjust model parameters based on feature importance and cross-validation results;

[0251] S4.2.6. Finally, the passenger flow prediction results based on the random forest algorithm are output.

[0252] In this implementation, it is characterized in that the step S5 specifically includes:

[0253] S5.1. Collect historical passenger flow dataset D train and validation dataset D val , construct training set and validation set, the data set contains time series data and related variables;

[0254] S5.2. Input the historical passenger flow data into the machine learning model, including the random forest model and the neural network model, and conduct preliminary training on the model. The training process uses the following loss function:

[0255]

[0256] Among them, L is the loss function, N is the number of training samples, y is the actual passenger flow, To predict passenger flow;

[0257] S5.3. Apply the back propagation algorithm and gradient descent optimization algorithm to update the weight matrix W and bias vector b of the neural network model and the parameters of the random forest model:

[0258]

[0259] Where η is the learning rate, and are the gradients of the loss function with respect to the weight matrix and the bias vector respectively;

[0260] S5.4. Use the validation dataset D val Verify the trained model and calculate the verification error and the mean square error of the verification error:

[0261]

[0262] Among them, M is the number of validation samples, y j is the actual passenger flow, Forecast passenger flow during verification;

[0263] S5.5. According to the verification results, adjust the hyperparameters of the model, including the learning rate α, the number of hidden layers, the number of neurons in each layer, and the number of trees M of the random forest model:

[0264]

[0265] Among them, γ and λ are weight parameters adjusted by learning rate, α and β are weight parameters adjusted by the number of hidden layers, representing the gradient influence on the validation set and training set respectively;

[0266] S5.6. Perform cross-validation, split the data set into multiple subsets, use each subset as a validation set in a loop, and use the remaining subsets as training sets, and calculate the average validation error:

[0267]

[0268] Among them, K is the number of cross-validation folds, D val,k and D train,k are the number of samples in the validation set and training set of the kth fold, μ is the adjustment weight parameter of the training set error to the cross-validation error, and y j,k and are the actual and predicted passenger flows in the k-th fold validation set, respectively, and y i,k and are the actual and predicted passenger flows in the k-th fold training set, respectively;

[0269] S5.7. Finally determine the optimized model parameters, train the final model, and output the passenger flow prediction results.

[0270] In this implementation, it is characterized in that the step S6 specifically includes:

[0271] S6.1. Predicted passenger flow output by the passenger flow prediction model Determine the expected passenger flow at each entrance and exit channel in the future time period and

[0272]

[0273] in, and are the capacity parameters of the i-th inlet and outlet channels, n and m are the numbers of inlet and outlet channels, respectively;

[0274] S6.2. Using real-time monitoring data real , monitor the actual passenger flow of the import and export channels, compare it with the predicted value, and calculate the error ∈:

[0275]

[0276]

[0277] S6.3, dynamically adjust the opening and closing status of the import and export channels according to the error ∈, the specific steps include:

[0278] S6.3.1, if ∈ entrance If the preset threshold value θ is exceeded, the number of open import and export channels will be increased, and the adjustment rule is:

[0279]

[0280] Among them, Δn entrance is the number of inlet channels to be added, α is the adjustment coefficient, represents the ceiling function;

[0281] S6.3.2, if ∈ exit If the preset threshold value θ is exceeded, the number of open exit channels will be increased, and the adjustment rule is:

[0282]

[0283] Among them, Δn exit is the number of outlet channels to be added, and β is the adjustment coefficient;

[0284] S6.4. Redistribute passenger flow paths and optimize passenger transfer and travel routes. The specific steps include:

[0285] S6.4.1. Use the Dijkstra algorithm to calculate the optimal transfer path for passengers based on the current passenger flow and channel status. The objective function of path optimization is:

[0286]

[0287] Among them, T i is the time consumption of the i-th path segment, C i is the congestion level of the ith path segment, S i is the safety factor of the i-th path segment;

[0288] S6.4.2. Update electronic display screens and intelligent guidance systems;

[0289] S6.5. Continuously monitor and record the adjusted passenger flow data and evaluate the effect of the adjustment. The specific steps include:

[0290] S6.5.1. Conduct real-time analysis of adjusted passenger flow data to evaluate the effects of channel adjustment and route optimization;

[0291] S6.5.2. Based on the evaluation results;

[0292] S6.5.3. Record the adjustment process and results.

[0293] In this implementation, it is characterized in that the step S7 specifically includes:

[0294] S7.1. Predicted passenger flow output by the passenger flow prediction model Determine the expected passenger flow at each transfer node in the future time period

[0295]

[0296] in, is the weight parameter of the i-th transfer node, n is the number of transfer nodes;

[0297] S7.2. Using real-time monitoring data real , monitor the actual passenger flow at the transfer node and compare it with the predicted value, and calculate the error ∈ node :

[0298]

[0299] S7.3, according to the error ∈ node Dynamically adjust transfer routes to optimize passengers' transfer time and distance. Specific steps include:

[0300] S7.3.1, if ∈ nodeIf the preset threshold value θ is exceeded, the optimal transfer path is recalculated and the adjustment rule is:

[0301]

[0302] Among them, P new New transfer path, T k is the time consumption of the kth path segment, C k is the congestion level of the kth path segment, S k is the safety factor of the kth path segment, λ and μ are weight parameters;

[0303] S7.3.2 Use intelligent guidance systems and electronic display screens to provide passengers with real-time transfer route suggestions:

[0304]

[0305] Among them, I path is the path guidance information, T total , C total and S total They are total time consumption, total congestion level and total safety factor respectively;

[0306] S7.4. Continuously monitor and record the adjusted transfer data and evaluate the effect of the adjustment. The specific steps include:

[0307] Conduct real-time analysis on the adjusted transfer data to evaluate the effect of route optimization:

[0308]

[0309] Among them, E opt To optimize the effect evaluation index, T before , C before and S before are the time consumption, congestion level and safety factor before optimization, T after , C after and S after They are respectively the optimized time consumption, congestion level and safety factor;

[0310] According to the optimization effect evaluation results, adjust the parameters of the path planning algorithm to increase the path capacity weight during peak hours, reduce the path capacity weight during off-peak hours, and increase the path safety weight during peak hours;

[0311] Record the specific process of each path adjustment, including the parameters used, adjustment steps, problems encountered and solutions, and analyze the effectiveness of the adjustment results;

[0312] Regularly update the passenger flow prediction model and route optimization algorithm, retrain the model every week, so that the model and algorithm can reflect the latest passenger flow changes and traffic conditions, adjust the model's training data set, add the latest passenger flow data, and delete outdated data;

[0313] Work closely with traffic management departments and system development teams to hold regular coordination meetings to discuss the implementation effects of route adjustment strategies, solve problems encountered during implementation, and jointly develop implementation plans and timetables for optimization measures.

[0314] In this embodiment, it is characterized in that step S8 specifically includes:

[0315] S8.1. Predicted passenger flow output by the passenger flow prediction model Determine the expected passenger flow of each mode of transportation in the future time period

[0316]

[0317] in, is the capacity parameter of the ith transportation vehicle, n is the number of transportation vehicles;

[0318] S8.2. Using real-time monitoring data real , monitor the actual passenger flow of each mode of transportation, compare it with the predicted value, and calculate the error ∈ vehicle :

[0319]

[0320] S8.3, according to the error ∈ vehicle Dynamically adjust the departure frequency and dispatching plan of transportation tools. The specific steps include:

[0321] S8.3.1, if ∈ vehicle If the preset threshold value θ is exceeded, the departure frequency of the vehicle will be increased, and the adjustment rule is:

[0322]

[0323] Among them, f new is the new departure frequency, f current is the current departure frequency, α is the adjustment coefficient, represents the ceiling function;

[0324] S8.3.2. Re-plan the transportation dispatch plan based on passenger flow changes and actual traffic conditions, and adjust the rules as follows:

[0325]

[0326] Among them, Dnew is the new scheduling scheme, T k is the time consumption of the kth path segment, C k is the congestion level of the kth path segment, S k is the safety factor of the kth path segment, λ and μ are weight parameters;

[0327] S8.4. Use intelligent traffic management systems and electronic display screens to provide passengers with real-time frequency and dispatch information, including:

[0328] S8.4.1. Update the departure frequency and dispatch information in the intelligent traffic management system;

[0329] S8.4.2. Provide real-time frequency and dispatch information to passengers through electronic display screens and mobile applications:

[0330]

[0331] Among them, I sch edule is the scheduling information, T total , C total and S total They are total time consumption, total congestion level and total safety factor respectively;

[0332] S8.5. Continuously monitor and record the adjusted transportation operation data and evaluate the adjustment effect. The specific steps include:

[0333] S8.5.1. Conduct real-time analysis of adjusted transportation operation data to evaluate the effectiveness of departure frequency and dispatch plans:

[0334]

[0335] Among them, E opt To optimize the effect evaluation index, T before , C before and S before are the time consumption, congestion level and safety factor before optimization, T after , C after and S after They are respectively the optimized time consumption, congestion level and safety factor;

[0336] S8.5.2. Further optimize the departure frequency and dispatching plan based on the optimization effect evaluation results;

[0337] S8.5.3. Record the adjustment process and results;

[0338] S8.5.4. Regularly update the passenger flow prediction model and scheduling optimization algorithm to reflect the latest passenger flow changes and traffic conditions;

[0339] S8.5.5. Work closely with the transportation operations department and system development team to hold regular coordination meetings to discuss the implementation effects of the dispatch strategy, solve problems encountered during implementation, and jointly develop an implementation plan and timetable for optimization measures.

[0340] 8. In this embodiment, it is characterized in that step S9 specifically includes:

[0341] S9.1. Establish an integrated information management system to integrate and process data from different data sources, including sensor data, camera data, and mobile device data. The specific steps include:

[0342] S9.1.1. Design a data integration module to receive and integrate various types of data to form a unified data structure;

[0343] S9.1.2. Establish a data cleaning and preprocessing module to clean and preprocess the integrated data to remove noise and outliers;

[0344] S9.2. Design a data sharing platform to achieve data sharing between different departments and systems. The specific steps include:

[0345] S9.2.1. Define data sharing protocols and interfaces to ensure seamless data transfer between different systems:

[0346] S9.2.2. Data sharing security and privacy protection through secure encryption and permission control:

[0347] D encrypted =E(D sh ared ,K);

[0348] Where E is the encryption function and K is the encryption key;

[0349] S9.3. Build a collaborative work module to improve the collaborative work capabilities among departments. The specific steps include:

[0350] S9.3.1. Establish a real-time communication and collaboration platform to support information exchange and collaborative decision-making among different departments:

[0351]

[0352] Among them, C collab For collaborative work efficiency, M i is the information exchange volume of the i-th department, T i is the information processing time of the i-th department;

[0353] S9.3.2. Develop collaborative workflow and task allocation systems to automatically generate optimization strategies and task allocation plans based on real-time data and forecast results:

[0354]

[0355] Among them, W task is the optimized task allocation scheme, T j is the time consumption of the jth task, C j is the resource consumption of the jth task, R j is the priority of the jth task, λ and μ are weight parameters;

[0356] S9.4. Design an information display and decision support system to provide managers with real-time data and analysis results to support scientific decision-making. The specific steps include:

[0357] S9.4.1. Develop a data visualization module to dynamically display passenger flow data and prediction results:

[0358] V data =g(D clean ,P visual );

[0359] Among them, V data is the visualization data set, g is the visualization function, P visual is the visualization parameter;

[0360] S9.4.2. Establish a decision support model to provide optimization suggestions and decision support based on real-time data and forecast results:

[0361]

[0362] Among them, D decision is the decision support data set, h is the decision support function, S policy is a policy parameter;

[0363] S9.5. Continuously monitor and evaluate the operation of the information management system. Specific steps include:

[0364] S9.5.1. Design a system performance monitoring module to monitor the system operation status and performance indicators in real time:

[0365]

[0366] Among them, P system is the system performance index, U k is the utilization rate of the kth module, R k is the response time of the kth module;

[0367] S9.5.2. Optimize system architecture and performance based on monitoring results:

[0368]

[0369] Among them, A optimized For the optimized system architecture, S l is the benefit of the lth optimization measure, C l is the cost of the lth optimization measure;

[0370] S9.5.3. Record the system’s operation logs and performance data.

[0371] Example:

[0372] In Beijing, a densely populated modern city, Xizhimen Integrated Transportation Hub is the core node of the city's transportation network, handling a large number of passengers every day, including transfers between subways, buses, taxis and private cars. Due to the large passenger flow and high mobility, traditional passenger flow control methods are difficult to effectively respond to real-time changes, resulting in frequent congestion and chaos in the transportation hub, which seriously affects the travel experience of passengers and the operational efficiency of the transportation hub.

[0373] To address this problem, we adopted the intelligent city integrated transportation hub passenger flow control optimization method of the present invention. By integrating multiple data sources and applying artificial intelligence and machine learning technologies, we achieved accurate prediction and real-time control of passenger flow, effectively improving the management efficiency of transportation hubs and passenger experience. The following are the specific implementation steps:

[0374] 1. Data collection: Sensors and cameras are installed at key locations in transportation hubs, and data from passengers’ mobile devices is used to collect passenger flow information in real time. Millions of data are collected every day, including the number of people passing through each channel, speed, and direction.

[0375] 2. Data cleaning and processing: Clean and process the collected raw data to remove noise and erroneous data. Through the data cleaning module, about 5% of the data is identified as noise and removed every day to ensure the accuracy and consistency of the data.

[0376] 3. Data analysis: Use data mining technology to analyze the cleaned data and identify the patterns and characteristics of passenger flow. The analysis results show that the morning peak hours 7:00-9:00 and the evening peak hours 17:00-19:00 are the periods with the largest passenger flow, and certain specific channels and transfer nodes are often congested.

[0377] 4. Establishment of passenger flow prediction model: Apply time series analysis, regression model and neural network learning algorithm to establish a high-precision passenger flow prediction model. The model training uses historical data from the past year, including tens of millions of records. Through repeated iterations and verification, the final model's prediction accuracy reached more than 85%.

[0378] 5. Model training and validation: Use historical data to train the passenger flow prediction model, and use validation data to evaluate the accuracy of the model. During the model training process, the cross-validation method is used to ensure that the model has good generalization ability.

[0379] 6. Dynamically adjust the entrance and exit channels: According to the prediction results, the opening and closing status of the entrance and exit channels of the transportation hub are adjusted in real time. When it is predicted that some entrance channels will be overloaded during the morning peak, the system automatically opens more entrance channels and notifies passengers to use the more idle channels through the electronic display screen and guidance system.

[0380] 7. Provide the best transfer route suggestions: Provide the best transfer route suggestions based on real-time passenger flow data and forecast results. The system detects that the passenger flow at the transfer node of Metro Line 1 and Line 2 is large, and immediately optimizes the transfer route, recommends passengers to use alternative channels, and pushes information in real time through the intelligent guidance system.

[0381] 8. Dynamically adjust transportation dispatch: Dynamically adjust the departure frequency and dispatch plan of buses and subways to ensure the transportation capacity demand during peak hours. When it is predicted that Metro Line 1 will usher in a large passenger flow, the system automatically increases the departure frequency of the line to one bus every 5 minutes to ensure that passengers can be evacuated in time.

[0382] 9. Establishment of an integrated information management system: Establish an integrated information management system to achieve data sharing and collaborative work between different departments and systems. Through the data sharing platform, the traffic management department, operating company and emergency response center can obtain passenger flow information in real time, coordinate work and respond quickly to emergencies. Time and place: This example is carried out in the transportation hub of the large city of Tokyo, and the time span is from January 1, 2023 to December 31, 2023.

[0383] The amount of passenger flow data collected every day reaches about 5 million, and about 250,000 noise data are eliminated every day, accounting for 5% of the total data volume. The prediction accuracy of the model reaches more than 85%, the average prediction error during the morning peak period is ±5%, and the average prediction error during the evening peak period is ±4%. The average congestion time during peak hours has been reduced by 30%, the utilization of different channels is more balanced, the gap between the highest utilization rate and the lowest utilization rate has been reduced by 20%, and the average transfer time of passengers has been reduced by 15%. According to the passenger questionnaire, satisfaction has increased by 10%, and the departure frequency of subways and buses during peak hours has increased by 20%. The capacity utilization rate has increased by 25%, effectively avoiding passenger backlogs, the response speed of emergencies has increased by 40%, and the efficiency of collaborative work between different departments has increased by 30%.

[0384] Table 1. Data collection and processing

[0385]

[0386] Table 2. Passenger flow prediction accuracy

[0387] Time Prediction accuracy (%) Morning peak forecast error (%) Evening peak forecast error (%) 2023-01 85.2 ±5.2 ±4.1 2023-02 85.5 ±5.0 ±4.3 2023-03 86.0 ±4.8 ±4.0 ... ... ... ... 2023-12 85.7 ±5.1 ±4.2

[0388] Table 3. Effect of import and export channel adjustment

[0389] Time Reduction in peak hour congestion time (%) Channel utilization gap reduction (%) 2023-01 28 18 2023-02 30 20 2023-03 29 19 ... ... ... 2023-12 30 20

[0390] Table 4. Transfer route optimization

[0391] Time Average transfer time reduction (%) Improvement in passenger satisfaction (%) 2023-01 14 9 2023-02 15 10 2023-03 15 10 ... ... ... 2023-12 15 10

[0392] Table 5. Transportation vehicle scheduling optimization

[0393] Time Increase in frequency of departures during peak hours (%) Increase in capacity utilization (%) 2023-01 18 24 2023-02 20 25 2023-03 19 25 ... ... ... 2023-12 20 25

[0394] Table 6. Information management system effects

[0395] Time Improved response speed to emergencies (%) Improvement in departmental collaboration efficiency (%) 2023-01 38 28 2023-02 40 30 2023-03 39 29 ... ... ... 2023-12 40 30

[0396] The smart city comprehensive transportation hub passenger flow control optimization method of the present invention realizes the intelligent, real-time and precise management of transportation hub passenger flow through the comprehensive application of multiple advanced technologies. It not only significantly improves the management efficiency and adaptability, but also greatly improves the travel experience of passengers. It has broad application prospects and significant social benefits.

[0397] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can replace or change the technical solution and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

[0398] It can improve the travel experience of customers and has broad application prospects and significant social benefits.

[0399] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can replace or change the technical solution and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for optimizing passenger flow control in a comprehensive transportation hub of a smart city, characterized in that: The steps include: S1. Collect passenger flow data inside and outside the transportation hub. Data sources include sensor data, camera data, and mobile device data. S2. Clean and process the collected raw data to remove noise and erroneous data; S3. Use data mining technology to analyze the cleaned data and identify the patterns and characteristics of passenger flow; S4. Apply machine learning algorithms to establish a passenger flow prediction model; S5. Use historical data to train the passenger flow prediction model and evaluate the accuracy of the model through validation data; S6. Dynamically adjust the opening and closing status of the import and export channels of the transportation hub according to the prediction results; S7. Provide the best transfer route suggestion; S8. Dynamically adjust the departure frequency and dispatching plan of buses and subways; S9. Establish a comprehensive information management system to enable data sharing and collaborative work between different departments and systems.

2. According to claim 1, a smart city comprehensive transportation hub passenger flow control optimization method is characterized in that , wherein step S4 specifically includes: S4.

1. Collect historical passenger flow data and real-time passenger flow data to construct a time series data set; S4.

2. Use machine learning algorithms based on large amounts of crowd data to conduct preliminary analysis of time series data sets to analyze basic trends and patterns in passenger flow changes; S4.

3. Apply the neural network model to conduct deep learning training on passenger flow data and build a passenger flow prediction model, which specifically includes the following steps: Construct the input feature matrix X and the target vector y, where X contains time series data and related variables, and the target vector y represents the actual passenger flow; Design a neural network architecture, including an input layer, several hidden layers, and an output layer. The structure of the neural network is as follows: in, is the predicted passenger flow, W i is the weight matrix of the i-th layer, b i is the bias vector of the i-th layer, and f is the activation function; The neural network is trained using the back-propagation algorithm and the gradient descent optimization algorithm. The update rules for updating the weight matrix W and the bias vector b are as follows: Among them, η is the learning rate, L is the loss function, and the loss function is as follows: Use cross-validation to verify the prediction performance of the neural network model and calculate the mean square error: According to the verification results, optimize the hyperparameters of the neural network model, such as learning rate, number of hidden layers, and number of neurons in each layer; Finally, the passenger flow prediction results based on the neural network model are output; S4.

4. Combining the prediction results of time series analysis and neural network model, conduct multi-dimensional analysis on passenger flow change trend and generate a future passenger flow prediction data set; S4.

5. Input the forecast data into the passenger flow control system of the transportation hub to dynamically adjust the opening and closing status of the import and export channels and optimize the transportation vehicle scheduling plan, which specifically includes the following steps: Adjust the opening and closing status of the import and export channels of the transportation hub in real time according to the predicted passenger flow; Use the control system to adjust the opening and closing of automatic doors, guide passenger flow, dynamically adjust channel signs and electronic display screen content, provide real-time transfer and passage information, and adjust the number of security check channels open; Dynamically adjust the departure frequency and dispatching plan of buses and subways according to the predicted peak passenger flow time period and flow rate. The specific steps include: Increase bus and subway frequency during peak hours; Optimize the dispatching plan of transportation tools to match the transportation capacity with the passenger flow demand; Use real-time traffic data to adjust the routes of vehicles; Establish a real-time feedback mechanism to continuously adjust and optimize strategies based on real-time monitoring data within the transportation hub. Specific steps include: Monitor passenger flow changes in real time and obtain real-time data through cameras and sensors; Use data analytics systems to analyze real-time data and identify potential congestion and safety issues; Adjust optimization strategies in a timely manner.

3. The method for optimizing passenger flow control in a comprehensive transportation hub of a smart city according to claim 1, characterized in that: The S42 specifically includes: S4.2.

1. Construct a feature vector X, which contains time series data and related variables such as weather, events, holidays, and transportation scheduling: S4.2.

2. Select the random forest algorithm to train the feature vector X and passenger flow data y. The training process of the random forest model is as follows: in, represents the predicted passenger flow, M is the number of decision trees, and f m is the mth decision tree, and the random forest integrates multiple decision trees; S4.2.

3. Calculate the importance of each feature to passenger flow prediction. Calculate the feature importance I(xi): Among them, Im(xi) represents the feature x in the mth decision tree i The importance of S4.2.

4. Use cross-validation to verify the prediction performance of the random forest model and calculate the mean square error: Where N is the number of samples, y i is the actual passenger flow, To predict passenger flow; S4.2.

5. Optimize the random forest model and adjust model parameters based on feature importance and cross-validation results; S4.2.

6. Finally, the passenger flow prediction results based on the random forest algorithm are output.

4. The method for optimizing passenger flow control in a comprehensive transportation hub of a smart city according to claim 1, characterized in that: The step S5 specifically includes: S5.

1. Collect historical passenger flow dataset D train and validation dataset D val , construct training set and validation set, the data set contains time series data and related variables; S5.

2. Input the historical passenger flow data into the machine learning model, including the random forest model and the neural network model, and conduct preliminary training on the model. The training process uses the following loss function: Among them, L is the loss function, N is the number of training samples, y is the actual passenger flow, To predict passenger flow; S5.

3. Apply the back propagation algorithm and gradient descent optimization algorithm to update the weight matrix W and bias vector b of the neural network model and the parameters of the random forest model: Where η is the learning rate, and are the gradients of the loss function with respect to the weight matrix and the bias vector respectively; S5.

4. Use the validation dataset D val Verify the trained model and calculate the verification error and the mean square error of the verification error: Among them, M is the number of validation samples, y j is the actual passenger flow, Forecast passenger flow during verification; S5.

5. According to the verification results, adjust the model's hyperparameters, including the learning rate α, the number of hidden layers, the number of neurons in each layer, and the number of trees m in the random forest model: Among them, γ and λ are weight parameters adjusted by learning rate, α and β are weight parameters adjusted by the number of hidden layers, representing the gradient influence on the validation set and training set respectively; S5.

6. Perform cross-validation, split the data set into multiple subsets, use each subset as a validation set in a loop, and use the remaining subsets as training sets, and calculate the average validation error: Among them, K is the number of cross-validation folds, D val,k and D train,k are the number of samples in the validation set and training set of the kth fold, μ is the adjustment weight parameter of the training set error to the cross-validation error, and y j,k and are the actual and predicted passenger flows in the k-th fold validation set, respectively, and y i,k and are the actual and predicted passenger flows in the k-th fold training set, respectively; S5.

7. Finally determine the optimized model parameters, train the final model, and output the passenger flow prediction results.

5. The method for optimizing passenger flow control in a comprehensive transportation hub of a smart city according to claim 1, characterized in that: The step S6 specifically includes: S6.

1. Predicted passenger flow output by the passenger flow prediction model Determine the expected passenger flow at each entrance and exit channel in the future time period and in, and are the capacity parameters of the i-th inlet and outlet channels, n and m are the numbers of inlet and outlet channels, respectively; S6.

2. Using real-time monitoring data real , monitor the actual passenger flow of the import and export channels, compare it with the predicted value, and calculate the error ∈: S6.3, dynamically adjust the opening and closing status of the import and export channels according to the error ∈, the specific steps include: S6.3.1, if ∈ entrance If the preset threshold value θ is exceeded, the number of open import and export channels will be increased, and the adjustment rule is: Among them, Δn entrance is the number of inlet channels to be added, α is the adjustment coefficient, represents the ceiling function; S6.3.2, if ∈ exit If the preset threshold value θ is exceeded, the number of open exit channels will be increased, and the adjustment rule is: Among them, Δn exit is the number of outlet channels to be added, and β is the adjustment coefficient; S6.

4. Redistribute passenger flow paths and optimize passenger transfer and travel routes. The specific steps include: S6.4.

1. Use the Dijkstra algorithm to calculate the optimal transfer path for passengers based on the current passenger flow and channel status. The objective function of path optimization is: Among them, T i is the time consumption of the i-th path segment, C i is the congestion level of the ith path segment, S i is the safety factor of the i-th path segment; S6.4.

2. Update electronic display screens and intelligent guidance systems; S6.

5. Continuously monitor and record the adjusted passenger flow data and evaluate the effect of the adjustment. The specific steps include: S6.5.

1. Conduct real-time analysis of adjusted passenger flow data to evaluate the effects of channel adjustment and route optimization; S6.5.

2. Select a strategy based on the evaluation results.

6. The method for optimizing passenger flow control in a comprehensive transportation hub of a smart city according to claim 1, characterized in that: The step S7 specifically includes: S7.

1. Predicted passenger flow output by the passenger flow prediction model Determine the expected passenger flow at each transfer node in the future time period in, is the weight parameter of the i-th transfer node, n is the number of transfer nodes; S7.

2. Using real-time monitoring data real , monitor the actual passenger flow at the transfer node and compare it with the predicted value, and calculate the error ∈ node : S7.3, according to the error ∈ node Dynamically adjust transfer routes to optimize passengers' transfer time and distance. Specific steps include: S7.3.1, if ∈ node If the preset threshold value θ is exceeded, the optimal transfer path is recalculated and the adjustment rule is: Among them, P new New transfer path, T k is the time consumption of the kth path segment, C k is the congestion level of the kth path segment, S k is the safety factor of the kth path segment, λ and μ are weight parameters; S7.3.2 Use intelligent guidance systems and electronic display screens to provide passengers with real-time transfer route suggestions: Among them, I path is the path guidance information, T total , C total and S total They are total time consumption, total congestion level and total safety factor respectively; S7.

4. Continuously monitor and record the adjusted transfer data and evaluate the effect of the adjustment. The specific steps include: Conduct real-time analysis on the adjusted transfer data to evaluate the effect of route optimization: Among them, E opt To optimize the effect evaluation index, T before , C before and S before are the time consumption, congestion level and safety factor before optimization, T after , C before and S before They are respectively the optimized time consumption, congestion level and safety factor; According to the optimization effect evaluation results, adjust the parameters of the path planning algorithm to increase the path capacity weight during peak hours, reduce the path capacity weight during off-peak hours, and increase the path safety weight during peak hours; Record the specific process of each path adjustment, including the parameters used, adjustment steps, problems encountered and solutions, and analyze the effectiveness of the adjustment results; Regularly update the passenger flow prediction model and route optimization algorithm, retrain the model every week, so that the model and algorithm can reflect the latest passenger flow changes and traffic conditions, adjust the model's training data set, add the latest passenger flow data, and delete outdated data; Work closely with traffic management departments and system development teams to hold regular coordination meetings to discuss the implementation effects of route adjustment strategies, solve problems encountered during implementation, and jointly develop implementation plans and timetables for optimization measures.

7. The method for optimizing passenger flow control in a comprehensive transportation hub of a smart city according to claim 1, characterized in that: Wherein step S8 specifically comprises: S8.

1. Predicted passenger flow output by the passenger flow prediction model Determine the expected passenger flow of each mode of transportation in the future time period in, is the capacity parameter of the ith transportation vehicle, n is the number of transportation vehicles; S8.

2. Using real-time monitoring data real , monitor the actual passenger flow of each mode of transportation, compare it with the predicted value, and calculate the error ∈ vehicle : S8.3, according to the error ∈ vehicle Dynamically adjust the departure frequency and dispatching plan of transportation tools. The specific steps include: S8.3.1, if ∈ vehicle If the preset threshold value θ is exceeded, the departure frequency of the vehicle will be increased, and the adjustment rule is: Among them, f new is the new departure frequency, f current is the current departure frequency, α is the adjustment coefficient, represents the ceiling function; S8.3.

2. Re-plan the transportation dispatch plan based on passenger flow changes and actual traffic conditions, and adjust the rules as follows: Among them, D new is the new scheduling scheme, T k is the time consumption of the kth path segment, C k is the congestion level of the kth path segment, S k is the safety factor of the kth path segment, λ and μ are weight parameters; S8.

4. Use intelligent traffic management systems and electronic display screens to provide passengers with real-time frequency and dispatch information, including: S8.4.

1. Update the departure frequency and dispatch information in the intelligent traffic management system; S8.4.

2. Provide real-time frequency and dispatch information to passengers through electronic display screens and mobile applications: Among them, I schedule is the scheduling information, T total , C total and S total They are total time consumption, total congestion level and total safety factor respectively; S8.

5. Continuously monitor and record the adjusted transportation operation data and evaluate the adjustment effect. The specific steps include: S8.5.

1. Conduct real-time analysis of adjusted transportation operation data to evaluate the effectiveness of departure frequency and dispatch plans: Among them, E opt To optimize the effect evaluation index, T before , C before and S before are the time consumption, congestion level and safety factor before optimization, T after , C after and S before They are respectively the optimized time consumption, congestion level and safety factor; S8.5.

2. Further optimize the departure frequency and dispatching plan based on the optimization effect evaluation results; S8.5.

3. Record the adjustment process and results; S8.5.

4. Regularly update the passenger flow prediction model and scheduling optimization algorithm to reflect the latest passenger flow changes and traffic conditions; S8.5.

5. Work closely with the transportation operations department and system development team to hold regular coordination meetings to discuss the implementation effects of the dispatch strategy, solve problems encountered during implementation, and jointly develop an implementation plan and timetable for optimization measures.

8. The method for optimizing passenger flow control in a comprehensive transportation hub of a smart city according to claim 1, characterized in that: Wherein step S9 specifically includes: S9.

1. Establish an integrated information management system to integrate and process data from different data sources, including sensor data, camera data, and mobile device data. The specific steps include: S9.1.

1. Design a data integration module to receive and integrate various types of data to form a unified data structure; S9.1.

2. Establish a data cleaning and preprocessing module to clean and preprocess the integrated data to remove noise and outliers; S9.

2. Design a data sharing platform to achieve data sharing between different departments and systems. The specific steps include: S9.2.

1. Define data sharing protocols and interfaces to ensure seamless data transfer between different systems: S9.2.

2. Data sharing security and privacy protection through secure encryption and permission control: D encrypted =E(D shared ,K); Where E is the encryption function and K is the encryption key; S9.

3. Build a collaborative work module to improve the collaborative work capabilities among departments. The specific steps include: S9.3.

1. Establish a real-time communication and collaboration platform to support information exchange and collaborative decision-making among different departments: Among them, C collab For collaborative work efficiency, M i is the information exchange volume of the i-th department, T i is the information processing time of the i-th department; S9.3.

2. Develop collaborative workflow and task allocation systems to automatically generate optimization strategies and task allocation plans based on real-time data and forecast results: Among them, W task is the optimized task allocation scheme, T j is the time consumption of the jth task, C j is the resource consumption of the jth task, R j is the priority of the jth task, λ and μ are weight parameters; S9.

4. Design an information display and decision support system to provide managers with real-time data and analysis results to support scientific decision-making. The specific steps include: S9.4.

1. Develop a data visualization module to dynamically display passenger flow data and prediction results: V data =g(D clean ,P visual ); Among them, V data is the visualization data set, g is the visualization function, P visual is the visualization parameter; S9.4.

2. Establish a decision support model to provide optimization suggestions and decision support based on real-time data and forecast results: Among them, D decision is the decision support data set, h is the decision support function, S policy is a policy parameter; S9.

5. Continuously monitor and evaluate the operation of the information management system. Specific steps include: S9.5.

1. Design a system performance monitoring module to monitor the system operation status and performance indicators in real time: Among them, P system is the system performance index, U k is the utilization rate of the kth module, R k is the response time of the kth module; S9.5.

2. Optimize system architecture and performance based on monitoring results: Among them, A optimized For the optimized system architecture, S l is the benefit of the lth optimization measure, C l is the cost of the lth optimization measure; S9.5.

3. Record the system’s operation logs and performance data.

Images