Public transportation hub passenger flow early warning system combining big data and machine learning

Abstract

The invention provides a public transport hub passenger flow early warning system combining big data and machine learning. The public transport hub passenger flow early warning system comprises a data acquisition module, a data preprocessing module, a data feature extraction module, a first passenger flow prediction unit, a second passenger flow prediction unit and a passenger flow early warning module. The data acquisition module is used for acquiring historical passenger flow data of the public transportation hub and constructing a passenger flow time-space information base of the public transportation hub according to the historical passenger flow data of the public transportation hub; the data preprocessing module is used for carrying out data preprocessing on the historical passenger flow data of the passenger station to obtain a preprocessed historical data set of the passenger station; importing the historical data set of the passenger station into a data feature extraction module; the data feature extraction module extracts data features from the historical passenger flow data of the passenger station, and constructs a historical passenger flow multi-source heterogeneous matrix.

Description

Technical Field

[0001] The present invention belongs to the field of early warning, and in particular relates to a public transportation hub passenger flow early warning system combining big data and machine learning. Background Art

[0002] At present, with the acceleration of urbanization, the passenger flow of public transportation hubs is increasing. How to effectively manage and predict passenger flow has become an important issue in urban traffic management. The public transportation hub passenger flow warning system combining big data and machine learning has emerged, aiming to improve the operating efficiency and safety of transportation hubs through real-time data analysis and intelligent prediction. This system usually uses information from different data sources, such as historical passenger flow data, weather conditions, holiday arrangements, social activity information, etc., and analyzes it through machine learning algorithms to predict future passenger flow, thereby providing decision support for traffic managers. However, the current technology still has many defects in practical applications, which limits its effectiveness and popularity.

[0003] The quality and integrity of data are the main issues facing the current passenger flow warning system. Although big data technology can process massive amounts of information, the diversity and inconsistency of data sources often lead to uneven data quality. In many cases, missing, incorrect or inaccurate data will have a negative impact on the training and prediction results of machine learning models. If there is a deviation in the input data, the prediction effect of the model will be seriously affected, resulting in erroneous warning information. In addition, the update frequency and accuracy of real-time data will directly affect the response speed of the system, which in turn affects the decision-making of traffic management.

[0004] The selection and tuning process of machine learning models also presents certain challenges. Although machine learning algorithms have advantages in dealing with complex problems, it is a complex task to select appropriate algorithms and models and optimize them for different passenger flow characteristics and patterns. In many cases, traffic management personnel may lack sufficient expertise and find it difficult to select the best model and parameter settings based on actual conditions. This technical threshold makes it difficult for some transportation hubs to implement passenger flow warning systems and cannot fully utilize the advantages of machine learning.

[0005] The existing system also needs to be improved in terms of real-time performance and response speed. During peak hours, passenger flow may change very quickly, but many early warning systems have certain delays in data processing and model prediction. This delay may result in the system failing to issue early warnings in abnormal passenger flow situations, affecting the response capabilities of traffic managers. In addition, the system's flexibility and adaptability in handling emergencies are also insufficient, and it is impossible to quickly adjust the prediction model to adapt to new passenger flow patterns.

[0006] Existing technologies are also relatively weak in terms of user interface and visualization. Although the system can provide a large amount of data analysis and prediction results, if the user interface is not friendly enough or the information presentation is not intuitive enough, traffic management personnel may find it difficult to quickly understand and apply this information. This usability problem limits the actual utility of the system, reducing the efficiency of traffic management and the scientific nature of decision-making.

[0007] Data privacy and security issues are also becoming increasingly prominent. When public transportation hubs collect and process passenger flow data, they often involve users' personal information and privacy. If effective security measures are not taken during data processing and storage, data leakage and abuse may occur, which may lead to user distrust and legal risks. Therefore, how to ensure user privacy and data security while ensuring data utilization efficiency is an important issue that needs to be solved by current technology.

[0008] The passenger flow warning system for public transportation hubs that combines big data and machine learning still has significant defects in data quality, model selection, real-time performance, user experience, and data security. These problems not only restrict the effectiveness and scope of application of the system, but also affect the overall efficiency of public transportation management. Therefore, in-depth research and improvement on these technical defects will help improve the performance of the passenger flow warning system, thereby better serving the needs of urban traffic management.

[0009] Therefore, there is an urgent need for a public transportation hub passenger flow warning system that combines big data and machine learning. Summary of the invention

[0010] The present invention proposes a public transportation hub passenger flow warning system that combines big data and machine learning, which solves the problems of inaccurate passenger flow prediction at public transportation hubs and failure to provide timely warnings for abnormal passenger flows. By combining big data and machine learning technology, the accuracy of passenger flow prediction and warning capabilities are improved.

[0011] The technical solution of the present invention is implemented as follows: a passenger flow warning system for a public transportation hub combining big data and machine learning, comprising a data acquisition module, a data preprocessing module, a data feature extraction module, a first passenger flow prediction unit, a second passenger flow prediction unit and a passenger flow warning module;

[0012] The data collection module is used to collect historical passenger flow data of the public transportation hub, and to construct a spatiotemporal passenger flow information database of the public transportation hub based on the historical passenger flow data of the public transportation hub;

[0013] The data preprocessing module is used to perform data preprocessing on the historical passenger flow data of the passenger station to obtain a preprocessed historical data set of the passenger station; and import the historical data set of the passenger station into the data feature extraction module;

[0014] The data feature extraction module extracts data features from the historical passenger flow data of the passenger station and constructs a multi-source heterogeneous matrix of historical passenger flow;

[0015] The first passenger flow prediction unit inputs the actual passenger flow data of the passenger station corresponding to different years in the passenger flow spatiotemporal information database into different passenger flow prediction models, outputs the passenger flow prediction results of the passenger station in different years, evaluates the passenger flow prediction according to the passenger flow prediction results of the passenger station in different years and the actual passenger flow data corresponding to different years, obtains the prediction accuracy of each passenger flow prediction model, and takes the passenger flow prediction model with the highest accuracy as the optimal passenger flow prediction model;

[0016] The second passenger flow prediction unit calculates the deviation degree of the passenger flow prediction feature of the passenger station, takes the feature whose deviation degree satisfies the preset deviation degree as the abnormal passenger flow prediction feature of the passenger station, and uses the optimal passenger flow prediction model to pre-process the abnormal passenger flow prediction feature of the passenger station to obtain passenger flow prediction data of the passenger station;

[0017] The passenger flow warning module warns of passenger flow at public transportation hubs based on passenger flow forecast data at passenger stations.

[0018] Compared with the existing technology, this public transportation hub passenger flow warning system combining big data and machine learning has shown significant advantages in data processing, prediction model selection and anomaly detection. Traditional passenger flow prediction systems usually rely on simple linear regression or historical average method for prediction, lack of in-depth analysis of complex spatiotemporal data, resulting in low prediction accuracy and difficulty in coping with the dynamic characteristics of passenger flow changes. However, this system first establishes a spatiotemporal information database of passenger flow at public transportation hubs through the data acquisition module to ensure the comprehensiveness and accuracy of input data. In the data preprocessing stage, the system can clean and organize historical passenger flow data, making subsequent analysis and modeling more reliable. Compared with traditional methods, the introduction of the data preprocessing module can effectively remove noise data and fill missing values, thereby improving the quality of the data set.

[0019] In terms of feature extraction, this system extracts multi-source heterogeneous features from historical passenger flow data through the data feature extraction module and constructs a multi-source heterogeneous matrix of historical passenger flow. This process is different from the traditional system that relies only on a single indicator or simple feature. It can more comprehensively reflect the changing laws of passenger flow and provide a rich information basis for subsequent predictions. Through the integration of multi-source data, the system can capture a variety of factors that affect passenger flow, such as weather, holidays, activities, etc., thereby improving the accuracy of predictions.

[0020] In the passenger flow prediction link, the system is divided into the first passenger flow prediction unit and the second passenger flow prediction unit, and multiple prediction models are used for passenger flow prediction. This design is significantly different from the application of a single model in the prior art. It can select the optimal model for prediction by comparing the prediction results of different models. The first passenger flow prediction unit ensures that the model used has the best performance by evaluating the prediction accuracy of different models. This model selection mechanism not only improves the accuracy of the prediction, but also enhances the adaptability of the system, and can adjust the prediction strategy according to different situations.

[0021] The second passenger flow prediction unit focuses on calculating the deviation of the passenger flow prediction characteristics of the passenger station, and uses the preset deviation to judge abnormal passenger flow. This mechanism enables the system to identify abnormal passenger flow in a timely manner, avoiding the delayed response that may occur in traditional methods, and improving the timeliness and accuracy of early warning. Traditional systems are often passive in monitoring abnormal passenger flow and lack effective real-time analysis methods. However, this system can monitor passenger flow in real time and issue early warnings in a timely manner by establishing an abnormal detection mechanism to help managers take necessary countermeasures.

[0022] The passenger flow warning module issues warnings based on forecast data and can provide real-time decision support for public transportation operation management. Existing technologies often lack a complete warning mechanism and cannot provide effective guidance for operation managers. This system helps managers predict passenger flow changes in advance through scientific forecasting and real-time monitoring, optimize resource allocation and scheduling, and improve the service quality and efficiency of public transportation. This public transportation hub passenger flow warning system, which combines big data and machine learning, overcomes the shortcomings of traditional methods through innovative designs in multiple links such as data collection, processing, feature extraction, model selection, and anomaly monitoring, and significantly improves the accuracy of passenger flow prediction and warning capabilities, and has broad application prospects.

[0023] As a preferred embodiment, the data acquisition module is built with a data acquisition unit, a data marking unit and a data integration unit. The data acquisition unit collects historical passenger flow data of the public transportation hub from a data server connected to the public transportation hub, and the data acquisition unit performs data acquisition through data acquisition instructions; the data marking unit marks the year to which the historical passenger flow data belongs in a one-to-one correspondence according to the data acquisition instructions; the data integration unit integrates the historical passenger flow data with the data marks in a data format to obtain the integrated historical passenger flow data; the collected historical passenger flow data includes the feature label, time label, train number label and station location label of the real-time passenger flow of the public transportation hub; the train number label and station location label include the start and end time and the starting station of the corresponding label.

[0024] As a preferred embodiment, the data preprocessing module includes a data cleaning unit, a preprocessing marking unit and a text classification unit. The data cleaning unit performs data cleaning on the data type and data value of the integrated historical passenger flow data; the preprocessing marking unit performs data labeling on the integrated historical passenger flow data to obtain a labeled data set; and the text classification unit converts the text data in the labeled data set into digital codes.

[0025] As a preferred implementation, the data feature extraction module extracts the month and date features of different years, extracts the features of weekdays and holidays, and extracts the starting and ending platform features; the data feature extraction module includes a feature extraction unit and a feature selection unit; the feature extraction unit extracts data features by analyzing the integrated historical passenger flow data; the feature selection unit selects the data features preferentially to construct a multi-source heterogeneous matrix of historical passenger flow.

[0026] As a preferred embodiment, the first passenger flow prediction unit includes a prediction model construction unit and a passenger flow prediction evaluation unit. The prediction model construction unit is used to normalize the historical passenger flow data of the bus station, construct a historical passenger flow data set, and use a machine learning method to construct a passenger flow prediction model based on the historical passenger flow multi-source heterogeneous matrix; the passenger flow prediction evaluation unit inputs the actual passenger flow data of the bus station corresponding to different years in the passenger flow spatiotemporal information library into the passenger flow prediction model, predicts the passenger flow data according to a preset time period, inputs the actual passenger flow data of the bus station corresponding to different years in the passenger flow spatiotemporal information library into different passenger flow prediction models, outputs the passenger flow prediction results of the bus station in different years, evaluates the passenger flow prediction according to the passenger flow prediction results of the bus station in different years and the corresponding actual passenger flow data in different years, obtains the prediction accuracy of each passenger flow prediction model, takes the maximum value of each prediction accuracy as the optimal prediction accuracy, and takes the passenger flow prediction model with the highest accuracy as the optimal passenger flow prediction model.

[0027] After adopting the above technical solution, the beneficial effect of the present invention is that the system ensures the basic data quality of passenger flow prediction through comprehensive data collection and preprocessing, thereby greatly improving the accuracy of the prediction. Accurate passenger flow prediction can help traffic management departments to reasonably arrange vehicles and personnel, reduce resource waste caused by passenger flow fluctuations, and improve the operational efficiency of public transportation. By establishing a multi-source heterogeneous matrix of historical passenger flow, the system can comprehensively reflect the various factors affecting passenger flow and enhance the ability to recognize complex passenger flow patterns. This comprehensive feature extraction capability enables the system to effectively predict passenger flow at different times and in different scenarios, further improving the flexibility and adaptability of public transportation services. In terms of model selection and anomaly detection, the system ensures that the model used has the best performance by introducing and evaluating multiple prediction models. This dynamic selection mechanism not only improves the accuracy of the prediction, but also enhances the adaptability of the system, and can adjust the prediction strategy in real time according to different situations.

[0028] By calculating the deviation of passenger flow prediction characteristics, the system can identify abnormal passenger flow in time, avoid the possible lag reaction in traditional methods, and improve the timeliness and accuracy of early warning. An effective abnormal monitoring mechanism can help traffic managers take measures in advance to avoid congestion or service quality degradation caused by a surge in passenger flow. The passenger flow early warning module issues early warnings based on forecast data, providing real-time decision support for public transportation operation management. This real-time monitoring and early warning mechanism can help managers predict passenger flow changes in advance, optimize resource allocation and scheduling, and improve the service quality and efficiency of public transportation. At the same time, the system can also provide passengers with timely information feedback to help them arrange travel time reasonably, reduce unnecessary waiting and congestion, and improve passengers' travel experience. The implementation of this system can also provide data support for the long-term development of public transportation, help managers analyze passenger flow trends, and formulate scientific transportation plans and policies. This data-driven decision-making method will help improve the overall operational efficiency and service level of public transportation and contribute to the sustainable development of the city. In summary, the implementation of this system not only improves the operational efficiency and service quality of public transportation, but also provides important technical support and guarantee for urban public transportation management through scientific prediction and real-time early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0030] Figure 1 It is a flow chart of the system of the present invention. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] Example:

[0033] like Figure 1 As shown, the public transportation hub passenger flow warning system combining big data and machine learning described in this application document is intended to improve the operational efficiency and safety of public transportation hubs, especially to timely warn of possible passenger congestion or abnormal conditions during peak hours or when emergencies occur. Its working principle and process can be elaborated in detail in specific implementation scenarios to understand the practical application of the system.

[0034] In an actual public transportation hub, such as a train station or subway station in a large city, the system first starts working through the data collection module. This module is responsible for collecting passenger flow data from various historical data sources, including ticket sales data, number of people entering and leaving the station, real-time surveillance camera data, social media information, etc. This data will be constructed into a passenger flow spatiotemporal information library, forming the basis for comprehensive passenger flow analysis of the public transportation hub. The key to this stage is to ensure the accuracy and comprehensiveness of the data for subsequent analysis and prediction.

[0035] Next, the data preprocessing module cleans and organizes the collected historical passenger flow data. Data preprocessing includes steps such as removing duplicate data, filling missing values, and standardizing data formats to ensure that the generated passenger station historical data set is clean and consistent. This process provides reliable basic data for subsequent data feature extraction, so that the extracted features can truly reflect the temporal and spatial changes of passenger flow.

[0036] In the data feature extraction module, meaningful features are extracted from the preprocessed historical passenger flow data set to construct a multi-source heterogeneous matrix of historical passenger flow. These features may include passenger flow in different time periods, weather factors, holiday impacts, surrounding activities, and other information. By constructing a multi-source heterogeneous matrix, the system can integrate multi-dimensional data, thereby improving the model's sensitivity and accuracy to passenger flow changes.

[0037] The first passenger flow prediction unit then inputs the actual passenger flow data of the bus station corresponding to different years into a variety of passenger flow prediction models for training and testing. These models may include time series prediction models, regression analysis models, and machine learning algorithms (such as random forests, neural networks, etc.). The system will output the passenger flow prediction results of the bus station in different years, and compare and analyze them with the corresponding actual passenger flow data to evaluate the prediction accuracy of each model. Finally, the system will select the passenger flow prediction model with the highest accuracy as the optimal model for use in subsequent predictions.

[0038] The second passenger flow prediction unit will calculate the deviation of the passenger flow prediction features of the passenger station to identify potential abnormalities. If the deviation of certain features exceeds the preset threshold, these features will be marked as abnormal passenger flow prediction features of the passenger station. Then, the system will use the optimal passenger flow prediction model selected earlier to pre-process these abnormal features to generate the corresponding passenger flow prediction data for the passenger station. This link is the core of the system, which can identify possible abnormalities in a timely manner and prevent potential passenger flow crises.

[0039] The passenger flow warning module monitors and warns the passenger flow of public transportation hubs in real time based on the passenger flow forecast data generated by the passenger station. When the system recognizes that the passenger flow in a certain period of time is about to exceed the safety threshold, it will issue a warning to prompt management personnel to take corresponding measures, such as increasing manpower, adjusting vehicle scheduling, and issuing passenger flow diversion information, in order to effectively deal with the upcoming passenger flow peak.

[0040] This public transportation hub passenger flow warning system, which combines big data and machine learning, has formed a complete passenger flow monitoring and warning mechanism through scientific data collection, preprocessing, feature extraction and the establishment of a prediction model. Its effectiveness and accuracy not only improve the operational efficiency of public transportation hubs, but also provide protection for passengers' travel safety. It has broad application prospects and important practical significance.

[0041] The data collection module is built with a data collection unit, a data marking unit and a data integration unit. The data collection unit collects historical passenger flow data of the public transportation hub from a data server connected to the public transportation hub. The data collection unit collects data through data collection instructions; the data marking unit marks the year to which the historical passenger flow data belongs in a one-to-one correspondence according to the data collection instructions; the data integration unit integrates the historical passenger flow data with data marks in data format to obtain integrated historical passenger flow data; the collected historical passenger flow data includes feature tags, time tags, train number tags and station location tags for real-time passenger flow at the public transportation hub; the train number tags and station location tags include the start and end times and the starting station of the corresponding tags.

[0042] In this technical solution, the combined design of the built-in data collection unit, data marking unit and data integration unit in the data collection module significantly improves the collection and processing efficiency of historical passenger flow data of public transportation hubs. Compared with the existing technology, the traditional data collection method is often isolated and lacks a systematic module design, resulting in independent data collection, marking and integration, and low information flow efficiency. However, this solution uses an integrated module design to closely connect the data collection, marking and integration processes to form an efficient workflow. The data collection unit obtains historical passenger flow data from the server through data collection instructions, and the data marking unit performs one-to-one corresponding marking according to the collection instructions and the data year, ensuring the accuracy and consistency of the data. The data integration unit formats the marked data and provides structured historical passenger flow data, covering information such as real-time passenger flow feature labels, time labels, train number labels and station location labels. This systematic processing method not only improves the availability of data, but also lays a solid foundation for subsequent analysis and modeling.

[0043] The data preprocessing module includes a data cleaning unit, a preprocessing marking unit and a text classification unit. The data cleaning unit cleans the data type and data value of the integrated historical passenger flow data; the preprocessing marking unit labels the integrated historical passenger flow data to obtain a labeled data set; and the text classification unit converts the text data in the labeled data set into digital codes.

[0044] In the data preprocessing module, the design of the data cleaning unit, preprocessing labeling unit and text classification unit reflects a significant difference from the existing technology. The traditional data preprocessing process usually lacks systematicity and often only focuses on a certain link in data cleaning or labeling, resulting in uneven data quality. However, this solution sets up a special cleaning unit to ensure the consistency of the data type and data value of the integrated historical passenger flow data, thereby improving the quality of the data from the source. Next, the preprocessing labeling unit labels the integrated data to ensure the readability and interpretability of the data set. Finally, the text classification unit converts the labeled text data into digital codes to facilitate subsequent algorithm processing. Through the close cooperation of these three links, this solution not only improves the data processing efficiency, but also enhances the quality and availability of the data, providing a more reliable foundation for subsequent analysis.

[0045] The data feature extraction module extracts the features of the month and date of different years, the features of working days and holidays, and the features of the starting and ending points and platforms respectively; the data feature extraction module includes a feature extraction unit and a feature selection unit; the feature extraction unit extracts data features by analyzing the integrated historical passenger flow data; the feature selection unit selects the data features preferentially and constructs a multi-source heterogeneous matrix of historical passenger flow. In the data feature extraction module, the design of the feature extraction unit and the feature selection unit enables the features of the month, date, working days and holidays of different years, as well as the features of the starting and ending points and platforms to be effectively extracted. The main difference between this method and the prior art is that the traditional data feature extraction is often limited to simple statistical features and lacks in-depth analysis of time and space, resulting in insufficient breadth and depth of feature extraction. However, this solution ensures that the extracted data features can truly reflect the law of passenger flow changes and construct a multi-source heterogeneous matrix of historical passenger flow through a systematic feature extraction and preferential selection mechanism. This extraction method that comprehensively considers multi-dimensional features not only improves the accuracy of data analysis, but also provides rich feature information for subsequent passenger flow prediction, so that the model can better adapt to complex passenger flow change scenarios.

[0046] The first passenger flow prediction unit includes a prediction model construction unit and a passenger flow prediction evaluation unit. The prediction model construction unit is used to normalize the historical passenger flow data of the passenger station, construct a historical passenger flow data set, and use a machine learning method to construct a passenger flow prediction model based on the historical passenger flow multi-source heterogeneous matrix; the passenger flow prediction evaluation unit inputs the actual passenger flow data of the passenger station corresponding to different years in the passenger flow spatiotemporal information library into the passenger flow prediction model, predicts the passenger flow data according to a preset time period, inputs the actual passenger flow data of the passenger station corresponding to different years in the passenger flow spatiotemporal information library into different passenger flow prediction models, outputs the passenger flow prediction results of the passenger station in different years, evaluates the passenger flow prediction according to the passenger flow prediction results of the passenger station in different years and the corresponding actual passenger flow data in different years, obtains the prediction accuracy of each passenger flow prediction model, takes the maximum value of each prediction accuracy as the optimal prediction accuracy, and takes the passenger flow prediction model with the highest accuracy as the optimal passenger flow prediction model.

[0047] In the first passenger flow prediction unit, the design of the prediction model construction unit and the passenger flow prediction evaluation unit shows the difference from the existing technology. Traditional passenger flow prediction methods usually rely on static models or simple linear regression, and lack in-depth analysis and processing of historical passenger flow data. However, this solution normalizes the historical passenger flow data, constructs a historical passenger flow data set, and uses machine learning methods to construct a prediction model based on a multi-source heterogeneous matrix, showing a higher level of intelligence. The passenger flow prediction evaluation unit inputs the actual passenger flow data of different years into the prediction model, conducts systematic predictions, and evaluates the accuracy of each model based on the comparison of the prediction results with the actual data. The automation and systematic design of this process not only improves the accuracy of the prediction, but also provides a scientific basis for selecting the optimal prediction model, ensuring the reliability of passenger flow prediction. This comprehensive use of modern data processing and machine learning technology has significantly improved the scientificity and practicality of passenger flow prediction, and overcomes the common problems of inaccurate predictions and poor model adaptability in the existing technology.

[0048] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A public transportation hub passenger flow warning system combining big data and machine learning, characterized in that: It includes a data acquisition module, a data preprocessing module, a data feature extraction module, a first passenger flow prediction unit, a second passenger flow prediction unit and a passenger flow warning module; The data collection module is used to collect historical passenger flow data of the public transportation hub, and to construct a spatiotemporal passenger flow information database of the public transportation hub based on the historical passenger flow data of the public transportation hub; The data preprocessing module is used to perform data preprocessing on the historical passenger flow data of the passenger station to obtain a preprocessed historical data set of the passenger station; and import the historical data set of the passenger station into the data feature extraction module; The data feature extraction module extracts data features from the historical passenger flow data of the passenger station and constructs a multi-source heterogeneous matrix of historical passenger flow; The first passenger flow prediction unit inputs the actual passenger flow data of the passenger station corresponding to different years in the passenger flow spatiotemporal information database into different passenger flow prediction models, outputs the passenger flow prediction results of the passenger station in different years, evaluates the passenger flow prediction according to the passenger flow prediction results of the passenger station in different years and the actual passenger flow data corresponding to different years, obtains the prediction accuracy of each passenger flow prediction model, and takes the passenger flow prediction model with the highest accuracy as the optimal passenger flow prediction model; The second passenger flow prediction unit calculates the deviation degree of the passenger flow prediction feature of the passenger station, takes the feature whose deviation degree satisfies the preset deviation degree as the abnormal passenger flow prediction feature of the passenger station, and uses the optimal passenger flow prediction model to pre-process the abnormal passenger flow prediction feature of the passenger station to obtain passenger flow prediction data of the passenger station; The passenger flow warning module warns of passenger flow at public transportation hubs based on passenger flow forecast data at passenger stations.

2. A public transportation hub passenger flow warning system combining big data and machine learning as claimed in claim 1, characterized in that: The data collection module has a built-in data collection unit, a data marking unit and a data integration unit. The data collection unit collects the historical passenger flow data of the public transportation hub from the data server connected to the public transportation hub. The data collection unit collects data through data collection instructions; the data marking unit marks the data collection instructions in a one-to-one correspondence with the year to which the historical passenger flow data belongs; The data integration unit integrates the historical passenger flow data with data tags in data format to obtain integrated historical passenger flow data; the collected historical passenger flow data includes feature tags, time tags, train number tags and station location tags for real-time passenger flow at public transportation hubs; the train number tags and station location tags include the start and end times and the starting station of the corresponding tags.

3. The public transportation hub passenger flow warning system combining big data and machine learning as claimed in claim 1, characterized in that: The data preprocessing module includes a data cleaning unit, a preprocessing marking unit and a text classification unit. The data cleaning unit cleans the data type and data value of the integrated historical passenger flow data; the preprocessing marking unit labels the integrated historical passenger flow data to obtain a labeled data set; and the text classification unit converts the text data in the labeled data set into digital codes.

4. The public transportation hub passenger flow warning system combining big data and machine learning as claimed in claim 1, characterized in that: The data feature extraction module extracts the features of months and dates of different years, extracts the features of working days and holidays, and extracts the features of starting and ending points and platforms; The data feature extraction module includes a feature extraction unit and a feature selection unit; The feature extraction unit extracts data features by analyzing the integrated historical passenger flow data; the feature selection unit selects the data features preferentially to construct a multi-source heterogeneous matrix of historical passenger flow.

5. The public transportation hub passenger flow warning system combining big data and machine learning as claimed in claim 1, characterized in that: The first passenger flow prediction unit includes a prediction model construction unit and a passenger flow prediction evaluation unit. The prediction model construction unit is used to normalize the historical passenger flow data of the passenger station, construct a historical passenger flow data set, and use a machine learning method to construct a passenger flow prediction model based on the historical passenger flow multi-source heterogeneous matrix; the passenger flow prediction evaluation unit inputs the actual passenger flow data of the passenger station corresponding to different years in the passenger flow spatiotemporal information library into the passenger flow prediction model, predicts the passenger flow data according to a preset time period, inputs the actual passenger flow data of the passenger station corresponding to different years in the passenger flow spatiotemporal information library into different passenger flow prediction models, outputs the passenger flow prediction results of the passenger station in different years, evaluates the passenger flow prediction according to the passenger flow prediction results of the passenger station in different years and the corresponding actual passenger flow data in different years, obtains the prediction accuracy of each passenger flow prediction model, takes the maximum value of each prediction accuracy as the optimal prediction accuracy, and takes the passenger flow prediction model with the highest accuracy as the optimal passenger flow prediction model.

Images