Method for determining congestion degree of rail transit train carriages and platform control equipment

Abstract

The invention discloses a method for determining the crowding degree of rail transit train carriages and platform control equipment, and the method comprises the steps: obtaining a video stream sent by a video collection device when a train stays at a current platform if detecting that the doors of the current carriage are all closed; identifying the video stream, and determining a first getting-on number and a first getting-off number of the current carriage at the current platform during the stop period; according to the initial number of people when the current compartment drives into the current platform, the first getting-on number of people and the first getting-off number of people, the existing number of people when the train drives away from the current platform is determined; the set neural network is applied, and the second get-off number of people of the current compartment at the next platform of the current platform in the train running direction is determined according to the historical get-off data; and determining the congestion degree of the current compartment when the train arrives at the next platform according to the existing number of passengers, the second number of passengers getting off and the fixed number of passengers in the current compartment. The accuracy of calculating the carriage congestion degree is improved, and the riding experience of the user is improved.

Description

Technical Field

[0001] The present application relates to the field of rail transit technology, and in particular to a method for determining the congestion degree of a rail transit train compartment and a platform control device. Background Art

[0002] As an important part of public transportation, rail transit promotes its development and provides great convenience for people's travel.

[0003] However, as the total number of trips continues to grow, the volume of rail transit transportation also continues to grow. There may be an imbalance in the use of carriage resources, such as some carriages are extremely crowded and some carriages are relatively spacious, which will reduce the carrying capacity of the train.

[0004] In the related technologies, the on-board computing equipment is generally used to predict the congestion situation by the change of train weight, which has a large error; or the number of people getting off the train is predicted by the disembarkation situation of the previous train, without considering the morning peak, evening peak and holidays, etc., resulting in a large difference between the prediction result and the actual result, which lacks scientificity and rationality. In addition, the on-board computing equipment transmits data to the ground equipment, which not only increases the transmission interface, but also may cause packet loss during the transmission process, resulting in low accuracy. Summary of the invention

[0005] In an exemplary embodiment of the present application, a method for determining the crowdedness of a rail transit train carriage and a platform control device are provided to improve the accuracy of calculating the crowdedness of the carriage, so as to guide users to refer to it when riding and improve the riding experience.

[0006] According to a first aspect of an exemplary embodiment, a method for determining the crowdedness of a carriage of a rail transit train is provided, which is applied to a platform control device, and includes:

[0007] When the train stops at the current platform, if it is detected that at least one door of the current carriage is closed, a video stream sent from a video acquisition device during the period when the train stops at the current platform is acquired; wherein the video acquisition device is arranged at the current platform;

[0008] Identify the video stream and determine the first number of people getting on and off the current carriage at the current platform during the stop;

[0009] Determine the number of passengers on the train when it leaves the current platform based on the initial number of passengers on the current carriage when it enters the current platform, the first number of passengers getting on the train, and the first number of passengers getting off the train;

[0010] A neural network is set up to determine the second number of passengers getting off the current carriage at the next station in the direction of train operation according to the historical getting-off data; wherein the historical getting-off data includes the historical getting-off data of each door corresponding to the current carriage; the historical getting-off data of each door includes the number of passengers getting off the current platform in the same time period corresponding to the current time within the set number of days from the train with the same train number;

[0011] The degree of crowding in the current carriage when the train arrives at the next platform is determined based on the number of passengers already on board, the number of passengers getting off the second train, and the capacity of the current carriage.

[0012] According to a second aspect of the exemplary implementation, there is provided a station control device, including a communication unit and a processor;

[0013] The communication unit is configured to perform:

[0014] When the train stops at the current platform, if it is detected that at least one door of the current carriage is closed, a video stream sent from a video acquisition device during the period when the train stops at the current platform is acquired; wherein the video acquisition device is arranged at the current platform;

[0015] A processor configured to execute:

[0016] Identify the video stream and determine the first number of people getting on and off the current carriage at the current platform during the stop;

[0017] Determine the number of passengers on the train when it leaves the current platform based on the initial number of passengers on the current carriage when it enters the current platform, the first number of passengers getting on the train, and the first number of passengers getting off the train;

[0018] A neural network is set up to determine the second number of passengers getting off the current carriage at the next station in the direction of train operation according to the historical getting-off data; wherein the historical getting-off data includes the historical getting-off data of each door corresponding to the current carriage; the historical getting-off data of each door includes the number of passengers getting off the current platform in the same time period corresponding to the current time within the set number of days from the train with the same train number;

[0019] The degree of crowding in the current carriage when the train arrives at the next platform is determined based on the number of passengers already on board, the number of passengers getting off the second train, and the capacity of the current carriage.

[0020] According to a third aspect of an exemplary embodiment, there is provided a device for determining the crowdedness of a carriage of a rail transit train, which is applied to a platform control device, comprising:

[0021] The video acquisition unit is used to: when the train stops at the current platform, if it is detected that at least one door of the current carriage is closed, acquire the video stream sent from the video acquisition device during the period when the train stops at the current platform; wherein the video acquisition device is set at the current platform;

[0022] A video recognition unit is used to: recognize the video stream and determine the first number of people getting on and the first number of people getting off in the current carriage at the current platform during the stop period;

[0023] A processing unit is used to determine the number of passengers present in the train when it leaves the current platform according to the initial number of passengers in the current carriage when it enters the current platform, the first number of passengers getting on the carriage, and the first number of passengers getting off the carriage;

[0024] The processing unit is further used to: apply a set neural network to determine the second number of passengers getting off the current carriage at the next platform in the train running direction at the current platform according to the historical getting off data; wherein the historical getting off data includes the historical getting off data of each door corresponding to the current carriage; the historical getting off data of each door includes the number of passengers getting off the current platform within the same time period corresponding to the current time within the set number of days of the train with the same train number;

[0025] The processing unit is also used to determine the congestion level of the current carriage when the train arrives at the next platform based on the existing number of people, the second number of people getting off and the capacity of the current carriage.

[0026] According to a fourth aspect of the exemplary implementation, a computer storage medium is provided, in which computer program instructions are stored. When the instructions are executed on a computer, the computer executes the method for determining the degree of congestion of a carriage of a rail transit train as described in the first aspect.

[0027] In the embodiment of the present application, the platform control device first calculates the number of passengers on the train when it leaves the current platform, and then uses a set neural network to predict the second number of passengers getting off the current car at the next platform in the direction of the train's operation, and then determines the crowdedness of the current car when the train arrives at the next platform based on the number of passengers on the train, the second number of passengers getting off, and the capacity of the current car. In this process, when calculating the number of passengers on the train when it leaves the current platform, if it is detected that at least one door of the current car is closed when the train stays at the current platform, the video stream sent from the video acquisition device set at the current platform during the train's stay at the current platform is obtained, the video stream is identified, and the first number of passengers getting on and the first number of passengers getting off the current car at the current platform during the stay is determined; in this way, the number of passengers on the train can be calculated based on the initial number of passengers on the train when it enters the current platform, the first number of passengers getting on, and the first number of passengers getting off. In the process of applying the set neural network prediction, taking into account the different travel habits in different time periods, the historical alighting data includes the historical alighting data of each door corresponding to the current carriage in the prediction process; the historical alighting data of each door includes the number of people who got off at the current platform in the same time period corresponding to the current time within the set number of days for the train with the same train number as the train, and then the set neural network is applied to determine the second number of people who got off at the next platform in the direction of the train's operation at the current platform according to the historical alighting data. Therefore, in this process, there is no need to rely on the on-board computing device, and there is no need to increase the train system interface, which avoids the problem of packet loss during the transmission from the on-board computing device to the ground device; in addition, considering the different riding habits at different times, the application of historical data to predict the number of people getting off is more accurate, and the calculated vehicle congestion is more accurate. Improve the accuracy of calculating the congestion of the carriage, so as to guide users to refer to it when riding, and improve the riding experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0029] Figure 1 A schematic diagram exemplarily shows an application scenario of a passenger information system provided by an embodiment of the present application;

[0030] Figure 2 The structure diagram of a passenger information system provided by an embodiment of the present application is exemplarily shown;

[0031] Figure 3 An application scenario diagram of a rail transit management system provided by an embodiment of the present application is exemplarily shown;

[0032] Figure 4 A flowchart of a method for determining the crowdedness of a rail transit train compartment provided by an embodiment of the present application is exemplarily shown;

[0033] Figure 5 A flowchart of determining the number of people getting on and getting off a bus by identifying a video stream provided by an embodiment of the present application is exemplarily shown;

[0034] Figure 6a A flowchart of a method for determining the initial number of people in a current carriage when entering a current platform provided by an embodiment of the present application is exemplarily shown;

[0035] Figure 6b A schematic diagram showing a prediction process of a BP neural network provided in an embodiment of the present application is exemplified;

[0036] Figure 7 A flowchart of a method for determining the second number of people getting off a bus provided in an embodiment of the present application is exemplarily shown;

[0037] Figure 8 A method flow chart of a BP neural network training process provided by an embodiment of the present application is exemplified;

[0038] Fig. 9 A flowchart of a complete method for determining the crowdedness of a rail transit train carriage provided by an embodiment of the present application is exemplarily shown;

[0039] Fig.10 An interactive flow chart of a method for determining the crowdedness of a rail transit train compartment provided by an embodiment of the present application is exemplarily shown;

[0040] Fig.11 A schematic diagram of the structure of a device for determining the crowdedness of a rail transit train compartment provided in an embodiment of the present application is exemplarily shown;

[0041] Fig.12 A schematic diagram of the structure of a station control device provided in an embodiment of the present application is exemplified. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0043] For ease of understanding, the terms involved in the embodiments of the present application are explained below:

[0044] (1) Closed circuit television (CCTV) is mainly composed of front-end equipment and central control system. The front-end equipment usually consists of cameras, manual or electric lenses, pan / tilt heads, protective covers, monitors, alarm detectors and multi-function decoders. It establishes corresponding connections with various devices in the central control system through wired, wireless or optical fiber transmission media, such as transmitting video / audio signals and control and alarm signals.

[0045] (2) Passenger Information System (PIS), with computer system as the core, station and on-board display terminals as the medium, and relying on multimedia network technology, provides comprehensive information services to the public traveling on rail transit.

[0046] (3) Parallax refers to the difference in direction when observing the same target from two points at a certain distance. The angle between the two points as seen from the target is called the parallax angle of the two points, and the line between the two points is called the baseline. As long as the parallax angle and the length of the baseline are known, the distance between the target and the observer can be calculated.

[0047] Figure 1 A schematic diagram of an application scenario of a passenger information system provided in an embodiment of the present application. When rail transit operations are normal, the passenger information system provides the public with real-time dynamic multimedia information such as boarding instructions, train arrival and departure times, train timetables, first and last train service times, operation notices, government announcements, and advertisements; in the event of an emergency, the passenger information system is responsible for providing the public with emergency auxiliary information such as evacuation information and transfer guidance information.

[0048] Figure 2 A schematic diagram of the structure of a passenger information system provided in an embodiment of the present application. The passenger information system mainly includes the first layer of the road network level, the second layer of the railway center level, the third layer of the station level and the fourth layer of the terminal equipment. Figure 2 Each level is described separately.

[0049] First layer road network level:

[0050] The road network layer is responsible for formulating the standards for the construction of the road network passenger information system, the road network access and testing of multi-line / route passenger information systems, the layout production of the passenger information system within the road network, the editing and distribution of playlists within the road network, the organization and coordination of road network operations, and the unified operation of the passenger information system in the event of emergency response to emergencies.

[0051] Second level railway center level:

[0052] The line center layer is responsible for the operation and management of multiple lines / line passenger information systems, including the unified production and editing of media information, the unified production and distribution of playlist formats, the unified distribution and withdrawal of operation messages, the unified collection and status monitoring of equipment information, the unified collection and processing of interface data, and the feedback of operation message playback status to the publisher.

[0053] The third station level:

[0054] The station layer includes the station center system and the on-board subsystem.

[0055] The station center system realizes the operation and management of the passenger information system equipment of the station, including monitoring the status of the terminal equipment of the station, sending operation messages to the equipment of the station, receiving information and instructions from the line center layer, issuing and executing data, and feeding back the operation message playback status to the publisher.

[0056] The on-board subsystem realizes the operation management of the train passenger information system equipment, including sending and canceling operation messages to the train, receiving and processing information from the line center layer.

[0057] The fourth layer terminal equipment:

[0058] The passenger information system terminal equipment includes liquid crystal display (LCD), light emitting diode (LED), inquiry machine, etc.

[0059] LCD devices are installed at the platform level, station hall level, entrances and exits, and on trains, and are responsible for outputting the passenger information system screen at the terminal. LED devices are installed at station entrances and exits, station halls, and other areas, and are responsible for displaying the passenger information system operation information. Query machine equipment is installed at the station hall level, and is responsible for providing passengers with query services for the passenger information system operation information.

[0060] In the field of rail transit, on-board computing equipment is generally used to predict congestion by changes in train weight, ignoring the differences in individual passenger size, resulting in large errors; or, the number of people getting off the train is predicted by the disembarkation situation of the previous train, without considering the morning peak, evening peak, and holidays, resulting in a large difference between the predicted results and the actual situation, lacking scientificity and rationality. In addition, the on-board computing equipment transmits data to the ground equipment, which not only increases the transmission interface, but also may cause packet loss during the transmission process, resulting in low accuracy.

[0061] After introducing the design ideas of the embodiments of the present application, the following briefly introduces the application scenarios to which the technical solutions of the embodiments of the present application can be applied. It should be noted that the application scenarios introduced below are only used to illustrate the embodiments of the present application and are not limited. In specific implementation, the technical solutions provided by the embodiments of the present application can be flexibly applied according to actual needs.

[0062] refer to Figure 3 , shows an application scenario diagram of a rail transit management system. Taking M stations (platforms) as an example, the platform control device analyzes the video stream collected by the video acquisition device of the platform, and sends the calculated car congestion to the public network interface server, so that the car congestion can be sent to the user's terminal device through the Internet. In addition, the car congestion is displayed on the display screen of the platform so that the user can decide whether to get on the train. The platform control device can be a server in the control center, or a general term for the server in the control center and the station server.

[0063] To further illustrate the technical solution provided by the embodiment of the present application, this is described in detail below in conjunction with the accompanying drawings and specific implementation methods. Although the embodiment of the present application provides the method operation steps shown in the following embodiments or drawings, more or fewer operation steps may be included in the method based on routine or no creative labor. In the steps where there is no necessary causal relationship logically, the execution order of these steps is not limited to the execution order provided in the embodiment of the present application.

[0064] Combine the following Figure 3 The application scenario shown is shown in Figure 4 A flowchart of a method for determining the crowdedness of a rail transit train compartment is shown to illustrate the technical solution provided by the embodiment of the present application. The method of the embodiment of the present application can be applied to a platform control device, which can be a server or other device with data processing function.

[0065] In the first case, the platform control device is the central server, that is, the central server calculates the crowdedness of the carriage based on the acquired video stream; in the second case, the platform control device is a general term for the central server and the station server, and the process of calculating the crowdedness is completed by the central server and the station server. For example, the station server calculates the number of people in the current carriage when the train leaves the current platform, and then the central server calculates the crowdedness of the carriage.

[0066] In the embodiment of the present application, the first case is taken as an example for explanation, and the process of each step in the second case is the same as that of the first case and will not be repeated.

[0067] S401: When the train stops at the current platform, if it is detected that at least one door of the current carriage is closed, a video stream sent from a video acquisition device during the period when the train stops at the current platform is obtained.

[0068] Among them, the video acquisition device is set at the current platform.

[0069] S402: Identify the video stream and determine the first number of people getting on and the first number of people getting off the current carriage during the stay.

[0070] S403: Determine the number of passengers present in the train when it leaves the current platform based on the initial number of passengers when the current carriage enters the current platform, the first number of passengers getting on the train, and the first number of passengers getting off the train.

[0071] S404: Apply the set neural network to determine the second number of passengers getting off the current carriage at the next platform in the train running direction based on the historical getting off data.

[0072] Among them, the historical alighting data includes the historical alighting data of each door corresponding to the current carriage; the historical alighting data of each door includes the number of people who got off at the current platform within the same time period corresponding to the current time within the set number of days and the same train number as the train.

[0073] S405: Determine the degree of crowding of the current carriage when the train arrives at the next platform based on the existing number of passengers, the second number of passengers getting off, and the capacity of the current carriage.

[0074] In the embodiment of the present application, the platform control device first calculates the number of passengers on the train when it leaves the current platform, and then uses a set neural network to predict the second number of passengers getting off the current car at the next platform in the direction of the train's operation, and then determines the crowdedness of the current car when the train arrives at the next platform based on the number of passengers on the train, the second number of passengers getting off, and the capacity of the current car. In this process, when calculating the number of passengers on the train when it leaves the current platform, if it is detected that at least one door of the current car is closed when the train stays at the current platform, the video stream sent from the video acquisition device set at the current platform during the train's stay at the current platform is obtained, the video stream is identified, and the first number of passengers getting on and the first number of passengers getting off the current car at the current platform during the stay is determined; in this way, the number of passengers on the train can be calculated based on the initial number of passengers on the train when it enters the current platform, the first number of passengers getting on, and the first number of passengers getting off. In the process of applying the set neural network prediction, taking into account the different travel habits in different time periods, the historical alighting data includes the historical alighting data of each door corresponding to the current carriage in the prediction process; the historical alighting data of each door includes the number of people who got off at the current platform in the same time period corresponding to the current time within the set number of days for the train with the same train number as the train, and then the set neural network is applied to determine the second number of people who got off at the next platform in the direction of the train's operation at the current platform according to the historical alighting data. Therefore, in this process, there is no need to rely on the on-board computing device, and there is no need to increase the train system interface, which avoids the problem of packet loss during the transmission from the on-board computing device to the ground device; in addition, considering the different riding habits at different times, the application of historical data to predict the number of people getting off is more accurate, and the calculated vehicle congestion is more accurate. Improve the accuracy of calculating the congestion of the carriage, so as to guide users to refer to it when riding, and improve the riding experience.

[0075] Regarding S401, since the congestion degree is calculated in units of carriages, in the embodiment of the present application, one of the carriages is taken as an example for explanation, for example, the current carriage X1 is taken as an example for explanation.

[0076] Assume that the train stops at platform Z1 and detects that at least one door of carriage X1 is closed. At this time, the train may leave platform Z1 immediately. At this time, the video stream collected by the video acquisition device installed at platform Z1 during the train's stay at platform Z1 is obtained.

[0077] Among them, the video acquisition device can be a monocular camera installed in the platform. The CCTV system in the existing rail transit is usually equipped with multiple cameras, and usually in accordance with the national standard requirements, the CCTV system needs to monitor the platform location in all directions, and the height of the platform camera is set between 1.8 meters and 2.5 meters in order to capture the overall view and facial features of the platform area. Therefore, in the embodiment of the present application, the camera in the CCTV system is directly used to obtain the video stream of the platform. On the one hand, there is no need to add additional equipment, reduce the interface with the vehicle system, and reduce costs; on the other hand, the acquisition is completed at the platform, which is completely independent of the data of the on-board equipment, reduces the risk of packet loss during data transmission, and improves transmission stability.

[0078] In S402, by identifying the video stream, the first number of people getting on and the first number of people getting off the current carriage at the current platform during the stop period are determined.

[0079] For example, the train stays at platform Z1 for 1 minute. The first number of people getting on board in carriage X1 is denoted by R1, and the first number of people getting off board is denoted by D1.

[0080] Figure 5 A flowchart of determining the number of people getting on and getting off a bus by identifying a video stream is provided in an embodiment of the present application. Figure 5 At least the following steps are included:

[0081] S402-1: Determine the video image represented by the video stream.

[0082] A carriage usually includes more than one door, and the video stream here refers to the video stream within a certain range of each door included in the current carriage. By dividing and extracting the video stream, a video image represented by the video stream can be obtained.

[0083] S402-2: Perform parallax calculation on the video image to obtain a three-dimensional human head portrait included in the video stream.

[0084] Among them, by performing parallax calculation on each video image, a three-dimensional image of a person's head in the video can be formed, that is, a three-dimensional head portrait included in the video stream.

[0085] S402-3: Perform trajectory analysis on the video image, and determine the first number of people getting on and the first number of people getting off the current carriage at the current platform based on the analysis result and the three-dimensional head portrait.

[0086] Exemplarily, the trajectory analysis algorithm is applied to analyze the video image, and combined with the obtained three-dimensional head portrait, the first number of people getting on R1 and the first number of people getting off D1 in the current carriage X1 at the current platform Z1 can be determined.

[0087] Involving S403, after determining the first number of people getting on R1 and the first number of people getting off D1 in the current carriage X1 at the current platform Z1, if you want to determine the number of people RL when the train leaves the current platform Z1, you also need to know the initial number of people RC when the current carriage enters the current platform. Then, the number of people RL when the train leaves the current platform is RC = RC + R1 - D1.

[0088] For example, Figure 6a A flow chart of a method for determining the initial number of people in a current carriage when entering a current platform provided in an embodiment of the present application, Figure 6a At least the following steps are included:

[0089] S403-1: Determine all platforms before the current platform in the running direction of the train.

[0090] Assume that the current platform Z1 is the fourth platform on the train route, and there are 10 platforms on the train route. Then, there are 3 platforms before the current platform in the direction of the train, which are represented by Z01, Z02 and Z03 respectively.

[0091] S403-2: For each of all platforms, determine the number of passengers who get on and off the train during the period when the train stays at the platform.

[0092] For each of the platforms Z01, Z02 and Z03, the number of people getting on and getting off the train during the period when the train stays at the corresponding platform is determined, so that three groups of people getting on and getting off the train can be determined.

[0093] S403-3: Determine the initial number of people in the current carriage when it enters the current platform based on the number of people getting on and off the train at all platforms.

[0094] Since the train is empty when it arrives at platform Z01, the initial number of people RC in the current carriage X1 when it enters the current platform is the sum of the three groups of boarding and alighting people in this example.

[0095] In addition, the initial number of people at each platform Z01, Z02 and Z03 can be determined respectively, and finally the initial number of people in the current car when entering the current platform can be determined according to the initial number of people at platform Z03, and the same result can be obtained, which will not be repeated here.

[0096] Regarding S404, in order to calculate the congestion degree of the current carriage Z1, the above embodiment has determined the number of passengers when the train leaves the current platform, and it is also necessary to determine the number of passengers who get off the current carriage Z1 after reaching the next platform.

[0097] For example, the number of train passengers is greatly affected by time and date, such as morning and evening rush hours, weekdays and holidays. The difference in the number of passengers usually affects the difference in the number of people getting off. Since the BP neural network has strong nonlinear mapping capabilities and flexible network structure, high learning efficiency and strong generalization ability, and low computing power requirements, in the embodiment of the present application, the application sets the neural network to determine the second number of people getting off at the next platform of the current carriage in the direction of train operation based on historical getting off data.

[0098] The process of using BP neural network to predict the number of people getting off the train is as follows: the number of people getting off at each door of the same train at the same platform and the same time period from the 1st working day to the nth working day is used to predict the number of people getting off at each door of the same train at the same platform and the same time period on the n+1th working day; and the number of people getting off at each door of the same train at the same platform and the same time period from the 2nd working day to the n+1th working day is used to predict the number of people getting off at each door of the same train at the same platform and the same time period on the n+2th working day; and this cycle is repeated using incremental learning, and the same applies to holidays.

[0099] Figure 6b A schematic diagram of the prediction process of a BP neural network provided in an embodiment of the present application. A total of 3 layers are set: an input layer, a hidden layer, and an output layer, w represents weight, b represents bias, and in this example, the data from the i-th day to the i+n-1th day are used to predict the data of the i+nth day. Since the number of network layers has less influence on the prediction results than the number of neurons, the neural network is set to 3 layers. The number of neurons in the hidden layer will affect the prediction accuracy. The more neurons there are, the longer the training time is required, which may cause the training results to be too accurate and the verification set to be inaccurate. When the number of neurons is too small, a reasonable weight relationship may not be established, resulting in a large model error. There is no definite standard for the number of neurons, which is generally determined by specific data simulation training and experience. According to the on-site historical data, different numbers of inputs and different numbers of hidden layer neurons can be set for training respectively, and the training speed and accuracy can be compared to select a suitable solution.

[0100] In the prediction process, the selection of historical data affects the accuracy of the prediction. In an embodiment of the present application, the historical disembarkation data includes the historical disembarkation data of each door corresponding to the current carriage; the historical disembarkation data of each door includes the number of people getting off at the current platform within the same time period corresponding to the current time within the set number of days for the train with the same train number as the train. For example, the current time is 08:49, and the current train number is Line 1. It can be seen from the operation schedule of Line 1 that Line 1 arrives at platform Z1 at 08:49 every day, and arrives at the next platform Z2 of the current platform Z1 at 08:57. The train is set to stay at each platform for 1 minute. Taking into account the error between the actual running time of the train and the train schedule, the time period can be 4 minutes, for example, then the same time period corresponding to the current time is 08:55-08:59.

[0101] In order to improve accuracy, taking into account the different travel habits of users on weekdays and holidays, if the current time is a weekday, the number of days is set to the 30 weekdays before the current day; if the current time is a holiday, the number of days is set to the 30 holidays before the current time.

[0102] In this example, assuming that the current time is 08:49 on October 24, 2023, the historical data is the number of people getting off at each door of carriage Z1 of Line 1 at platform Z2 from 08:55 to 08:59 in the past 30 working days.

[0103] Figure 7 A flow chart of a method for determining the second number of people getting off the bus provided in an embodiment of the present application, Figure 7 At least the following steps are included.

[0104] S404-1: For each door, the historical alighting data belonging to the door is used as a training sample, and a set neural network is applied to determine the number of passengers who will alight at the next station of the door at the current platform in the direction of train operation.

[0105] Since a carriage includes more than one door, the number of people getting off at the next stop is determined for each door. Specifically, for each door, the historical getting-off data belonging to the door is used as a training sample to determine the number of people getting off at the next stop at the current platform in the direction of train operation.

[0106] S404-2: Determine the sum of the number of people getting off the bus corresponding to each door as the second number of people getting off the bus.

[0107] The BP neural network model predicts the number of people getting off the bus corresponding to each door. Therefore, the second number of people getting off the bus can be determined by calculating the sum of the number of people getting off the bus corresponding to each door.

[0108] In order to make the technical solution of this application more perfect, the training process of setting the neural network is described below. Figure 8 A method flow chart of a BP neural network training process provided in an embodiment of the present application, Figure 8 At least the following steps are included:

[0109] S801: Determine a set of training sample pairs.

[0110] The training sample pair set includes historical data of a target door, and the target door is any door of the current carriage.

[0111] S802: Normalize the set of training samples.

[0112] In order to improve the accuracy of training, the training samples can be normalized to the [0,1] interval.

[0113]

[0114] Among them, X i0 For X i The normalized value, X i is the i-th sample in the training sample, X min is the minimum value in the training sample, X max is the maximum value in the training sample.

[0115] S803: Based on the set of training sample pairs, the network parameters are adjusted continuously by applying the momentum gradient descent method in an iterative manner to train the set neural network to be trained until the error between the output value of the output layer and the expected value is less than the error threshold, and the set neural network outputted in the last round is used as the set neural network.

[0116] The neural network is set to include an input layer, a hidden layer and an output layer; the hidden layer adopts the Sigmoid function; the output layer adopts the linear function purelin function; and the training function of back propagation in the training process is the momentum gradient descent function traingdm function.

[0117] For example, the Sigmoid function takes traingdm has the function of network training and faster convergence speed. It introduces a momentum term to effectively avoid the local minimum problem that may occur in BP neural network.

[0118] Still refer to Figure 6bDuring the training process, the number of passengers getting off at each door of the same train at the same platform and in the same time period from the ith working day to the i+n-1th working day is used as input, and the number of passengers getting off at each door of the same train at the same platform and in the same time period on the n+1th working day is used as the expected output; the number of passengers getting off at each door of the same train at the same platform and in the same time period from the ith holiday to the i+n-1th holiday is used as input, and the number of passengers getting off at each door of the same train at the same platform and in the same time period on the n+1th holiday is used as the expected output. Incremental learning is used for training. The error between the output layer output and the expected output is judged. The error function is: Among them, y i is the number of passengers getting off at the i-th door of the carriage, D i is the number of people who actually got off the i-th door. Backward propagation: If the error E is greater than the preset standard E standard , the network weights and thresholds are continuously adjusted according to the momentum gradient descent algorithm, and training is continued to reduce the error between the actual output and the expected output until the error is less than the preset standard E standard When the error E is less than the preset standard E standard , it is considered that the model training is complete and can be used to predict the number of passengers getting off at the platform on the same day.

[0119] It should be noted that the detailed training process can be found in the training process of the BP neural network in the related art, which will not be described here. In addition, as computing power increases, algorithms such as machine learning or deep learning can also be used to predict the number of people getting off the bus.

[0120] In S405, the degree of crowding of the current carriage when the train arrives at the next platform is determined based on the existing number of people RC, the second number of people getting off ER and the capacity of the current carriage Q.

[0121] An exemplary process of calculating the degree of congestion in a carriage may be seen in steps A1-A2:

[0122] A1: Determine the difference between the existing number of people RC and the second number of people getting off ER as the estimated number of people in the current car RC-ER.

[0123] A2: Determine the ratio of the estimated number of people RC-ER to the number of people Q as the crowdedness of the current carriage when the train arrives at the next platform, where crowdedness J = (RC-ER) / Q.

[0124] After the congestion degree is calculated, passenger flow guidance can be achieved, for example, by the following method 1 or method 2:

[0125] Method 1: The degree of congestion and the number of the current carriage are sent to a display screen set up at the next platform in the direction of the train operation, so that passengers waiting at the next platform can decide whether to get on the train based on the degree of congestion.

[0126] Method 2: Send the congestion degree and the number of the current car to the public network interface server, so that the public network interface server sends the congestion degree and / or the number of the current car to the user's terminal device via the Internet, so that the user can understand the congestion situation in advance and arrange his own trip.

[0127] In order to make the technical solution of the present application more perfect, the following is a complete flowchart to illustrate the method for determining the crowdedness of a rail transit train in an embodiment of the present application, with reference to Fig. 9 , including at least the following steps.

[0128] S901: When the train stops at the current platform, if it is detected that at least one door of the current carriage is closed, a video stream sent from a video acquisition device during the period when the train stops at the current platform is obtained.

[0129] The implementation method of this step may refer to step S401 and will not be described in detail here.

[0130] S902: Determine the video image represented by the video stream.

[0131] The implementation method of this step may refer to step S402-1, which will not be described in detail here.

[0132] S903: Perform parallax calculation on the video image to obtain a three-dimensional human head portrait included in the video stream.

[0133] The implementation method of this step may refer to step S402-1, which will not be described in detail here.

[0134] S904: Perform trajectory analysis on the video image, and determine the first number of people getting on and the first number of people getting off the current carriage at the current platform based on the analysis result and the three-dimensional head portrait.

[0135] The implementation method of this step may refer to step S402-2, which will not be described in detail here.

[0136] S905: Determine the number of passengers present in the train when it leaves the current platform based on the initial number of passengers when the current carriage enters the current platform, the first number of passengers getting on the train, and the first number of passengers getting off the train.

[0137] The implementation method of this step may refer to step S402-3, which will not be described in detail here.

[0138] S906: For each door, the historical alighting data belonging to the door is used as a training sample, and a set neural network is applied to determine the number of passengers who will alight at the next station of the door at the current platform in the direction of train operation.

[0139] The implementation method of this step may refer to step S404-1, which will not be described in detail here.

[0140] S907: Determine the sum of the number of people getting off the bus corresponding to each door as the second number of people getting off the bus.

[0141] The implementation method of this step may refer to step S404-2, which will not be described in detail here.

[0142] S908: Determine the difference between the existing number of people and the second number of people getting off as the estimated number of people in the current carriage.

[0143] The implementation method of this step can refer to step A1 and will not be repeated here.

[0144] S909: Determine the ratio of the estimated number of people to the capacity number of people as the degree of crowding of the current carriage when the train arrives at the next platform.

[0145] The implementation method of this step can refer to step A2 and will not be repeated here.

[0146] S910: Send the degree of congestion and / or the number of the current carriage to a display screen provided at the next platform of the current platform in the direction of train movement.

[0147] The implementation method of this step can refer to method 1, which will not be repeated here.

[0148] S911: Send the degree of congestion and / or the number of the current car to the public network interface server, so that the public network interface server sends the degree of congestion and / or the number of the current car to the user's terminal device via the Internet.

[0149] The implementation method of this step can refer to method 2, which will not be repeated here.

[0150] In the embodiment of the present application, the monocular camera installed in the platform is used to collect passenger flow and complete the passenger flow analysis and calculation of each carriage of the train, and the BP neural network is used to predict the number of people getting off each carriage at the next station. After the control center integrates and analyzes various information, the congestion information of each carriage at the next station is given, and pushed to the platform display screen for waiting passengers to view. At the same time, a public network interface is provided in the control center, and various travel application software (application, APP) can obtain train location and carriage congestion information in real time through this interface, which is convenient for passengers to view in real time on the mobile terminal. It can greatly improve the passenger's riding experience and facilitate the operation to reasonably guide passenger flow.

[0151] The above embodiment is for the introduction of the first case. Fig.10 The second situation is introduced. Fig.10 An interactive flow chart of a method for determining the crowdedness of a rail transit train compartment provided in an embodiment of the present application.

[0152] S1001: When the train stops at the current platform, if the station server detects that at least one door of the current carriage is closed, it obtains the video stream sent from the video acquisition device during the period when the train stops at the current platform.

[0153] Among them, the video acquisition device is set at the current platform.

[0154] S1002: The station server identifies the video stream and determines the first number of people getting on and the first number of people getting off the current carriage during the stop.

[0155] S1003: The station server determines the number of passengers present when the train leaves the current platform based on the initial number of passengers when the current carriage enters the current platform, the first number of passengers getting on the train, and the first number of passengers getting off the train.

[0156] S1004: The station server sends the number of passengers on the train when it leaves the current platform to the central server.

[0157] S1005: The central server applies a neural network to determine the second number of passengers getting off the current carriage at the next platform in the direction of train operation based on historical getting off data.

[0158] S1006: The central server determines the congestion level of the current carriage when the train arrives at the next platform based on the existing number of passengers, the second number of passengers getting off, and the capacity of the current carriage.

[0159] The process of each step in the second case is the same as that in the first case and will not be described in detail.

[0160] like Fig.11 As shown, based on the same inventive concept, an embodiment of the present application provides a device for determining the crowdedness of a carriage of a rail transit train, which is applied to a platform control device. The device includes a video acquisition unit 111, a video recognition unit 112 and a processing unit 113.

[0161] The video acquisition unit 111 is used to: when the train stops at the current platform, if it is detected that at least one door of the current carriage is closed, obtain a video stream sent from a video acquisition device during the period when the train stops at the current platform; wherein the video acquisition device is set at the current platform;

[0162] The video recognition unit 112 is used to: recognize the video stream and determine the first number of people getting on and the first number of people getting off the current carriage at the current platform during the stop period;

[0163] The processing unit 113 is used to determine the number of passengers present in the train when it leaves the current platform according to the initial number of passengers in the current carriage when it enters the current platform, the first number of passengers getting on the carriage, and the first number of passengers getting off the carriage;

[0164] The processing unit 113 is further used to: apply a set neural network to determine the second number of passengers getting off the current carriage at the next platform in the train running direction at the current platform according to the historical getting off data; wherein the historical getting off data includes the historical getting off data of each door corresponding to the current carriage; the historical getting off data of each door includes the number of passengers getting off the current platform within the same time period corresponding to the current time within the set number of days of the train with the same train number;

[0165] The processing unit 113 is further used to determine the congestion degree of the current carriage when the train arrives at the next platform according to the existing number of people, the second number of people getting off and the capacity of the current carriage.

[0166] In an optional implementation manner, the processing unit 113 is specifically configured to:

[0167] Determine all platforms before the current platform in the running direction of the train;

[0168] For each of all platforms, determine the number of passengers who board and get off the train during the train's stay at the platform;

[0169] According to the number of people getting on and off the train at all platforms, the initial number of people in the current carriage when it enters the current platform is determined.

[0170] In an optional implementation manner, the processing unit 113 is specifically configured to:

[0171] For each door, the historical alighting data belonging to the door is used as a training sample, and a neural network is set to determine the number of alighting passengers at the next station of the door at the current platform in the direction of train operation;

[0172] The sum of the number of people getting off at each door is determined as the second number of people getting off.

[0173] In an optional implementation, after determining the congestion degree of the current carriage, the processing unit 113 is further configured to:

[0174] Sending the degree of congestion and / or the number of the current carriage to a display screen provided at the next platform of the current platform in the direction of train movement; or

[0175] The degree of congestion and / or the number of the current carriage are sent to the public network interface server, so that the public network interface server sends the degree of congestion and / or the number of the current carriage to the user's terminal device via the Internet.

[0176] In an optional implementation, the processing unit 113 is further configured to obtain a set neural network in the following manner:

[0177] Determine a training sample pair set; wherein the training sample pair set includes historical data of a target door; the target door is any door of the current carriage;

[0178] Normalize the set of training samples;

[0179] Based on the set of training sample pairs, the momentum gradient descent method is applied in an iterative manner to continuously adjust the network parameters to train the set neural network until the error between the output value of the output layer and the expected value is less than the error threshold, and the set neural network output from the last round is used as the set neural network.

[0180] In an optional implementation, the neural network is set to include an input layer, a hidden layer, and an output layer;

[0181] The hidden layer uses Sigmoid function;

[0182] The output layer uses the linear function purelin function;

[0183] The training function of back propagation in the training process is the momentum gradient descent function traingdm function.

[0184] In an optional implementation, the video recognition unit 112 is specifically configured to:

[0185] determining a video image represented by a video stream;

[0186] Perform parallax calculation on the video image to obtain a three-dimensional human head portrait included in the video stream;

[0187] Perform trajectory analysis on the video image, and determine the first number of people getting on and off the current carriage at the current platform based on the analysis results and the three-dimensional head portrait.

[0188] In an optional implementation manner, the processing unit 113 is specifically configured to:

[0189] Determine the difference between the existing number of people and the second number of people getting off as the estimated number of people in the current carriage;

[0190] The ratio of the estimated number of passengers to the capacity is determined as the crowdedness of the current carriage when the train arrives at the next platform.

[0191] In an optional implementation, the number of days is set to N days before the current time;

[0192] If the current time is a working day, then N days are working days;

[0193] If the current time is a holiday, then N days will be holidays.

[0194] Since the device is the device in the method in the embodiment of the present application, and the principle of solving the problem by the device is similar to that of the method, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0195] like Fig.12 As shown, based on the same inventive concept, an embodiment of the present application provides a station control device, including a communication unit 121 and a processor 122 .

[0196] The communication unit 121 is configured to execute:

[0197] When the train stops at the current platform, if it is detected that at least one door of the current carriage is closed, a video stream sent from a video acquisition device during the period when the train stops at the current platform is acquired; wherein the video acquisition device is arranged at the current platform;

[0198] The processor 122 is configured to execute:

[0199] Identify the video stream and determine the first number of people getting on and off the current carriage at the current platform during the stop;

[0200] Determine the number of passengers on the train when it leaves the current platform based on the initial number of passengers on the current carriage when it enters the current platform, the first number of passengers getting on the train, and the first number of passengers getting off the train;

[0201] A neural network is set up to determine the second number of passengers getting off the current carriage at the next station in the direction of train operation according to the historical getting-off data; wherein the historical getting-off data includes the historical getting-off data of each door corresponding to the current carriage; the historical getting-off data of each door includes the number of passengers getting off the current platform in the same time period corresponding to the current time within the set number of days from the train with the same train number;

[0202] The degree of crowding in the current carriage when the train arrives at the next platform is determined based on the number of passengers already on board, the number of passengers getting off the second train, and the capacity of the current carriage.

[0203] In an optional implementation, the processor 122 is further configured to determine the initial number of people in the current carriage when entering the current platform in the following manner:

[0204] Determine all platforms before the current platform in the running direction of the train;

[0205] For each of all platforms, determine the number of passengers who board and get off the train during the train's stay at the platform;

[0206] According to the number of people getting on and off the train at all platforms, the initial number of people in the current carriage when it enters the current platform is determined.

[0207] In an optional implementation, the processor 122 is specifically configured to:

[0208] For each door, the historical alighting data belonging to the door is used as a training sample, and a neural network is set to determine the number of alighting passengers at the next station of the door at the current platform in the direction of train operation;

[0209] The sum of the number of people getting off at each door is determined as the second number of people getting off.

[0210] In an optional implementation, after determining the congestion level of the current carriage, the processor 122 is further configured to:

[0211] Sending the degree of congestion and / or the number of the current carriage to a display screen provided at the next platform of the current platform in the direction of train movement; or

[0212] The degree of congestion and / or the number of the current carriage are sent to the public network interface server, so that the public network interface server sends the degree of congestion and / or the number of the current carriage to the user's terminal device via the Internet.

[0213] In an optional implementation, the processor 122 is further configured to obtain a set neural network in the following manner:

[0214] Determine a training sample pair set; wherein the training sample pair set includes historical data of a target door; the target door is any door of the current carriage;

[0215] Normalize the set of training samples;

[0216] Based on the set of training sample pairs, the momentum gradient descent method is applied in an iterative manner to continuously adjust the network parameters to train the set neural network until the error between the output value of the output layer and the expected value is less than the error threshold, and the set neural network output from the last round is used as the set neural network.

[0217] In an optional implementation, the neural network is set to include an input layer, a hidden layer, and an output layer;

[0218] The hidden layer uses Sigmoid function;

[0219] The output layer uses the linear function purelin function;

[0220] The training function of back propagation in the training process is the momentum gradient descent function traingdm function.

[0221] In an optional implementation, the processor 122 is specifically configured to:

[0222] determining a video image represented by a video stream;

[0223] Perform parallax calculation on the video image to obtain a three-dimensional human head portrait included in the video stream;

[0224] Perform trajectory analysis on the video image, and determine the first number of people getting on and off the current carriage at the current platform based on the analysis results and the three-dimensional head portrait.

[0225] In an optional implementation, the processor 122 is specifically configured to:

[0226] Determine the difference between the existing number of people and the second number of people getting off as the estimated number of people in the current carriage;

[0227] The ratio of the estimated number of passengers to the capacity is determined as the crowdedness of the current carriage when the train arrives at the next platform.

[0228] In an optional implementation, the number of days is set to N days before the current time;

[0229] If the current time is a working day, then N days are working days;

[0230] If the current time is a holiday, then N days will be holidays.

[0231] An embodiment of the present application also provides a computer storage medium, in which computer program instructions are stored. When the instructions are executed on a computer, the computer executes the steps of the above-mentioned method for determining the congestion degree of a rail transit train compartment.

[0232] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0233] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0234] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0235] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0236] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for determining the crowdedness of a rail transit train compartment, characterized in that: Applied to station control equipment, including: When the train stops at the current platform, if it is detected that at least one door of the current carriage is closed, a video stream sent from a video acquisition device during the period when the train stops at the current platform is acquired; wherein the video acquisition device is arranged at the current platform; Identify the video stream, and determine the first number of passengers getting on and the first number of passengers getting off the current carriage at the current platform during the stay; Determine the number of passengers already on the train when it leaves the current platform based on the initial number of passengers on the current carriage when it enters the current platform, the first number of passengers getting on the train, and the first number of passengers getting off the train; A neural network is set to determine the second number of passengers getting off the current carriage at the next platform at the current platform in the direction of train operation according to historical getting-off data; wherein the historical getting-off data includes historical getting-off data of each door corresponding to the current carriage; the historical getting-off data of each door includes the number of passengers getting off the current platform within the same time period corresponding to the current time within a set number of days of the train with the same train number as the train; The degree of congestion of the current carriage when the train arrives at the next platform is determined based on the existing number of people, the second number of people getting off, and the capacity of the current carriage.

2. The method according to claim 1, characterized in that: The initial number of people in the current carriage when entering the current platform is determined by: Determine all platforms before the current platform in the running direction of the train; For each of the platforms, determining the number of passengers who board and get off the train during the train's stay at the platform; The initial number of people in the current carriage when entering the current platform is determined according to the number of people getting on and off the train corresponding to all the platforms.

3. The method according to claim 1, characterized in that The application sets a neural network to determine the second number of passengers getting off the current carriage at the next platform at the current platform in the train running direction according to the historical getting off data, including: For each door, the historical alighting data belonging to the door is used as a training sample, and the set neural network is applied to determine the number of passengers who will alight at the next station of the door at the current platform in the running direction of the train; The sum of the number of people getting off the bus corresponding to each door is determined as the second number of people getting off the bus.

4. The method according to claim 1, characterized in that After determining the congestion degree of the current carriage, the method further includes: Sending the degree of congestion and / or the number of the current carriage to a display screen provided at the next station of the current station in the direction of the train running; or The degree of congestion and / or the number of the current carriage are sent to a public network interface server, so that the public network interface server sends the degree of congestion and / or the number of the current carriage to a user's terminal device via the Internet.

5. The method according to any one of claim 1, characterized in that: The setting neural network is obtained by the following method: Determine a training sample pair set; wherein the training sample pair set includes historical data of a target door; the target door is any door of the current compartment; Normalize the set of training samples; Based on the set of training sample pairs, the network parameters are continuously adjusted in an iterative manner using the momentum gradient descent method to train the set neural network until the error between the output value of the output layer and the expected value is less than an error threshold, and the set neural network outputted in the last round is used as the set neural network.

6. The method according to claim 5, characterized in that The neural network is set to include an input layer, a hidden layer and an output layer; The hidden layer adopts Sigmoid function; The output layer adopts a linear function purelin function; The training function of back propagation in the training process is the momentum gradient descent function traingdm function.

7. The method according to claim 1, characterized in that The identifying the video stream and determining the first number of passengers boarding and the first number of passengers getting off the current carriage at the current platform during the stay period includes: Determining a video image represented by the video stream; Performing parallax calculation on the video image to obtain a three-dimensional human head portrait included in the video stream; Perform trajectory analysis on the video image, and determine the first number of people getting on and the first number of people getting off the current carriage at the current platform based on the analysis result and the three-dimensional head portrait.

8. The method according to claim 1, characterized in that The determining, based on the existing number of people, the second number of people getting off the train, and the number of people allowed to get off the train in the current carriage, the crowdedness of the current carriage when the train arrives at the next platform includes: Determine the difference between the existing number of people and the second number of people getting off as the estimated number of people in the current carriage; The ratio of the estimated number of people to the capacity number of people is determined as the congestion degree of the current carriage when the train arrives at the next platform.

9. The method according to any one of claims 1 to 8, characterized in that: The set number of days is N days before the current time, where N is a positive integer; If the current time is a working day, then the N days are working days; If the current time is a holiday, then the N days are holidays.

10. A platform control device, characterized in that: including a communication unit and a processor; The communication unit is configured to execute: When the train stops at the current platform, if it is detected that at least one door of the current carriage is closed, a video stream sent from a video acquisition device during the period when the train stops at the current platform is acquired; wherein the video acquisition device is arranged at the current platform; The processor is configured to execute: Identify the video stream, and determine the first number of passengers getting on and the first number of passengers getting off the current carriage at the current platform during the stay; Determine the number of passengers already on the train when it leaves the current platform based on the initial number of passengers on the current carriage when it enters the current platform, the first number of passengers getting on the train, and the first number of passengers getting off the train; A neural network is set to determine the second number of passengers getting off the current carriage at the next platform at the current platform in the direction of train operation according to historical getting-off data; wherein the historical getting-off data includes historical getting-off data of each door corresponding to the current carriage; the historical getting-off data of each door includes the number of passengers getting off the current platform within the same time period corresponding to the current time within a set number of days of the train with the same train number as the train; The degree of congestion of the current carriage when the train arrives at the next platform is determined based on the existing number of people, the second number of people getting off, and the capacity of the current carriage.

Images