Artificial intelligence-based ticket checking system and method for people flow distribution in station

Abstract

The invention discloses a ticket checking system and method for station people flow distribution based on artificial intelligence, and relates to the technical field of distribution ticket checking. According to the artificial intelligence-based ticket checking method for station people flow distribution, ticket business information of people to be checked is obtained, a plurality of people to be checked of a plurality of train numbers are obtained through division, security check image data of each person to be checked is obtained at the same time, and an initial congestion index is obtained through analysis; the method comprises the steps of obtaining video stream data of a waiting area, inputting the video stream data of the waiting area into a pre-trained target behavior recognition model to obtain a behavior congestion correction index of each train number, obtaining station facility data, performing analysis to obtain a facility congestion correction index, and performing comprehensive analysis to obtain a passenger flow distribution index of each train number; according to the invention, intelligent shunting and ticket checking are carried out on the to-be-checked personnel of the corresponding train number through the passenger flow shunting index of each train number, so that passengers can carry out ticket checking according to the most appropriate sequence and mode, the ticket checking efficiency is improved, and efficient and smooth passenger flow of a station is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of diversion ticket checking, and in particular to an artificial intelligence-based ticket checking system and method for passenger flow diversion at a station. Background Art

[0002] With the acceleration of urbanization and the continuous growth of public transportation demand, especially in large transportation hubs such as railway stations, the density of passenger flow has reached an unprecedented level. Especially during peak hours, the passenger flow in the station has become particularly complicated. The traditional manual ticket checking, diversion and queue management methods have been unable to meet the increasing passenger demand. This not only increases operating costs, but also causes passengers to wait for too long, seriously affecting the travel experience and traffic efficiency, and may even bring safety hazards. Therefore, how to efficiently and intelligently divert passengers and improve the operational efficiency and safety of the station has become an urgent problem to be solved in railway stations. However, in recent years, the rapid development of artificial intelligence technology, especially breakthroughs in computer vision, data analysis and intelligent control, has provided new solutions for passenger flow management in stations.

[0003] Prior art includes the automatic ticket checking method, device, and system disclosed in the patent application with publication number CN117789316A. The method comprises: using an automatic ticket vending machine (TVM) to capture passenger images during the process of providing tickets, and using a first image capture device to capture images of passengers passing through, the first image capture device being deployed in the target station where the automatic ticket vending machine (TVM) is located; storing the captured passenger images in a local database at the target station; using a second image capture device to capture surveillance images at a preset distance from the entrance of the automatic ticket checking machine (AGM); matching the passenger image of the target passenger from the local database based on the surveillance image captured by the second image capture device, the target passenger being the passenger in the surveillance image; authenticating the target passenger using the matched passenger image, and controlling the automatic ticket checking machine (AGM) to allow the target passenger to pass after authentication. This application solves the technical problem of slow pedestrian flow at gates.

[0004] Based on the above solution, it is found that the limitations of the existing technology include at least the following problems: the existing technology is difficult to fully consider the dynamic behavior changes of passengers in the station, and it is easy to ignore the congestion problems caused by passenger flow and behavior, resulting in low ticket checking efficiency in the station during high-traffic periods, which in turn easily leads to queuing congestion, affecting passengers' riding experience and the station's operational efficiency. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides an artificial intelligence-based ticket checking system and method for station passenger flow diversion, which solves the problem that the existing technology does not fully consider the dynamic behavior of passengers and ignores the congestion caused by flow, thereby affecting passenger experience and station operations.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a ticket checking method for station passenger flow diversion based on artificial intelligence, comprising the following steps: after a number of people waiting for ticket checking at the station to be diverted undergo security checks, ticket information is obtained, and train number division is performed to obtain a number of people waiting for ticket checking for a number of train numbers at the station to be diverted; security inspection image data of each person waiting for ticket checking for a number of train numbers at the station to be diverted is obtained, and image analysis is performed to obtain the initial congestion index of each train number at the station to be diverted; and at the same time, the congestion index of each train number at the station to be diverted is obtained. The video stream data of the waiting area is input into the pre-trained target behavior recognition model for comprehensive analysis to obtain the behavior congestion correction index of each train at the station to be diverted; the facility data of the station to be diverted is obtained and data analysis is performed to obtain the facility congestion correction index of the station to be diverted, and a comprehensive analysis is performed based on the initial congestion index and behavior congestion correction index of each train to obtain the passenger flow diversion index of each train at the station to be diverted; based on the passenger flow diversion index of each train at the station to be diverted, intelligent diversion and ticket checking are carried out for the corresponding trains.

[0007] Furthermore, the specific formula for calculating the passenger flow diversion index of each train at the station to be diverted is as follows: ;in, The first station to be diverted The passenger flow diversion index of each train trip, The first station to be diverted The initial congestion index of each train, is the initial congestion adjustment coefficient stored in the database, The first station to be diverted The behavioral congestion correction index of each trip, is the behavior congestion adjustment coefficient stored in the database, is the facility congestion correction index of the station to be diverted, is the facility congestion adjustment coefficient stored in the database, is the collaborative congestion correction adjustment coefficient stored in the database, 1, 2, 3, ..., , The number of train trips.

[0008] Furthermore, the security inspection image data specifically includes the pixel value and two-dimensional coordinates of each pixel point in the security inspection image, and the specific steps for obtaining the initial congestion index of each train at the station to be diverted are as follows: edge recognition processing is performed on the pixel value of each pixel point in the security inspection image of each person to be ticketed at each train at the station to be diverted, and a number of luggage edge pixel points and a number of person edge pixel points in the security inspection image of each person to be ticketed at each train at the station to be diverted are obtained; the two-dimensional coordinates of each luggage edge pixel point and the two-dimensional coordinates of each person edge pixel point in the security inspection image of each person to be ticketed at each train at the station to be diverted are read, and a comprehensive analysis is performed to obtain the initial congestion index of each train at the station to be diverted.

[0009] Furthermore, the waiting area video stream data includes several frames of waiting area image data, and each frame of waiting area image data includes several waiting pixel points of the waiting area image and the waiting pixel value and waiting two-dimensional coordinates of each waiting pixel point. The target behavior recognition model is specifically a target detection and tracking model, and the specific steps of obtaining the behavior congestion correction index of each train at the station to be diverted are as follows: the waiting pixel value and waiting two-dimensional coordinates of each waiting pixel point of each frame of the waiting area image of each train at the station to be diverted are input into the pre-trained target detection and tracking model for predictive analysis to obtain a set of congestion behavior indexes for each train at the station to be diverted, namely, spatial balance index, spacing index, retention index, and behavior anomaly index; the spatial balance index, spacing index, retention index, and behavior anomaly index of each train at the station to be diverted are standardized; and the congestion behavior index set of each train at the station to be diverted after the standardized processing is comprehensively analyzed to obtain the behavior congestion correction index of each train at the station to be diverted.

[0010] Furthermore, the specific formula for calculating the behavioral congestion correction index of each train at the station to be diverted is as follows: ;in, The first station to be diverted The behavioral congestion correction index of each trip, The first station to be diverted after standardization The spatial balance index of train trips, is the balance adjustment coefficient stored in the database, The first station to be diverted after standardization The interval index of each train trip, is the spacing adjustment coefficient stored in the database, is the smoothing coefficient stored in the database, The first station to be diverted after standardization The detention index of each train trip, is the detention adjustment factor stored in the database, The first station to be diverted after standardization The behavioral abnormality index of each train trip, is the behavioral anomaly adjustment coefficient stored in the database, 1, 2, 3, ..., , The number of train trips.

[0011] Furthermore, the target detection and tracking model includes an input layer, a target detection layer, a target tracking layer, and a temporal behavior output layer, and the specific steps of obtaining the congestion behavior index set of each train at the station to be diverted are as follows: in the input layer of the target detection and tracking model, several frames of waiting area image data of each train at the station to be diverted are received and preprocessed; in the target detection layer of the target detection and tracking model, target recognition processing is performed on the preprocessed several frames of waiting area image data of each train at the station to be diverted, and each frame of waiting area image data of each train at the station to be diverted is obtained. The bounding box coordinates of several people waiting for ticket inspection are obtained; in the target tracking layer of the target detection and tracking model, the bounding box coordinates of each person waiting for ticket inspection in each frame of the waiting area image of each train at the station to be diverted are tracked; in the temporal behavior output layer of the target detection and tracking model, the bounding box coordinates of several people waiting for ticket inspection in each frame of the waiting area image of each train at the station to be diverted after multi-target tracking processing are subjected to multi-task analysis processing to obtain the spatial balance index, spacing index, retention index, and behavior abnormality index of each train at the station to be diverted, that is, the congestion behavior index set.

[0012] Furthermore, the facility data includes the number value of ticket checking machines, the channel structure index, and the equipment efficiency index, and the specific steps for obtaining the facility congestion correction index of the station to be diverted are as follows: obtain the equipment failure index and environmental factors of the station to be diverted, and normalize them in combination with the number value of ticket checking machines, the channel structure index, and the equipment efficiency index; and perform a comprehensive analysis based on the normalized number value of ticket checking machines, the channel structure index, the equipment efficiency index, the equipment failure index, and the environmental factors of the station to be diverted to obtain the facility congestion correction index of the station to be diverted.

[0013] Furthermore, the specific formula for calculating the facility congestion correction index of the station to be diverted is as follows: ;in, is the facility congestion correction index of the station to be diverted, is the normalized number of ticket checking machines at the station to be diverted, The adjustment factor for the number of machines stored in the database, is the channel structure index of the station to be diverted after normalization, is the structural adjustment coefficient stored in the database, is the equipment efficiency index of the station to be diverted after normalization, is the equipment efficiency adjustment coefficient stored in the database, is the collaborative adjustment coefficient stored in the database, is the smoothing coefficient stored in the database, is the equipment failure index of the station to be diverted after normalization, is the fault adjustment factor stored in the database, is the environmental factor of the station to be diverted after normalization, It is the environmental adjustment coefficient stored in the database.

[0014] Furthermore, the specific steps for conducting intelligent diversion and ticket checking for passengers waiting for ticket checking on the corresponding train based on the passenger flow diversion index of each train at the station to be diverted are as follows: the passenger flow diversion index of each train at the station to be diverted is judged and analyzed with the preset passenger flow diversion index threshold range; if the passenger flow diversion index of each train at the station to be diverted is lower than the lower limit of the preset passenger flow diversion index threshold range, the first diversion and ticket checking measure is taken; if the passenger flow diversion index of each train at the station to be diverted is within the preset passenger flow diversion index threshold range, the second diversion and ticket checking measure is taken; if the passenger flow diversion index of each train at the station to be diverted is higher than the upper limit of the preset passenger flow diversion index threshold range, the third diversion and ticket checking measure is taken.

[0015] The ticket checking system for passenger flow diversion at stations based on artificial intelligence includes: a data acquisition and division module, which is used to obtain ticket information after several passengers waiting for ticket inspection at the station to be diverted undergo security inspection, and divide the train numbers to obtain several passengers waiting for ticket inspection for several train numbers at the station to be diverted; a security inspection image acquisition and analysis module, which is used to obtain security inspection image data of each passenger waiting for ticket inspection for several train numbers at the station to be diverted, and perform image analysis to obtain the initial congestion index of each train number at the station to be diverted; a video acquisition and analysis module, which is used to obtain video stream data of the waiting area of ​​each train number at the station to be diverted , and input it into the pre-trained target behavior recognition model for comprehensive analysis to obtain the behavioral congestion correction index of each train at the station to be diverted; the comprehensive diversion analysis module is used to obtain the facility data of the station to be diverted, and perform data analysis to obtain the facility congestion correction index of the station to be diverted, and combine the initial congestion index and the behavioral congestion correction index of each train to perform comprehensive analysis to obtain the passenger flow diversion index of each train at the station to be diverted; the intelligent diversion ticket checking module is used to perform intelligent diversion and ticket checking on the passengers waiting for ticket checking of the corresponding train based on the passenger flow diversion index of each train at the station to be diverted.

[0016] The present invention has the following beneficial effects: (1) The ticket checking method for the station flow diversion based on artificial intelligence obtains ticket information and divides the trains after security checks on the ticket check personnel, thereby identifying the specific ticket check personnel for each train. Then, based on the corresponding security check image data, image analysis technology is used to obtain the initial congestion index, which is combined with the video stream data of the waiting area, further adjusted and corrected through the behavior recognition model, and then analyzed together with the facility data to finally generate the diversion index for each train. Passengers can check their tickets in the most appropriate order and method, thereby improving the ticket checking efficiency, especially during high-traffic periods, which significantly reduces queuing and congestion in the station, optimizes the overall operation of the station, and ensures efficient and smooth passenger flow in the station.

[0017] (2) The AI-based ticket checking method for passenger flow diversion at stations provides an intelligent and user-friendly diversion solution for stations by acquiring and analyzing security inspection image data, video streams of waiting areas, and facility data. With the help of AI technology, it can accurately analyze the actual congestion situation of each train while ensuring safety and efficiency, and make dynamic adjustments, thereby helping to improve the station's traffic capacity and safety management level, and thus ensure that every passenger in the station can complete ticket checking through the optimal channel at the right time, thereby reducing passenger dissatisfaction and improving passengers' travel experience.

[0018] (3) The AI-based station passenger flow diversion and ticket checking method generates an accurate passenger flow diversion index for each train through comprehensive analysis, and based on this, performs intelligent diversion and ticket checking on ticket inspectors, so that the station can accurately dispatch passenger flow according to actual conditions, thereby improving the scientificity and accuracy of passenger flow distribution, and effectively reducing the formation of overcrowding and high-risk areas, while ensuring the safety and comfort of passengers. This makes the station operation more flexible and efficient, able to cope with various emergencies, and improves the service quality of the station.

[0019] (4) The AI-based ticket checking system for passenger flow diversion at stations integrates multiple modules for intelligent analysis and processing, effectively improving the accuracy and automation level of the station ticket checking process. For example, the data acquisition and division module quickly completes the train division by obtaining the ticket information after security inspection, thereby ensuring that passengers waiting for ticket inspection on different trains in the station are clearly identified and accurately diverted. The combination of the security inspection image acquisition and analysis module and the video acquisition and analysis module enables the station to grasp the actual situation of the waiting area in real time, thereby evaluating the initial congestion situation and dynamically optimizing the passenger flow distribution through the behavior correction index. The comprehensive diversion analysis module comprehensively considers the facility conditions and further optimizes the station's congestion correction strategy, thereby significantly improving the station's operational efficiency and safety, thereby improving the system's response speed, and then ensuring the efficient operation of the station and being able to cope with the ticket checking needs during large passenger flows and peak hours.

[0020] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The present invention is a flow chart of the ticket checking method for station passenger flow diversion based on artificial intelligence.

[0022] Figure 2 The present invention is a flowchart of the specific steps of obtaining the behavior congestion correction index of each train at the station to be diverted in the ticket checking method for station passenger flow diversion based on artificial intelligence.

[0023] Figure 3 This is a block diagram of the ticket checking system for station passenger flow diversion based on artificial intelligence of the present invention. DETAILED DESCRIPTION

[0024] See also Figure 1The embodiment of the present invention provides a technical solution: an artificial intelligence-based ticket checking method for station passenger flow diversion, comprising the following steps: after a number of people waiting for ticket checking at a station to be diverted (such as a railway station) undergo security checks, obtaining ticket information, and dividing the train numbers to obtain a number of people waiting for ticket checking for a number of train numbers at the station to be diverted; obtaining security check image data of each person waiting for ticket checking for a number of train numbers at the station to be diverted, and performing image analysis to obtain an initial congestion index for each train number at the station to be diverted; and obtaining a waiting time index for each train number at the station to be diverted. The regional video stream data is input into the pre-trained target behavior recognition model for comprehensive analysis to obtain the behavior congestion correction index of each train at the station to be diverted; the facility data of the station to be diverted is obtained and data analysis is performed to obtain the facility congestion correction index of the station to be diverted, and the initial congestion index and behavior congestion correction index of each train are combined for comprehensive analysis to obtain the passenger flow diversion index of each train at the station to be diverted; and based on the passenger flow diversion index of each train at the station to be diverted, intelligent diversion and ticket checking are performed on the people waiting for ticket inspection of the corresponding train.

[0025] Among them, the specific steps for obtaining a number of people waiting for ticket inspection for a number of trains at the station to be diverted are as follows: perform natural language processing on the ticket information of a number of people waiting for ticket inspection at the station to be diverted (first, perform text preprocessing on the ticket information of the station to be diverted, such as removing irrelevant information, word segmentation, etc., and then use named entity recognition technology to identify key information including the train serial number from the ticket text. Usually, the train number or other related fields will be clearly marked in the ticket information. Through pattern matching and rule recognition, the train serial number is extracted, classified and integrated. Finally, the train serial number corresponding to each person waiting for ticket inspection is obtained), and the train serial numbers of the people waiting for ticket inspection at the station to be diverted are obtained; and statistical analysis is performed to obtain a number of people waiting for ticket inspection for a number of trains at the station to be diverted.

[0026] The specific formula for calculating the passenger flow diversion index for each train at the station to be diverted is as follows: ;in, The first station to be diverted The passenger flow diversion index of each train trip, The first station to be diverted The initial congestion index of each train, is the initial congestion adjustment coefficient stored in the database, The first station to be diverted The behavioral congestion correction index of each trip, is the behavior congestion adjustment coefficient stored in the database, is the facility congestion correction index of the station to be diverted, is the facility congestion adjustment coefficient stored in the database, is the collaborative congestion correction adjustment coefficient stored in the database, 1, 2, 3, ..., , The number of train trips.

[0027] It needs to be explained that the formula It is used to adjust the superimposed correction effect of the behavior congestion correction index and the facility congestion correction index to avoid the crowd diversion index being too high or too low.

[0028] 、 、 、 It can be obtained through the following steps: using historical data, combined with the initial congestion index, behavioral congestion correction index, and facility congestion correction index, statistical regression analysis is performed to quantify the specific impact of each factor on the crowd diversion index, thereby fitting the initial weight value. Secondly, the sensitivity analysis method is used to adjust the value range of each coefficient, observe its impact on the crowd diversion evaluation results, and ensure the stability and rationality of the model.

[0029] Specifically, the security inspection image data specifically includes the pixel value and two-dimensional coordinates of each pixel point in the security inspection image, and the specific steps for obtaining the initial congestion index of each train at the station to be diverted are as follows: edge recognition processing is performed on the pixel value of each pixel point in the security inspection image of each person to be checked for each train at the station to be diverted, and a number of luggage edge pixel points and a number of person edge pixel points in the security inspection image of each person to be checked for each train at the station to be diverted are obtained; the two-dimensional coordinates of each luggage edge pixel point and the two-dimensional coordinates of each person edge pixel point in the security inspection image of each person to be checked for each train at the station to be diverted are read, and a comprehensive analysis is performed (i.e., the luggage area and the person area to be checked are calculated based on the polygon area formula, and weighted processing is performed, and then summation processing is performed based on the weighted processing results) to obtain the initial congestion index of each train at the station to be diverted.

[0030] Among them, the edge recognition processing adopts Canny edge processing, that is, the original security inspection image is converted into a grayscale image and then Gaussian filtering is applied to eliminate noise interference, and potential edges are located by calculating the pixel gradient amplitude and direction; the edge contour is refined using non-maximum suppression technology, and then combined with double threshold detection to screen out strong edge pixels and connect weak edges to form a continuous edge line. After completing edge recognition, the two-dimensional coordinate set of all pixels marked as luggage edges and edges of people to be inspected in the image is extracted, that is, several luggage edge pixels and several people edge pixels in the security inspection image of each person to be inspected for each train at the station to be diverted.

[0031] In this implementation scheme, through the Canny edge processing technology, the edges of the luggage and people waiting for ticket inspection in the security inspection image can be accurately identified, noise interference can be eliminated, and high-precision image data can be obtained, thereby making the congestion assessment more accurate and ensuring that the calculation of the initial congestion index of each train is more reliable. Secondly, the two-dimensional coordinates and polygon area formula in the image data are used for weighted processing and summation, so that the crowd diversion work can be more automated and intelligent, and by comprehensively analyzing the edge information of the luggage and people in the image, the congestion index of each train can be dynamically calculated, thereby helping the station to carry out refined management. Finally, the station can obtain the preliminary congestion situation of each train in real time and make timely adjustments, thereby improving the ticket checking speed, especially in high-traffic periods or emergency situations, helping the station to ensure the safety and efficient passage of passengers and improve overall operational efficiency.

[0032] Specifically, if Figure 2 As shown, the waiting area video stream data includes several frames of waiting area image data, and each frame of waiting area image data includes several waiting pixels of the waiting area image and the waiting pixel value and waiting two-dimensional coordinates of each waiting pixel. The target behavior recognition model is specifically a target detection and tracking model, and the specific steps of obtaining the behavior congestion correction index of each train at the station to be diverted are as follows: the waiting pixel value and waiting two-dimensional coordinates of each waiting pixel of each frame of the waiting area image of each train at the station to be diverted are input into the pre-trained target detection and tracking model for prediction analysis to obtain a congestion behavior index set of each train at the station to be diverted, namely, a spatial balance index, a spacing index, a retention index, and a behavior anomaly index; the spatial balance index, spacing index, retention index, and behavior anomaly index of each train at the station to be diverted are standardized; and the congestion behavior index set of each train at the station to be diverted after the standardized processing is comprehensively analyzed to obtain the behavior congestion correction index of each train at the station to be diverted.

[0033] The specific formula for calculating the behavioral congestion correction index of each train at the station to be diverted is as follows: ;in, The first station to be diverted The behavioral congestion correction index of each trip, The first station to be diverted after standardization The spatial balance index of train trips, is the balance adjustment coefficient stored in the database, The first station to be diverted after standardization The interval index of each train trip, is the spacing adjustment coefficient stored in the database, is the smoothing coefficient stored in the database (to avoid the denominator being 0), and in this implementation example, it is set to 0.1. The first station to be diverted after standardization The detention index of each train trip, is the detention adjustment factor stored in the database, The first station to be diverted after standardization The behavioral abnormality index of each train trip, is the behavioral anomaly adjustment coefficient stored in the database, 1, 2, 3, ..., , The number of train trips.

[0034] What needs to be explained is that 、 、 、 It can be obtained through the following steps: using historical data, combined with the spatial balance index, spacing index, detention index, and behavioral anomaly index, to conduct statistical regression analysis, quantify the specific impact of each factor on the behavioral congestion correction index, and thus fit the initial weight value; secondly, using the sensitivity analysis method, adjust the value range of each coefficient, observe its impact on the behavioral congestion correction evaluation results, ensure the stability and rationality of the model, and based on the characteristics of the train and the actual situation, correct and optimize the preliminary fitted coefficients, and finally determine the coefficient value applicable to the specific train.

[0035] In this implementation, a target detection and tracking model is used to predict and analyze each frame of the waiting area image, thereby monitoring and capturing the behavioral characteristics of passengers in the waiting area in real time, thereby accurately identifying multiple key factors such as spatial balance, spacing, retention, and behavioral anomalies, thereby providing detailed data support for congestion correction, thereby ensuring that the station can dynamically adjust based on the actual situation of passenger behavior. Secondly, by standardizing the spatial balance index, spacing index, retention index, and behavioral anomaly index, and combining them with a pre-set adjustment coefficient, it is ensured that the behavioral congestion correction index of each train is derived based on actual conditions and historical data, and in this process, the specific differences between different trains are taken into account, making station management more flexible and efficient, and thus being able to dynamically optimize the flow distribution based on real-time data. Finally, by combining historical data, sensitivity analysis method, and coefficient correction process, by quantifying the impact of various factors on the behavioral congestion correction index, it can help stations more accurately predict and control the congestion in the waiting area, thereby effectively improving overall operational efficiency and passenger travel experience.

[0036] Specifically, the target detection and tracking model includes an input layer, a target detection layer, a target tracking layer, and a temporal behavior output layer, and the specific steps of obtaining the congestion behavior index set of each train at the station to be diverted are as follows: in the input layer of the target detection and tracking model, several frames of waiting area image data of each train at the station to be diverted (i.e., several waiting pixels and the waiting pixel value and waiting two-dimensional coordinates of each waiting pixel) are received and preprocessed; in the target detection layer of the target detection and tracking model, target recognition processing is performed on the preprocessed several frames of waiting area image data of each train at the station to be diverted (based on the called convolutional neural network to extract the image data). The basic features in the image, such as edges, textures, etc., are detected by target detection algorithms such as YOLO, and the bounding box coordinates of each target are output. Then, the target category in the image, i.e., the person waiting for ticket inspection, is identified. A confidence score is then calculated for each target to indicate the accuracy of the recognition. Then, duplicate detection frames are removed and the optimal detection result is retained). The bounding box coordinates of several persons waiting for ticket inspection in each frame of the waiting area image of each train at the station to be diverted are obtained. In the target tracking layer of the target detection and tracking model, the bounding box coordinates of each person waiting for ticket inspection in each frame of the waiting area image of each train at the station to be diverted are tracked (i.e., Each detected target is assigned a unique ID and its initial bounding box position is recorded. Then, the Kalman filter is used to predict the next frame position of the target, and the target of the current frame is associated with the target of the previous frame through a target matching algorithm, such as the Hungarian algorithm. Then, based on the association result, the state information of each target, such as position, speed, etc., is updated. Then, the trajectory of the target is corrected using motion prediction and the model to ensure the accuracy of tracking. Finally, the tracking information of each target is obtained, including its bounding box coordinates and ID in each frame); in the temporal behavior output layer of the target detection and tracking model, each frame of each train at the station to be diverted after multi-target tracking processing is processed. The bounding box coordinates of several people waiting for ticket inspection in the waiting area image are subjected to multi-task analysis (first, the motion state is analyzed based on the tracking information of each target to obtain the speed, acceleration, position change, direction change, etc. of each target. Then, based on this basic information, it is input into the corresponding task classification layer for analysis. For example, the spatial balance index can be used to calculate the density distribution in the waiting area by analyzing the spatial distribution of people in each frame of the image. The clustering algorithm is used to analyze the target position and calculate the uniformity of the target position. The uniformity of the spatial distribution is quantified by spatial dispersion or standard deviation. Higher values ​​indicate congestion and lower values ​​indicate uniformity.The spacing index calculates the distance between objects based on the bounding box position of each object. The spacing between adjacent objects is calculated using methods such as Euclidean distance or Manhattan distance, and the average spacing for each frame is calculated. This yields the spatial balance index, spacing index, retention index (reflecting the time passengers waiting for ticket inspection spend in waiting areas for that train, such as ticket gates and waiting areas), and behavior anomaly index (indicating behavior that is clearly out of character, such as sudden large-scale gatherings or frequent movement, indicating abnormal speed or acceleration), for each train at the station to be diverted.

[0037] Among them, the input layer is used to receive and obtain the image data of the waiting area and perform image preprocessing, such as scaling, denoising, and enhancement, to ensure that the data is suitable for subsequent processing.

[0038] The target detection layer is used to perform target recognition on each input frame image and detect the position of the person to be checked in the image.

[0039] The target tracking layer is used to track multiple targets based on the bounding box information provided by the target detection layer, and assigns a unique target ID to each target to ensure that the same target is continuously tracked in subsequent frames.

[0040] The temporal behavior output layer is used to analyze the target's position change, speed change, residence time and other information between consecutive frames, and output a set of congestion behavior indexes.

[0041] The pre-training process of the target detection and tracking model is as follows: Obtain a labeled dataset and collect images or video data of waiting areas containing people waiting for ticket inspection. This data usually comes from waiting areas at different stations and different time periods to ensure data diversity and perform annotation. Target annotation involves marking the locations of people waiting for ticket inspection in the image (bounding boxes) and assigning a unique label to each target. Behavior annotation involves also labeling behavior labels if there are behavioral analysis targets, such as whether there is a delay or congestion.

[0042] The dataset is divided into a training set and a validation set, usually 70%-80% is used for training and 20%-30% is used for validation.

[0043] Initialize the target detection and tracking model. Select a pre-designed framework or architecture, such as YOLO, Faster R-CNN, or DeepSORT. These architectures have been proven effective for target detection and tracking tasks. Initialize the optimizer, i.e., select an optimization algorithm (e.g., Adam, SGD) and set parameters such as the learning rate and momentum. Then, set the loss function. Common loss functions used in target detection include bounding box regression loss and classification loss, while in target tracking, you may need to define ID loss or similarity loss.

[0044] Training is performed based on the training set, and the number of training cycles is set (for example, training 50 times or 100 times). In each training cycle, forward propagation is performed (i.e., target detection layer: predicting the position of each target in each image, such as bounding box and category; target tracking layer, tracking the bounding box of each target and predicting its position in future frames; timing analysis, performing timing analysis on each target, combining the target's motion information, such as speed, acceleration, etc., to further optimize the prediction), calculating loss (target detection loss: classification loss, such as cross entropy loss, measuring the difference between the category predicted by the model and the true category, regression loss, such as mean square error loss, measuring the difference between the bounding box predicted by the model and the true bounding box; target tracking loss: ID loss, measuring the consistency of the target ID during the tracking process, motion loss, including target position prediction error, speed and acceleration prediction error, etc.), back propagation (back propagation calculates the gradient of each layer, updates the network weights, the loss function obtains the gradient through back propagation, and the optimizer adjusts the weights according to these gradients).

[0045] After each training cycle, an evaluation analysis is performed based on the validation set to calculate the loss value of each index in the predicted congestion behavior index set and the actual congestion behavior index set. If the loss value increases, it may indicate overfitting. At this time, it may be necessary to adjust the model structure, learning rate, training strategy, etc.

[0046] The model is evaluated. During the training process, the model's effectiveness is usually evaluated through indicators such as precision, recall rate, and F1 score. In target detection, the commonly used evaluation indicator is mAP (mean Average Precision), while in target tracking tasks, evaluation indicators such as MOTA (Multi-Object Tracking Accuracy) and MOTP (Multi-Object Tracking Precision) are used to evaluate tracking quality.

[0047] When the model evaluation results achieve the expected effect, the training process ends and a trained network model is obtained.

[0048] In this implementation, the target detection layer extracts features and identifies targets from the waiting area image by calling a convolutional neural network (such as YOLO), thereby ensuring that each person waiting for ticket inspection can be accurately located and obtain bounding box coordinates. The target tracking layer uses Kalman filtering and target matching algorithms (such as the Hungarian algorithm) to track each target in consecutive frames, thereby ensuring continuous tracking and accurate positioning of the same target, and avoiding the problem of target loss caused by occlusion or complex environment, thereby making behavior analysis more reliable and providing a solid data foundation for the subsequent calculation of congestion behavior index. Secondly, the motion state of the target between consecutive frames is analyzed through the temporal behavior output layer. The model can not only calculate the target The model can not only collect information such as speed and acceleration, but also obtain indicators such as spatial balance, spacing, and retention based on this information, so as to further refine the assessment of the congestion status of the waiting area. Finally, through the collection and training of a large amount of labeled data, the model can learn and adapt to various station environments based on historical data, thereby improving its adaptability. During the training process, the model will be evaluated based on indicators such as precision, recall rate, and F1 score to ensure the efficiency and stability of the final model, thereby providing the station with an efficient, intelligent and reliable congestion monitoring and crowd diversion solution. The station can also grasp the dynamic changes of the waiting area in real time, adjust the crowd distribution according to real-time data, and optimize the flow of people in congested areas.

[0049] Specifically, the facility data includes the number of ticket checking machines, the channel structure index, and the equipment efficiency index, and the specific steps for obtaining the facility congestion correction index of the station to be diverted are as follows: obtain the equipment failure index and environmental factors of the station to be diverted, and normalize them in combination with the number of ticket checking machines, the channel structure index, and the equipment efficiency index; and conduct a comprehensive analysis based on the normalized number of ticket checking machines, the channel structure index, the equipment efficiency index, the equipment failure index, and the environmental factors of the station to be diverted to obtain the facility congestion correction index of the station to be diverted.

[0050] The number of ticket checking machines is the sum of the number of self-service ticket checking gates and manual ticket checking gates in the station, and the number of self-service ticket checking gates and manual ticket checking gates can be obtained through regular reports of equipment management stored in the database.

[0051] The channel structure index is a measure of the capacity of the station ticket checking channel. It can be obtained by obtaining the channel width, buffer area, and effective length of the guide strip of each channel, and performing normalization processing. Based on the normalization processing, weighted processing is performed and average processing is performed. The result is the channel structure index. The channel width, buffer area, and effective length of the guide strip can all be obtained through the data in the design drawings stored in the database.

[0052] The equipment efficiency index measures the working efficiency of relevant equipment in the station (such as self-service ticket gates, security inspection equipment, etc.). It can be obtained by obtaining the single opening and closing time of each ticket gate (the time it takes for the self-service ticket gate to complete a working cycle, such as completing the ticket inspection process for one passenger) and the throughput rate (the number of passengers or luggage that the security inspection machine can handle per unit time, usually expressed as "number of passengers processed per hour"), and performing standardization. Based on the standardization, weighted processing is performed and average processing is performed. The result is the equipment efficiency index. The single opening and closing time can be obtained by obtaining several historical single opening and closing times through the equipment logs stored in the database, and the single opening and closing time is obtained by performing average processing. The throughput rate can be obtained through the counter in the device.

[0053] The specific steps for obtaining the equipment failure index are as follows: obtain the usage time value of each ticket machine (which can be obtained through the equipment log stored in the database), the fault repair time value (that is, the average repair time for each fault, and the repair time for each fault can be obtained through the equipment log stored in the database), and perform standardization. Based on the standardization, weighted processing is performed, and the resulting structure is the equipment failure index.

[0054] The specific steps for obtaining environmental factors are as follows: obtain the ground friction coefficient of each ticket gate (which can be measured by a friction testing instrument and the results uploaded to the database), light intensity (which can be measured by a light meter and the results uploaded to the database), and perform standardization. Based on the standardization, weighted processing is performed, and the result obtained is the environmental factor.

[0055] The specific formula for calculating the facility congestion correction index of the station to be diverted is as follows: ;in, is the facility congestion correction index of the station to be diverted, is the normalized number of ticket checking machines at the station to be diverted, The adjustment factor for the number of machines stored in the database, is the channel structure index of the station to be diverted after normalization, is the structural adjustment coefficient stored in the database, is the equipment efficiency index of the station to be diverted after normalization, is the equipment efficiency adjustment coefficient stored in the database, is the collaborative adjustment coefficient stored in the database, is the smoothing coefficient stored in the database (to avoid the denominator being 0), and in this implementation example, it is set to 0.1. is the equipment failure index of the station to be diverted after normalization, is the fault adjustment factor stored in the database, is the environmental factor of the station to be diverted after normalization, It is the environmental adjustment coefficient stored in the database.

[0056] It needs to be explained that the formula Used to adjust the positive synergistic effect of the number of ticket checking machines, channel structure index, and equipment efficiency index.

[0057] 、 、 、 、 、 It can be obtained through the following steps: Based on historical data, determine the initial impact weight of each variable (number of ticket checking machines, channel structure index, equipment efficiency index, equipment failure index, environmental factors) on the facility congestion correction index through statistical regression analysis. Then, use the sensitivity analysis method to adjust the value range of the coefficient to evaluate the stability and applicability of these parameters to the formula output. Next, further fit the weights through model optimization (such as machine learning algorithm or multi-objective optimization) to ensure that the formula can accurately reflect the congestion correction status of the actual station facilities.

[0058] The following is an implementation example of calculating the facility congestion correction index of the station to be diverted. The following data are available: the number of ticket checking machines, channel structure index, equipment efficiency index, equipment failure index, and environmental factors. The specific data examples are shown in Table 1: Table 1 Example of facility data for stations to be diverted Normalize the data in Table 1 to obtain Table 2: Table 2 Example of facility data of stations to be diverted after normalization Machine quantity adjustment coefficient stored in the database Approximately: 0.318; Structural adjustment coefficients stored in the database Approximately: 0.247; Equipment efficiency adjustment coefficients stored in the database Approximately: 0.284; Co-adjustment coefficients stored in the database Approximately: 0.663; Failure adjustment factors stored in the database Approximately: 0.572; Environmental adjustment factors stored in the database Approximately: 0.384; Smoothing coefficients stored in the database is 0.1; Substituting the data in Table 2 and the above coefficients into the specific formula for calculating the facility congestion correction index of the station to be diverted, we obtain: The facility congestion correction index of the station to be diverted = (ln (1 + 0.617 0.318 ×0.573 0.247 ×0.614 0.284 ) / (0.663+0.1))×(1 / (1+0.237)) 0.572 ×exp(-0.384×0.317)≈0.516.

[0059] In this implementation plan, through a comprehensive analysis of key facility indicators such as the number of ticket checking machines, channel structure, and equipment efficiency, the station's capacity can be accurately quantified. In particular, the calculation of the channel structure index takes into account multiple factors such as channel width, buffer area, and effective length of the guide belt, thereby enabling a comprehensive assessment of the station's channel capacity, thereby helping the station to reasonably divert personnel during peak hours and avoid congestion. Secondly, the analysis of the equipment failure index and environmental factors provides real-time monitoring of the facility status. The equipment failure index can evaluate the health status of the equipment based on the equipment's usage time and fault repair time, and promptly detect and handle possible equipment failures. The comprehensive evaluation of environmental factors, such as light intensity and ground friction coefficient, can help stations optimize environmental factors and improve passengers' travel experience, thereby ensuring that the station always maintains efficient operation. Finally, through the standardization, weighting and averaging of facility data, the impact of various indicators can be comprehensively evaluated, and the facility congestion correction index can be calculated based on the comprehensive results, which provides a comprehensive operational optimization reference for the station. It can help station management accurately understand the current operating status of facilities and adjust or optimize resources according to actual needs, further improve the station's operational efficiency, reduce equipment failures and congestion, and provide passengers with a smoother travel experience.

[0060] Specifically, the specific steps of intelligently diverting and checking tickets for the passengers waiting for ticket inspection on the corresponding train based on the passenger flow diversion index of each train at the station to be diverted are as follows: the passenger flow diversion index of each train at the station to be diverted is judged and analyzed with the preset passenger flow diversion index threshold range respectively; if the passenger flow diversion index of each train at the station to be diverted is lower than the lower limit of the preset passenger flow diversion index threshold range, then (for the passengers waiting for ticket inspection on this train) the first diversion and ticket inspection measure is taken (i.e., providing relevant personnel with the advice to check tickets in accordance with the normal process, i.e., the ticket inspection windows and personnel configuration are carried out as usual, and the station APP pushes to the passengers waiting for ticket inspection the navigation path of the ticket gate preset when they purchased the ticket); if the passenger flow diversion index of each train at the station to be diverted is within the preset passenger flow diversion index threshold range, then (for the passengers waiting for ticket inspection on this train) the second diversion and ticket inspection measure is taken (i.e., proposing to relevant personnel to push ticket inspection suggestions in different time periods, such as splitting the train ticket inspection time into 3 Virtual time periods with minute intervals, such as 8:00-8:03 for Group A and 8:03-8:06 for Group B, will automatically push the exclusive time period and designated ticket gate number to the app of all passengers on that train in the order of ticket purchase, such as "Your time period: 8:00-8:03, please go to Gate 5 for ticket inspection." If the passenger flow diversion index of each train at the station to be diverted exceeds the upper limit of the preset passenger flow diversion index threshold range, the third diversion and ticket inspection measure will be implemented (for passengers waiting for ticket inspection on that train) (i.e., proposing to relevant personnel the physical division of the waiting area and pre-authorization of electronic wristbands, such as using movable barriers to divide the queue into five independent units in front of the ticket gate, with a limit of 30 people per unit, and releasing the next unit after the previous unit is 80% cleared; and issuing disposable electronic wristbands with built-in NFC chips to passengers who have passed security checks. Passengers only need to hold the wristbands close to the gate sensing area to pass through, and the wristbands will automatically expire after leaving the station, thereby avoiding lag caused by network delays or code scanning failures during peak hours).

[0061] In this implementation plan, by dynamically adjusting the ticket checking process in real time according to the passenger flow diversion index, concentrated ticket checking during peak hours can be avoided, congestion and waiting time in a single time period can be reduced, and exclusive time periods and ticket gate recommendations can be provided based on the passenger flow of different trains, so as to avoid all passengers queuing at the same ticket gate and effectively disperse the passenger flow. Secondly, the ticket checking time of the train is divided into virtual time periods, and personalized ticket checking time periods and slogans are automatically pushed, which can make the ticket checking process more orderly and smooth, reduce long queues at the ticket gate due to passenger concentration, and issue pre-authorized electronic wristbands. Using NFC technology, passengers only need to place the wristband close to the gate to complete the passage, which greatly improves the passage speed and avoids delays caused by traditional scanning or ticketing problems. Finally, the ticket checking strategy is dynamically adjusted according to the passenger flow index of different trains. When there are fewer people, the regular ticket checking process is maintained. When the number of people is moderate, a time-divided strategy is adopted. During peak passenger flow periods, physical segmentation and electronic wristbands are used to ensure passage, thereby ensuring efficient and smooth ticket checking and passage.

[0062] See also Figure 3 The embodiment of the present invention provides a technical solution: an artificial intelligence-based ticket inspection system for station passenger flow diversion, comprising: a data acquisition and division module, for obtaining ticket information after a number of passengers waiting for ticket inspection at the station to be diverted undergo security inspection, and performing train division to obtain a number of passengers waiting for ticket inspection for a number of trains at the station to be diverted; a security inspection image acquisition and analysis module, for obtaining security inspection image data of each passenger waiting for ticket inspection for a number of trains at the station to be diverted, and performing image analysis to obtain an initial congestion index for each train at the station to be diverted; a video acquisition and analysis module, for obtaining the waiting time of each train at the station to be diverted The video stream data of the train area is input into the pre-trained target behavior recognition model for comprehensive analysis to obtain the behavioral congestion correction index of each train at the station to be diverted; the comprehensive diversion analysis module is used to obtain the facility data of the station to be diverted, and perform data analysis to obtain the facility congestion correction index of the station to be diverted, and combine the initial congestion index and the behavioral congestion correction index of each train to perform comprehensive analysis to obtain the passenger flow diversion index of each train at the station to be diverted; the intelligent diversion ticket checking module is used to perform intelligent diversion and ticket checking on the passengers waiting for ticket checking of the corresponding train based on the passenger flow diversion index of each train at the station to be diverted.

[0063] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0064] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. The ticket checking method for passenger flow diversion at a station based on artificial intelligence is characterized by: The following steps are involved: After the security check is completed for several persons waiting for ticket inspection at the station to be diverted, ticket information is obtained and train numbers are divided to obtain several persons waiting for ticket inspection for several train numbers at the station to be diverted; Obtain security inspection image data of each passenger waiting for ticket inspection on several trains at the station to be diverted, and perform image analysis to obtain the initial congestion index of each train at the station to be diverted; At the same time, the video stream data of the waiting area of ​​each train at the station to be diverted is obtained and input into the pre-trained target behavior recognition model for comprehensive analysis to obtain the behavior congestion correction index of each train at the station to be diverted; Obtain the facility data of the station to be diverted and perform data analysis to obtain the facility congestion correction index of the station to be diverted. Combined with the initial congestion index and behavior congestion correction index of each train, a comprehensive analysis is performed to obtain the passenger flow diversion index of each train at the station to be diverted; Based on the passenger flow diversion index of each train at the station to be diverted, intelligent diversion and ticket checking are carried out for the passengers waiting for ticket checking on the corresponding train.

2. The ticket checking method for passenger flow diversion at a station based on artificial intelligence according to claim 1 is characterized in that: The specific formula for calculating the passenger flow diversion index for each train at the station to be diverted is as follows: ； in, 、 、 The stations to be diverted are The passenger flow diversion index, initial congestion index, and behavioral congestion correction index of each train trip are is the facility congestion correction index of the station to be diverted, 、 、 、 They are the initial congestion adjustment coefficient, behavior congestion adjustment coefficient, facility congestion adjustment coefficient, and collaborative congestion correction adjustment coefficient stored in the database. 1, 2, 3, ..., , The number of train trips.

3. The ticket checking method for passenger flow diversion at a station based on artificial intelligence according to claim 1 is characterized in that: The security inspection image data specifically includes the pixel value and two-dimensional coordinates of each pixel point in the security inspection image, and the specific steps for obtaining the initial congestion index of each train at the station to be diverted are as follows: Perform edge recognition processing on the pixel value of each pixel point in the security inspection image of each person waiting for ticket inspection for each train at the station to be diverted, and obtain a number of luggage edge pixel points and a number of person edge pixel points in the security inspection image of each person waiting for ticket inspection for each train at the station to be diverted; The two-dimensional coordinates of each luggage edge pixel point and the two-dimensional coordinates of each person waiting for ticket inspection in the security inspection image of each train at the station to be diverted are read, and a comprehensive analysis is performed to obtain the initial congestion index of each train at the station to be diverted.

4. The ticket checking method for passenger flow diversion at a station based on artificial intelligence according to claim 1 is characterized in that: The waiting area video stream data includes several frames of waiting area image data, and each frame of waiting area image data includes several waiting pixels of the waiting area image, a waiting pixel value of each waiting pixel, and a waiting two-dimensional coordinate of each waiting pixel. The target behavior recognition model is specifically a target detection and tracking model, and the specific steps of obtaining the behavior congestion correction index of each train at the station to be diverted are as follows: The waiting pixel value and two-dimensional coordinates of each waiting pixel point in each frame of the waiting area image of each train at the station to be diverted are input into the pre-trained target detection and tracking model for prediction analysis, and a set of congestion behavior indices for each train at the station to be diverted, namely, spatial balance index, spacing index, retention index, and behavior anomaly index, are obtained. Standardize the spatial balance index, spacing index, retention index, and behavioral abnormality index of each train at the diversion station; A comprehensive analysis is then conducted on the standardized congestion behavior index set of each train at the station to be diverted to obtain the behavioral congestion correction index of each train at the station to be diverted.

5. The ticket checking method for passenger flow diversion at a station based on artificial intelligence according to claim 4 is characterized in that: The specific formula for calculating the behavioral congestion correction index of each train at the station to be diverted is as follows: ； in, The first station to be diverted The behavioral congestion correction index of each trip, 、 、 、 The stations to be diverted after standardization are The spatial balance index, spacing index, detention index, and behavioral abnormality index of each train trip, 、 、 、 The following are the balance adjustment coefficient, spacing adjustment coefficient, retention adjustment coefficient, and abnormal behavior adjustment coefficient stored in the database: is the smoothing coefficient stored in the database, 1, 2, 3, ..., , The number of train trips.

6. The ticket checking method for passenger flow diversion at a station based on artificial intelligence according to claim 4 is characterized in that: The target detection and tracking model includes an input layer, a target detection layer, a target tracking layer, and a temporal behavior output layer. The specific steps of obtaining the congestion behavior index set for each train at the station to be diverted are as follows: In the input layer of the target detection and tracking model, several frames of waiting area image data of each train number at the station to be diverted are received and preprocessed; In the target detection layer of the target detection and tracking model, target recognition processing is performed on the pre-processed waiting area image data of several frames for each train at the station to be diverted, and the bounding box coordinates of several people waiting for ticket inspection in each frame of the waiting area image of each train at the station to be diverted are obtained; In the target tracking layer of the target detection and tracking model, the bounding box coordinates of each person waiting for ticket inspection in each frame of the waiting area image of each train at the diversion station are tracked; In the temporal behavior output layer of the target detection and tracking model, multi-task analysis is performed on the bounding box coordinates of several people waiting for ticket inspection in each frame of the waiting area image of each train at the station to be diverted after multi-target tracking processing. The spatial balance index, spacing index, retention index, and behavioral anomaly index of each train at the station to be diverted are obtained, that is, the congestion behavior index set.

7. The ticket checking method for passenger flow diversion at a station based on artificial intelligence according to claim 1 is characterized in that: The facility data includes the number of ticket checking machines, the channel structure index, and the equipment efficiency index. The specific steps for obtaining the facility congestion correction index of the station to be diverted are as follows: Obtain the equipment failure index and environmental factors of the station to be diverted, and perform normalization processing based on the number of ticket checking machines, channel structure index, and equipment efficiency index; A comprehensive analysis is conducted based on the normalized number of ticket checking machines, channel structure index, equipment efficiency index, equipment failure index, and environmental factors at the stations to be diverted to obtain the facility congestion correction index of the stations to be diverted.

8. The ticket checking method for passenger flow diversion at a station based on artificial intelligence according to claim 7 is characterized in that: The specific formula for calculating the facility congestion correction index of the station to be diverted is as follows: ； in, is the facility congestion correction index of the station to be diverted, 、 、 、 、 They are the number of ticket checking machines, channel structure index, equipment efficiency index, equipment failure index, and environmental factors at the station to be diverted after normalization. 、 、 、 、 、 The following are the machine quantity adjustment coefficient, structure adjustment coefficient, equipment efficiency adjustment coefficient, coordination adjustment coefficient, failure adjustment coefficient, and environment adjustment coefficient stored in the database. is the smoothing coefficient stored in the database.

9. The ticket checking method for passenger flow diversion at a station based on artificial intelligence according to claim 1, characterized in that: The specific steps for intelligently diverting and checking passengers for the corresponding trains based on the passenger flow diversion index of each train at the station to be diverted are as follows: The passenger flow diversion index of each train at the station to be diverted is compared with the preset passenger flow diversion index threshold range; If the passenger flow diversion index of each train at the station to be diverted is lower than the lower limit of the preset passenger flow diversion index threshold range, the first diversion and ticket checking measure is adopted; If the passenger flow diversion index of each train at the station to be diverted is within the preset passenger flow diversion index threshold range, the second diversion and ticket checking measure is adopted; If the passenger flow diversion index of each train at the station to be diverted is higher than the upper limit of the preset passenger flow diversion index threshold range, the third diversion ticket checking measure will be taken.

10. An artificial intelligence-based ticket checking system for passenger flow diversion at a station, applying the artificial intelligence-based ticket checking method for passenger flow diversion at a station according to any one of claims 1 to 9, characterized in that: include: The data acquisition and division module is used to obtain ticket information after several people waiting for ticket inspection at the station to be diverted have undergone security inspection, and to divide the train numbers to obtain several people waiting for ticket inspection for several train numbers at the station to be diverted; The security inspection image acquisition and analysis module is used to obtain the security inspection image data of each person waiting for ticket inspection on several trains at the station to be diverted, and perform image analysis to obtain the initial congestion index of each train at the station to be diverted; The video acquisition and analysis module is used to obtain the video stream data of the waiting area of ​​each train at the station to be diverted, and input it into the pre-trained target behavior recognition model for comprehensive analysis to obtain the behavior congestion correction index of each train at the station to be diverted; The comprehensive diversion analysis module is used to obtain the facility data of the station to be diverted and perform data analysis to obtain the facility congestion correction index of the station to be diverted. It also conducts a comprehensive analysis based on the initial congestion index and behavior congestion correction index of each train to obtain the passenger flow diversion index of each train at the station to be diverted; The intelligent diversion and ticket checking module is used to perform intelligent diversion and ticket checking on passengers waiting for ticket checking on the corresponding train based on the passenger diversion index of each train at the station to be diverted.

Images