Dynamic passenger flow track analysis method and device, electronic equipment and medium

Abstract

The invention provides a dynamic passenger flow track analysis method and device, electronic equipment and a medium, and relates to the technical field of data analysis. Comprising the following steps: mapping passenger flow data of every preset time slice into each grid, and generating a plurality of OD matrixes every day; analyzing a flow direction line of the OD matrix of each time slice; connecting a plurality of flow direction lines end to end according to the slicing time to obtain a line dynamic track; searching a plurality of hot tracks based on the line dynamic track to form a hot track set; and optimizing the hot track set by using a full-connection neural network FNN. The real-time hot passenger flow route is generated by updating the passenger flow data in real time. Any threshold does not need to be set in advance, construction can be carried out adaptively according to data characteristics, and the hot route generation precision is improved. The data is processed by setting rules, the hot route generation precision is improved, the influence of noise data is effectively reduced, and the reflecting capacity of the model to the actual situation is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and particularly to a method, device, electronic device and medium for analyzing dynamic passenger flow trajectories. Background Art

[0002] With the acceleration of the global urbanization process, the urban population density has been increasing continuously, and the traffic pressure has been growing day by day. Urban managers need effective tools to monitor urban passenger flow in order to optimize traffic management, alleviate congestion, and improve the utilization efficiency of public resources.

[0003] The development of big data, artificial intelligence and Internet of Things technologies has made the collection and analysis of passenger flow data more convenient and efficient. By mining and analyzing massive data, the passenger flow patterns and trends hidden behind the data can be revealed, providing a scientific basis for decision-making.

[0004] The research on dynamic passenger flow trajectories has important application values in modern traffic management and tourism planning. By analyzing the dynamic passenger flow trajectories, the allocation of traffic resources can be optimized, the efficiency of the urban traffic system can be improved, and the development of the tourism industry can be promoted. However, the current research on dynamic passenger flow trajectories is relatively limited, especially in the tourism field.

[0005] Currently, the research on dynamic passenger flow trajectories mainly focuses on the traffic field, and the OD matrix (Origin-Destination Matrix) is used to analyze the characteristic distribution of traffic trajectories. Among them, O represents the starting point of the trajectory, and D represents the ending point of the trajectory. The OD matrix studies the characteristic distribution of traffic trajectories by recording the starting point and ending point of the passenger flow. These studies use the dynamic OD matrix for traffic flow monitoring and passenger flow characteristic analysis, and the data sources include traditional survey data and modern positioning data, especially the OD data acquisition technology based on mobile phone positioning.

[0006] Generally speaking, the existing research on dynamic trajectories mainly focuses on the research of the distribution characteristics of hot spots in the OD matrix, and the application fields are still concentrated in traffic scenarios. Although dynamic trajectory monitoring has become a research hotspot, most studies do not conduct further iterative research on the trajectory itself, but mainly focus on the characteristic research between hot spots.

[0007] If it is possible to rely on location big data to dynamically monitor the OD matrix and continuously track the trend of passenger flow trajectories in terms of time and space, so as to conduct a more in-depth and comprehensive analysis of dynamic passenger flow trajectories, assist scenic area managers in making reasonable plans, and take corresponding preventive measures, the occurrence of risk accidents can be reduced, and tourists can be ensured to enjoy the tourism experience in a safe and orderly environment. This method can not only be applied to the traffic field, but also be extended to other fields such as tourism, providing new ideas and methods for the analysis of dynamic passenger flow trajectories in multiple fields. Summary of the Invention

[0008] Objective of the Invention: To propose a dynamic passenger flow trajectory analysis method, device, electronic device and medium to effectively solve the above problems existing in the prior art.

[0009] In the first aspect of the present invention, a dynamic passenger flow trajectory analysis method is proposed, including the following steps:

[0010] Map the passenger flow data sliced at every predetermined time into each grid, and generate several OD matrices every day;

[0011] Analyze the flow direction lines of the OD matrix of each time slice;

[0012] Connect the several flow direction lines head to tail according to the slice time to obtain the line dynamic trajectory;

[0013] Based on the line dynamic trajectory, search out several popular trajectories to form a set of popular trajectories;

[0014] Use the fully connected neural network FNN to optimize the set of popular trajectories to generate an optimized set of trajectories.

[0015] In a further embodiment of the first aspect, the i-th row in the OD matrix is the departure grid, the j-th column is the arrival grid, and the expression of the OD matrix is as follows:

[0016]

[0017] In the formula, A is the OD matrix; A ij is the passenger flow from grid i to grid j within the time slice; m is the total number of departure grids; n is the total number of arrival grids.

[0018] In a further embodiment of the first aspect, within the time slice t, the passenger flow from grid i to grid j is A ij , and each passenger flow trajectory is represented as a sequence T k :

[0019] T k ={(x1,y1,t1),(x2,y2,t2),…(x p ,y p ,t p ),…,(x l ,y l ,t l )}

[0020] Wherein, (x p ,y p ) is the coordinate of the trajectory point p in the grid, obtained by linear interpolation:

[0021]

[0022] t p is the timestamp of the trajectory point p; l is the trajectory length;

[0023] The calculation formula for the timestamp of the trajectory point p is as follows:

[0024]

[0025] In the formula, t start and t end are respectively the start time and the end time of the time slice t.

[0026] In a further embodiment of the first aspect, a plurality of flow lines are connected end to end according to the start and end times of the slices to obtain a line dynamic trajectory, which specifically includes:

[0027] There are K time slices, and the trajectory set within each time slice is where T k,t is the k-th trajectory within the time slice t, and N t is the total number of trajectories within the time slice t;

[0028] For adjacent time slices t and t + 1, if the end point of a trajectory T k,t and the start point of another trajectory T m,t+1 belong to the same grid or adjacent grids in space, then they are connected into a dynamic trajectory D s :

[0029] D s ={T k1,t1 , T k2,t2 , …, T ki,ti , …, T kr,tr}

[0030] In the formula, T ki,ti is the ki-th trajectory within the time slice ti; r is the length of the dynamic trajectory;

[0031] All dynamic trajectories are represented as a set D:

[0032] D={D s |s = 1, 2, …, S}

[0033] In the formula, S is the total number of dynamic trajectories.

[0034] In a further embodiment of the first aspect, based on the line dynamic trajectory, popular trajectories are searched, which specifically includes:

[0035] Count the occurrence frequency f s of each dynamic trajectory D s :

[0036]

[0037] Select the trajectories with frequencies higher than the threshold τ as popular trajectories to form a set of popular trajectories

[0038] In a further embodiment of the first aspect, the structure of the fully connected neural network FNN can be expressed as:

[0039]

[0040] where f model is a neural network model, Θ is the parameter of the model, and X s is the feature vector input into the model;

[0041] Define the loss function of the fully connected neural network FNN

[0042]

[0043] where represents the loss function with the goal of shortening time; represents the loss function with the goal of reducing distance; α and β are their respective weights;

[0044]

[0045] where represents the optimized time; t target represents the target time; represents the optimized distance; d target represents the target distance.

[0046] In a further embodiment of the first aspect, use the trained fully connected neural network FNN to optimize the popular trajectories to generate optimized trajectories

[0047]

[0048] where Θ * is the parameter of the trained model.

[0049] In the second aspect of the present invention, a dynamic passenger flow trajectory analysis device is proposed. The device includes:

[0050] A first module for mapping the passenger flow data sliced every predetermined time into each grid to generate a number of OD matrices every day;

[0051] A second module for parsing the flow direction lines of the OD matrix of each time slice;

[0052] The third module is used to connect a number of flow lines in sequence according to the start and end times of the slices to obtain the dynamic track of the line;

[0053] The fourth module searches for a number of popular tracks based on the dynamic track of the line to form a set of popular tracks;

[0054] The fifth module uses the fully connected neural network FNN to optimize the set of popular tracks to generate an optimized set of tracks.

[0055] In the third aspect of the present invention, an electronic device is proposed, including: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the dynamic passenger flow track analysis method described in the first aspect is implemented.

[0056] In the fourth aspect of the present invention, a computer-readable storage medium is proposed, in which at least one executable instruction is stored. When the executable instruction runs on an electronic device, the electronic device executes the dynamic passenger flow track analysis method described in the first aspect.

[0057] Beneficial effects: Use the OD matrix to construct the passenger flow grid. Use real-time updated passenger flow data to generate real-time popular passenger flow routes. It is not necessary to preset any thresholds and can be adaptively constructed according to the data characteristics. Further improve the accuracy of generating popular routes. By setting rules to process the data, further improve the accuracy of generating popular routes, effectively reduce the influence of noise data, and improve the ability of the model to reflect the actual situation. Description of the Drawings

[0058] Figure 1 It is a flowchart of the dynamic passenger flow track analysis method.

[0059] Figure 2 It is a schematic diagram of the OD passenger flow matrix in the embodiment.

[0060] Figure 3 It is a visualization diagram of the OD passenger flow matrix in the embodiment. Detailed Embodiment

[0061] In the following description, a large number of specific details are given to provide a more thorough understanding of the present invention. However, it is obvious to those skilled in the art that the present invention can be implemented without one or more of these details. In other examples, some well-known technical features are not described to avoid confusion with the present invention.

[0062] This invention will use location big data as the main data source and, based on technical routes such as data, algorithms, and practice, conduct in-depth research on dynamic passenger flow trajectories to complement the research on passenger flow across the entire urban area. This invention proposes a dynamic passenger flow trajectory analysis algorithm. The dynamic passenger flow trajectory analysis algorithm conducts passenger flow analysis through the concept of the TOPn popular grid flow trajectories.

[0063] Before elaborating on the embodiments, some terms that will appear later are first explained.

[0064] OD matrix: The OD matrix (Origin-Destination Matrix) is a way to describe travel flow, used to record the number of trips or travel distances from an origin to a destination. The OD matrix is usually a two-dimensional table, where the rows and columns represent the origins and destinations respectively, and the matrix elements represent the number of trips or travel distances from the origin to the destination.

[0065] TOPn popular grid flow trajectories: Refers to the n most popular grid flow trajectories.

[0066] A dynamic passenger flow trajectory analysis method disclosed in this embodiment has a flow schematic diagram as Figure 1 shown, and the specific technical implementation steps are as follows:

[0067] Step1: Draw the OD matrix

[0068] Relying on passenger flow location data, map the passenger flow data at each moment into each grid. Taking the starting moment of each time slice as the standard, the passenger flow in each grid at this moment is the number of passengers in this time slice. The time slice of this project is 1 hour.

[0069] The expression of the OD matrix is as follows:

[0070]

[0071] A is the OD matrix; A ij is the passenger flow from grid i to grid j within the time slice; m is the total number of departure grids; n is the total number of arrival grids.

[0072] In the OD matrix, the i-th row is the departure grid, the j-th column is the arrival grid, and the value of Aij is the number of person-times from the index grid in the i-th row to the column grid in the j-th column within the 1-hour time section. The research group constructs a passenger flow OD matrix every hour according to the time slice and generates several dynamic OD matrices every day to prepare the basic data for subsequent popular trajectory lines. Figure 2 is an example of the OD passenger flow matrix. Figure 3 is Figure 2Visual display of the matrix. It can be seen from the figure that the number of people on the diagonal is the largest, indicating that there is no large passenger flow movement in this grid within the 1-hour time slice.

[0073] Step2: Search for the flow direction of popular grids

[0074] Within the time slice t, the passenger flow from grid i to grid j is A ij , and each passenger flow trajectory is represented as a sequence T k :

[0075] T k = {(x1,y1,t1),(x2,y2,t2),…9x p ,y p ,t p ),…,(x l ,y l ,t l )}

[0076] (x p ,y p ) are the coordinates of the trajectory point p in the grid, obtained by linear interpolation:

[0077]

[0078] t p is the timestamp of the trajectory point p; l is the trajectory length;

[0079] The calculation formula for the timestamp of the trajectory point p is as follows:

[0080]

[0081] t start and t end are the start time and end time of the time slice t, respectively.

[0082] In this embodiment, according to the above OD matrix drawing method and the 1-hour time slice, the research group took Jinghu District of Wuhu City as the test object and tracked the passenger flow trajectories between the grids in this area. From 6:30 in the morning to 21:30 in the evening, one grid was drawn every hour, and 16 grids were obtained.

[0083] Taking the OD matrix sliced at 12:30 on January 18, 2023 as an example, the passenger flow matrix is as shown in Table 1 below. Among them, the red number 246 indicates that within the 12:30 time slice, there are 246 people still staying in the wts4jr grid without moving, and 117 indicates that within 12:30, 117 people flowed into the wts4jr grid from the wts4m2 grid. Overall, at this time point, the number of passenger flow movements in each grid of this area is relatively concentrated.

[0084] Table 1: OD Passenger Flow Matrix Data Table within the 12:30 Time Slice

[0085]

[0086]

[0087] Next, query the TOPn short routes for the OD matrix of each time slice. Taking the OD matrix at 12:30 as an example above, when n is taken as 20, the TOP20 popular flow routes of this matrix are shown in Table 2 below. It can be seen from the table that 75% of the popular short routes are within the grid. On the one hand, it indicates that the flow rate of tourists slows down at this time point. On the other hand, the staying time of tourists in this grid exceeds 1 hour, which will serve as an important data support for subsequent passenger flow early warning, etc.

[0088] Table 2: TOP20 Popular Flow Short Routes within the 12:30 Time Slice

[0089] TOPn Outflow grid Inflow grid 1 wts4jr wts4jr 2 wts4m3 wts4m3 3 wts4mq wts4mq 4 wts4jx wts4jx 5 wts4jr wts4m2 6 wts4jz wts4jz 7 wts4m2 wts4m2 8 wts4mw wts4mw 9 wts4m2 wts4jr 10 wts4jw wts4jw 11 wts4mb wts4mb 12 wts4mc wts4mc 13 wts4m6 wts4m6 14 wts4jv wts4jv 15 wts4jx wts4jr 16 wts4md wts4md 17 wts4mf wts4mf 18 wts4m2 wts4m3 19 wts4ms wts4ms 20 wts4m3 wts4m2

[0090] And so on, conduct the search for the TOP20 popular short routes for the OD matrices of the remaining 15 time slices respectively to establish a data foundation for subsequent dynamic trajectories.

[0091] Step3: Search for Popular Dynamic Trajectories

[0092] Connect several flow routes end to end according to the start and end times of the time slices to obtain the dynamic trajectory of the route. There are K time slices, and the trajectory set within each time slice is where T k,t is the kth trajectory within the time slice t, and N t is the total number of trajectories within the time slice t;

[0093] For adjacent time slices t and t+1, if the end point of a trajectory T k,t is in the same grid or adjacent grids in space as the start point of another trajectory T m,t+1 , then connect them into a dynamic trajectory D s :

[0094] D s ={T k1,t1 ,T k2,t2 ,…,T ki,ti ,…,T kr,tr}

[0095] T ki,ti is the kith trajectory within the time slice ti; r is the length of the dynamic trajectory;

[0096] Represent all dynamic trajectories as a set D:

[0097] D = {D s | s = 1, 2, …, S}

[0098] where S is the total number of dynamic trajectories.

[0099] Based on the line dynamic trajectories, search for popular trajectories, and count the occurrence frequency f s of each dynamic trajectory D s :

[0100]

[0101] Select the trajectories with a frequency higher than the threshold τ as popular trajectories to form a set of popular trajectories

[0102] In this embodiment, relying on the TOP20 short lines of the above-mentioned time slices, splicing is performed according to the slice time, and the popular flow directions of different slice times of the media are traversed and connected end to end to obtain the popular line. Taking Example 1 in Table 3 as an example, the dynamic trajectory is expressed as: from the 6:30 wts4jv grid → 7:30 wts4jw grid → 8:30 wts4jr grid → 9:30 wts4jx grid → 10:30 wts4jw grid → 11:30 wts4jr grid → 12:30 wts4jx grid → 13:30 wts4jw grid → 14:30 wts4jr grid → 15:30 wts4m2 grid → 16:30 wts4m3 grid → 17:30 wts4jr grid → 18:30 wts4jx grid → 19:30 wts4jw grid → 20:30 wts4jr grid → 21:30 wts4jx grid.

[0103] Table 3: Set of popular trajectories based on each time slice on January 18, 2023

[0104] Example 1 2 3 4 5 6 202301180630 wts4jv wts4jv wts4jv wts4jv wts4jv wts4jv 202301180730 wts4jw wts4jw wts4jw wts4jw wts4jw wts4jw 202301180830 wts4jr wts4jr wts4jr wts4jr wts4jr wts4jr 202301180930 wts4jx wts4jx wts4jx wts4jx wts4jx wts4jx 202301181030 wts4jw wts4jw wts4jw wts4jw wts4jw wts4jw 202301181130 wts4jr wts4jr wts4jr wts4jr wts4jr wts4jr 202301181230 wts4jx wts4jx wts4jx wts4jx wts4jx wts4jx 202301181330 wts4jw wts4jw wts4jw wts4jw wts4jw wts4jw 202301181430 wts4jr wts4jr wts4jr wts4jr wts4jr wts4jr 202301181530 wts4m2 wts4m3 wts4m3 wts4m3 wts4m3 wts4m3 202301181630 wts4m3 wts4m2 wts4m2 wts4m2 wts4m2 wts4m2 202301181730 wts4jr wts4jr wts4jr wts4jr wts4jr wts4jr 202301181830 wts4jx wts4jw wts4jw wts4jx wts4jx wts4m3 202301181930 wts4jw wts4jx wts4jx wts4jw wts4jw wts4m2 202301182030 wts4jr wts4jr wts4jr wts4jr wts4jr wts4jr 202301182130 wts4jx wts4jw wts4m2 wts4jx wts4m2 wts4jw

[0105] The line is not just a straight trajectory, but there are some circular dynamic trajectories. During the test, it is found that there are many problems with the trajectory circulating between grids. The main reasons are: one is that the grids for circling are all tourist-related places. For example, a large scenic area covers multiple grids, and tourists only move in different areas of the scenic area, but it is reflected as moving between different grids in the grid; the other is related to the time slice. The time slice is short, which also causes the passenger flow to circle back and forth in a small range of grids. To solve the above problems, by adjusting the value of n and intervening in the circular interaction grids, for example, prohibiting the passenger flow from flowing back within m time slices, that is, deleting the trajectories of the line flowing back within m time slices, so as to continuously optimize the dynamic trajectory.

[0106] During the process of optimizing the dynamic trajectory, a fully connected neural network FNN can be introduced to achieve it, and its structure can be expressed as:

[0107]

[0108] f model is a neural network model, Θ are the parameters of the model, and X s is the feature vector input to the model;

[0109] Define the loss function of the fully connected neural network FNN

[0110]

[0111] represents the loss function with the goal of shortening time; represents the loss function with the goal of reducing distance; α, β are their respective weights;

[0112]

[0113] represents the optimized time; t target represents the target time; represents the optimized distance; d target represents the target distance.

[0114] Use the trained fully connected neural network FNN to optimize the popular trajectories and generate optimized trajectories

[0115]

[0116] Θ * are the trained model parameters.

[0117] In this embodiment, a dynamic passenger flow trajectory analysis device is also disclosed. The device includes a first module, a second module, a third module, a fourth module, and a fifth module. The first module is used to map the passenger flow data sliced at each predetermined time into each grid and generate several OD matrices every day. The second module is used to analyze the flow direction lines of the OD matrix of each time slice. The third module is used to connect the several flow direction lines end to end according to the slice time to obtain the line dynamic trajectory. The fourth module searches out several popular trajectories based on the line dynamic trajectory to form a popular trajectory set. The fifth module uses the fully connected neural network FNN to optimize the popular trajectory set and generate an optimized trajectory set.

[0118] The dynamic passenger flow trajectory analysis device can automatically execute all the processes of the dynamic passenger flow trajectory analysis method disclosed in the above embodiment, and will not be elaborated here.

[0119] In addition, the logical ideas behind the methods and apparatuses disclosed in the above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs.

[0120] When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0121] As described above, although the present invention has been shown and described with reference to specific preferred embodiments, it should not be construed as a limitation on the present invention itself. Various changes can be made in its form and details without departing from the spirit and scope of the present invention defined by the appended claims.

Claims

1. A method for analyzing dynamic passenger flow trajectories, characterized in that, It includes the following steps: Map the passenger flow data sliced at every predetermined time into each grid, and generate several OD matrices every day; Analyze the flow lines of the OD matrix for each time slice; Connect the several flow lines head to tail according to the slice time to obtain the line dynamic trajectory; Based on the line dynamic trajectory, search out several popular trajectories to form a set of popular trajectories; Use the fully connected neural network FNN to optimize the set of popular trajectories and generate an optimized set of trajectories.

2. The dynamic passenger flow trajectory analysis method according to claim 1, wherein In the OD matrix, the i-th row is the departure grid and the j-th column is the arrival grid. The expression of the OD matrix is as follows: where A is the OD matrix; A ij is the passenger flow from grid i to grid j within the time slice; m is the total number of origin grids; n is the total number of destination grids.

3. The dynamic passenger flow trajectory analysis method according to claim 2, wherein During the time slice t, the passenger flow from grid i to grid j is A ij , and each passenger flow trajectory is represented as a sequence T k : T k = {(x1,y1,t1),(x2,y2,t2),…(x p ,y p ,t p ),…,(x l ,y l ,t l )} where (x p , y p ) are the coordinates of the trajectory point p in the grid, obtained by linear interpolation: t p is the timestamp of the trajectory point p; l is the trajectory length; The calculation formula for the timestamp of the trajectory point p is as follows: where t start and t end are the start time and end time of the time slice t, respectively.

4. The dynamic passenger flow trajectory analysis method according to claim 3, wherein Connect the several flow lines head to tail according to the slice time to obtain the line dynamic trajectory, which specifically includes: There are K time slices, and the trajectory set within each time slice is where T k,t is the k-th trajectory within time slice t, and N t is the total number of trajectories within time slice t; For adjacent time slices t and t+1, if the end point of a trajectory T k,t is spatially in the same grid or an adjacent grid as the start point of another trajectory T m,t+1 , then they are connected into a dynamic trajectory D s : D s = {T k1,t1 , T k2,t2 , …, T ki,ti , …, T kr,tr} where T ki,ti is the ki -th trajectory within the time slice ti; r is the length of the dynamic trajectory; Represent all dynamic trajectories as a set D: D = {D s | s = 1, 2, …, S} In the formula, S is the total number of dynamic trajectories.

5. The dynamic passenger flow trajectory analysis method according to claim 1, wherein Based on the line dynamic trajectory, search out popular trajectories, which specifically includes: Count the occurrence frequency f of each dynamic trajectory D s s :​ Select the trajectories with frequencies higher than the threshold τ as popular trajectories to form a set of popular trajectories 6. The dynamic passenger flow trajectory analysis method according to claim 1, wherein The fully connected neural network FNN, whose structure can be expressed as: where f model is the neural network model, Θ is the parameter of the model, and X s is the feature vector input to the model; Define the loss function of the fully connected neural network FNN wherein, represents a loss function with the goal of shortening time; represents a loss function with the goal of reducing distance; α and β are their respective weights; In the formula, represents the optimized time; t target represents the target time; represents the optimized distance; d target represents the target distance.

7. The dynamic passenger flow trajectory analysis method according to claim 6, wherein Optimize the popular trajectory using the trained fully connected neural network FNN to generate an optimized trajectory where Θ * are the model parameters after training.

8. A dynamic passenger flow trajectory analysis device, characterized in that, It includes: The first module is used to map the passenger flow data sliced at every predetermined time into each grid and generate several OD matrices every day; The second module is used to analyze the flow lines of the OD matrix for each time slice; The third module is used to connect the several flow lines head to tail according to the slice time to obtain the line dynamic trajectory; The fourth module, based on the line dynamic trajectory, searches out several popular trajectories to form a set of popular trajectories; The fifth module uses the fully connected neural network FNN to optimize the set of popular trajectories and generate an optimized set of trajectories.

9. An electronic device, characterized in that, It includes: A processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the dynamic passenger flow trajectory analysis method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, At least one executable instruction is stored in the storage medium. When the executable instruction runs on an electronic device, the electronic device executes the dynamic passenger flow trajectory analysis method according to any one of claims 1 to 7.

Images