WO2017133627A1

WO2017133627A1 - User commuter track management method, device and system

Info

Publication number: WO2017133627A1
Application number: PCT/CN2017/072697
Authority: WO
Inventors: 韦薇; 谢思远; 范贤友; 王启贵
Original assignee: 中兴通讯股份有限公司
Priority date: 2016-02-03
Filing date: 2017-01-25
Publication date: 2017-08-10
Also published as: CN107038168A

Abstract

A method, a device and a system for user commuter track management, the method comprising: acquiring, by means of a positioning system, travel data of each user , the travel data including a stay point and a time stamp corresponding to the stay point (310); calculating, according to the travel data of each user, a user regularity track of each user (320); and outputting the user regularity track of each user to a commuter road management system (330).

Description

User commute trajectory management method, device and system

Technical field

The present disclosure relates to the field of user commuting, for example, to a user commute trajectory management method, apparatus, and system.

Background technique

Commuting commuting is the focus of urban transportation. In order to improve the traffic experience of users commuting to work, the concept of commuter transportation is proposed. Commuting traffic connects family position and working position with strong regularity. The travel rules of commuting traffic of relevant urban residents usually adopt macroscopic surveys and research. For example, the questionnaires are used, and the travel rules of commuting traffic of relevant urban residents are not timely, the sample size is insufficient, and the user's commuting cannot be fully understood and controlled.

Therefore, those skilled in the art urgently need to provide a user commute trajectory management method to solve the technical problem that the user travel rules can not fully understand and control the user's commuting by using macro investigation and research.

Summary of the invention

The present disclosure provides a user commute trajectory management method, device and system, which can fully understand and control user commuting.

The present disclosure provides a user commute trajectory management method, including:

Obtaining travel data of each user by using a positioning system, wherein the travel data includes a time stamp corresponding to the stay point and the stay point;

Calculate the user's regular trajectory for each user based on the travel data of each user;

The regularity track of each user's user is output to the commuter road management system.

Optionally, calculating a regularity track of each user according to each user's travel data includes:

The commute feature is calculated based on the travel data of each user, and the commute features include a home location, a work location, and a commute time period;

According to the commute time of the commute feature, the stay point in the travel data, and the time stamp corresponding to the stop point, The travel data belonging to the commute time period is filtered, and the stay points in the filtered travel data are sorted by time stamp to generate a commute trajectory;

Calculating a trajectory similarity between each two commute trajectories of each user, and clustering the commute trajectories of each user's two commute trajectories with a trajectory similarity greater than the first threshold to generate a cluster cluster, and determining each Representative clustering trajectories;

Selecting a frequent frequency of the commute trajectory from the stay points of all the commutation trajectories of each cluster cluster according to the selection strategy, and generating a navigation path according to the time stamp corresponding to the home location, the working position, the frequency of the commute trajectory, and the frequent points of the commute trajectory;

Calculating the trajectory similarity between the navigation path and the representative commute trajectory. If the trajectory similarity between the navigation path and the representative commute trajectory is greater than the second threshold, the navigation path is used as the commuting regular trajectory of the cluster cluster, according to the cluster cluster The regular trajectory of commuting generates a regular trajectory of the user.

Optionally, the commute time includes a commute time and a commute time, and the commute trajectory includes a commute trajectory and a commute trajectory, and the navigation path is generated according to a time stamp corresponding to the home location, the work location, the frequent commute trajectory, and the frequent commute trajectory. : If the cluster cluster is generated by the commute trajectory clustering, the work navigation route starting from the user's home position, the work position as the end point, and the frequent frequent commute trajectory is generated; and if the cluster cluster is clustered by the commute trajectory The generation generates a work-by-work navigation path starting from the user's work position, the home position as the end point, and frequent frequent commute trajectories.

Optionally, the method further includes: if the trajectory similarity between the navigation path and the representative commute trajectory is less than the second threshold, reselecting the frequent trajectory trajectory and generating a new navigation path, calculating and determining the new navigation path and representing the commute Whether the trajectory similarity between the trajectories is greater than the second threshold, if the trajectory similarity between the new navigation path and the representative commute trajectory is greater than the second threshold, the new navigation path is used as the commuting rule of the cluster cluster a trajectory, if the trajectory similarity between the new navigation path and the representative commute trajectory is not greater than the second threshold, performing a step of reselecting frequent points of the commute trajectory and generating a new navigation path until the new navigation path The trajectory similarity to the representative commute trajectory is greater than the second threshold.

Optionally, the method further includes: calculating a navigation estimation duration of the navigation path, and determining whether the navigation estimation duration of the navigation path and the representative commute are performed when determining whether the trajectory similarity between the navigation path and the representative commutation trajectory is greater than a second threshold. Whether the difference of the duration of the trajectory is smaller than the third threshold; and if the trajectory similarity between the navigation path and the representative commute trajectory is greater than the second threshold, and the navigation estimation duration and representative of the navigation path If the duration difference of the commute trajectory is less than the third threshold, the navigation path is taken as the commuting regular trajectory of the cluster cluster.

Optionally, if the clustering cluster is multiple, generating a user regular trajectory according to the commuting regular trajectory of the clustering cluster comprises: calculating a proportion of the number of commuting trajectories of each cluster cluster, and occupying a number of commuting trajectories of the cluster cluster Compared with the weight of the commuting regular trajectory as the clustering cluster, the set of commuting regular trajectories of each clustering cluster is taken as the user regular trajectory; and if the clustering cluster is one, the commuting regular trajectory according to the clustering cluster Generating the user's regular trajectory includes: using the commuting regular trajectory of the cluster cluster as the user's regular trajectory.

Optionally, if the clustering cluster is multiple, the method further includes: deleting the cluster cluster whose occupation trajectory number is smaller than the fourth threshold.

Optionally, determining a representative commute trajectory of each cluster cluster includes: calculating, by one by one, a sum of trajectory similarities between each commute trajectory and other commute trajectories in the cluster cluster, and a commuting trajectory having the largest sum of similar trajectories As a representative commuter track.

Optionally, the calculation of the trajectory similarity between the trajectories comprises: calculating the trajectory similarity between the trajectory A and the trajectory B into all the locations of the trajectory A path, and the minimum distance from the point of the trajectory B path is less than the fifth threshold The number a'; the trajectory similarity between the calculated trajectory B and the trajectory A is the number b' of the distance minimum from the point of the trajectory A path among all the points of the trajectory B path; and the calculated trajectory A and the trajectory B The trajectory similarity between (a'+b')/(a+b), a is the number of locations of the trajectory A pathway, b is the number of locations of the trajectory A pathway, and / represents the division operation.

Optionally, after generating the commutation trajectory, the method further includes: calculating, by using an outlier detection algorithm, a trajectory anomaly coefficient, and deleting a commutation trajectory whose trajectory anomaly coefficient is greater than a sixth threshold.

Optionally, before the generating the commuter trajectory, the method further includes: selecting, from all the users, the commute users that are identifiable by the home location and the work location, and generating the traversable trajectory of the identifiable commute user.

Optionally, the commuting user that can be identified from all the users by the home location and the work location includes: calculating a discrete entropy of the travel location of each user according to the travel data of each user, and the discrete entropy of the travel location is less than a seventh threshold. The user acts as a commute user; and obtains and according to the travel data of each commute user, identifies the home location and work location of each commute user, and deletes at least one unrecognizable commuter user in the home location and the work location.

The present disclosure provides a user commute trajectory management apparatus, including:

Obtaining a module, configured to obtain, by using a positioning system, travel data of each user, where the travel data includes a time stamp corresponding to the stay point and the stay point;

a processing module configured to calculate a regularity trajectory of each user according to each user's travel data;

The output module is configured to output a regular trajectory of each user to the commuter road management system.

Optionally, the processing module includes:

The commute feature calculation sub-module is configured to calculate a commute feature according to each user's travel data, and the commute feature includes a home location, a work location, and a commute time period;

The commute trajectory generation sub-module is configured to filter the travel data belonging to the commute time period according to the commute time of the commute feature, the stay point in the travel data, and the time stamp corresponding to the stop point, and select the stay point in the filtered travel data according to the time. Poke sorting generates a commute trajectory;

The commuter cluster management sub-module is configured to calculate a trajectory similarity between each two commute trajectories of each user, and a commute trajectory of each user's two commute trajectories with a trajectory similarity greater than a first threshold The class generates a cluster of clusters and determines a representative commute trajectory for each cluster of clusters;

The navigation path generation sub-module is configured to select a frequent commutation trajectory from the stay points of all the commutation trajectories of each cluster cluster according to the selection strategy, according to the home location, the work location, the frequent points of the commute trajectory, and the time stamp corresponding to the frequent points of the commute trajectory Generate a navigation path;

The commute regular track sub-module is set to calculate the trajectory similarity between the navigation path and the representative commute trajectory. If the trajectory similarity between the navigation path and the representative commute trajectory is greater than the second threshold, the navigation path is used as the clustering group for commuting The regular trajectory generates a user regular trajectory according to the commuting regular trajectory of the cluster cluster.

Optionally, the commute time includes a commute time and an off-duty commute time, the commute trajectory includes a commute trajectory and an off-duty commute trajectory, and the navigation path generation sub-module is set to generate a user if the cluster cluster is generated by the commute trajectory of the commute The home position is the starting point, the working position is the ending point, and the commuting navigation path frequently passing through the commute trajectory; and if the clustering cluster is generated by the commuting trajectory trajectory clustering, the generation starts from the user's working position, the family position is the ending point, and The commute navigation path that frequently passes through the commute trajectory in turn.

Optionally, the commute regular trajectory sub-module is further configured to: if the trajectory similarity between the navigation path and the representative commute trajectory is less than a second threshold, triggering the navigation path generating sub-module to re-select the commuting trajectory frequently Pointing and generating a new navigation path, calculating and determining whether the trajectory similarity between the new navigation path and the representative commute trajectory is greater than a second threshold, if the trajectory similarity between the new navigation path and the representative commute trajectory is greater than the The second threshold is used as a commuting regular trajectory of the cluster cluster. If the trajectory similarity between the new navigation path and the representative commute trajectory is not greater than the second threshold, the loop performs reselection. The step of frequently commuting the trajectory and generating a new navigation path until the trajectory similarity between the new navigation path and the representative commute trajectory is greater than a second threshold.

Optionally, the navigation path generation sub-module is further configured to calculate a navigation estimation duration of the navigation path, and the commute regular trajectory sub-module is further configured to: when determining whether the trajectory similarity between the navigation path and the representative commute trajectory is greater than a second threshold, Determining whether the difference between the navigation estimation duration of the navigation path and the duration of the representative commutation trajectory is less than a third threshold; and if the trajectory similarity between the navigation path and the representative commute trajectory is greater than a second threshold, and the navigation estimation duration of the navigation path is If the duration difference representing the commute trajectory is less than the third threshold, the navigation path is used as the commuting regular trajectory of the cluster cluster.

Optionally, if there are multiple clustering clusters, the commute regular trajectory sub-module is set to calculate the proportion of commuting trajectories of each cluster cluster, and the proportion of the commuting trajectories of the cluster clusters is used as the commuting regularity of the cluster clusters. The weight of the trajectory, the set of commuting regular trajectories of each cluster cluster is taken as the user regular trajectory; and if the cluster cluster is one, the commute regular trajectory sub-module is set to use the commuting regular trajectory of the cluster cluster as the user Regular trajectory.

Optionally, if the clustering cluster is multiple, the commuting regular trajectory sub-module is further configured to delete the cluster cluster whose occupation trajectory number is smaller than the fourth threshold.

Optionally, the commuter cluster management sub-module is configured to calculate the sum of the trajectory similarities between each commute trajectory and other commute trajectories in the cluster cluster one by one, and the commute trajectory with the largest sum of trajectories is used as the representative commuter trajectory .

Optionally, the processing module further includes a trajectory similarity calculation sub-module, configured to calculate that the trajectory similarity between the trajectory A and the trajectory B is the trajectory A path, and the minimum distance from the point of the trajectory B path is less than the fifth threshold. The number a'; the trajectory similarity between the calculated trajectory B and the trajectory A is the number b' of the distance minimum from the point of the trajectory A path among all the locations of the trajectory B path; and the calculated trajectory A and The trajectory similarity between the trajectories B is (a'+b')/(a+b), a is the number of locations of the trajectory A pathway, b is the number of locations of the trajectory A pathway, and / represents a division operation.

Optionally, after the commute trajectory is generated, the commute trajectory generating sub-module is further configured to calculate a trajectory abnormality coefficient by using an outlier detection algorithm, and delete a commute trajectory whose trajectory abnormal coefficient is greater than a sixth threshold.

Optionally, before the commute trajectory is generated, the commute trajectory generating sub-module is further configured to screen the commute users that are identifiable from the home location and the working location from all the users, and generate a commensurate commute trajectory of the commute user.

Optionally, the commute trajectory generation sub-module is configured to calculate a discrete entropy of each user's travel location according to the travel data of each user, and use a user whose travel location discrete entropy is less than a seventh threshold as a commute user; and obtain and according to each commute The travel data of the user identifies the home location and work location of each commute user, and deletes at least one unrecognizable commuter user of the home location and the work location.

The present disclosure provides a user commute trajectory management system, including: a positioning system, a commuter road management system, and a user commute trajectory management apparatus provided by the present disclosure; wherein the positioning system is configured to monitor user travel data, and the travel data includes a stay point and The time stamp corresponding to the stay point; the user commute trajectory management device is configured to acquire the travel data of each user through the positioning system, calculate the regularity trajectory of each user according to the travel data of each user, and output the user rule of each user The trajectory to the commuter road management system; and the commuter road management system is set to manage the commuting road according to the user's regular trajectory. The present disclosure also provides a non-transitory computer readable storage medium storing computer executable instructions arranged to perform the user commute trajectory management method described above.

The present disclosure also provides an electronic device, including:

At least one processor;

a memory communicatively coupled to the at least one processor; wherein

The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform the method described above.

The present disclosure provides a user commute trajectory management method, which acquires travel data of a user through a positioning system, such as a positioning device or a communication base station on a user terminal, calculates a user regular trajectory according to user travel data, and outputs, in the process, Without user research, the management of user commuting is enhanced, and a more comprehensive understanding and control of user commuting can be achieved, improving the user experience.

DRAWINGS

1 is a schematic structural diagram of a user commute trajectory management system according to a first embodiment of the present disclosure;

2 is a schematic structural diagram of a processing module in a first embodiment of the present disclosure;

FIG. 3 is a flowchart of a user commute trajectory management method according to a second embodiment of the present disclosure;

4 is a flow chart showing the steps of processing travel data in the second embodiment of the present disclosure;

FIG. 5 is a flowchart of a user commute trajectory management method according to a third embodiment of the present disclosure;

6 is a schematic diagram of calculation of trajectory similarity in a third embodiment of the present disclosure;

FIG. 7 is a schematic structural diagram of hardware of an electronic device according to an embodiment of the present disclosure.

detailed description

Alternative explanations of the present disclosure are now made by way of specific embodiments in conjunction with the accompanying drawings. The features of the following embodiments and embodiments may be arbitrarily combined with each other without conflict.

First embodiment

FIG. 1 is a schematic structural diagram of a user commute trajectory management system according to a first embodiment of the present disclosure. As shown in FIG. 1 , in the embodiment, the user commute trajectory management system provided by the present disclosure includes: user commute trajectory management provided by the present disclosure. The device 1, the positioning system 2, and the commuter road management system 3. The positioning system 2 is configured to monitor the travel data of the user, and the travel data includes a time stamp corresponding to the stay point and the stay point. The positioning system 2 may include a base station, a Global Positioning System (GPS), a Beidou positioning system, and a Wireless Fidelity (WiFi) or Bluetooth based positioning system. The user commute trajectory management device 1 is configured to acquire travel data of each user through the positioning system, calculate a regularity trajectory of each user according to travel data of each user, and output a regular trajectory of each user to commute road management. System 3. The commuter road management system 3 is arranged to manage commuting roads according to the user's regular trajectory.

In some embodiments, as shown in FIG. 1 , the user commute trajectory management apparatus 1 provided by the present disclosure may include an acquisition module 11 , a processing module 12 , and an output module 13 .

The obtaining module 11 is configured to acquire travel data of each user through a positioning system, wherein the travel data includes a time stamp corresponding to the stay point and the stay point.

The processing module 12 is arranged to calculate a user regularity trajectory for each user based on the travel data of each user.

The output module 13 is arranged to output a user trajectory of each user to the commuter road management system.

In some embodiments, as shown in FIG. 2, the processing module 12 in the foregoing embodiment may include: a commute feature calculation sub-module 121, a commute trajectory generation sub-module 122, a commuter cluster management sub-module 123, and a navigation path generation sub-module. 124 and the commuter regular track sub-module 125.

The commute feature calculation sub-module 121 is configured to calculate a commute feature based on travel data for each user, wherein the commute feature includes a home location, a work location, and a commute time period.

The commute trajectory generation sub-module 122 is configured to filter the travel data belonging to the commute time period according to the commute period of the commute feature, the stay point in the travel data, and the time stamp corresponding to the stay point, and select the stay point in the filtered travel data according to the time. The stamp ordering generates a commute trajectory.

The commuter cluster management sub-module 123 is configured to calculate a trajectory similarity between each two commute trajectories of each user, and gather a commute trajectory with a trajectory similarity between each two commute trajectories of each user that is greater than a first threshold. The class generates a cluster of clusters and determines the representative commuter trajectory of the cluster cluster.

The navigation path generation sub-module 124 is configured to select a frequent point of the commute trajectory from the stay points of all the commutation trajectories of the cluster cluster according to the selection strategy, and generate navigation according to the time position corresponding to the home location, the work position, the frequent points of the commute trajectory, and the frequent points of the commute trajectory. path.

The commute regular trajectory sub-module 125 is configured to calculate a trajectory similarity between the navigation path and the representative commute trajectory. If the trajectory similarity between the navigation path and the representative commute trajectory is greater than the second threshold, the navigation path is used as a clustering cluster for commuting The regular trajectory generates a user regular trajectory according to the commuting regular trajectory of the cluster cluster.

In some embodiments, the commute time period in the above embodiment may include a commute time and an off-duty commute period, the commute trajectory includes a commute trajectory and an off-duty commute trajectory, and the navigation path generation sub-module 124 may be configured to: if the cluster cluster is from work The commutation trajectory clustering generation generates a work navigation path starting from the user's home position, the work position as the end point, and the frequent frequent commute trajectory; and if the cluster cluster is generated by the commute trajectory clustering, the user is generated. The working position is the starting point, the family position is the ending point, and the off-duty navigation path that is frequently followed by the commute trajectory.

In some embodiments, the commute regular trajectory sub-module 125 in the above embodiment may further be configured to trigger the navigation path generation sub-module to re-select the commute trajectory if the trajectory similarity between the navigation path and the representative commute trajectory is less than the second threshold. Frequently generating a new navigation path, calculating and determining whether the trajectory similarity between the new navigation path and the representative commute trajectory is greater than a second threshold, if the trajectory similarity between the new navigation path and the representative commute trajectory is greater than The second threshold is used, and the new navigation path is used as a commuting regular trajectory of the cluster cluster. If the trajectory similarity between the new navigation path and the representative commute trajectory is not greater than the second threshold, the loop execution is performed again. The step of selecting a frequent commute trajectory and generating a new navigation path until the trajectory similarity between the new navigation path and the representative commute trajectory is greater than a second threshold.

In some embodiments, the navigation path generation sub-module 124 in the above embodiment may further be configured to calculate a navigation estimation duration of the navigation path, and the commute regular trajectory sub-module 125 may also be configured to determine between the navigation path and the representative commute trajectory. If the trajectory similarity is greater than the second threshold, determine whether the difference between the navigation estimation duration of the navigation path and the duration of the representative commutation trajectory is less than a third threshold; and if the trajectory similarity between the navigation path and the representative commutation trajectory is greater than the second The threshold value and the difference between the navigation estimation duration of the navigation path and the duration of the representative commutation trajectory are smaller than the third threshold, and the navigation path is used as the commuting regular trajectory of the cluster cluster.

In some embodiments, if there are multiple cluster clusters in the foregoing embodiment, the commute regular trajectory sub-module 125 is configured to calculate the proportion of the number of commute trajectories of each cluster cluster, and the proportion of the commutation trajectories of the cluster clusters As the weight of the clustering commute regular trajectory, the set of commuting regular trajectories of each cluster cluster is taken as the user regular trajectory; and if the clustering cluster is one, the commuting regular trajectory sub-module 125 is set to cluster The regular trajectory of the commute of the cluster serves as the regular trajectory of the user.

In some embodiments, if there are multiple cluster clusters in the foregoing embodiment, the commute regular trajectory sub-module 125 is further configured to delete the cluster cluster whose occupation trajectory number is smaller than the fourth threshold.

In some embodiments, the commuter cluster management sub-module 123 in the above embodiment is configured to calculate the sum of the trajectory similarities between each commute trajectory and other commute trajectories in the cluster cluster one by one, and maximize the sum of trajectory similarities. The commuter trajectory serves as a commuter trajectory.

In some embodiments, as shown in FIG. 2, the processing module 12 in the above embodiment may further include a trajectory similarity calculation sub-module 126 configured to calculate the trajectory similarity between the trajectory A and the trajectory B as all locations of the trajectory A path. The minimum distance from the point of the trajectory B path is less than the number a' of the fifth threshold; the trajectory similarity between the calculated trajectory B and the trajectory A is the smallest distance from the point of the trajectory A path in all the locations of the trajectory B path The value b' is less than the fifth threshold; and the trajectory similarity between the calculated trajectory A and the trajectory B is (a'+b')/(a+b), where a is the number of locations of the trajectory A path, b For the number of locations in the trajectory A route, / indicates the division operation.

In some embodiments, after the commute trajectory is generated, the commute trajectory generating sub-module 121 in the foregoing embodiment may further be configured to calculate a trajectory abnormal coefficient by an outlier detection algorithm, and delete a commute whose trajectory abnormal coefficient is greater than a sixth threshold. Track.

In some embodiments, the commute trajectory generation sub-module 121 in the foregoing embodiment may further be configured to filter out commute users whose home location and work location are identifiable from all users before generating the commute trajectory, and generate an identifiable commute. User's commute trajectory.

In some embodiments, the commute trajectory generation sub-module 121 in the above embodiment is configured to calculate a discrete entropy of the travel location of each user according to the travel data of each user, and use the user whose travel location discrete entropy is less than the sixth threshold as the commute user. And obtaining and determining, according to the travel data of each commute user, a home location and a work location of each commute user, and deleting at least one unrecognizable commuter user of the home location and the work location.

Second embodiment

FIG. 3 is a flowchart of a method for managing a user commute trajectory according to a second embodiment of the present disclosure.

In step 310, the travel data of each user is obtained by the positioning system, wherein the travel data includes a time stamp corresponding to the stay point and the stay point.

In step 320, the user's regularity trajectory for each user is calculated based on the travel data for each user.

In step 330, the user's regular trajectory for each user is output to the commuter road management system.

In some embodiments, as shown in FIG. 4, calculating the user regularity trajectory of each user according to the travel data of each user in the above embodiment may include the following steps.

In step 410, a commute feature is calculated based on the travel data, wherein the commute feature includes a home location, a work location, and a commute time period.

In step 420, according to the commute time of the commute feature, the stay point in the travel data, and the timestamp corresponding to the stop point, the travel data belonging to the commute time period is filtered, and the stay points in the filtered travel data are sorted according to the time stamp. Commuter trajectory.

In step 430, the trajectory similarity between each two commute trajectories of each user is calculated, the commute trajectory with the trajectory similarity greater than the first threshold is clustered to generate a cluster cluster, and the representative commute trajectory of the cluster cluster is determined.

In step 440, the frequent points of the commute trajectory are selected from the stay points of all the commutation trajectories of the cluster cluster according to the selection strategy, and the navigation path is generated according to the time stamp corresponding to the home position, the working position, the frequent points of the commute trajectory and the frequent points of the commute trajectory.

In step 450, the trajectory similarity between the navigation path and the representative commuter trajectory is calculated. If the trajectory similarity between the navigation path and the representative commute trajectory is greater than the second threshold, the navigation path is used as the commuting regular trajectory of the cluster cluster. According to the regularity trajectory of the clustering cluster, the user's regular trajectory is generated.

In some embodiments, the commute time period in the above embodiment includes commuting time and commuting to work. During the time period, the commute trajectory includes a commute trajectory and an off-duty commute trajectory, and step 440 includes: if the cluster cluster is generated by the commuting trajectory of the commute, the generation starts from the user's home position, the working position is the end point, and the commute trajectory is frequently followed. The work navigation path; and if the cluster cluster is generated by the commute trajectory clustering, the work navigation route starting from the user's work position, the family position as the end point, and the frequent commute trajectory in turn is generated.

In some embodiments, the step 450 in the above embodiment may further include: if the trajectory similarity between the navigation path and the representative commute trajectory is less than the second threshold, reselecting the frequent points of the commute trajectory and generating a new navigation path, and calculating Determining whether the trajectory similarity between the new navigation path and the representative commute trajectory is greater than a second threshold. If the trajectory similarity between the new navigation path and the representative commute trajectory is greater than the second threshold, the new navigation path is used as a cluster cluster. The commuting regularity trajectory, if the trajectory similarity between the new navigation path and the representative commute trajectory is not greater than the second threshold, the loop performs the steps of reselecting the frequent points of the commute trajectory and generating a new navigation path until the new navigation path and The trajectory similarity between the representative commute trajectories is greater than the second threshold.

In some embodiments, the method in the foregoing embodiment may further include: calculating a navigation estimation duration of the navigation path, and determining whether the navigation path is greater than whether the trajectory similarity between the navigation path and the representative commutation trajectory is greater than a second threshold. Whether the difference between the estimated duration and the duration of the commute trajectory is less than a third threshold; and if the trajectory similarity between the navigation path and the representative commute trajectory is greater than a second threshold, and the navigation estimation duration of the navigation path and the representative commute trajectory If the time difference is smaller than the third threshold, the navigation path is used as the commuting regular track of the cluster.

In some embodiments, if the clustering clusters in the foregoing embodiment are multiple, generating a user regular trajectory according to the commuting regular trajectory of the clustering cluster may include: calculating a proportion of the number of commuting trajectories of each cluster cluster, The number of commuting trajectories of the cluster clusters is the weight of the regular trajectory of the commuting clusters, and the set of commuting regular trajectories of each cluster cluster is taken as the regular trajectory of the user; and if the clustering cluster is one, according to The regularity trajectory of the commuting regularity trajectory of the clustering cluster includes the regular trajectory of the commuting of the clustering cluster as the regular trajectory of the user.

The number of commute trajectories of each cluster cluster is a ratio of the number of commute trajectories in each cluster cluster to the number of all commute trajectories.

In some embodiments, if the clustering clusters in the foregoing embodiment are multiple, the user commute trajectory management method may further include: deleting cluster clusters whose occupation trajectory number is smaller than a fourth threshold.

In some embodiments, the representative commute trajectory for determining each cluster cluster in the above embodiment may be packaged. Including: calculating the sum of the trajectories similarity between each commute trajectory and other commuting trajectories in the cluster cluster one by one, and taking the commute trajectory with the largest sum of trajectories similarity as the representative commuter trajectory.

In some embodiments, the calculation of the trajectory similarity between the trajectories in the above embodiment may include: calculating the trajectory similarity between the trajectory A and the trajectory B as the distance from the point of the trajectory B path among all the locations of the trajectory A path The minimum value is less than the number of the fifth threshold a'; the trajectory similarity between the calculated trajectory B and the trajectory A is the number b' of the distance minimum from the point of the trajectory A path among all the locations of the trajectory B path And the trajectory similarity between the calculated trajectory A and the trajectory B is (a'+b')/(a+b), where a is the number of locations of the trajectory A pathway, and b is the number of locations of the trajectory A pathway, / Indicates the division operation.

In some embodiments, after the method of generating the commutation trajectory, the method in the foregoing embodiment may further include: calculating a trajectory abnormality coefficient by using an outlier detection algorithm, and deleting a commutation trajectory whose trajectory abnormal coefficient is greater than a sixth threshold.

In some embodiments, the method in the foregoing embodiment may further include: selecting, from all users, a commute user that is identifiable by the home location and the work location, and generating a commensurable commute trajectory of the commute user.

In some embodiments, the commuting user that can identify the home location and the work location from all the users in the foregoing embodiment may include: calculating the discrete entropy of each user's travel location according to the travel data of each user, and traveling a user whose location discrete entropy is less than a sixth threshold is used as a commute user; and obtains and identifies a home location and a work location of each commute user according to travel data of each commute user, and if not recognized, deletes the home location and the work At least one unrecognizable commuter user in the location.

Third embodiment

The relevant commuter survey method is to use the questionnaire. This method has poor timeliness, insufficient sample size, and poor commutation statistics.

This embodiment provides a user commute trajectory management system, which may include: a storage device (or a distributed storage device), a collection device, a positioning system, a user commute trajectory management device provided by the present disclosure, and a commuter road management system.

The obtaining of the commute information by the positioning system involved in the present application may include: statistic of commuting characteristics, such as home position, working position, commute duration, and radius of gyration, by means of GPS, Beidou positioning system or mobile network positioning data; Card (Integrated Circuit Card, IC) The data obtains the commuter travel characteristic information of the residents, such as commuter travel time, commuter travel distance and transfer characteristics, etc.; or, obtains commute information by monitoring the live broadcast of the camera at the early peak and the late peak hours.

Optionally, the collecting device is configured to collect GPS data, data of a Beidou positioning system, mobile network location signaling data, or call bill data, and extract a user identifier, a time stamp, and a location field. For GPS data and Beidou positioning system data, the location field may be longitude and latitude. For mobile network location signaling data or call CDR data, the location field may be a location area number and a base station cell number. Each piece of position data is the place or stop point in the trip, collectively referred to as the stop point. The data collected by the collection device can be pre-processed, including de-privacy processing, and the privacy-removal process can replace the user identifier with a meaningless identifier. In the preprocessing of data, base stations with close spherical distances can also be combined by base station clustering.

The storage device (or distributed storage device) can store a variety of data, including data and calculation results of the collection device, the user commute trajectory management device, and the commuter road management system.

The user commute trajectory management device can calculate the discrete entropy of each user's travel location based on the travel data of each user. Entropy is a measure of chaos and disorder. The smaller the entropy (closer to 0), the higher the stability and the stronger the predictability. The travel location discrete entropy is the discrete optional entropy of the user's travel location. The discrete entropy of the travel location can be analyzed from the time and location data of each user's travel, and the discrete entropy of the travel location per hour and 7 days per user is calculated, and the calculation result is saved in the storage device.

The commute users proposed by the present disclosure may be typical office workers with small discrete entropy during office hours and nighttime sleep periods on weekdays, such users typically having stable home locations and work locations. Calculate the regular commute trajectory of such users, and calculate the commute characteristics of each commute user for each commute user. The commute features may include family location, work location, commute time and working hours, commute time and off-duty time. The calculation result is saved to the storage device.

Optionally, for each commute user, the travel data of each commute user from the family location to the work location during the commute time of the working day is extracted, and each commuter user is extracted from the working position during the commute time of the working day. Travel data for home location. Time-stamped travel data for the same user can form a sequence of commute travel trajectories.

The user commute trajectory management device can process the commute travel trajectory sequence of each commute user, wherein the user commute trajectory management device can separately process the commute commute data and the commute commute data.

Optionally, the trajectory abnormality coefficient is calculated by the outlier detection algorithm, and the commutation trajectory whose trajectory abnormal coefficient is greater than the sixth threshold is deleted. The trajectory anomaly coefficient indicates the difference between the trajectory and other travel trajectory data of the user. The commute trajectory with the trajectory anomaly coefficient greater than the sixth threshold may be identified, and the identified commute trajectory may be regarded as an irregular abnormal trajectory, and the irregularity abnormal trajectory is discarded in the preprocessing of the commuter trajectory clustering. The commute trajectory from which the irregularity trajectory is removed is recorded as set A, and the trajectory similarity between each two commute trajectories in the set A is calculated, and the commute trajectory with the trajectory similarity between each two commute trajectories being greater than the first threshold is performed. Cluster and generate cluster cluster Ci. One or more cluster clusters Ci form a merged set C, and calculate a frequent point of the commute trajectory of each cluster cluster Ci in each of C, and each commute trajectory is frequently accompanied by a commute trajectory number and an hour tag of the frequent trajectory of the commute trajectory, and Calculate the proportion of the number of commute trajectories for each cluster cluster. For each commute track tr in each cluster cluster Ci, calculate the sum of the trajectory similarities between the commute trajectories other than the commute trajectory tr in Ci, tr_dist, and set the commute trajectory with the largest tr_dist as the cluster cluster Ci's representative commuter track Cij.

The frequent points of the commute trajectory may be a position that the user frequently passes and stays in the commute trajectory, and the frequent selection strategy of the commute trajectory may be that the number of frequent points is equal to the average number of stay points of each commute trajectory in the cluster cluster. The value, the frequent point may also be a stay point where the number of commute tracks accounts for more than 0.2. The hour tag of the frequent commute track may be the time stamp of the commute track at the stay point, the hour tag may be accurate to the hour, and if there are multiple timestamps, the average is averaged and the average is accurate to the hour.

According to the sequence of the time-of-day trajectory of each cluster cluster Ci in C, according to the sequence formed by the time stamp of the frequent trajectory of the commute trajectory, a navigation path Di is generated which takes the family position as the starting point and the working position as the end point, and sequentially passes through each commute trajectory frequently. . The GPS coordinates of the stay point of the navigation path Di and the navigation estimation duration of the navigation path can be extracted, the trajectory similarity between the representative commute trajectory Cij of the cluster cluster Ci and the corresponding navigation path Di, and the duration of the commute trajectory Cij can be calculated. The difference between the estimated length of the navigation and the navigation path Di. The navigation path Di and the navigation estimation duration can be obtained through a path planning and navigation function application programming interface (API) provided by the location service provider's map open platform. The duration representing the commute trajectory Cij is the time difference between the start time stamp and the end point time stamp representing the commute trajectory Cij. If the trajectory similarity between the commute trajectory Cij and the navigation path Di is greater than the second threshold, and the difference between the duration of the commute trajectory Cij and the navigation estimation duration of the navigation path Di is less than the third threshold, the navigation may be The path Di serves as a commute regular trajectory for the user. The second threshold and the third threshold may be set according to the accuracy requirements of the user or the developer. Because there may be multiple clusters in the merged set C, there may be multiple navigation paths, and there are multiple commuter regular trajectories. Extract the GPS coordinates of the navigation path Di route point, where, extract The number of the navigation path Di route points may be an integer part of the average value of the number of track stay points in the cluster cluster Ci, and the extracted navigation path Di route point is a point with a certain interval between the frequent points of the commute track.

The regularity of commuting can also be determined by comprehensive similarity. The comprehensive similarity F = f (x, y), x is the trajectory similarity, y is the difference of the commute duration, and f (x, y) is a function of the two variables x and y. F = f (x, y) can be a binary linear function, then F = f (x, y) = ax + by + c, where a, b and c are constants. Among the two variables, the weight of the influence of x on F is greater than the influence of y on F. If F is in the threshold range, it indicates that the merged track and the merged track have a high degree of coincidence of the navigation path, and the navigation path is a regular track of the user. If F is not within the threshold range, it indicates that the navigation path corresponding to the merged track and the merged track has a large difference, and the frequent points of the commutation trajectory of the cluster cluster can be re-selected and a new navigation path is generated, and F is recalculated. If the number of commuting trajectories of the cluster cluster is too small (for example, less than 0.1), the cluster cluster trajectory may be discarded.

If there is only one cluster cluster in the above merged set C, the commute regular trajectory of the obtained cluster cluster is the user's regular trajectory of the user. If there are multiple clusters in the merged set C, calculate the proportion of the number of commuting trajectories of each cluster cluster, and the proportion of the number of commuting trajectories of the cluster clusters is used as the weight of the commuting regular trajectory of the cluster cluster, and each will be The set of commuting regular trajectories of the clustering clusters serves as the user's regular trajectory.

As shown in FIG. 5, the present disclosure also proposes a user commute trajectory management method.

In step 510, data preprocessing is performed.

Data pre-processing can be to process the collected raw user travel location data into a specific format. Taking the mobile network user data as an example, the attribute data including the anonymously processed user identifier, the user location (the GPS coordinates of the base station cell number to be converted into the base station location), and the talk time may be extracted from the original data. The original call data is usually massive and redundant. By preprocessing and filtering out certain data, the amount of data can be reduced and the efficiency of subsequent processing can be improved. In data preprocessing, base stations with very close spherical distances can be combined, and smooth handover can be used to suppress frequent handover of base stations, reduce data volume, and enhance data validity.

In step 520, a user travel stability feature is calculated.

Optionally, when the commuting users whose home location and work location are identifiable are selected from all users, the discrete entropy of each user's travel location may be calculated according to the travel data of each user, and the user's travel stability is measured by using discrete entropy. The discrete entropy of the travel location is defined as follows:

Where p(R _i ) is the empirical probability of the user at the location R _i , and for the mobile network user data, R _i may be the base station number. The greater the discrete entropy of the travel location, the lower the regularity of the motion of the identified user. Calculate the discrete entropy of the user's travel location on weekdays, and calculate the discrete entropy of each user's travel location on weekdays by hour.

In step 530, the city commute time period is extracted.

Calculate the characteristics of the user's travel location. Define a data structure path-pattern=<h_1,R_1><h_2,R_2>...<h_n,R_n>, where h_i represents the ith time slice (preset to divide the 24 hours of the day into 24 time slices , each time slice 1 hour), i = 1, 2, ..., 24, R_i represents the user's position of the i-th time slice, and the data structure path-pattern can represent the user's movement track.

Optionally, a two-dimensional array of 24 rows and 7 columns, Pattern[24][7], describes the characteristics of the user's travel location, and Pattern[24][7] indicates the most frequent stopover time of 24 hours per day per week. It is called an array of hourly stops. The most frequent stopover location can be the location with the longest stay time during that time period.

Optionally, the two-dimensional array Pattern_go[24][7] is used to describe the change of the user's stop location. When the current hour is the same as the previous hour, the hourly element value is 0, and the current hour is different from the previous hour. , the hour element value is 1. Analyze the two-dimensional array Pattern_go[24][7] to find out which days of the week the workday is, and to know the commute time, that is, the morning peak and the late peak commute travel time. Optionally, the statistical value distribution of the 24 hours of working days in Pattern_go[24][7] presents a double peak, and the double peak periods correspond to the morning peak and the late peak commuting time respectively. The morning and evening peak commute times are the city commuting hours.

In step 540, the commuter user is screened.

A user who can define the travel date of the workday with a discrete entropy less than a certain threshold is a commute user.

In step 550, the user commute feature is extracted.

The user commute feature description can be a four-tuple of {home location, work location, time taken to work, time taken to work}.

For the commute user selected in the above step 540, the OD (Origin-Destination) can be calculated, and the commute OD between the home location and the work location is a travel mode.

Taking the call data as an example, the call data T of each user is divided into two sets, Tday and Tnight, Representing the day and night call data, respectively, where T={<mobile number, call base station, talk time>}, {} represents at least one call data of a user, and each call data is a mobile phone number, a call base station, and The triple sequence of the call time; the Tday and Tnight call data are respectively divided according to the call base station, and each call record is equivalent to a positioning record; the base stations are arranged from large to small according to the number of calls, and the spherical surface is clustered by the base station. The base stations that are close to each other are merged to form a new call location point; according to the call periodicity, the call frequency of each location point is calculated; according to the frequency of the call, the location points are filtered to delete the location where the call is sparse; And the location point where the call frequency is the most frequent among the Tnight data is D and O, that is, the work position and the home position; and the commute OD of each user is output as the home position and the work position in the user commute feature. Call frequency refers to the percentage of calls that the user has made at that location as a percentage of the total number of calls for that user.

Remove at least one unrecognizable user from the home location and work location of the commuter user.

The time taken to work on the road may refer to the average value of the time taken by the commuter user from the home position to the work position during the commute time on the working day from the pre-processed data; the time taken after the work is taken from the pre-processed The data is used to calculate the average time taken by the commuter user from the working position to the home position during the commute time of the working day.

In step 560, an irregularity trajectory is detected.

For the same user, a small number of irregular trajectories can be detected and eliminated in the trajectory of the user from the home position to the work position during the commute time.

t is the trajectory to be tested, T is the set of trajectories to be compared, m is the number of runs, ψ is the number of samples per run, array n[m] is used to count the value obtained for each run, and array n[m] is initialized. Is 0.

The processing steps include: n[m]=0; for i=1to m do;

Take the n[m] array of each t as an intermediate result and perform the operation:

S=2 ^{-E((n(t))/C(N))} ;

Where C(N)=2H(N-1)-2(N-1)/N; ← denotes an assignment.

H(i)=ln(i)+0.57721566, where 0.57721566 is the Euler constant and / represents the division operation.

After this operation, the higher the S value, the more deviation from the mainstream commuter trajectory. S is also referred to as a trajectory anomaly coefficient, wherein the commute trajectory in which the trajectory anomaly coefficient S exceeds the sixth threshold is an abnormal trajectory, and the abnormal trajectory can be deleted.

In step 570, commutation trajectory clustering is performed.

The commute trajectory except the regular trajectory obtained by the above-mentioned irregular trajectory detection is recorded as set A, and the trajectory similarity between each two commute trajectories in the set A is calculated and the trajectories with high trajectory similarity are combined and merged. Set C calculates the frequent points of the commute trajectory of each cluster cluster in C and the number of commute trajectories of each cluster cluster.

The algorithm description for calculating the trajectory similarity in the commute trajectory to work includes:

Calculating the trajectory similarity Tdis(A, B, limit_len) of the trajectory A and the trajectory B, that is, the minimum value of the Euclidean distance of the location passing through the trajectory B is less than the number a' of the fifth threshold limit_len, among all the locations where the A passes, a' is less than or equal to the number of elements of A;

Calculate the trajectory similarity Tdis(B, A, limit_len) of the trajectory B and the trajectory A, that is, among all the locations where the B passes, the minimum Euclidean distance of the point passing through the trajectory A is smaller than the number b' of the limit_len, b' is smaller than Or the number of elements equal to B;

The trajectory similarity between the calculated trajectory A and the trajectory B is: (Tdis(A, B, limit_len) + Tdis(B, A, limit_len))/(a+b), and the trajectory similarity ranges from 0. Between 1 and 1. In extreme cases, the A and B tracks are coincident or very similar, and the trajectory similarity is 1.

Where a is the number of trajectory A passing locations (including non-adjacent identical locations), b is the number of trajectory B passing locations (including non-adjacent identical locations), limit_len is the distance threshold, and / represents the division operation.

For example, as shown in FIG. 6, the trajectory A is recorded as <a1, a2, a3, a4>, passing 4 locations along the way; the trajectory B is recorded as <b1, b2, b3, b4, b5>, passing 5 locations along the way ;limit_len is set to 2. A1 is closest to the distance b1 on the trajectory B, and the distance between the a1 and b1 points is 1; a2 is the distance from the b2 point on the trajectory B Recently, the distance between a2 and b2 is 2; a3 is closest to b3 on track B, the distance between a3 and b3 is 3; a4 is closest to b3 on track B, and the distance between a4 and b3 is 2; then Tdis (A, B, limit_len) = 1 + 1 + 0 + 1 = 3. B1 is closest to the distance a1 on the trajectory A, the distance between the b1 and the a1 point is 1; b2 is the closest distance from the a1 point on the trajectory A, the distance between the b2 and the a1 point is 1; b3 is the closest distance from the a4 point on the trajectory A , b3 and a4 point distance is 2; b4 is closest to the distance a4 on the track A, the distance between b4 and a4 is 8; b5 is closest to the distance a4 on the track A, and the distance between b5 and a4 is 3; Then Tdis(B, A, limit_len) = 1+1 + 1 + 0 + 0 = 3. The trajectory similarity between the trajectory A and the trajectory B is (3+3) / (4 + 5) = 0.667.

In step 580, the commute trajectory map is matched.

According to the frequent trajectory of the commuting trajectory, the home position and the working position of each cluster cluster in step C, the navigation path is generated, the GPS coordinates of the navigation path approach point are extracted, and the navigation estimation duration of the navigation path is calculated, and the cluster cluster is calculated. Represents the trajectory similarity and duration difference between the commute trajectory and the corresponding navigation path. If the trajectory similarity between the commute trajectory and the navigation path is greater than the second threshold, and the difference between the duration of the commute trajectory and the navigation estimation duration of the navigation path is less than the third threshold, the navigation path is the commuting rule of the user. Sexual trajectory; representing the trajectory similarity between the commute trajectory and the navigation path, and the difference between the duration representing the commute trajectory and the navigation estimation duration of the navigation path does not satisfy the above condition, and the C clustering in step 570 can be reselected. The commuter trajectory of the cluster is frequent. If the number of commuting trajectories of the cluster cluster is too small (for example, less than 0.1), the cluster cluster trajectory may be discarded.

In step 590, a user regular commute trajectory is generated.

When there is only one cluster cluster in the above merged set C, the commute regular trajectory of the obtained cluster cluster may be the user's regular trajectory of the user. If there are multiple cluster clusters in the merged set C, calculate the proportion of the number of commuting trajectories of each cluster cluster, and the proportion of the number of commuting trajectories of the cluster clusters as the weight of the commuting regular trajectory of the cluster clusters, The set of commuting regular trajectories of each cluster cluster is taken as the user regular trajectory.

The application of the present disclosure will now be described in connection with scenario 1 and scenario 2.

In scenario 1, after collecting the mobile network location signaling data (including the periodic location update signaling data), the solution is: the traffic of several common lines going to work in the commute users in the A location and the working location in the B location. how many.

The embodiment provides a solution for solving the problem in the scenario 1, and the solution includes: calculating a commute feature of the user, screening a user of the commute user whose work position is A, and working position B, and calculating The number of users; the commute trajectory of each user in the collection user is extracted, the commute trajectory is an ordered sequence of the base station of the working route, and the GPS coordinates of the user location are obtained by the mapping relationship between the cell number of the base station and the GPS coordinates of the base station location; Each user in the user uses the outlier detection algorithm to detect the irregular commute trajectory and eliminates it; for each user Ui in the set users, the trajectory clustering and map matching are performed on the commute trajectory except the regular commute trajectory. Obtaining one or more regular trajectories Tra_i with weights for each user, the set of the regular trajectories Tra_i is recorded as Tra_users; according to the above method, trajectory clustering and map matching are performed on the Tra_users trajectory set to obtain weighted One or more commute regular trajectories Tra_users_j, the weight of the commute regular trajectory Tra_users_j is the sum of the weights of the commuting regular trajectories in the corresponding cluster clusters, and the set of commuting regular trajectories Tra_users_j is the line set CC from A to B Each line CCi in the CC is Tra_users_j, and the traffic of each line is the weight of Tra_users_j.

In scene 2, the GPS data of a large number of vehicles is collected to solve the problem: the number of vehicles passing through a section r to work, the proportion of traffic per hour, extracting user information and regular commute information, such as the distribution of family locations, working position What is the distribution, and what is the habitual route of commuting. r may be a highway, or it may be a bridge or other connected section.

This embodiment provides a solution for solving the problem in the scenario 2, the solution includes: screening the road segment set R related to the road segment r, calculating the latitude and longitude range of the R Region_R; calculating the user commute feature; extracting the user commute trajectory, each commute trajectory Corresponding to one hour time label, the user trajectory containing the Region_R and the user in the commute trajectory are filtered, and the user trajectory set is recorded as T_R, and the user set is recorded as Users. For each user Ui in Users, the eligible commuting trajectory is recorded. The weight Ui_num is not followed by the users in Users. All the commute tracks of each user Ui in the user set Users are recorded as T; after counting, the number of vehicles per hour can be obtained; and each user Ui in Users is extracted. The commute feature can derive specific user information; obtain regular commute information from the user. The obtaining the regular commute information of the user may include: detecting, for each Ui, an irregular commute trajectory of the user Ui for the user set User and the commute trajectory T; and clustering clusters and map matching of the user's commute trajectory, Get one or more commuter regular trajectories with weights from the user.

The present disclosure provides a user commute trajectory management method, which acquires travel data of a user through a positioning system, such as a positioning device or a communication base station on a user terminal, calculates a user regular trajectory according to user travel data, and outputs, in the process, Without user research, the management of user commuting is enhanced, and the user's commuting can be more fully understood and controlled, improving the user experience.

The disclosure also collects the user travel location data, extracts the user commute trajectory, detects the irregular commute trajectory by the outlier detection algorithm, and performs the clustering and map matching on the commute trajectory of the regular commute trajectory, thereby realizing the user regularity. The calculation system and method of the trajectory. The technical solution provided by the embodiments of the present disclosure is beneficial to the fine management of the commuter road section, and is convenient for grasping some key traffic information, such as the load level of each road section during the peak of the commute, such as one road section closure, regulation or restriction, etc. How many citizens travel, and the distribution of these citizens.

Embodiments of the present disclosure also provide a non-transitory computer readable storage medium storing computer executable instructions arranged to perform the method of any of the above embodiments.

The embodiment of the present disclosure further provides a hardware structure diagram of an electronic device. Referring to FIG. 7, the electronic device includes:

At least one processor 70, which is exemplified by a processor 70 in FIG. 7; and a memory 71, may further include a communication interface 72 and a bus 73. The processor 70, the communication interface 72, and the memory 71 can complete communication with each other through the bus 73. Communication interface 72 can be used for information transfer. Processor 30 may invoke logic instructions in memory 31 to perform the methods of the above-described embodiments.

In addition, the logic instructions in the memory 71 described above may be implemented in the form of a software functional unit and sold or used as a stand-alone product, and may be stored in a computer readable storage medium.

The memory 71 is a computer readable storage medium and can be used to store a software program, a computer executable program, a program instruction or a module corresponding to the method in the embodiment of the present disclosure. The processor 70 executes the functional application and the data processing by executing a software program, an instruction or a module stored in the memory 71, that is, implementing the method in the above method embodiment.

The memory 71 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application required for at least one function; the storage data area may store data created according to usage of the terminal device, and the like. Further, the memory 71 may include a high speed random access memory, and may also include a nonvolatile memory.

The technical solution of the present disclosure may be embodied in the form of a software product stored in a storage medium, including one or more instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) Performing all or part of the steps of the method of the embodiments of the present disclosure. The foregoing storage medium may be a non-transitory storage medium, including: a USB flash drive, a mobile hard disk, a read-only memory (ROM), and a random storage memory (Random-Access). A medium that can store program code, such as a memory, a RAM, or an optical disk, or a transient storage medium.

Industrial applicability

The user commute trajectory management method, device and system provided by the disclosure obtain the travel data of the user through the positioning system, calculate the regular trajectory of the user according to the travel data of the user, and output, without user research, and enhance the management of the user's commuting, A more comprehensive understanding and control of user commuting has improved the user experience.

Claims

A user commute trajectory management method includes:

Obtaining, by the positioning system, travel data of each user, where the travel data includes a stay point and a time stamp corresponding to the stay point;

Calculating a regularity trajectory of the user of each user according to the travel data of each user;

The user's regular trajectory of each user is output to the commuter road management system.
The method of claim 1, wherein the calculating the user regularity trajectory of each user according to the travel data of each user comprises:

Calculating a commute feature according to the travel data of each user, the commute feature including a home location, a work location, and a commute time period;

And selecting, according to the commute period of the commute feature, the stay point in the travel data, and the timestamp corresponding to the stay point, the travel data that belongs to the commute time period, and the selected travel data in the travel data The stop points are sorted by time stamp to generate a commute trajectory;

Calculating a trajectory similarity between each two commute trajectories of each user, and generating a cluster cluster by commuting trajectory clustering between each two commute trajectories of each user that is greater than a first threshold And determining the representative commute trajectory of each cluster cluster;

Selecting, according to the selection strategy, a frequent point of the commute trajectory from the stay points of all the commutation trajectories of each cluster cluster, according to the home location, the working position, the frequent points of the commute trajectory, and the frequent points of the commute trajectory The timestamp generates a navigation path;

Calculating a trajectory similarity between the navigation path and the representative commute trajectory, and if the trajectory similarity between the navigation path and the representative commute trajectory is greater than a second threshold, using the navigation path as the convergence The commuting regularity trajectory of the clusters generates the regular trajectory of the user according to the commuting regular trajectory of the cluster cluster.
The method of claim 2 wherein said commuting time comprises commuting time to work and During the commute time, the commute trajectory includes a commute trajectory and an off-duty commute trajectory, and the navigation path is generated according to the home location, the working location, the frequent trajectory of the commute trajectory, and the time stamp corresponding to the frequent trajectory of the commute trajectory The method includes: if the clustering cluster is generated by the commuting trajectory trajectory, generating a working navigation path starting from the family position of the user, the working position being an end point, and frequently passing through the commuting trajectory; And if the clustering cluster is generated by the off-duty commute trajectory clustering, generating an off-duty navigation path starting from the working position of the user, the family location as an end point, and frequently passing through the commuting trajectory in turn.
The method of claim 2, further comprising: if the trajectory similarity between the navigation path and the representative commute trajectory is less than the second threshold, reselecting the frequent trajectory of the commute trajectory and generating a new navigation path, calculating and determining whether a trajectory similarity between the new navigation path and the representative commute trajectory is greater than the second threshold, if a trajectory similarity between the new navigation path and the representative commute trajectory If the second threshold is greater than the second threshold, the new navigation path is used as a commute regular trajectory of the cluster cluster, if the trajectory similarity between the new navigation path and the representative commute trajectory is not greater than the second The threshold is performed by repeating the step of reselecting the frequent points of the commute trajectory and generating a new navigation path until the trajectory similarity between the new navigation path and the representative commute trajectory is greater than the second threshold.
The method of claim 2, the method further comprising: calculating a navigation estimation duration of the navigation path, determining whether a trajectory similarity between the navigation path and the representative commute trajectory is greater than the second And determining, by the threshold, whether a difference between the navigation estimation duration of the navigation path and the duration of the representative commute trajectory is less than a third threshold; and if the trajectory similarity between the navigation path and the representative commute trajectory is greater than the second And a threshold value, and the difference between the navigation estimation duration of the navigation path and the duration of the representative commute trajectory is less than a third threshold, and the navigation path is used as a commuting regular trajectory of the cluster cluster.
The method according to claim 2, wherein if the cluster cluster is plural, the generating the user regularity trajectory according to the commuting regular trajectory of the cluster cluster comprises: calculating each cluster cluster The number of commute trajectories is proportional, the number of commuting trajectories of the cluster clusters is taken as the weight of the commuting regular trajectory of the cluster clusters, and the set of commuting regular trajectories of each cluster cluster is taken as the user regularity. a trajectory; and if the cluster cluster is one, the generating the user regular trajectory according to the commuting regular trajectory of the cluster cluster comprises: using the commutation regular trajectory of the cluster cluster as the user rule Sexuality.
The method according to claim 6, wherein if the cluster of clusters is plural, the method further comprises: deleting cluster clusters in which the number of commutation trajectories is less than a fourth threshold.
The method of claim 2, wherein said determining a representative commute trajectory of each cluster cluster comprises: calculating a sum of trajectory similarities between each commute trajectory and other commute trajectories in said cluster cluster one by one, The commute trajectory having the largest sum of the trajectories of the trajectories is taken as the representative commute trajectory.
The method of claim 2 wherein the calculation of the trajectory similarity between the trajectories comprises:

Calculating the trajectory similarity between the trajectory A and the trajectory B is the number a' of the distance minimum from the point of the trajectory B path among all the locations of the trajectory A path;

Calculating the trajectory similarity between the trajectory B and the trajectory A is the number b' of the minimum value of the distance from the point of the trajectory A path among all the points of the trajectory B path;

The trajectory similarity between the calculated trajectory A and the trajectory B is (a'+b')/(a+b), a is the number of locations of the trajectory A pathway, b is the number of locations of the trajectory A pathway, and / represents division Operation.
The method of claim 2, after generating the commute trajectory, the method further comprising: calculating a trajectory anomaly coefficient by an outlier detection algorithm, and deleting a commute trajectory of the trajectory anomaly coefficient greater than a sixth threshold.
The method according to any one of claims 2 to 10, before the generating a commute trajectory, the method further comprises: selecting, from all users, the commuter user that is identifiable by the home location and the work location, generating a location The traversable trajectory of the identifiable commuter user.
The method of claim 11, wherein the screening of the commute users identifiable from the home location and the work location from all the users comprises: calculating the discrete entropy of the travel location of each of the users according to the travel data of each user And identifying, as a commute user, a user whose travel location has a discrete entropy smaller than a seventh threshold; and obtaining and identifying, according to travel data of each commute user, a home location and a work location of each commute user, deleting the home location and the At least one unrecognizable commuter user in the work location.
A user commute trajectory management device includes:

Obtaining a module, configured to acquire travel data of each user by using a positioning system, where the travel data includes a stay point and a time stamp corresponding to the stay point;

a processing module, configured to calculate a user regularity trajectory of each user according to the travel data of each user;

The output module is configured to output the regularity trajectory of each user of the user to the commuter road management system.
The apparatus of claim 13 wherein said processing module comprises:

a commute feature calculation sub-module, configured to calculate a commute feature according to the travel data of each user, where the commute feature includes a home location, a work location, and a commute time period;

The commuting trajectory generating sub-module is configured to: according to the commute period of the commute feature, the stay point in the travel data, and the timestamp corresponding to the stay point, filter the travel data belonging to the commute period, and filter The obtained staying points in the travel data are sorted by time stamp to generate a commute trajectory;

a commuting cluster management sub-module, configured to calculate a trajectory similarity between each two commute trajectories of each user, and the trajectory similarity between each two commute trajectories of each user is greater than a first threshold The commute trajectory clustering generates cluster clusters and determines representative commuter trajectories of each cluster cluster;

a navigation path generating submodule, configured to select a frequent point of the commute trajectory from the stay points of all the commuting trajectories of each cluster cluster according to the selection policy, according to the home location, the working location, the frequent points of the commute trajectory, and Generating a navigation path by a timestamp corresponding to a frequent point of the commute trajectory;

a commutation regular trajectory sub-module, configured to calculate a trajectory similarity between the navigation path and the representative commute trajectory, if a trajectory similarity between the navigation path and the representative commute trajectory is greater than a second threshold, The navigation path is used as a commuting regular trajectory of the cluster cluster, and the user regular trajectory is generated according to the commuting regular trajectory of the cluster cluster.
The device of claim 14, wherein the commute period includes a commute time and an off-duty commute, the commute trajectory includes a commute trajectory and an off-duty trajectory, and the navigation path generation sub-module is set to, if Generating cluster clusters by the commuting trajectory trajectory, generating a work navigation path starting from the user's home position, the work position being an end point, and frequently passing through the commute trajectory; and The cluster is generated by the off-duty commute trajectory clustering, and the off-duty navigation path is generated starting from the working position of the user, the home location as the end point, and the frequent passing point through the commute trajectory.
The apparatus of claim 14, wherein the commute regular trajectory sub-module is further configured to trigger the navigation if a trajectory similarity between the navigation path and the representative commute trajectory is less than the second threshold The path generation sub-module reselects the frequent points of the commute trajectory and generates a new navigation path, and calculates and determines whether the trajectory similarity between the new navigation path and the representative commute trajectory is greater than the second threshold, if the new If the trajectory similarity between the navigation path and the representative commute trajectory is greater than the second threshold, the new navigation path is used as a commuting regular trajectory of the cluster cluster, if If the trajectory similarity between the new navigation path and the representative commute trajectory is not greater than the second threshold, the step of reselecting the frequent points of the commute trajectory and generating a new navigation path is performed cyclically until the new navigation path and the location The trajectory similarity between the representative commute trajectories is greater than the second threshold.
The apparatus according to claim 14, wherein the navigation path generation sub-module is further configured to calculate a navigation estimation duration of the navigation path, and the commute regular trajectory sub-module is further configured to determine the navigation path and the location Determining whether the difference between the navigation estimation duration of the navigation path and the duration of the representative commute trajectory is less than a third threshold when the trajectory similarity between the commute trajectories is greater than the second threshold; and if the navigation is And the trajectory similarity between the path and the representative commute trajectory is greater than a second threshold, and the difference between the navigation estimation duration of the navigation path and the duration of the representative commute trajectory is less than a third threshold, and the navigation path is used as The commuting regular trajectory of the cluster cluster.
The apparatus according to claim 14, wherein if the clustering cluster is plural, the commuting regular trajectory sub-module is configured to calculate a proportion of the number of commuting trajectories of each cluster cluster, and the clustering cluster The number of commuting trajectories is used as the weight of the commuting regular trajectory of the cluster cluster, and the set of the commuting regular trajectories of each cluster cluster is taken as the user regular trajectory; and if the cluster cluster is one, The commute regular trajectory sub-module is configured to use the commuting regular trajectory of the cluster cluster as the user regular trajectory.
The apparatus according to claim 18, wherein if the clustering cluster is plural, the commuting regular trajectory sub-module is further configured to delete a cluster cluster whose occupation trajectory number is smaller than a fourth threshold.
The apparatus according to claim 14, wherein said commuter cluster management sub-module is configured to calculate a sum of trajectory similarities between each commute trajectory and other commute trajectories in said cluster cluster one by one, said trajectory The commute trajectory with the largest sum of similarities serves as the representative commute trajectory.
The apparatus of claim 14, wherein the processing module further comprises a trajectory similarity calculation sub-module configured to calculate a trajectory similarity between the trajectory A and the trajectory B as all the trajectory A paths In the point, the minimum distance from the point of the trajectory B path is smaller than the number a' of the fifth threshold; the trajectory similarity between the calculated trajectory B and the trajectory A is the distance from the point of the trajectory A path among all the locations of the trajectory B path The minimum value is less than the number b' of the fifth threshold; and the trajectory similarity between the calculated trajectory A and the trajectory B is (a'+b')/(a+b), where a is the number of locations of the trajectory A path, b is the number of locations of the trajectory A route, and / represents the division operation.
The apparatus according to claim 14, wherein the commute trajectory generation sub-module is configured to calculate a trajectory abnormality coefficient by an outlier detection algorithm after the commutation trajectory is generated, and delete a commute trajectory whose trajectory anomaly coefficient is greater than a sixth threshold .
The apparatus according to any one of claims 14 to 22, wherein the commute trajectory generation sub-module is configured to filter out, from all users, the home location and the work location to be identifiable before generating a commute trajectory The commuter user generates a commute trajectory of the identifiable commute user.
The apparatus of claim 23, wherein the commute trajectory generation sub-module is configured to calculate, according to each user's travel data, a discrete entropy of the travel location of each user, and a user whose travel location discrete entropy is less than a seventh threshold As a commuter user; and obtaining and according to the travel data of each commute user, identifying the home location and the work location of each commute user, deleting at least one unrecognizable commuter user of the home location and the work location.
A user commute trajectory management system, comprising: a positioning system, a commuter road management system, and the user commute trajectory management device according to any one of claims 13 to 24; wherein the positioning system is configured to monitor user travel data, The travel data includes a stay point and a time stamp corresponding to the stay point; the user commute trajectory management device is configured to acquire travel data of each user through the positioning system, and calculate the travel data according to the travel data of each user. Describe the user regularity trajectory of each user and output the user regularity trajectory of each user to the commute road management system; and the commuter road management system is configured to manage the commute road according to the user regular trajectory.
A non-transitory computer readable storage medium storing computer executable instructions arranged to perform the user commute trajectory management method of any of claims 1-12.