CN115238834B - User group space-time abnormal pattern recognition method based on trajectory data - Google Patents
User group space-time abnormal pattern recognition method based on trajectory data Download PDFInfo
- Publication number
- CN115238834B CN115238834B CN202211155565.8A CN202211155565A CN115238834B CN 115238834 B CN115238834 B CN 115238834B CN 202211155565 A CN202211155565 A CN 202211155565A CN 115238834 B CN115238834 B CN 115238834B
- Authority
- CN
- China
- Prior art keywords
- user
- time
- aggregation
- space
- area
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
A user group space-time abnormal pattern recognition method based on trajectory data comprises the following steps: the method comprises the steps of realizing preliminary processing of mass trajectory data, calculating the adjacent aggregation degree of each user in each time sequence state, extracting time-space aggregation areas of a starting point, a terminal point and a traveling point of the user on the basis of the initial processing, upgrading the periodically-appearing time-space aggregation areas into time-space co-occurrence areas, identifying time-space behavior patterns of user groups in each area, and finally finding abnormal aggregation behaviors of the user according to the historical group aggregation state. The invention is convenient to inquire the aggregation state of each time node, further knows the cluster dynamic information of the urban population and enhances the utilization of the information; the abnormal space-time mode of the urban group is effectively extracted, and the abnormal behavior of the group is judged in an auxiliary manner; the spatial space-time aggregation area and space-time co-occurrence area information obtained by the invention can realize the map visual display of the information.
Description
Technical Field
The invention relates to the field of geographic big data identification, in particular to a track data-based user group space-time abnormal pattern identification method, which can be used for constructing a multi-user space-time aggregation area and discovering the user space-time co-occurrence area based on the cluster analysis of user track positioning data of a GPS and a communication base station on user stop points and aggregation areas, thereby obtaining a group space-time behavior pattern of users and realizing the identification of abnormal aggregation behaviors of groups.
Background
With the continuous development of positioning technology, terminals of many electronic devices can acquire real-time position information more quickly and accurately. Positioning technologies are configured on mobile devices such as mobile phones, automobiles, airplanes and the like, and the purpose is to obtain real-time positions of the mobile devices in time at a terminal.
The traces are connected by the points at the real-time locations in chronological order. For example, in the time when a vehicle starts from a place and travels to a destination, the terminal acquires the current position of the vehicle every few seconds, and points at the current position are connected in chronological order to form a track. The method has a very wide application scene for researching the track abnormity detection. For example, if a region has a large number of intersections of user trajectories in the same time period, it is very necessary to identify a time-space anomaly if the intersection belongs to a normal state of urban group aggregation.
In the technology of correlation analysis and anomaly detection of location data of mobile users of lie, an anomaly behavior detection algorithm firstly detects anomaly characteristics, and track data to be measured, a user behavior mode and a user group diagram structure need to be prepared; and then judging the abnormal type and checking the evolved abnormal behavior, wherein the method constructs an abnormal behavior detection multi-classification model through a training sample, so that various abnormal phenomena can be better detected without meeting specific abnormal detection requirements, and then the specific method is adopted for abnormal detection. However, this method generally detects only specific types of anomalies and cannot classify anomalies.
In li jian user track abnormity detection research facing to the mobile terminal device, for track abnormity detection, track segmentation is firstly carried out, the whole track is segmented into a plurality of track line segments, and the analysis characteristic of the local track replaces the global track characteristic, so that the analysis speed of the track can be accelerated. Secondly, measuring the similarity of the moving track, and finally realizing the abnormal detection of the moving track. But the method ignores the time law analysis in human activities, and the user behavior recognition capability is weak.
In the research OF a track clustering and anomaly detection algorithm facing GPS data OF Zhang thunder, for an abnormal track detection method based on track spatio-temporal feature clustering, firstly GPS data acquisition is carried out, then track preprocessing and data cleaning are carried out, then the track field is calculated, the track OF (p) is calculated, whether the initial time is more than Threadhold or not is judged, if yes, the time track is abnormal, and if not, the track is normal. However, the accuracy of the method is basically in direct proportion to the data volume, so that the data volume is usually large, the calculation speed is slow and the cost is high when the better calculation accuracy needs to be achieved.
In addition, most algorithms in the field of track clustering and track anomaly detection and analysis are based on a certain application background, and the generality is not strong. A track clustering method based on track GPS spatio-temporal feature extraction divides abnormal conditions into time feature abnormality, space feature abnormality and global abnormality through abnormal property description. There are still many areas of improvement in the ability to classify clusters into normal and abnormal situations by setting statistical features.
Therefore, how to standardize the system to obtain the abnormal mode of the user space-time cluster becomes a technical problem which needs to be solved urgently to solve various defects in the prior art.
Disclosure of Invention
The invention aims to provide a user space-time cluster pattern abnormity identification method based on trajectory data, which can effectively extract the abnormal space-time pattern of an urban cluster, carry out comprehensive information push aiming at the condition of larger difference with a historical user gathering area, assist the manual judgment of the abnormal behavior of the cluster, and realize the map visual display of the information by using the method.
In order to achieve the purpose, the invention adopts the following technical scheme:
a user group space-time abnormal pattern recognition method based on trajectory data comprises the following steps:
trajectory data preprocessing step S110:
acquiring user track positioning data, converting the abstracted user track positioning data into a data set which can be used for analysis, wherein the method comprises the steps of grouping original data, sequencing the original data according to a time sequence, realizing spatialization through latitude conversion, then denoising the data, extracting characteristic point information in track data, classifying the characteristic point information into a starting point, an end point and a traveling point, and storing the characteristic point information as a completely stored user track;
the neighboring user concentration calculation step S120:
according to the current position of the user, calculating the user concentration degree aiming at three states of a starting point, a terminal point and a travel point in the track information characteristic points;
a space-time aggregation area calculation step S130:
the data of the three states of the starting point, the end point and the advancing point of the user are further sorted, a space-time aggregation area is calculated according to the aggregation values of the elements, the starting place aggregation area, the destination aggregation area and the traffic aggregation area of the user group are further calculated respectively according to the tendency of low density aggregation to high density aggregation existing in the group aggregation, and the occurrence time and duration time of each aggregation area are obtained;
spatio-temporal co-occurrence region calculating step S140:
according to the current known user group aggregation areas, further aggregating and comparing the time sequences of the aggregation areas, firstly performing space-time superposition analysis on a time range and an area boundary, and then performing similarity calculation according to time frequency, user number and an area center on the basis of primarily acquiring the aggregation areas at the same position to judge the relevance of each attribute value among the aggregation areas;
spatio-temporal behavior pattern recognition step S150:
and judging a group behavior mode of the co-occurrence area according to POI, AOI, news and activities in the co-occurrence area, calculating a maximum influence boundary based on the current space-time co-occurrence area boundary, calculating the number of various POI and the area of the AOI in the coverage area, judging the type mode of the POI and the area of the AOI, and further explaining the group aggregation behavior mode by attaching news and activity labels.
Optionally, the method further includes an abnormal group behavior discovery step S160:
aiming at the special condition found by the daily aggregation behavior of the group, the space-time behavior pattern recognition is carried out and the forming reason is mined.
Optionally, in step S110,
the user track positioning data comprises GPS positioning data and communication base station positioning data, and discrete point detection denoising is adopted.
Alternatively, step S120 includes the following sub-steps,
defining a spatiotemporal range substep S121: taking the current position of a user as a center, and converting time into a space distance for data processing;
user quantity sub-step S122: counting the number of users in a specified space-time range;
three-state aggregation degree calculating sub-step S123: and calculating a starting point data set, an end point data set and a traveling point data set of the track data, and identifying and analyzing the starting place, the destination and the road traffic state of the user.
Optionally, step S130 specifically includes:
1) Retrieving the peripheral adjacent user j of each user i in the space-time range of 10km, and acquiring the attribute of the user i with respect to the user number j (ID) and the user adjacent number j (count);
2) Calculating a minimum distance dist (i, j) between i and j according to the geographic space position j (get), and constructing an ij vector when j (count) > i (count);
3) When the user i does not have j (count) > i (count) in the space-time range, the space-time range is expanded, the user j of j (count) > i (count) is further searched, min (dist (i, j)) is calculated, and an ij vector is constructed, so that the user i belongs to the cluster of the user j in the space-time range larger than 10 km;
4) Setting a distance threshold, deleting vectors exceeding the distance threshold, grouping users of reserved vector connection relation into one class to obtain a user cluster, and defining an outer convex bag (namely an external polygon) of all elements of the same class as an aggregation area;
5) According to 1) to 4), respectively calculating a starting place gathering area, a destination gathering area and a traffic gathering area, and counting the time range, the time frequency, the area center, the area boundary and the number of users of each cluster.
Optionally, in step S140, correlation between the attribute values of the aggregation areas is determined by using pearson correlation test.
Optionally, step S160 specifically includes:
firstly, identifying a current group aggregation area according to the latest user position, searching whether the current group aggregation area is in a sequence of known co-occurrence areas, verifying the attribution of the current aggregation area, and judging the special situation as a sudden space-time aggregation abnormal mode when no user space-time co-occurrence area is matched with the current group aggregation area; when the time-space co-occurrence areas are matched, if the period of the latest crowd gathering number does not accord with the period of the historical crowd number, the special condition is judged to be an abnormal mode of 'fluctuation of the number of users in the periodic time-space gathering area', wherein the specific condition is that the number of people increases suddenly or the number of people decreases suddenly.
The invention further discloses a storage medium for storing computer executable instructions, and the computer executable instructions, when executed by a processor, execute the method for identifying the user group space-time abnormal pattern based on the trajectory data.
Optionally, the space-time aggregation area and the space-time co-occurrence area obtained by the user group space-time abnormal pattern recognition method can be visually displayed on a map.
Drawings
FIG. 1 is a flow chart of a method for user population spatiotemporal anomaly pattern recognition based on trajectory data according to a specific embodiment of the present invention;
FIG. 2 is a diagram of neighboring user aggregations spatiotemporal ranges, according to an embodiment of the present invention;
FIG. 3 is a schematic diagram illustrating a cluster distribution of users at a certain time interval according to an embodiment of the present invention;
fig. 4 is a schematic diagram of a distribution of a time-series change of the number of users in a cluster according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some structures related to the present invention are shown in the drawings, not all of them.
The invention mainly comprises the following steps: clustering analysis is carried out on the user stopping point and the gathering area based on the user track positioning data of the GPS and the communication base station, a multi-user space-time gathering area is constructed, and the user space-time co-occurrence area is found, so that a group space-time behavior mode of the user is obtained, and the abnormal gathering behavior of the group is recognized.
Specifically, the original data is subjected to geometric linearization, characteristic dotting and analysis meeting the follow-up steps. On the basis of the spatialized user position information, the number of users in each space-time range is calculated to form a space-time aggregation degree index of the users, and analysis is carried out on three states of the users at a starting point, an end point and a traveling point. And clustering according to the space-time aggregation values of the users, further acquiring the space-time aggregation areas of the users, and analyzing the initial aggregation area, the destination aggregation area and the traffic aggregation area of the group on the results of the starting point, the end point and the travel point. And carrying out time sequence aggregation comparison processing on the currently known user group aggregation areas, carrying out primary extraction on the co-occurrence areas according to the time range and the area boundary of each aggregation area, carrying out similarity calculation on the related aggregation areas, and upgrading the periodic same-area group aggregation areas into space-time co-occurrence areas. And judging the group aggregation behavior pattern of the site according to POI, AOI, news and activities in the co-occurrence area, and deeply mining main group aggregation areas such as residential areas, accommodation areas, shopping areas, working areas and the like. Finally, the discovery is carried out aiming at special situations of daily population aggregation, such as 'occurrence of a sudden space-time aggregation area', 'abnormal number of users in a periodic space-time aggregation area', and the like. And analyzing the abnormal region according to the space-time behavior pattern recognition to generate a group abnormal region image, and finally, manually judging the abnormal mode.
The invention mainly aims at the abnormal recognition method of the user space-time cluster mode of the trajectory data, but the invention is not limited by the method, and can realize the abnormal mode recognition of the group space-time cluster aiming at various trajectory data.
Specifically, referring to fig. 1, a flowchart of a user population spatiotemporal anomaly pattern recognition method based on trajectory data according to an embodiment of the present invention is shown, which includes the following steps:
trajectory data preprocessing step S110:
the method comprises the steps of obtaining user track positioning data, converting the abstracted user track positioning data into a data set which can be used for analysis, grouping original data, sequencing the original data according to time sequence, achieving spatialization through latitude conversion, then denoising the data, extracting characteristic point information in track data, classifying the characteristic point information into a starting point, a terminal point and a traveling point, storing the characteristic point information into a complete user track, and exemplarily and finally storing the user track in a database.
In this step, the purpose of the trajectory data preprocessing is to convert trajectory information obtained from the raw data into a data set that can be used for analysis. The following steps can be roughly divided. Spatialization and tracking, noise removal and characteristic point (starting point and end point) extraction.
Wherein the user trajectory location data comprises GPS location data and communication base station location data.
Taking GPS positioning data as an example, since the initial raw data consists of records collected by GPS devices and one file contains many data, it is necessary to first group the data, filter the corresponding device GPS records by device ID, and sort them in time order. And the data is converted into space data according to the longitude and latitude, so that spatialization is realized. The GPS track data has the problems of large partial offset of space noise, large data volume, unstable signals, inaccurate positioning and the like, so the data needs to be further denoised.
In the invention, a Hadoop frame can be used for processing data, the track is divided or the required sub-track is extracted according to the requirement, the data is denoised in a segmented mode, and then the cleaned track is stored in a track database. The denoising method adopts discrete point detection (Outlier detection), the core of the discrete point detection is deleting detection data and completely eliminating the influence of the detection data, the principle is detecting unreasonable points which obviously deviate from a route, defining the unreasonable points as discrete points, and then clearing the discrete points as noise data.
Later, the crowd aggregation is analyzed and processed subsequently to further analyze the characteristics of the crowd track, and the characteristic point information in the track data needs to be extracted, so that the GPS information needs to be classified into a starting point, an end point and a traveling point in order to store the complete user track. The starting point and the ending point of the user can be judged according to the change of the user state, and the key track points are reserved so as to achieve the purpose of data processing.
The processed data should be saved, for example, in a database.
The neighboring user concentration calculation step S120:
and according to the current position of the user, calculating the user concentration degree according to the three states of the starting point, the end point and the travel point in the track information characteristic points.
Specifically, the step includes the following substeps
Defining a spatiotemporal range substep S121:
and taking the current position of the user as the center, and converting the time into a space distance for data processing.
The number of user points within a specified distance range is calculated by setting 60km/h (1 km/min) with the vehicle speed inside the city as a reference. Taking the spatio-temporal scope diagram (fig. 2) as an example, the target user in the range of 1km is calculated by taking the plus sign point as the center, and the data of the spatio-temporal cube (represented by the five-pointed star point + the round dot) in the user coordinate range of 1km can be obtained based on the spatial index, and the data (represented by the five-pointed star point) in the spatio-temporal sphere in the user coordinate range of 1km can also be obtained based on the distance calculation.
User quantity sub-step S122:
and counting the number of users in a specified space-time range.
Taking database storage as an example:
when the user has an additional quantity field, such as the number of people in a taxi, the location information may be weighted appropriately. As shown in table 1, the count field is added to the original table to record the aggregation level data.
Table 1:1km space-time sphere user number centralized meter
Three-state aggregation degree calculating sub-step S123:
and calculating a starting point data set, an end point data set and a traveling point data set of the track data, and identifying and analyzing the starting place, the destination and the road traffic state of the user.
Wherein each state user aggregation calculation is only calculated in the corresponding data set. For example, the start point state only takes part in the calculation of the start point data set, and the end point state only takes part in the calculation of the end point data set. The tri-state aggregation value determines the aggregation zone identification of S130.
A space-time aggregation area calculation step S130:
the data of the three states of the starting point, the end point and the advancing point of the user are further sorted, the space-time aggregation areas are calculated according to the aggregation values of the elements, the starting place aggregation area, the destination aggregation area and the traffic aggregation area of the user group are further calculated respectively according to the tendency that the group aggregation is from low density to high density, and the occurrence time and duration time of each aggregation area are obtained.
In a particular embodiment of the present invention,
referring to fig. 3, the improved spatio-temporal clustering algorithm is adopted to analyze the group trajectory data of multiple spaces, multiple time sequences and multiple users. The method comprises the following specific steps:
based on the user aggregation calculation result in step S120, retrieving an adjacent user j in the space-time range of 10km of each user i, obtaining attributes of a number j (ID) and a user adjacent number j (count), calculating a minimum distance dist (i, j) between i and j according to a geographic space position j (get), and constructing an ij vector when j (count) > i (count), wherein the user i belongs to an aggregation area of the user j; when the user i does not have j (count) > i (count) in the space-time range, the space-time range is expanded, the user j of j (count) > i (count) is further retrieved, for example, the change density field can be used as an index sequence, min (dist (i, j)) is calculated, an ij vector is constructed, the user i belongs to the cluster of the user j in the space-time range larger than 10km, and the iteration is carried out until the vector relation of all users is obtained, wherein the vector relation of max (j (count)) is 0. Because the trend vectors of the users in the high space-time density areas within the corresponding time are obtained through the calculation, and the trend vectors between the areas with higher density are far, different distance thresholds are set according to the analysis requirements of different visual scales, and the vector relation is deleted (cut) in a distance-based grading manner. For example, at the 10km level scale, all vectors with dist (i, j) >10km are deleted (clipped). And (3) gathering the users with the reserved vector connection relation into one type to obtain a user cluster, and defining the convex packet (external polygon) of all elements of the same type as an aggregation area.
And respectively calculating a starting place gathering area, a destination gathering area and a traffic gathering area according to the mode, and counting the time range, the time frequency, the area center, the area boundary and the number of users of each cluster on the basis.
TABLE 2 aggregation zone Attribute statistics
In a specific implementation, the cluster ID is a unique identifier of the aggregation area, the time range mainly counts the earliest starting time and the latest ending time of the current cluster, and the time frequency is the frequency of user presentation within 10 minutes, and is stored in a json format. The method comprises the steps of recording a core space-time point of an aggregation area at the center of an area, recording influence boundaries of the area boundaries at different time intervals at the boundary of the area, and storing the influence boundaries in a wkt geographic information format. The number of users is the number of users in total occurrence, and the final category represents a starting place gathering area, a destination gathering area and a traffic gathering area by 1, 2 and 3 respectively.
The json format of the time frequency is described as follows, in ": "is split, the former is the timing ID, the latter is the number of users:
{1：1，2：3，3：3，······，61：2}
a spatio-temporal co-occurrence region calculation step S140:
in this step, the spatio-temporal co-occurrence region is a time-series aggregation comparison process for the currently known user group aggregation region. And preliminarily extracting the co-occurrence areas according to the time range and the area boundary of each aggregation area, and calculating the similarity of the related aggregation areas. In the aspect of similarity calculation, pearson correlation test is mainly performed based on the comparison characteristics of the categories, such as time frequency, region center and user number. And upgrading the space-time gathering areas with high similarity, which are presented periodically in the geographic area, into space-time co-occurrence areas, namely the user group co-occurrence areas in a specified time.
Thus, the steps are:
and further aggregating, comparing and processing the time sequence of each aggregation area according to the currently known user group aggregation areas. Firstly, performing space-time overlapping analysis on a time range and a region boundary, and then performing similarity calculation according to time frequency, user number and region center on the basis of initially acquiring the aggregation regions at the same position to judge the relevance among the attribute values of the aggregation regions.
The degree of acquaintance calculation may employ pearson correlation tests to determine the correlation between attribute values between aggregation zones.
In one particular embodiment:
firstly, performing space-time overlapping analysis on the time range and the region boundary, referring to fig. 4, taking the region boundary of each time point as overlapping basis, taking time as sequence interval, and acquiring user aggregation areas with overlapping area exceeding 80% and time interval length being 70% similar. Here, the time range does not indicate the same time on the same day, but mainly indicates the same period of time on different days, for example, the user aggregation areas of the same location area at 2016 year 08, month 01 08 and at 2016 year 08, month 02, day 08.
On the basis of preliminarily obtaining the aggregation areas at the same position, similarity calculation is carried out according to time frequency, the number of users and the area center, pearson correlation test is adopted in specific calculation, and the correlation among the attribute values of the aggregation areas is judged.
In order to define the space-time co-occurrence region, the aggregation regions in which the same region periodically appears three times or more in one month are collectively referred to as the space-time co-occurrence region. The period is in a one-day and one-week form, and if special conditions exist, special period processing is performed according to the day, such as holidays and the like.
Spatio-temporal behavior pattern recognition step S150:
and mainly judging group behavior patterns of the place according to POI, AOI, news and activities in the co-occurrence area, such as accommodation behaviors of residential aggregation areas and shopping behaviors of commercial aggregation areas. Specifically, based on the current spatio-temporal co-occurrence area boundary, the number of POI inside and the AOI area ratio are calculated, the type mode of the POI is judged, and news and activity labels are attached to further explain the behavior mode of group aggregation.
The method comprises the steps of obtaining space boundaries of each time sequence of a space-time co-occurrence area, calculating a maximum influence boundary, and counting the number of various POIs and the areas of various AOIs in a coverage area, wherein for example, when the number of POIs and the area weighted value of a residential community are larger, the residential community is endowed with lodging behavior attributes. The method is also suitable for identifying other aggregation areas.
After identifying the aggregation areas, the method can be further used for discovering abnormal group behaviors, namely: abnormal group behavior discovery step S160:
abnormal group behavior discovery is mainly performed aiming at special conditions gathered in daily groups. Specific special cases include "a sudden space-time accumulation area occurs", and "the number of users in the periodic space-time accumulation area is abnormal". And (4) performing space-time behavior pattern recognition according to the discovered abnormal group behaviors, and mining the cause of a special situation.
Namely, comprehensive information retrieval such as POI \ AOI \ news \ activity and the like which affect the behaviors of special abnormal groups, and related data are handed to manual processing and research to accelerate the emergency management speed of urban group events.
The steps further include:
firstly, identifying a current group aggregation area according to the latest user position, retrieving whether the current group aggregation area is in a sequence of a known co-occurrence area, verifying the attribution of the current aggregation area by adopting a space-time superposition analysis method of S140 by a specific matching method, and judging the phenomenon as a sudden space-time aggregation abnormal mode when no user space-time co-occurrence area is matched with the current aggregation area; when the time-space co-occurrence areas are matched, if the period of the latest crowd gathering quantity does not accord with the period of the historical crowd quantity, the phenomenon is judged to be an abnormal mode of fluctuation of the number of users in the periodic time-space gathering area, wherein the specific situation is that the number of people increases suddenly or the number of people decreases suddenly.
One specific example is shown below:
in order to further analyze the space-time characteristics of the group abnormal phenomena, a space-time behavior pattern recognition method of S150 is adopted to collect POI \ AOI \ news \ activities and other information in the gathering area respectively. Form the cluster image of the current abnormal area. For example: in 2016, echo-Carnival music in 8 months and 13 days, held in the Shanghai world expo garden, causes the abnormal high-value crowd gathering in the area. The extraction results were as follows:
{
time 2016: 8/13/08
Area: { POI: wkt (Shanghai world exposition), AOI: wkt (Shanghai world exposition Area) }
Event echo Carnival music festival
News:[ http://sh.bendibao.com/tour/2016520/161956.shtm,······]
Other:{cluster_ID}
}
And finally, the information auxiliary management of the urban population is realized by manually checking the abnormal cluster image.
The invention further discloses a storage medium for storing computer executable instructions, which is characterized in that:
the computer executable instructions, when executed by a processor, perform the above-described method for user population spatiotemporal anomaly pattern recognition based on trajectory data.
Further, the space-time aggregation area and the space-time co-occurrence area obtained by the method can be visually displayed on a map.
In summary, the invention has the following advantages:
1. the method is favorable for realizing the aggregative analysis of the trajectory data in urban activities and acquiring the space-time aggregation areas of user activity groups;
2. the method is convenient for the city to inquire the aggregation state of each time node, helps city management personnel to further know the cluster dynamic information of city groups, and enhances the utilization of the information;
3. the abnormal space-time mode of the urban group can be effectively extracted, comprehensive information push is carried out aiming at the condition that the abnormal space-time mode of the urban group is greatly different from the condition of the historical user gathering area, and the judgment of the abnormal behavior of the group by manpower is assisted.
4. The obtained information of the spatial space-time gathering area and the space-time co-occurrence area is beneficial to realizing the map visual display of the information in the subsequent computer software development.
While the invention has been described in further detail with reference to specific preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (7)
1. A user group space-time abnormal pattern recognition method based on trajectory data is characterized by comprising the following steps:
trajectory data preprocessing step S110:
acquiring user track positioning data, converting the abstracted user track positioning data into a data set which can be used for analysis, grouping original data, sequencing the original data according to a time sequence, realizing spatialization through latitude conversion, then denoising the data, extracting characteristic point information in track data, classifying the characteristic point information into a starting point, an end point and a traveling point, and storing the characteristic point information as a complete user track;
the neighboring user concentration calculation step S120:
according to the current position of the user, calculating the user concentration degree aiming at three states of a starting point, a terminal point and a travel point in the track information characteristic points;
a space-time aggregation area calculation step S130:
the data of the three states of the starting point, the end point and the advancing point of the user are further sorted, a space-time aggregation area is calculated according to the aggregation values of the elements, the starting place aggregation area, the destination aggregation area and the traffic aggregation area of the user group are further calculated respectively according to the tendency of low density aggregation to high density aggregation existing in the group aggregation, and the occurrence time and duration time of each aggregation area are obtained;
a spatio-temporal co-occurrence region calculation step S140:
according to the current known user group aggregation areas, further aggregating and comparing the time sequences of the aggregation areas, firstly performing space-time superposition analysis on a time range and an area boundary, and then performing similarity calculation according to time frequency, user number and an area center on the basis of primarily acquiring the aggregation areas at the same position to judge the relevance of each attribute value among the aggregation areas;
spatio-temporal behavior pattern recognition step S150:
judging a group behavior mode of the co-occurrence area according to POI, AOI, news and activities in the co-occurrence area, calculating a maximum influence boundary based on the current space-time co-occurrence area boundary, calculating the number of various POI and the area of the AOI in the coverage area, judging the type mode of the POI and the area of the AOI, and further explaining the behavior mode of group aggregation by adding news and activity labels;
the step S120 includes the sub-steps of,
defining a spatiotemporal range substep S121: taking the current position of a user as a center, and converting time into a space distance for data processing;
user quantity sub-step S122: counting the number of users in a specified space-time range;
three-state aggregation degree calculating sub-step S123: calculating a starting point data set, an end point data set and a traveling point data set of the track data, and identifying and analyzing the starting place, the destination and the road traffic state of the user;
step S130 specifically includes:
1) Retrieving the peripheral adjacent user j of each user i in the space-time range of 10km, and acquiring the user number and the attribute of the user adjacent number j (count);
2) Calculating a minimum distance dist (i, j) between i and j according to the geographic spatial position j (get), and constructing an ij vector when the user adjacent quantity j (count) > i (count);
3) When the user i does not have j (count) > i (count) in the space-time range relative to the user adjacent quantity, the space-time range is expanded, the user j of j (count) > i (count) is further retrieved, min (dist (i, j)) is calculated, and an ij vector is constructed, so that the user i belongs to the cluster of the user j in the space-time range larger than 10 km;
4) Setting a distance threshold, deleting vectors exceeding the distance threshold, grouping users of reserved vector connection relation into one class to obtain a user cluster, and defining an outer convex packet of all elements of the same class as a gathering area;
5) According to 1) to 4), respectively calculating a starting place gathering area, a destination gathering area and a traffic gathering area, and counting the time range, the time frequency, the area center, the area boundary and the number of users of each cluster.
2. The method of claim 1, wherein the user group spatiotemporal anomaly pattern recognition method,
the method further comprises an abnormal group behavior discovery step S160:
aiming at the special condition found by the daily aggregation behavior of the group, the space-time behavior pattern recognition is carried out and the forming reason is mined.
3. The method of claim 2, wherein the user group spatiotemporal anomaly pattern recognition method,
in the step S110, in the step S,
the user track positioning data comprises GPS positioning data and communication base station positioning data, and discrete point detection denoising is adopted.
4. The method of claim 2, wherein the user group spatiotemporal anomaly pattern recognition method,
in step S140, correlation between the attribute values of the aggregation areas is determined by pearson correlation test.
5. The method of claim 2, wherein the user group spatiotemporal anomaly pattern recognition method,
step S160 specifically includes:
firstly, identifying a current group aggregation area according to the latest user position, searching whether the current group aggregation area is in a sequence of known co-occurrence areas, verifying the attribution of the current aggregation area, and judging the special situation as a sudden space-time aggregation abnormal mode when no user space-time co-occurrence area is matched with the current group aggregation area; when the time-space co-occurrence areas are matched, if the period of the latest crowd gathering number does not accord with the period of the historical crowd number, the special condition is judged to be an abnormal mode of fluctuation of the number of users in the periodic time-space gathering area, wherein the specific condition is that the number of people increases suddenly or the number of people decreases suddenly.
6. A storage medium storing computer-executable instructions, characterized in that:
the computer-executable instructions, when executed by a processor, perform the method of trajectory-data-based user population spatiotemporal anomaly pattern recognition according to any one of claims 1 to 5.
7. The storage medium of claim 6,
the space-time aggregation area and the space-time co-occurrence area obtained by the user group space-time abnormal pattern recognition method can be visually displayed on a map.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211155565.8A CN115238834B (en) | 2022-09-22 | 2022-09-22 | User group space-time abnormal pattern recognition method based on trajectory data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211155565.8A CN115238834B (en) | 2022-09-22 | 2022-09-22 | User group space-time abnormal pattern recognition method based on trajectory data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115238834A CN115238834A (en) | 2022-10-25 |
| CN115238834B true CN115238834B (en) | 2022-12-09 |
Family
ID=83667328
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211155565.8A Active CN115238834B (en) | 2022-09-22 | 2022-09-22 | User group space-time abnormal pattern recognition method based on trajectory data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115238834B (en) |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170111245A1 (en) * | 2015-10-14 | 2017-04-20 | International Business Machines Corporation | Process traces clustering: a heterogeneous information network approach |
| CN111860383B (en) * | 2020-07-27 | 2023-11-10 | 苏州市职业大学 | Group abnormal behavior identification method, device, equipment and storage medium |
| CN112463901B (en) * | 2020-11-18 | 2022-05-03 | 浙江工业大学 | Microscopic visual analysis method for high-density colony track data |
| CN113378891B (en) * | 2021-05-18 | 2022-03-29 | 东北师范大学 | Urban area relation visual analysis method based on track distribution representation |
-
2022
- 2022-09-22 CN CN202211155565.8A patent/CN115238834B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN115238834A (en) | 2022-10-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106600960B (en) | Travel origin-destination point identification method based on space-time clustering analysis algorithm | |
| CN110738856B (en) | Mobile clustering-based urban traffic jam fine identification method | |
| Wang et al. | Extracting trips from multi-sourced data for mobility pattern analysis: An app-based data example | |
| CN106323301B (en) | Method and device for acquiring road information | |
| CN109684384B (en) | Trajectory data space-time density analysis system and analysis method thereof | |
| CN111078818B (en) | Address analysis method and device, electronic equipment and storage medium | |
| CN108256560B (en) | Parking identification method based on space-time clustering | |
| CN111475746B (en) | Point-of-interest mining method, device, computer equipment and storage medium | |
| CN108733774A (en) | A kind of unemployment dynamic monitoring method based on big data | |
| CN111291216B (en) | Method and system for analyzing foothold based on face structured data | |
| CN114530038B (en) | Travel interest area extraction method and system based on space-time data clustering | |
| CN109614450A (en) | The method and system of personnel's whereabouts analysis based on multidimensional data | |
| Rodrigues et al. | Impact of crowdsourced data quality on travel pattern estimation | |
| CN112579915B (en) | Analysis method and device for trip chain | |
| CN114442623A (en) | Agricultural machinery operation track field segmentation method based on space-time diagram neural network | |
| Tanner et al. | Anomalies in the Sky: Experiments with traffic densities and airport runway use. | |
| CN111444286B (en) | Long-distance traffic node relevance mining method based on trajectory data | |
| CN115238834B (en) | User group space-time abnormal pattern recognition method based on trajectory data | |
| CN115979215B (en) | Floor recognition method and device and computer readable storage medium | |
| CN116013084A (en) | Traffic management and control scene determining method and device, electronic equipment and storage medium | |
| CN113516850B (en) | Pipeline traffic flow data acquisition method based on space syntactic analysis | |
| Zhao et al. | Study on urban road network traffic district division based on clustering analysis | |
| CN109448379A (en) | A kind of identification of sporadic traffic events of social media data and localization method | |
| Elleuch et al. | Collection and exploration of GPS based vehicle traces database | |
| CN110515961B (en) | Basic geographic information updating demand evaluation method based on space-time big data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |