CN102607553B - Travel track data-based stroke identification method - Google Patents

Abstract

The invention discloses a travel identification method based on travel trajectory data. The trajectory points are divided by speed, and the trajectory points whose speed is lower than a certain speed threshold are combined into candidate staying positions, and then the distance threshold and time threshold are used to identify the candidate The stay positions are merged to determine the real stay point. The above method solves the problems of positioning drift and jitter in the positioning track data of the mobile phone, and the accuracy of the travel recognition is high. When the time threshold is 300 seconds and the distance threshold is 1100m, the recall rate is 87.66%, and the precision rate is 81.56%. At the same time, the method can also realize the analysis of GPS positioning track data by adjusting the distance and time threshold.

Description

A Travel Recognition Method Based on Travel Trajectory Data

technical field

The invention relates to a travel identification method based on travel trajectory data. This method is mainly applied to obtain the travel trajectory by analyzing the mobile phone positioning trajectory data, but it can also be applied to obtain the travel trajectory by analyzing the GPS positioning trajectory data.

Background technique

The traditional behavioral activity information acquisition is mainly based on residents' travel survey and activity log survey, which is carried out through manual survey. Resident travel surveys and activity log surveys have formed a complete set of survey procedures and norms. They have been used for many years at home and abroad, but they have been plagued by several problems, such as heavy burden on the victims, low survey accuracy, and huge costs. wait.

In recent years, with the development of wireless communication technology and Internet technology, devices with spatial positioning capabilities such as mobile phones and portable GPS have been popularized. After setting, they can automatically record the spatial longitude and latitude information of continuous moments. The collected data can form a complete activity trajectory, which greatly enriches and enhances the restoration effect of travel activities, and becomes another effective way to obtain residents' activity behavior data. But at the same time, these passive travel data collection methods often lack corresponding attribute information as a reference, and travel information such as travel endpoints, travel time, travel mode, and travel purpose cannot be obtained directly from trajectory points. Therefore, data processing and mining analysis methods for trajectory information have become a hot spot in the field of residents' travel research.

At present, there are mainly two ways to collect the trajectory of urban active objects: GPS positioning and mobile phone network positioning. The trajectory data obtained by positioning technology only includes the latitude and longitude of each trajectory point and its corresponding time information, and the characteristic information of the activity behavior cannot be directly obtained through the data itself, such as travel time, travel mode, travel purpose, and deeper activity rules, etc. . The basis of the statistical analysis of the above information is to identify two types of activities of travelers, namely, the stay phase and the movement phase of the activity. Therefore, the task of trip recognition is to convert the incomprehensible trajectory into recognizable stop locations and movements between them.

As shown in Figure 1, itinerary recognition is to divide the spatially discrete trajectory points into two categories: stay active points and moving active points. Obtain the position or position range of the person during the stay activity through the stay activity point, and generate the movement path of the person by moving the activity point. In this way, information such as length of stay and purpose of stay activities can be mined and analyzed from the information of the stay location, and information such as travel mode, travel time, and travel distance can be extracted from the movement path.

At present, the itinerary recognition algorithm starts from identifying the stop point, and mainly includes two categories: exploratory method and clustering method. Exploratory methods also include the following:

(1) Based on record gap

The GPS devices used in early vehicle travel surveys did not have batteries and needed to be powered by a running vehicle. After the vehicle is started, the GPS device is powered on to start recording track data, and when the vehicle is turned off, the device is powered off to stop recording. Therefore, the obtained vehicle trajectories have gaps in time, which can be used to distinguish parking behaviors. During the normal driving process, the vehicle will also have a short flameout behavior, and it is necessary to set a time threshold to distinguish it from the vehicle flameout during the stop activity.

(2) Based on static point

This method mainly starts from the speed, and examines the obtained track point data. When the speed of the track point is 0, the track point is considered as a static point, and a stop point can be judged if it is continuously gathered. Due to the existence of positioning error and drift, the speed of the track point during the stay is not always 0, and a speed threshold needs to be set, usually a speed that is lower than the walking speed as the standard.

(3) Based on missing points

In the case of underground or building occlusion, the GPS device will not be able to receive GPS signals, resulting in data loss, that is, the time interval between two adjacent track points collected is greater than the set time interval. In this case, by calculating the velocity between these two points and comparing it with the velocity of several track points before and after the two points, it is judged whether the stay activity or the movement activity occurs when the track point is missing.

(4) Based on direction features

Certain short-term stay activities will change direction at the beginning or end, such as parking to pick up people, pick up and deliver goods, etc., and the stay can be judged by identifying the direction changes of several track points. Du divides stays into long-term confirmed stays and short-term suspected stays, and investigates the direction changes of suspected stays to further verify whether stays occur.

(5) Based on road network

Calculate the distance between the trajectory point and the road network, obtain the trajectory points that deviate from the road network, and make further judgments on these points based on the length of stay.

Trajectory data is generally collected at equal time intervals. Therefore, when dwelling activities occur, a large number of trajectory points will gather near a certain location, so clustering methods can be used for identification. Clustering methods also include the following methods:

(1) Based on K-means clustering

The method first needs to determine two parameters: the minimum number of trajectory points n forming a cluster and the clustering radius d. Starting from the first track point, calculate the maximum distance between any two points in the n track points, if it is less than d, these track points form a cluster, that is, a stop position. After that, calculate the distance between the next trajectory point and the center point of the cluster, if it is less than d/2, the trajectory point will be added to the cluster, otherwise the clustering process of the cluster will end. The above process is repeated until all track points are processed, so that several staying positions are clustered.

(2) Clustering based on DBSCAN

Similar to K-means clustering, it is also necessary to determine the number of trajectory points n and the clustering radius d. Calculate the number of trajectory points within the range of each point d, if it is less than n, the point is considered to be noise, otherwise these trajectory points form a cluster. If there is overlap between the clusters, the intersecting clusters are merged to form several dwell positions at last. This method assumes that trajectory points are always recorded isochronously and is susceptible to missing data.

The exploratory method is based on the researcher's understanding of the travel trajectory data and the experience of personal travel rules, and sets and continuously optimizes multiple identification rules and parameters to achieve the purpose of travel identification. This method is close to the experience of the real world, the required rules and parameters are intuitive and clear, and reasonable settings can achieve good recognition results. However, the exploratory method is relatively specific to the data. If the acquisition method and characteristics of the trajectory data change, the existing methods will no longer be applicable. The clustering algorithm has a weak dependence on the existing knowledge experience and the characteristics of the data itself, and has good adaptability, but whether it is K-means or DBSCAN clustering, the processing effect on noise and long-distance drift is poor, and it is easy to stop at one time. Divided into multiple stays, the recognition accuracy is not high.

The above-mentioned exploratory method is mainly aimed at the travel trajectory data obtained by GPS positioning, and it is difficult to apply to the travel identification analysis of mobile phone positioning trajectory data. The trajectory data obtained by GPS positioning and mobile phone positioning are very different in terms of trajectory feature positioning accuracy. GPS positioning trajectory data has high positioning accuracy and is not prone to long-distance drift, while mobile phone positioning trajectory data positioning accuracy and drift The characteristics all depend on the distribution density of mobile base stations. In areas far away from the city center, the drift distance is relatively large, and sometimes there may even be positioning drift and jitter of one or two kilometers; most people's mobile phones generally do not turn off during the day or some people The method based on recording gaps is greatly limited if the phone is not turned off all day; the mobile phone signal coverage in the city is relatively comprehensive, and it is not easy to miss the track in the building, and the usability of the method of judging the stay by the missing point is reduced. Therefore, the identification method for the GPS trajectory is not applicable to the positioning trajectory of the mobile phone.

Contents of the invention

The invention designs and develops a travel identification method based on travel trajectory data. The present invention divides track points by speed, and merges track points whose speed is lower than a certain speed threshold into candidate stay positions, and then uses distance and time thresholds to merge candidate stay positions, thereby determining the real stay point. The above method solves the problem of positioning drift and jitter in the positioning track data of the mobile phone, and has high accuracy of travel recognition; at the same time, the method can also realize the analysis of the GPS positioning track data by adjusting the distance and time thresholds.

The technical scheme provided by the invention is:

A method for identifying a trip based on travel trajectory data, comprising the following steps:

Step 1. Calculate the velocity of the track point;

Step 2. Merge a plurality of adjacent track points whose speeds are all below the speed threshold into a candidate stay position, wherein the stay time of the candidate stay position is from the first track point to the last track point in the plurality of track points the time interval between a track point;

Step 3. When the distance between the centers of multiple candidate stay positions and any one of the multiple candidate stay positions is less than the distance threshold, and when the first candidate stay among the multiple candidate stay positions When the time interval between the start moment of the stay duration of the position and the end moment of the stay duration of the last candidate stay position is greater than the time threshold, the center of the multiple candidate stay positions is determined as a stay point;

Step 4: The time interval between the start time of the dwell time of the first candidate stay position and the end time of the stay time of the last candidate stay position among the plurality of candidate stay positions is the stay time at the stay point.

Preferably, in the itinerary identification method based on travel trajectory data, the third step is realized in the following manner,

(1) taking the first candidate stay position among all candidate stay positions to be determined as a stay sequence, wherein the first candidate stay position is the center of the stay sequence,

(2) When the distance between the first candidate stay position at the rear of the stay sequence and the center of the stay sequence is less than the distance threshold, place the first candidate stay position at the rear of the stay sequence enter the dwell sequence, re-center the dwell sequence,

(3) Repeat (2) until the distance between the first candidate stay position behind the stay sequence and the center of the stay sequence is greater than the distance threshold, when the stay of the first candidate stay position in the stay sequence When the time interval between the start time of the duration and the end time of the last candidate stay position is greater than the time threshold, the center of the stay sequence in (2) is the stay point.

Preferably, in the itinerary identification method based on travel trajectory data, in the second step, the plurality of adjacent trajectory points whose speeds are all below the speed threshold are merged into a candidate stop position by the following method achieved,

(1) Calculate the average coordinates (X _{(i, i+1)} , y _{(i, i+1)} ) of two adjacent track points in the candidate stay position in turn,

(2) Calculate in turn the time interval Δt _{(i, i+1)} between the two adjacent trajectory points, and the ratio weight _{(i, i+ 1)} ,

(3) Calculate the coordinates (Stay'.x, Stay'.y) of the candidate stay position: ${\mathrm{stay}}^{\′}.x=\frac{1}{\mathrm{no}-1}\underset{1}{\overset{\mathrm{no}-1}{\mathrm{\Σ}}}{\mathrm{weight}}_{(i,i+1)}\&Center\; Dot;x_{(i,i+1)},$ ${\mathrm{stay}}^{\′}.\mathrm{the\; y}=\frac{1}{\mathrm{no}-1}\underset{1}{\overset{\mathrm{no}-1}{\mathrm{\Σ}}}{\mathrm{weight}}_{(i,i+1)}\·{\mathrm{the\; y}}_{(i,i+1)}.$

Preferably, in the described trip identification method based on travel trajectory data, in the step 3, the center of the stay sequence in (2) is realized in the following manner,

(1) Calculate the stay duration Stay' _i .Δt of the first candidate stay position at the rear of the stay sequence and the start time of the stay duration of the first candidate stay position in the stay sequence to the last candidate stay position The ratio weight _i between the time interval Sq.Δt between the end moments of the dwell time,

(2) Calculate the central coordinates (Sq.x, Sq.y) of the stay sequence:

Sq.x=weight _i Stay' _i .x+(1-weight _i ) Sq.x,

Sq.y=weight _i ·Stay' _i .y+(1-weight _i )·Sq.y.

Preferably, in the described itinerary identification method based on travel trajectory data, it also includes

Step 5. In the step 3, when the distance between the first candidate stay position behind the stay sequence in (3) and the center of the stay sequence is greater than the distance threshold, when the first candidate stay position in the stay sequence When the time interval between the start moment of the length of stay of the stay position and the end moment of the length of stay of the last candidate stay position is less than the time threshold, then the length of stay of the first candidate stay position in the stay sequence described in (2) All trajectory points between the start time and the end time of the last candidate stay position are determined as moving points, and the end time of the last candidate stay position in the stay sequence is to the end of the stay sequence All the track points between the start time of the dwell time of the first candidate stay position are judged as moving points.

Preferably, in the itinerary identification method based on the travel trajectory data, in the step 4, the starting time of the stay duration of the first candidate stay position in the plurality of candidate stay positions is set to the last candidate stay position All track points between the end of the dwell time are deleted.

Preferably, in the method for identifying a trip based on travel trajectory data, in the first step, the average speed of the trajectory points is calculated.

Preferably, in the described trip identification method based on travel trajectory data, in the first step, the calculation of the average speed of the trajectory points is achieved in the following manner,

Select at least one track point in front of and behind the current track point, calculate the straight-line distance from the first track point to the last track point, and calculate the time interval between the first track point and the last track point, The average speed of the current track point is obtained by dividing the linear distance from the first track point to the last track point by the time interval between the first track point and the last track point.

Preferably, in the described trip identification method based on travel trajectory data, in the first step, the calculation of the average speed of the trajectory points is achieved in the following manner,

Select at least one track point in front of and behind the current track point, respectively calculate the straight-line distance between two adjacent track points, and calculate the distance between the straight-line distances between all two adjacent track points in the selected track point and, calculate the time interval between the first track point and the last track point, the average speed of the current track point is divided by the sum of the straight-line distances between all two adjacent track points in the selected track point The time interval between the first track point and the last track point is obtained.

Preferably, in the trip identification method based on travel trajectory data, the time threshold is 300 seconds, and the distance threshold is 1100 meters.

The itinerary identification method based on the travel trajectory data of the present invention divides the trajectory points by speed, and merges the trajectory points whose speed is lower than a certain speed threshold into candidate stop positions, and then uses the distance threshold and time threshold to identify the candidate stop points. The positions are merged to determine the true stop point. The above method solves the problems of positioning drift and jitter in the positioning track data of the mobile phone, and the accuracy of the travel recognition is high. When the time threshold is 300 seconds and the distance threshold is 1100 meters, the recall rate is 87.66%, and the precision rate is 81.56%. At the same time, the method can also realize the analysis of GPS positioning track data by adjusting the distance and time threshold.

Description of drawings

Figure 1 is a schematic diagram of stroke recognition;

Fig. 2 is the velocity calculation schematic diagram of track point;

Fig. 3 is the schematic diagram that the candidate stay point in the trajectory point is merged into the candidate stay position;

Figure 4 is a flow chart of the stay point recognition algorithm

Fig. 5 is the three-dimensional histogram of the recall rate based on different distance thresholds and time thresholds in the itinerary recognition method based on travel trajectory data of the present invention;

Fig. 6 is a three-dimensional histogram of precision based on different distance thresholds and time thresholds in the itinerary recognition method based on travel trajectory data of the present invention;

Fig. 7 is a schematic diagram of visualization of individual traveler's itinerary recognition results, where Fig. 7(a) is the spatio-temporal path obtained from the original trajectory data. (b) is the space-time path drawn after itinerary identification, the straight line part marked by brackets indicates the identified staying stage, and the bending part between the two straight line parts indicates the identified moving stage.

Figure 8 is a conceptual model hierarchy based on spatio-temporal trajectory data;

Fig. 9 is the relationship mapping diagram in the logical model based on spatio-temporal trajectory data;

Fig. 10 is a visual expression diagram of a single traveler's trajectory in the prototype system for visual analysis and mining of residents' travel trajectories;

Fig. 11 is a visual expression diagram of multiple traveler trajectories in the prototype system for visual analysis and mining of resident travel trajectories;

Figure 12 is the visual expression of multiple date trajectories of a single traveler in the prototype system for visual analysis and mining of resident travel trajectories;

Figure 13 shows the distribution of trajectories of multiple travelers on the same date in the prototype system for visual analysis and mining of resident travel trajectories.

Detailed ways

The present invention will be further described in detail below in conjunction with the accompanying drawings, so that those skilled in the art can implement it with reference to the description.

The present invention provides a travel identification method based on travel trajectory data, comprising the following steps:

Step 1. Calculate the velocity of the track point;

Step 2. Merge a plurality of adjacent track points whose speeds are all below the speed threshold into a candidate stay position, wherein the stay time of the candidate stay position is from the first track point to the last track point in the plurality of track points the time interval between a track point;

Step 3. When the distance between the centers of multiple candidate stay positions and any one of the multiple candidate stay positions is less than the distance threshold, and when the first candidate stay among the multiple candidate stay positions When the time interval between the start moment of the stay duration of the position and the end moment of the stay duration of the last candidate stay position is greater than the time threshold, the center of the multiple candidate stay positions is determined as a stay point;

Step 4: The time interval between the start time of the dwell time of the first candidate stay position and the end time of the stay time of the last candidate stay position among the plurality of candidate stay positions is the stay time at the stay point.

In the itinerary identification method based on travel trajectory data, the third step is realized in the following manner,

(1) taking the first candidate stay position among all candidate stay positions to be determined as a stay sequence, wherein the first candidate stay position is the center of the stay sequence,

(2) When the distance between the first candidate stay position at the rear of the stay sequence and the center of the stay sequence is less than the distance threshold, place the first candidate stay position at the rear of the stay sequence enter the dwell sequence, re-center the dwell sequence,

(3) Repeat (2) until the distance between the first candidate stay position behind the stay sequence and the center of the stay sequence is greater than the distance threshold, when the stay of the first candidate stay position in the stay sequence When the time interval between the start time of the duration and the end time of the last candidate stay position is greater than the time threshold, the center of the stay sequence in (2) is the stay point.

In the itinerary identification method based on travel trajectory data, in the second step, the plurality of adjacent trajectory points whose speeds are all below the speed threshold are merged into one candidate stop position, which is realized in the following manner,

(1) Calculate the average coordinates (X _{(i, i+1)} , y _{(i, i+1)} ) of two adjacent track points in the candidate stay position in turn,

(2) Calculate in turn the time interval Δt _{(i, i+1)} between the two adjacent trajectory points, and the ratio weight _{(i, i+ 1)} ,

(3) Calculate the coordinates (Stay'.x, Stay'.y) of the candidate stay position: ${\mathrm{stay}}^{\′}.x=\frac{1}{\mathrm{no}-1}\underset{1}{\overset{\mathrm{no}-1}{\mathrm{\Σ}}}{\mathrm{weight}}_{(i,i+1)}\&Center\; Dot;x_{(i,i+1)},$ ${\mathrm{stay}}^{\′}.\mathrm{the\; y}=\frac{1}{\mathrm{no}-1}\underset{1}{\overset{\mathrm{no}-1}{\mathrm{\Σ}}}{\mathrm{weight}}_{(i,i+1)}\&Center\; Dot;{\mathrm{the\; y}}_{(i,i+1)}.$

In the described itinerary identification method based on travel trajectory data, in the step 3, the center of the stay sequence in (2) is realized in the following manner,

(1) Calculate the stay duration Stay' _i .Δt of the first candidate stay position at the rear of the stay sequence and the start time of the stay duration of the first candidate stay position in the stay sequence to the last candidate stay position The ratio weight _i between the time interval Sq.Δt between the end moments of the dwell time,

(2) Calculate the central coordinates (Sq.x, Sq.y) of the stay sequence:

Sq.x=weight _i Stay' _i .x+(1-weight _i ) Sq.x,

Sq.y=weight _i ·Stay' _i .y+(1-weight _i )·Sq.y.

In the itinerary recognition method based on the travel trajectory data, it also includes step 5. In the step 3, in (3), the first candidate stop position at the rear of the stop sequence arrives at the center of the stop sequence When the distance is greater than the distance threshold, when the time interval between the start moment of the stay duration of the first candidate stay position in the stay sequence and the end moment of the stay duration of the last candidate stay position is less than the time threshold, then (2) All track points between the start time of the first candidate stay time in the stay sequence and the end time of the last candidate stay time are determined as moving points, and the last stay in the stay sequence All track points between the end of the dwell time of a candidate stay position and the start of the stay time of the first candidate stay position behind the stay sequence are determined as moving points.

In the itinerary recognition method based on travel trajectory data, in the step 4, the starting moment of the length of stay at the first candidate stay position in the plurality of candidate stay positions to the length of stay at the last candidate stay position All track points between the end moments are deleted.

In the itinerary identification method based on travel trajectory data, in the first step, the average speed of the trajectory points is calculated.

In the described trip identification method based on the travel trajectory data, in the step 1, calculating the average speed of the trajectory point is realized in the following manner, at least one trajectory point is selected respectively in front of and behind the current trajectory point, and the calculation is performed from The straight-line distance between the first track point and the last track point, calculate the time interval between the first track point and the last track point, the average speed of the current track point passes from the first track point to the last track point The linear distance between one track point is divided by the time interval between the first track point and the last track point.

In the itinerary recognition method based on the travel trajectory data, in the step 1, calculating the average speed of the trajectory point is realized in the following manner, at least one trajectory point is selected respectively in front of and behind the current trajectory point, and calculated respectively The straight-line distance between two adjacent track points, and calculate the sum of the straight-line distances between all two adjacent track points in the selected track point, and calculate the time between the first track point and the last track point Interval, the average speed of the current track point is divided by the time interval between the first track point and the last track point by the sum of the straight-line distances between all two adjacent track points in the selected track point get.

In the itinerary identification method based on travel trajectory data, the time threshold is 300 seconds, and the distance threshold is 1100 meters.

The method of the present invention can be divided into three parts: (1) speed calculation; (2) candidate stop position generation; (3) stop point recognition.

1. Speed calculation

The obtained original trajectory data does not contain speed information. The first step of the algorithm needs to calculate the traveler's speed at each trajectory point according to the longitude, latitude and time information recorded by the trajectory point. The calculation of the instantaneous speed in the strict sense is difficult and complicated, so consider using the average speed on a section of the track where the track point is located instead.

For GPS positioning track data, its positioning accuracy is high, and long-distance drift is not easy to occur. The speed of the track point is replaced by the average speed on the path composed of the track point and the two track points connected to it, as shown in the figure The trajectory point p ₃ in 2, its speed calculation formula is as follows:

${\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; D.</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2,3</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3,4</span>}<span\; class="notranslate">\; )</span>}}{{\mathrm{<span\; class="notranslate">\; \&Delta;t</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2,3</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&Delta;t</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3,4</span>}<span\; class="notranslate">\; )</span>}}$

In the formula, p ₃ .v——velocity of track point p ₃ ;

D _{(i, j)} - the distance between the trajectory point p _i and the trajectory point p _j ;

Δt _{(2, 3)} ——the time interval between the trajectory point p _i and the trajectory point p _j .

For mobile phone positioning trajectory data, its positioning accuracy and drift characteristics depend on the distribution density of mobile base stations. In areas far away from the city center, the drift distance is relatively large, and sometimes positioning drift and jitter of one or two kilometers may occur. To solve the above problems, this paper tries to use the straight-line distance between track points instead of the track path distance to participate in the speed calculation. As shown in Figure 2, when calculating the velocity of the trajectory point p ₃ , the sum of the adjacent distances between the trajectory points p ₁ , p ₂ , p ₃ , p ₄ , and p ₅ is no longer calculated, and the trajectory points p ₁ , The straight-line distance between p ₅ is taken as the path length traveled by the traveler at time t ₁ and time t ₅ . Calculated as follows:

${\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; D.</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1,5</span>}<span\; class="notranslate">\; )</span>}}{{\mathrm{<span\; class="notranslate">\; \&Delta;t</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1,2</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&Delta;t</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2,3</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&Delta;t</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3,4</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&Delta;t</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4,5</span>}<span\; class="notranslate">\; )</span>}}$

Comparing the calculation results with the actual travel situation, the speed characteristics calculated by this method are consistent with the moving speed of the actual travel activities, and can well improve the influence of positioning jitter on the speed calculation results.

2. Candidate stop location generation

According to the calculated speed, the trajectory points are divided into two types: candidate stay points and candidate moving points, and the continuous candidate stay points are merged into candidate stay positions for the next step of stay judgment. Specifically, it includes two aspects of work: the merging of candidate stop points and the calculation of the coordinates of candidate stop positions.

(1) Candidate stop point merge

The division of candidate staying points and candidate moving points mainly depends on the set speed threshold. The speed threshold generally takes the lowest speed limit in residents’ travel mode, that is, the speed limit of walking mode. The walking speed of normal people is generally between 3-6 km/h, that is to say, the slowest walking speed is about 0.8m/s. Considering the influence of speed calculation errors and the previous experiments, this paper takes 1 m/s as the speed threshold to classify trajectory points into two categories: candidate staying points and candidate moving points. Afterwards, more than two consecutive candidate stay points ps are merged into a candidate stay position Stay'. The candidate stay position is the position where travelers may have stay activities. Whether it is determined as a stay point requires further processing in the follow-up work. The schematic diagram of this process is shown in Figure 3.

(2) Coordinate calculation of the candidate stop position

Candidate stay points merged into candidate stay positions cannot completely overlap in space, and it is necessary to calculate a stay center that can represent each candidate stay position, that is, the coordinates of the candidate stay positions, based on the coordinates of these candidate stay positions. On the premise that the trajectory data is recorded at equal time intervals, the coordinates of the stop center can be obtained by calculating the average coordinates of each candidate stop point, but in practice, there will be missing data, which will affect the calculation results. Therefore, this paper adopts a time-weighted method to calculate the coordinates of the candidate stop positions.

First, calculate the average coordinates (X _{(i, i+1) ,} y _{(i, i+1)} ) of two consecutive candidate stay points ps _i , ps i+1 in turn:

${\mathrm{<span\; class="notranslate">\; x</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; ps</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; ps</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; x</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}$

${\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; ps</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; ps</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}$

In the formula, ps _i .x—longitude coordinates of candidate stop point ps _i ;

ps _i .y—latitude coordinates of candidate stop point ps _i ;

After that, the ratio of the time interval Δt _{(i, i+1)} between the two candidate stay points to the length of stay Stay′.Δt of the entire candidate stay position is used as the weight weight _{(i, i+1)} of the average coordinate:

${\mathrm{<span\; class="notranslate">\; weight</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&Delta;t</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>}}{{\mathrm{<span\; class="notranslate">\; stay</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; \&Delta;t</span>}}$

Finally, calculate the coordinates of the candidate dwell locations:

${\mathrm{<span\; class="notranslate">\; stay</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}\underset{\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; weight</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; \&Center\; Dot;</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>}$

${\mathrm{<span\; class="notranslate">\; stay</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}\underset{\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; weight</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; \&Center\; Dot;</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>}$

In the formula, Stay′.x—the longitude coordinate of the candidate stay position Stay′;

Stay'.y—the latitude coordinate of the candidate stay location Stay'.

3. Recognition of stop points

After the candidate stay points are merged into candidate stay locations, further analysis is required to obtain the stay points where the traveler's activities occur. The selection and combination of candidate stop positions is mainly done by examining the two factors of distance and time. The algorithm flow chart is shown in Figure 4. The algorithm steps are described in detail as follows:

Step 1. Read the first candidate stay position Stay' ₁ , put it into the stay sequence Sq, and use the coordinates of Stay' ₁ as the center coordinates of the stay sequence Sq.

Step 2, judge whether there are unread candidate stay positions, if so, read the next candidate stay position Stay' _i , calculate the distance D _{(i, Sq)} between the center coordinates of Stay' _i and the Sq center coordinates, and turn Go to step 3; if no, go to step 4.

Step 3, whether D _{(i, Sq)} is less than the set distance threshold Td, if yes, put Stay' _i into the stay sequence Sq, recalculate the center coordinates of the stay sequence Sq, and go to step 2; if not, go to Step 4.

Step 4. Calculate the time interval Sq.Δt between the stay start time Sq.st of Sq and the stay end time Sq.et. The time interval Sq.Δt here is the start time of the stay duration of the first candidate stay position in the stay sequence The time interval until the end of the dwell duration of the last candidate stop position.

Step 5. Whether Sq.Δt is greater than the set time threshold Tt, if yes, the candidate stay positions in Sq are merged into stay points: (1) include the trajectory between the time Sq.st and the time Sq.et of Sq Point deletion (2) the coordinates of all track points between the time Sq.st and the time Sq.et are all replaced with the central coordinates of Sq, and go to step 6; if not, the candidate stop position in Sq does not constitute a stop point, Then all the trajectory points between the time Sq.st and the time Sq.et are judged as moving points, judge whether Sq contains the last candidate stop position, if yes, end this stay point judgment, if not, go to step 6 .

- Step 6. Empty the candidate stay positions in Sq, put Stay' _i into Sq, use the stay center coordinates of Stay' _i as the center coordinates of stay sequence Sq, and go to step 2. When the last candidate stay position is not included in Sq, it is necessary to continue to determine the candidate stay position to identify a new stay point, so the candidate moving point between Stay' _i-1 and Stay' _i is also determined to move the point.

When recalculating the central coordinates of the stay sequence Sq in the above step 2, the method of time weighting is adopted, and the calculation formula is as follows:

${\mathrm{<span\; class="notranslate">\; weight</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; stay</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; \&Delta;t</span>}}{\mathrm{<span\; class="notranslate">\; Sq</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; et</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; Sq</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; st</span>}}$

Sq.x＝weight _i ·Stay′ _i .x+(1-weight _i )·Sq.x

Sq.y＝weight _i ·Stay′ _i .y+(1-weight _i )·Sq.y

In the formula, weight _i - the weight value of the candidate stay position Stay′ _i ;

Sq.st——the stay start time of the stay sequence Sq (the start moment of the stay duration of the first candidate stay position);

Sq.et——the end time of the stay sequence Sq (the end moment of the stay duration of the last candidate stay position).

Sq.x——longitude coordinates of the central point of the stay sequence Sq;

Sq.y——latitude coordinates of the center point of the stay sequence Sq.

4. Evaluation indicators

The indicators for evaluating the recognition effect of the itinerary recognition algorithm mainly include the recall rate and the precision rate. The definitions of the two concepts are as follows:

recall $R=\frac{\mathrm{DS}}{S}$

Precision $P=\frac{\mathrm{DS}}{\mathrm{D.}}$

Among them, DS is the number of identified real stay points, S is the actual number of real stay points, and D is the number of identified stay points.

5. Recognition results

The mobile phone positioning trajectory data of 8 volunteers for a total of 20 days was collected as the research object of mobile phone positioning trajectory data travel recognition. After obtaining the data, volunteers are asked to recall the itinerary of the day, and fill in the activity log based on the travel trajectory to determine the real number of stop points.

The design of the time threshold and distance threshold first refers to the experience obtained from existing research, starting with 120 seconds and 200 meters, and then increasing in units of 60 seconds and 100 meters, and the maximum increases to 600 seconds and 1500 meters. In this way, there are 9 time thresholds and 14 distance thresholds in the experiment, and 126 sets of threshold combinations can be obtained. The stop points identified by each combination are compared with the actual stop points, and the recall rate and precision of each combination are calculated separately. The three-dimensional histograms of recall and precision are obtained as shown in Figure 5 and Figure 6.

It can be seen from Figure 5 and Figure 6 that with the increase of time threshold and distance threshold, the recall rate of stay point recognition decreases and the precision rate increases, the main reason is that a smaller threshold will split a stay into multiple times, while larger thresholds combine more recent stops in time or space into one stop. Through comparative analysis, when the time threshold is 300 seconds and the distance threshold is 1100 meters, the travel recognition of mobile phone positioning trajectory data can achieve better results, with a recall rate of 87.66% and a precision rate of 81.56%.

For GPS positioning trajectory data, the time threshold can be selected as 300 seconds, and the distance threshold can be selected as 200 meters. The effect of travel recognition is better, the recall rate is 88.14%, and the precision rate is 83.25%.

Fig. 7 is a visual representation of the recognition results of the individual traveler's itinerary. Figure 7(a) is the space-time path obtained from the original trajectory data, and Figure 7(b) is the space-time path drawn after travel recognition, the straight line part represents the identified staying stage, and the bending part between two straight line parts represents The identified mobile phase.

The method of the invention has a good effect on overcoming the large-scale drift produced by the mobile phone positioning mode during stationary activities.

Mobile phone positioning trajectory data was obtained by installing location recording software on the mobile phones of recruited volunteers. The installed software is the Mobiletrack software developed by Antfoot (http://mymobiletrack.com/), which supports WinCE, Symbian S60 and Android and other smart phone systems, and adopts the Cell-ID positioning method. A total of 20 days of mobile phone positioning trajectory data was collected from 8 volunteers, and the collection time interval was also 1 minute. The text data in kml format was obtained after the base station information was processed on its website. It mainly includes the time, longitude, and latitude information of the location record. It is also necessary to add record keywords, volunteer number and other information to distinguish the location record.

The traveler's trajectory is a kind of "spatial-temporal trajectory". For the storage of this trajectory data, its spatio-temporal characteristics should be fully considered to ensure the correlation between spatial information, time information and attribute information. The best way is to design and establish a A spatio-temporal data model is used for the storage of trajectories.

Each element in the mathematical form used for trajectory expression is represented by a hierarchical structure diagram (as shown in Figure 8), thereby constructing a conceptual model.

The purpose of designing the logical model is to convert from the conceptual model to the database logical structure that the relational database can handle in the emerging stage, so that these structures can meet the user's requirements in terms of function, performance, integrity, consistency and scalability. FIG. 9 is a relational mapping diagram of a logical model obtained after converting a conceptual model.

After conversion, there are mainly four types of entity objects in the model: traveler, trip, moving track point, and stay. The traveler (PERSON) object represents a traveler in the survey and its basic personal information, including attributes: traveler code (PID), traveler name (PName), gender (Sex), income (Salary), home address (HAddress), driver's license status (PLic), work place (WAddress) and occupation (Occupation). Among them, PID is the main keyword, which is used to identify different travelers. The trip (TRIP) object represents a certain travel activity of the traveler and the relevant attributes of the trip, mainly including: travel code (TID), travel date (Date), traveler keyword (TPID), and travel sequence number (TSequence) , travel purpose (TType) and travel mode (TMode). Among them, TID is the main keyword, which is used to identify each travel activity. The track point (TRACK_POINT) object represents a certain track point experienced by travel activities and its spatio-temporal attribute information, mainly including: track point code (TpID), travel keyword (TpTID), track point time (TpTime), track point longitude coordinates (TpLon) and latitude coordinates (TpLat). Among them, TpID is the main keyword, which is used to identify different trajectory points. The stay (STAY) object represents an important activity place and its space-time and attribute information, mainly including: stay code (SID), traveler keyword (SPID), stay start time (SSTime), stay end time (SETime), stay point Longitude coordinates (SLon), stay point latitude coordinates (SLat), stay activity type (SType). Among them, LID is the main keyword, which is used to mark different places.

After converting the conceptual model, each data table needs to be designed according to the selected database. Oracle is selected as the database for trace storage. According to the designed logic model and the obtained traveler trajectory data, four data tables are designed for the trajectory data: the trajectory point data table of each day, the traveler information table, the stay activity table of each day and the travel activity table of each day.

(1) Track point data table. This table mainly stores the position information of all track points in a day, including position point number, traveler number, time, longitude, latitude, distance from the previous point, speed, direction and travel activity number and other information. The field design of the table is shown in Table 1.

Table 1 Track point data table

(2) Traveler information form. This table mainly stores the relevant information of the traveler, mainly including the traveler number, traveler's name, gender, income, residence information, residence coordinates, driver's license status, occupation, work place and work place coordinates, etc. The field design of the table is shown in Table 2.

Table 2 Traveler Information Form

(3) stay activity table. This table mainly stores information about stay activities, including traveler number, stay start time, stay end time, location point number corresponding to stay start time, stay activity type, etc. The field design of the table is shown in Table 3.

Table 3 stay activity table

(4) Travel activity list. This table mainly stores information about travel activities, mainly including traveler number, travel activity number, travel purpose and travel mode, etc. The field design of the table is shown in Table 4.

Table 4 Travel Activity List

Through the above design, a database for spatio-temporal trajectory data is established. Further, by using the C# programming language combined with the secondary development based on ArcGIS Engine, the prototype system of visual analysis and mining of residents' travel trajectories is realized. The prototype system for visual analysis and mining of residents' travel trajectories realizes the visualization of residents' travel trajectories in various situations, as shown in Figure 10-13.

In the axSceneControl scene of ArcGIS Engine, the three-dimensional mainly represents the longitude, latitude and height (x, y, z) in the geographic space, and the height z is used to represent the time t, then the three-dimensional scene in the geographic space can be transformed into a time-space A three-dimensional scene, representing longitude, latitude, and time (x, y, t) respectively.

In the specific implementation, ArcGIS three-dimensional type points are created according to the longitude, latitude and time records of the track points obtained from the database, and are added to the axSceneControl scene through the AddElement method provided by the axSceneControl scene control to realize the display of track points. However, the trajectory of the traveler is continuous in time and space. Whether it is to connect the trajectory points into a space-time path, or to relabel the space-time path by dynamic segmentation, it is necessary to interpolate the known trajectory points to obtain other trajectory points. coordinate of. For example, to obtain the space-time coordinates (x _i , y _i , t _i ) of an unknown track point, two known track points (x _m , y _m , t _m ), (x _n , y _n , t _n ) , the space-time coordinates of t _m <t _i <t _n are calculated, generally using the linear interpolation formula:

${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\frac{{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}}{{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}$

${\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\frac{{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}}{{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}}$

The prototype system first provides four basic functions, including trajectory data management and query, trajectory loading and display, plane movement and trajectory time slice display. Subsequent data mining, result display, etc. can be expanded and implemented on top of these basic functions.

The management and query function of trajectory data provides basic operations such as import, maintenance and deletion of trajectory original text data in the form of a graphical interface. At the same time, as the core part of the database operation, it also encapsulates the SQL statement used for trajectory data query, which simplifies the query command and facilitates the calling and development of other functions.

Track loading and display This function realizes the visual display of tracks in 3D scenes. Through the dialog box, select the number and date of the traveler whose trajectory needs to be displayed. The traveler number can be selected through the number list (single selection or multiple selection), and the traveler number of interest can also be manually collected. Similarly, the date can specify a specific date, or choose to display the track of all dates of the traveler after specifying a certain traveler. Once selected, the track data will be displayed in the main form scene.

The time axis movement function of the base map plane can make the base map plane move in the time axis direction, so as to observe the spatial position of the track at a certain moment. In the "Move Timeline" toolbar, select the basemap plane layer to be moved and the time to move to. After the setting is completed, the selected basemap plane will move to the corresponding moment position.

The track point time slice display function displays the spatial positions of all travelers at a certain moment in the scene, which is convenient for users to observe and analyze the spatial distribution and regularity of the crowd. Select the basemap planar layer, data date, and slice moment in the Time Slicing toolbar. After the setting is completed, all the track points at the selected moment will be plotted on the basemap plane.

Although the embodiment of the present invention has been disclosed as above, it is not limited to the use listed in the specification and implementation, it can be applied to various fields suitable for the present invention, and it can be easily understood by those skilled in the art Therefore, the invention is not limited to the specific details and examples shown and described herein without departing from the general concept defined by the claims and their equivalents.

Claims (8)

1. A method for identifying a trip based on travel trajectory data, characterized in that, comprising the following steps: Step 1. Calculate the velocity of the track point; Step 2. Merge a plurality of adjacent track points whose speeds are all below the speed threshold into a candidate stay position, wherein the stay time of the candidate stay position is from the first track point to the last track point in the plurality of track points The time interval between a track point, the multiple adjacent track points whose speed is below the speed threshold are merged into a candidate stop position, which is realized in the following way, (1) Calculate the average coordinates (x _{(i, i+1)} , y _{(i, i+1)} ) of two adjacent track points in the candidate stay position in turn, (2) Calculate in turn the time interval Δt _{(i, i+1)} between the two adjacent track points and the ratio weight _{(i, i+1)} of the length of stay of the candidate stay position Stay·Δt ₎ (3) Calculate the coordinates (Stay' x, Stay' y) of the candidate stay position: Step 3. When the distance between the centers of multiple candidate stay positions and any one of the multiple candidate stay positions is less than the distance threshold, and when the first candidate stay among the multiple candidate stay positions When the time interval between the start moment of the stay duration of the position and the end moment of the stay duration of the last candidate stay position is greater than the time threshold, the center of the multiple candidate stay positions is determined as a stay point, and the step 3 is Achieved by: (1) using the first candidate stay position in all candidate stay positions to be determined as a stay sequence, wherein the first candidate stay position is the center of the stay sequence, (2) When the distance between the first candidate stay position at the rear of the stay sequence and the center of the stay sequence is less than the distance threshold, place the first candidate stay position at the rear of the stay sequence Enter the dwell sequence, re-determine the center of the dwell sequence, (3) Repeat (2) until the distance between the first candidate stay position behind the stay sequence and the center of the stay sequence is greater than the distance threshold, when the stay of the first candidate stay position in the stay sequence When the time interval between the start moment of the duration and the end moment of the duration of the last candidate stay position is greater than the time threshold, then the center of the stay sequence described in (2) is a stay point; Step 4: The time interval between the start time of the dwell time of the first candidate stay position and the end time of the stay time of the last candidate stay position among the plurality of candidate stay positions is the stay time at the stay point. the 2. the itinerary identification method based on travel trajectory data as claimed in claim 1, is characterized in that, in described step 3, the center of stay sequence described in (2) is realized by the following way, (1) Calculate the stay duration Stay' _i Δt of the first candidate stay position at the rear of the stay sequence and the start time of the stay duration of the first candidate stay position in the stay sequence to the last candidate stay position The ratio weight _i between the time interval Sq·Δt between the end moments of the dwell time, (2) Calculate the central coordinates (Sq x, Sq y) of the stay sequence: Sq·x=weight _i ·Stay′i·x+(1-weight _i )·Sq·x Sq·y=weight _i ·Stay′i·y+(1-weight _i )·Sq·y. 3. the itinerary identification method based on travel trajectory data as claimed in claim 1, is characterized in that, also comprises Step 5. In the step 3, when the distance between the first candidate stay position behind the stay sequence in (3) and the center of the stay sequence is greater than the distance threshold, when the first candidate stay position in the stay sequence When the time interval between the start moment of the length of stay of the stay position and the end moment of the length of stay of the last candidate stay position is less than the time threshold, then the length of stay of the first candidate stay position in the stay sequence described in (2) All trajectory points between the start time and the end time of the last candidate stay position are determined as moving points, and the end time of the last candidate stay position in the stay sequence is to the end of the stay sequence All the track points between the start time of the dwell time of the first candidate stay position are judged as moving points. the 4. the itinerary identification method based on travel track data as claimed in claim 1, is characterized in that, in described step 4, the start moment of the length of stay of first candidate stay position in described multiple candidate stay positions to All track points between the end of the dwell time of the last candidate stay position are deleted. the 5. The itinerary identification method based on travel trajectory data as claimed in claim 1, characterized in that, in said step one, the average speed of the trajectory points is calculated. the 6. the itinerary recognition method based on travel locus data as claimed in claim 5, is characterized in that, in described step 1, calculates the average speed of locus point, realizes by following way, Select at least one track point in front of and behind the current track point, calculate the straight-line distance from the first track point to the last track point, and calculate the time interval between the first track point and the last track point, The average speed of the current track point is obtained by dividing the linear distance from the first track point to the last track point by the time interval between the first track point and the last track point. the 7. the itinerary recognition method based on travel locus data as claimed in claim 5, is characterized in that, in described step 1, calculates the average speed of locus point, realizes by following way, Select at least one track point in front of and behind the current track point, respectively calculate the straight-line distance between two adjacent track points, and calculate the distance between the straight-line distances between all two adjacent track points in the selected track point and, calculate the time interval between the first track point and the last track point, the average speed of the current track point is divided by the sum of the straight-line distances between all two adjacent track points in the selected track point The time interval between the first track point and the last track point is obtained. the 8. The itinerary identification method based on travel trajectory data according to claim 1, wherein the time threshold is 300 seconds, and the distance threshold is 1100 meters. the