A method for identifying road segment detour behavior based on bicycle trajectory data

Abstract

The application provides a kind of road section bypass behavior identification method based on bicycle trajectory data, comprising the following steps: 1) bicycle trajectory data acquisition and cleaning, the trajectories needing cleaning have low sampling rate and abnormal speed trajectories;2) map matching: project each trajectory to the corresponding road network, realize map matching;3) one-way road bypass determination: consider the bypass trajectory characteristics, determine the bypass behavior according to the bypass form in geometry, on this basis, add road attribute factors to make further determination;4) bypass behavior frequency grade division: count the frequency of bypass events of each section in a week, and divide three grades with frequency window, mark the section on OSM map platform with different colors to represent the bypass severity.The application determines bypass behavior through map matching and geometric method, effectively improves the accuracy of bypass detection.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation information processing technology, and relates to a method for identifying road detour behavior based on bicycle trajectory data, including three aspects: data cleaning, map matching, and detour determination. Background Technology

[0002] As a convenient, green, and popular mode of transportation, bicycles have greatly facilitated people's lives. With the emergence of shared bicycles, bicycle travel in my country has experienced explosive growth after a period of decline, and it has been very helpful in solving the "last mile" problem. However, the "shortcomings" in the cycling environment are also quite obvious. For example, detours greatly affect the convenience and accessibility of travel.

[0003] By reviewing relevant domestic and international studies on detour behavior on road sections, it was found that the current research focuses mainly on macro-regional path selection and path preference analysis. It mainly analyzes the differences between the selected path and the shortest path from the perspective of regional land use attributes, and cannot yet reflect the cycling behavior of bicycles on road sections.

[0004] Therefore, how to provide a method for identifying road detour behavior based on bicycle trajectory data is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] To address the shortcomings of existing research, this invention proposes a method for identifying road detour behavior based on bicycle trajectory data. Its key feature is that it eliminates the need for a geographic information system (GIS) platform to identify detour behavior; instead, it uses map matching and geometric methods to determine detour behavior, effectively improving the accuracy of detour behavior identification. Detour behavior is specifically defined as the continuous riding of the same vehicle in both upward and downward directions on the same road.

[0006] The specific solution adopted in this invention is as follows:

[0007] Step 1: Acquisition and cleaning of bicycle trajectory data, mainly cleaning trajectory data with low sampling rate and abnormal speed;

[0008] Step 2: Map matching, projecting each trajectory onto the corresponding road segment;

[0009] Step 3: On a one-way road, determine the bicycle's detour behavior;

[0010] Step 4: Classify the frequency of detour behavior.

[0011] in:

[0012] The process of acquiring and cleaning bicycle trajectory data in step 1 is as follows:

[0013] Step 1.1: Bicycle trajectory data comes from shared bicycles, and shared bicycle trajectory data comes from the rider's corresponding APP. The shared bicycle trajectory data is obtained based on the shared bicycle order data and dynamic data, including order number, trajectory timestamp, latitude and longitude coordinates.

[0014] Step 1.2: Clean the trajectory data with a low sampling rate. Trajectory data with a low sampling rate refers to trajectory data that is lost due to GPS system failure or mobile communication network being affected by tall buildings.

[0015] Step 1.3: Clean up trajectory data with abnormal speeds. Trajectory data with abnormal speeds refers to trajectories where the travel speed is higher than 20km / h. This may be because the cyclist did not lock the bike after finishing the ride, and their walking or public transportation trajectory was still mistakenly identified as a cycling trajectory.

[0016] The map matching process in step 2 is as follows:

[0017] Step 2.1: Road network data preparation.

[0018] Step 2.2: Index the trajectory sub-chains and trajectory points in the sub-chains according to the time series.

[0019] Step 2.3: After each trajectory point is matched with the map, four attributes will be added: road segment, road level, road length, and one-way / two-way identification.

[0020] Step 2.4: Delete invalid matching results.

[0021] The process for determining detour behavior in step 3 is as follows:

[0022] The specific definition of detour behavior is: riding along the same road in both directions. Steps 3.1, 3.2, and 3.3 below first determine the detour behavior from the perspective of the trajectory chain, without considering road attributes, based on the geometric detour form.

[0023] Step 3.1: Smooth the trajectory points with abnormal orientation angles, as well as the start and end points, in the trajectory chain. Taking true north as 0°, set an orientation angle interval every 90° in a clockwise direction, and set the abnormal points, start and end points of the orientation angle intervals to be consistent with the orientation angle intervals of the neighboring points according to the set conditions.

[0024] Step 3.2: Set the sliding window according to the number of trajectory points in the trajectory chain, as shown below:

[0025]

[0026] Where n is the number of trajectory points in a trajectory chain, and w is the number of sliding windows.

[0027] Step 3.3: Determine the bicycle's detour behavior by analyzing the difference in directional angles generated on both sides of the one-way road based on the cycling trajectory. The formula is as follows:

[0028] When n>15:

[0029]

[0030] When n < 15:

[0031]

[0032] Where distance(a,b) represents the distance between point a and point b, i represents the i-th trajectory point in the trajectory chain, and j represents the j-th trajectory point in the trajectory chain.

[0033] Step 3.5: To further optimize the above scheme, based on the above steps, road attribute factors are added to further identify detour behavior.

[0034] Step 3.6: Combine and match road segments in the cleaned road network data (removing two-way roads and branch roads) (combine and match the corresponding road segments on both sides of one-way roads in sequence). Two-way roads refer to lanes that allow two-way traffic, and branch roads refer to roads within a region.

[0035] Step 3.7: Determine the trajectory chain detour behavior:

[0036]

[0037] Where i represents the i-th trajectory point in the trajectory chain; roadgroup represents road pairing data; i∈(roadgroup) means that the trajectory chain is located in the paired road segment.

[0038] n i d represents the number of trajectory points i; i This represents the distance between trajectory point i and trajectory point i+1; Indicates the direction angle of trajectory point i; θ represents the direction angle of road j. The allowed floating value is 20°.

[0039] Step 3.8: Determine the detour distance. The detour distance in the trajectory chain must be greater than 20 meters.

[0040] The process of classifying the frequency levels of detour behavior in step 4 is as follows:

[0041] The frequency of bicycle detour events on each road segment within a week is counted, and three levels are divided using an appropriate frequency pane. The road segments are marked on the GIS map platform with green, orange, and red colors to indicate the severity of detours, so as to observe the detour behavior status at the road network level.

[0042] This invention first cleans massive amounts of shared bicycle trajectories; secondly, it performs map matching on the riding trajectories; and then, using geometric knowledge and considering road attribute factors, it determines whether detour behavior has occurred. It boasts high accuracy and reliability, filling the gap in existing technologies for determining detour behavior at the microscopic level, and plays a crucial role in promoting the development of public bicycle systems. This invention determines detour behavior through map matching and geometric methods, and uses detour behavior detection to assess the detour behavior status at the road network level, demonstrating significant innovation. Attached Figure Description

[0043] Figure 1 This is a flowchart illustrating a method for identifying road detour behavior based on bicycle trajectory data.

[0044] Figure 2 A schematic diagram of the study area (Xicheng District, Beijing);

[0045] Figure 3 This is a schematic diagram showing the location of the boundary points of the study area;

[0046] Figure 4 This is a schematic diagram of the trajectory drift points;

[0047] Figure 5 This is a diagram illustrating the missing trajectory points.

[0048] Figure 6 This is a diagram illustrating the detour route.

[0049] Figure 7 This is a schematic diagram of paired road segments for a one-way road;

[0050] Figure 8 A schematic diagram illustrating the division of the detour behavior frequency window. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0052] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0053] The flowchart of the method for identifying road segment detour behavior based on bicycle trajectory data is as follows: Figure 1 As shown, it includes the following steps:

[0054] Step 1: Acquisition and cleaning of bicycle travel trajectory data. Trajectories that need to be cleaned include those with low sampling rates and those with abnormal speeds.

[0055] Step 2: Map matching;

[0056] Step 3: Determine whether bicycles will detour on a one-way road;

[0057] Step 4: Classify the frequency of detour behavior.

[0058] The following are application examples of the present invention.

[0059] The determination of shared bicycle detour behavior within the Xicheng District of Beijing, such as Figure 2 As shown.

[0060] (1) Acquisition and cleaning of bicycle trajectory data

[0061] 1) Bicycle trajectory data comes from shared bicycles. The shared bicycle trajectory data is obtained based on shared bicycle order data and dynamic data, including the order number, time, latitude and longitude coordinates of each trajectory point.

[0062] 2) Clean up trajectory chains with fewer than 7 trajectory points.

[0063] 3) Clean trajectory chains with a length of less than 400m.

[0064] 4) Cleaning trajectory points outside the area boundary (in this embodiment, the area is set as Xicheng District, Beijing). The boundary location data and road network data of the study area were obtained through the open-source OSM (OpenStreetMap) map platform. Boundary data was mainly obtained through data cleaning to delineate the shared bicycle riding trajectories within the study area. For example... Figure 3 As shown, these are the locations of the boundary points in Xicheng District, with each boundary point corresponding to a set of latitude and longitude data.

[0065] 5) Calculate the direction angle, time difference, and speed of the switch lock point and adjacent trajectory points.

[0066] 6) Clean up the drift points in the trajectory chain. Figure 4 As shown, a drift point is defined as a point on the track where the instantaneous velocity exceeds the maximum riding speed; that is, the position of the track point exceeds the maximum riding distance at normal speed. The velocity of a track point is obtained by dividing the distance between two adjacent points by the time interval between them. The drift point determination principle is as follows:

[0067] distance(p i -p i-1 )>k×time(p i -p i-1 )

[0068] Where, p i Let represent the i-th trajectory point on the travel chain, and k be the upper limit of cycling speed. By plotting the velocity box plot of the trajectory point, the upper limit k = 20 km / h is obtained. distance(p i -p i-1 ) represents the trajectory point p i -p i-1 The cycling distance between them, time(p) i -p i-1 ) represents the trajectory point p i -p i-1 The cycling time between.

[0069] 7) Missing point handling: When the time interval between two adjacent points in a trajectory chain is too large, it is determined that there is a missing point between the two trajectory points. A diagram illustrating missing points is shown below. Figure 5 As shown. The presence of missing points reduces the accuracy of the trajectory chain, especially significantly affecting the identification of some micro-cycling behaviors. Therefore, the trajectory chain is truncated at points with excessively large spacing, forming multiple trajectory sub-chains. The missing point determination principle is as follows:

[0070] time(P i -P i-1 )>t

[0071] Where t represents the point time-distance outlier. By plotting the point time-distance box plot, the threshold for the point time-distance outlier is found to be 60s.

[0072] (2) Map matching

[0073] 1) Road Network Preparation: Road network data within the study area was obtained from the OSM platform, and internal roads and two-way roads, which are not within the study scope, were removed. Furthermore, considering the current road conditions in Beijing, four road types were retained: expressway auxiliary roads, arterial roads, secondary arterial roads, and local roads. To ensure the stability of detour frequency, one-way roads with high cycling volume were selected as the study subjects.

[0074] 2) Road network data: Road network data consists of road segments and nodes. Each road segment contains four elements: road sign, road grade, road direction, and road azimuth. Each node contains three elements: road number, longitude, and latitude.

[0075] 3) Indexing: Indexing the trajectory chain and the trajectory points within the chain according to the time series;

[0076] 4) Matching: The IVMM map matching algorithm considers the spatiotemporal network topology information of GPS trajectories and designs an interactive voting strategy to simulate the weighted mutual influence between trajectory points, thus solving the problem of incorrect matching of neighboring roads caused by trajectory data drift in complex road networks. This scheme improves upon the IVMM map matching algorithm by removing roads that are not accessible to bicycles, such as highways and expressways; changing the road topology from a directed graph to an undirected graph, disregarding road direction information; and using offset, i.e., the projected distance between the trajectory point and the centerline of the road segment. Parameter calibration includes the maximum number of candidate road segments and the neighborhood query radius. Finally, after map matching, each trajectory point gains four attributes: road ID, road class, road length, and one-way / two-way identifier.

[0077] 5) Deletion of invalid matching results: Some trajectories appear in areas not covered by the road network (such as residential areas). When these trajectories are projected onto adjacent road segments, it will cause a large projection deviation that exceeds the threshold. Other trajectories meander near roads, causing irregular changes in the projected travel distance. These two types of trajectories are invalid and need to be deleted.

[0078] (3) Determination of bicycle detour behavior on one-way roads

[0079] 1) The specific definition of detour behavior is: the behavior of the same vehicle continuously riding in the up or down direction on the same road, such as... Figure 6 As shown.

[0080] The following steps first start from the perspective of the trajectory chain, without considering road attributes, and determine the detour behavior based on the geometric detour form.

[0081] 2) Smoothing: Smooth the trajectory points, start points and end points with abnormal orientation angles in the trajectory chain to avoid misjudgment of the results due to local orientation angle abnormalities.

[0082] With true north as 0°, a direction angle interval is set every 90° in a clockwise direction for smoothing. Smoothing means that the direction angle intervals of several adjacent trajectory points in a chain are consistent. If the interval of one trajectory point differs from the intervals of the others, the different trajectory point direction angle intervals are set to the same interval. The requirements for smoothing abnormal trajectory points are as follows:

[0083]

[0084] Among them, C i This represents the directional angle interval of the i-th trajectory point in a trajectory chain.

[0085] The requirements for smoothing the initial and final points are as follows:

[0086]

[0087] Where C0 represents the direction angle interval of the initial trajectory points of a trajectory chain, n represents the number of trajectory points in a trajectory chain, and c n-1 This represents the range of direction angles of the final trajectory point in a trajectory chain.

[0088] Abnormal trajectory points are smoothed according to probability values ​​as follows:

[0089]

[0090] in, This represents the number of trajectory points within the same angular interval in a trajectory chain. This represents the maximum number of trajectory points within the same angular interval, where λ represents the λth trajectory point.

[0091] 3) Set the sliding window according to the number of trajectory points in the trajectory chain. Divide a trajectory chain into five intervals according to the number of trajectory points, and set the corresponding number of sliding windows, as shown below:

[0092]

[0093] Where n is the number of trajectory points in a trajectory chain, and w is the number of sliding windows.

[0094] 4) Detour Determination: The detour behavior of a bicycle can first be determined by the difference in the direction angle generated by the bicycle track chain on both sides of the one-way road, as shown in the following formula:

[0095] When the number of trajectory points n > 15, detour determination is made based on the difference between the number of trajectory points and the number of sliding windows:

[0096]

[0097] When the number of trajectory points n < 15, detour determination is made based on the difference between the number of trajectory points and the number of sliding windows:

[0098]

[0099] Here, distance(a,b) represents the distance between point a and point b.

[0100] To further optimize the above scheme, road attribute factors are added to the above steps to further identify detour behavior.

[0101] 5) Road segment pairing: The cleaned road network data (excluding two-way roads and secondary roads) is then used to pair road segments (the corresponding road segments on both sides of a one-way road are paired sequentially), such as... Figure 7 As shown, the blue sections represent the paired road sections on both sides of the road.

[0102] 6) Determining the behavior of trajectory chains:

[0103]

[0104] Where i represents the i-th trajectory point in the trajectory chain; roadgroup represents road pairing data; i∈(roadgroup) means that the trajectory chain is located in the paired road segment;

[0105] n i d represents the number of trajectory points i; i This represents the distance between trajectory point i and trajectory point i+1; Indicates the direction angle of trajectory point i; Indicates the direction angle of road j; indicates the angle range (-20°, 20°).

[0106] 8) Detour distance judgment: The detour distance in the trajectory chain is required to be greater than 20m.

[0107] (4) Classification of detour behavior frequency levels

[0108] 1) Count the frequency of bicycle detour events on each road segment within a week, and divide them into three levels using an appropriate frequency pane. Mark the road segments on the GIS map platform with green, orange, and red colors to indicate the severity of detours, so as to observe the detour behavior status at the road network level. Figure 8 As shown.

[0109] Matters not covered in this invention are common knowledge.

[0110] The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All equivalent changes or modifications made in accordance with the spirit and essence of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for identifying road segment detour behavior based on bicycle trajectory data, characterized in that, Determining detour behavior through map matching and geometric methods, the identification method includes the following steps: Step 1: Obtaining and cleaning bicycle travel trajectory data; Step 2: Map matching, projecting each bicycle's trajectory onto the corresponding road; Step 3: Determine bicycle detours on a one-way road; determine bicycle detours on a one-way road based on the trajectory chain, as follows: Step 31: Smooth the trajectory points with abnormal direction angles, as well as the start and end points, in the trajectory chain; Step 32: Set the sliding window according to the number of trajectory points in the trajectory chain, as shown below: ； in The number of trajectory points in a trajectory chain. This represents the number of sliding windows; Step 33: The bicycle's detour behavior is determined by the difference in direction angle generated by the bicycle's trajectory chain on both sides of the one-way road, using the following formula: When n>15: ； When n < 15: ； in, In order to seek Click Distance between points; The method for determining bicycle detours on one-way roads based on road attributes is as follows: Step 311: Incorporate road attribute factors to further determine detour behavior; Step 312: Remove bidirectional road data and branch road data, and combine and pair the corresponding road segments on both sides of the cleaned road network data. Step 313: Determine the trajectory chain detour behavior: ； in, This represents the i-th trajectory point in the trajectory chain; This represents road pairing data; This indicates that the trajectory chain is located in the paired road segment; Represents trajectory points The number of; Represents trajectory points With trajectory points The distance between them; Represents trajectory points Direction angle; Indicates road The direction angle; indicating the angle range (-20°, 20°); Step 4: Classify bicycle detour behavior into levels based on the frequency of bicycle detour events.

2. The method for identifying road segment detour behavior based on bicycle trajectory data according to claim 1, characterized in that, The trajectory data acquisition and cleaning process in step 1 is as follows: Step 11: Bicycle trajectory data comes from shared bicycles. The shared bicycle trajectory data is obtained based on the shared bicycle order data and dynamic data. The shared bicycle trajectory data includes the order number, trajectory point timestamp, and latitude and longitude coordinates. Step 12: Clean the trajectory data with a low sampling rate. Trajectory data with a low sampling rate refers to trajectory data that is lost due to GPS system failure or the influence of tall buildings on the mobile communication network. Step 13: Clean up trajectory data with abnormal speed. Trajectory data with abnormal speed refers to trajectories whose travel speed is higher than a predetermined threshold.

3. The method for identifying road segment detour behavior based on bicycle trajectory data according to claim 1, characterized in that, The method for map matching in step 2 is as follows: Step 21: Road network data preparation; Step 22: Index the trajectory sub-chains and trajectory points within the sub-chains according to the time series; Step 23: After each trajectory point is matched with the map, five attributes will be added: road segment ID, road level, road length, and one-way / two-way identification. Step 24: Delete invalid match results.

4. The method for identifying road segment detour behavior based on bicycle trajectory data according to claim 1, characterized in that, The method for classifying bicycle detour behavior in step 4 is as follows: count the frequency of bicycle detour events on each road segment within a week, divide the road segment into three levels using a predetermined frequency pane, and mark the road segment with three colors on the GIS map platform to indicate the severity of detours, thereby displaying the detour behavior status at the road network level.

Images