Urban road network toughness sequence parameter measurement method based on GPS trajectory data

Abstract

The invention provides an urban road network toughness sequence parameter measurement method based on GPS trajectory data, which is based on floating car GPS trajectory data, and comprises the following steps: determining a research area range, and constructing a research area road network topology; performing data cleaning on the original floating car GPS trajectory data according to the road network topology; on the basis of road network topology, track map matching is conducted through the cleaned data, and the average travel speed of each track is calculated; and S3, repeating the steps S2 and S3 in each research time window, respectively establishing empirical distribution of the average travel speed of the track, and extracting a mean value and 85% quantile of the distribution as measurement indexes of the road network toughness sequence parameters. According to the method, from the perspective of travelers, floating car GPS track data are utilized, and the mean value and 85% quantile of track average travel speed distribution are adopted as sequence parameter indexes, so that the running state changes of a road network under various disturbance events are more comprehensively and accurately captured; and refined scientific basis and decision support are provided for urban traffic planning, management, emergency response and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic engineering, and in particular to a method for measuring order parameters of urban road network resilience based on GPS trajectory data. Background Art

[0002] The urban transportation system is an important lifeline system of the city, and the urban road network is a key component of it. Its normal operation is crucial to the city's economic development, residents' lives, emergency rescue, etc. However, in the daily operation process, the urban road network will inevitably be affected by various disturbance events, such as traffic accidents, major failures, natural disasters, major events, etc. These disturbance events will not only cause traffic paralysis on local sections of the road, but also easily cause significant damage to the operating efficiency and functional integrity of the entire road network through the dynamic propagation characteristics of traffic flow.

[0003] Urban road network resilience refers to the ability of an urban road network to maintain its structural and functional stability, recover quickly and adapt to changes after being subjected to various disturbances and damages. By accurately measuring road network resilience, we can deeply analyze the performance of road networks in complex dynamic environments, and provide an indispensable key basis for traffic planning, management decisions and the formulation of emergency response strategies.

[0004] In the past, most methods for measuring urban road network resilience were model-driven, and these methods were mainly based on the "route selection-traffic load" paradigm. However, in reality, the complexity and dynamics of the traffic system make it difficult for the above traditional models to fully fit and adapt to the actual road network conditions. Their assumptions may not reflect the actual traffic behavior, resulting in a large deviation between the evaluation results and the actual situation.

[0005] In addition, there are also some data-driven studies that use data from various fixed detectors to monitor road network resilience. However, due to the low coverage of fixed detectors, these methods are difficult to reflect the overall operating conditions of the road network, and the constructed indicators often lack actual physical meaning. Most of them can only be used for abnormal state identification, and it is difficult to effectively characterize the road network operation status from the perspective of travelers' intuitive feelings, such as speed changes, travel time changes, etc.

[0006] In summary, the existing urban road network resilience measurement methods have deficiencies in terms of the authenticity of theoretical models, the extensiveness of data, and the practicality of indicators, and cannot accurately meet the needs of modern urban traffic management and development for road network resilience assessment. Therefore, it is urgent to use high-coverage road network operation data to develop a new, scientific and reasonable road network resilience order parameter to achieve accurate measurement of network resilience. Summary of the invention

[0007] The purpose of the present invention is to solve the above problems and provide a method for measuring the resilience of urban road networks with higher universality and accuracy.

[0008] To achieve the above object, the present invention proposes a method for measuring the resilience order parameter of an urban road network based on GPS trajectory data, which includes the following steps based on the GPS trajectory data of a floating vehicle:

[0009] S1: Determine the scope of the study area and construct the road network topology of the study area;

[0010] S2: Cleaning the original floating vehicle GPS trajectory data according to the road network topology;

[0011] S3: Based on the road network topology, the cleaned data is used to perform trajectory map matching, so as to obtain the travel distance and travel time of each trajectory, and calculate the average travel speed of each trajectory;

[0012] S4: Repeat steps S2 and S3 within each research time window to establish the empirical distribution of the average travel speed of the trajectory, and extract the mean and 85% quantile of the distribution as the measurement index of the road network resilience order parameter. The order parameter index can be used to monitor the evolution of network-level resilience after abnormal events, identify the start and end times of the impact of abnormal events on urban road network traffic, and provide refined scientific basis and decision-making support for urban traffic planning, management, and emergency response.

[0013] Furthermore, step S1 is specifically as follows: obtaining map data of the study area, taking intersections as nodes, road sections as edges, and road section lengths as weights, and constructing a directed network G=(N,A), wherein N represents a node set and A represents a road section set.

[0014] Furthermore, map data of the study area were obtained through the OpenStreetMap platform.

[0015] Furthermore, in step S2, the data cleaning includes time and space range limitation, trajectory denoising and trip segmentation.

[0016] Furthermore, the method of limiting the time and space range is: based on the road network topology and the research time window, excluding GPS track points that are more than 100 meters away from any road network element, and excluding GPS track points that are not within the research time window;

[0017] The trajectory denoising method is as follows: the speed of the current GPS trajectory point is defined as the sum of the distances between the current GPS trajectory point and the adjacent GPS trajectory point divided by the time difference between the adjacent GPS trajectory points. If the value exceeds 120 km / h, or the angle between the current GPS trajectory point and the adjacent GPS trajectory point is less than 30°, the current GPS trajectory point is regarded as an abnormal point and removed;

[0018] The segmentation rule of the itinerary is as follows: if the distance between more than 5 consecutive GPS track points is less than 20 meters, and the total time interval between these GPS track points exceeds 1 minute, then the point closest to the center among these GPS track points is selected as the stop point, and this is used as the basis for segmenting the itinerary.

[0019] Furthermore, in step S3, trajectory map matching is performed based on the hidden Markov algorithm to obtain the travel distance and travel time of each trajectory and calculate the average travel speed of each trajectory; specifically, it includes:

[0020] S3.1: Based on each GPS track after cleaning Create GPS track points Corresponding candidate road segment set The observation probability is calculated by the following formula

[0021] in, Represents at time t i (i=1,2,…,n) observed GPS track point coordinates; σ x is the GPS measurement standard deviation, For section a j Upper distance The nearest projection point, ||·|| Euclidean is the straight-line distance between two points;

[0022] S3.2: From GPS track point Corresponding candidate road segment a j Move to GPS track point Corresponding candidate road segment a k The probability of is defined as the transition probability, and the transition probability is calculated by the following formula:

[0023]

[0024]

[0025] Where β is the distribution parameter, ||·|| Route is the path distance between two points; and GPS track points and Corresponding to a j and a k The closest projection point on the

[0026] S3.3: Based on the observation probability matrix and the transition probability matrix, the Viterbi algorithm is used to calculate and generate the most likely segment sequence ε = {a j ,ak ,…,a m} and the corresponding projection point sequence Calculate the trajectory travel time T X , track distance D X and the average travel speed of the trajectory Pace X .

[0027] Furthermore, in step S3.3, the trajectory travel time T X , track distance D X and the average travel speed of the trajectory Pace X The calculation formulas are:

[0028] Trajectory travel time T X The calculation formula is: X =t n -t1;

[0029] Track distance D X The calculation formula is:

[0030]

[0031] Average travel speed of track Pace X The calculation formula is:

[0032] Among them, l(·) is the length of the segment, head(·) and tail(·) are the head node and tail node of the segment respectively.

[0033] Furthermore, step S4 specifically includes:

[0034] S4.1: Establish a research time window set TW = {tw1, tw2, ..., tw n}, for each time window tw i Execute steps S2 and S3 respectively to establish each time window tw in turn. i Empirical distribution of average travel speed of trajectories under Among them, X(tw i ) is the time window tw i All trajectories under |X(tw i )| is the number of trajectories, I(·) is the indicator function;

[0035] S4.2: Define each time window tw i The average travel speed of the trajectory under and 85% quantile Pace(tw i ) 85％ , the calculation formula is as follows:

[0036]

[0037]

[0038] in, To set the time window tw i Average travel speed of all tracks within Pace X Sort in ascending order value.

[0039] Compared with the prior art, the advantages of the present invention are:

[0040] 1. Compared with the traditional method that relies on data from fixed detectors, the present invention uses the GPS trajectory data of floating vehicles that are more widely distributed in the urban road network to obtain more comprehensive road network operation information, making the characterization of the resilience of the urban road network more comprehensive and realistic, avoiding the problem of only reflecting local operating conditions due to insufficient coverage of fixed detectors, thereby more accurately capturing the changes in the operating status of the road network under various disturbance events, and ensuring the reliability of subsequent analysis and decision-making.

[0041] 2. Starting from the perspective of travelers and closely focusing on the actual travel experience of travelers, the present invention establishes the order parameter of urban road network resilience from the probability distribution of the average travel speed of the trajectory to characterize the evolution process of network-level resilience after disturbance, and establishes the order parameter index based on the average travel speed of the trajectory, which has physical meaning and can thus more clearly and intuitively reflect the changes in the operation status of the road network.

[0042] 3. The present invention uses the 85% quantile of the average travel speed of the trajectory to focus on the extreme travel conditions (i.e., long-tail risks) after the disturbance occurs, which helps traffic management departments and emergency response agencies to promptly and accurately identify the most severely affected areas and time periods so as to quickly initiate emergency measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A flow chart of a method for measuring the resilience order parameters of an urban road network based on GPS trajectory data proposed in an embodiment of the present invention;

[0044] Figure 2 A schematic diagram of the road network topology of the research area in an embodiment of the present invention;

[0045] Figure 3 Schematic diagram of the probability distribution of the average travel speed of the trajectory between 7:00 and 8:00 and between 8:00 and 9:00 on three days in an embodiment of the present invention;

[0046] Figure 4 It is a schematic diagram of the change of the mean value of the average travel speed distribution of the trajectory at 7:00-8:00 and 8:00-9:00 every day in an embodiment of the present invention;

[0047] Figure 5 It is a schematic diagram of the 85% quantile change of the average travel speed distribution of the trajectory from 7:00 to 8:00 and 8:00 to 9:00 every day in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be further described below.

[0049] The background was the accident in Shanghai on May 23, 2016 where the Central Ring Road was crushed by a truck. The accident resulted in the closure of the inner circle of the Shanghai Central Ring Elevated Road from Gulang Road to Wanrong Road. The closure period was from May 23 to June 6, 2016.

[0050] Taking the above-mentioned 2016·5·23 Shanghai Central Ring Road accident as an example, this embodiment proposes a method for measuring the order parameters of urban road network resilience based on floating vehicle GPS trajectory data. The method is based on the floating vehicle GPS trajectory data of 19 working days from May 9 to June 7, 2016, with the morning rush hour of 7:00-8:00 and 8:00-9:00 as the time windows. By establishing the average travel speed probability distribution of travelers from massive GPS trajectory data, the distribution mean and 85% quantile are used as the road network resilience order parameters, and the constructed road network resilience order parameters are used to characterize the network-level resilience evolution process after the disturbance event.

[0051] like Figure 1 As shown, the method of this embodiment includes the following steps:

[0052] S1) Determine the scope of the study area and construct the road network topology of the study area;

[0053] The map data of the study area was obtained through the OpenStreetMap platform, and a directed network G = (N, A) was constructed with intersections as nodes, road sections as edges, and road section lengths as weights, where N represents the node set and A represents the road section set.

[0054] S2) According to the constructed road network topology, the original floating vehicle GPS trajectory data is cleaned. The data cleaning method includes time and space range restriction, trajectory denoising and trip segmentation, specifically:

[0055] S201, data cleaning by time and space range restriction: based on the road network topology and research time window constructed in step S1, GPS track points that are more than 100 meters away from any road network element are eliminated, and GPS track points that are not within the research time window are eliminated;

[0056] S202, data cleaning by trajectory denoising: define the speed of the current GPS trajectory point as the sum of its distances to adjacent GPS trajectory points divided by the time difference between adjacent GPS trajectory points. If the value exceeds 120 km / h, or the angle between the current GPS trajectory point and the adjacent GPS trajectory point is less than 30°, the current GPS trajectory point is regarded as an abnormal point and removed;

[0057] S203. Data cleaning by itinerary segmentation: Considering that the original floating vehicle GPS trajectory data mainly comes from the online car-hailing platform, it usually includes multiple passenger trips of a vehicle. In order to obtain the average travel speed of each trip, it is necessary to segment the original trajectory. The segmentation rule is: if the distance between more than 5 consecutive GPS trajectory points is less than 20 meters, and the total time interval between these GPS trajectory points exceeds 1 minute, then the point closest to the center among these GPS trajectory points is selected as the stop point, and this is used as the basis for segmenting the itinerary.

[0058] S3) performing trajectory map matching based on the constructed road network topology and the cleaned GPS trajectory data based on the hidden Markov algorithm, thereby obtaining the travel distance and travel time of each trajectory and calculating the average travel speed of each trajectory;

[0059] S301, based on each GPS track after cleaning Create GPS track points Corresponding candidate road segment set The observation probability is calculated by the following formula

[0060] in, Represents at time t i (i=1,2,…,n) observed GPS track point coordinates; σ x is the GPS measurement standard deviation, For section a j Upper distance The nearest projection point, ‖·‖ Euclidean is the straight-line distance between two points;

[0061] S302, from the GPS track point Corresponding candidate road segment a j Move to GPS track point Corresponding candidate road segment a k The probability is defined as the transition probability, and the GPS trajectory point and Corresponding to a j and a k The closest projection points are and The transition probability p(a) is calculated by the following formula k ∣a j ):

[0062]

[0063]

[0064] Where β is the distribution parameter, ||·|| Route is the path distance between two points;

[0065] S303, according to the observation probability matrix and the transition probability matrix, the Viterbi algorithm is used to calculate and generate the most likely road segment sequence ε = {a j ,a k ,…,a m} and the corresponding projection point sequence Calculate the trajectory travel time T X , track distance D X and the average travel speed of the trajectory Pace X , the calculation process is:

[0066] Trajectory travel time T X The calculation formula is: X =t n -t1;

[0067] Track distance The calculation formula is:

[0068]

[0069] Average travel speed of track Pace X The calculation formula is:

[0070] Among them, l(·) is the length of the segment, head(·) and tail(·) are the head node and tail node of the segment respectively.

[0071] S4) Repeat steps S2 and S3 in each research time window to establish the empirical distribution of the average travel speed of the trajectory, and extract the mean and 85% quantile of the distribution as measurement indicators of the road network resilience order parameter.

[0072] S401, establish a research time window set TW= { tw1,tw2,…,tw n} , for each time window tw i Execute steps S2 and S3 respectively to establish each time window tw in turn. i Empirical distribution of average travel speed of trajectories under Among them, X(tw i ) is the time window tw i All trajectories under |X(tw i )| is the number of trajectories, I(·) is the indicator function;

[0073] S402, define each time window tw i The average travel speed of the trajectory under and 85% quantile Pace(tw i ) 85％ , the calculation formula is as follows:

[0074]

[0075]

[0076] in, To set the time window tw i Average travel speed of all tracks within Pace X Sort in ascending order value.

[0077] In this embodiment, when constructing the road network topology of the study area in step S1, the study area is first delineated according to the accident location of the Central Ring Road. The area is centered on the accident location and is mainly bounded by the Outer Ring Expressway, the North-South Elevated Road, the Central Ring Expressway, the Yixian Elevated Road, the Inner Ring Expressway and the Shanghai-Nanjing Expressway. Considering that there are obvious differences in peak-hour traffic conditions within and outside the Central Ring Road under normal circumstances, the Central Ring Expressway is used as the boundary to divide it into study area 1 (area outside the Central Ring Road) and study area 2 (area within the Central Ring Road), so that the measurement of the order parameter of road network resilience is more refined in space. Then, the national map data for 2016 is obtained through the OpenStreetMap platform, and the map is cropped to the size of the study area using the Osmconvert software. Finally, the OSMnx library is used to establish the road network topology, and a connectivity check is performed to correct the topology. The following is obtained: Figure 2 The diagram shows the road network topology diagram of the Shanghai Middle Ring Road accident study area on May 23, 2016.

[0078] In step S2, when cleaning the floating vehicle GPS track data, the average sampling interval of the floating vehicle GPS track data used in this embodiment is 3.5 seconds, so there is no need to perform the track densification step; if the sampling interval exceeds 10 seconds, the interpolation method can be used for densification. Since the floating vehicle GPS track data used in this embodiment does not have passenger information, it is necessary to identify the stop point to segment the itinerary (if the GPS track data with passenger information is used, the itinerary can be segmented by the passenger field).

[0079] In this embodiment, in step S3, in order to reduce the impact of GPS trajectory data noise on the calculation results of the road network resilience order parameter, in the process of performing trajectory map matching and calculating the average travel speed of each trajectory, trajectories with a travel distance of more than 1 km and a travel time of more than 5 minutes are selected from the trajectories that have completed map matching for calculating the average travel speed of the trajectory. The average travel speed distribution of the trajectory in the time windows of 7:00-8:00 and 8:00-9:00 for a total of 3 days before the Central Ring Road accident (May 19), on the day of the accident (May 23), and after the accident (May 30) is compared respectively. The results are as follows: Figure 3 As shown, (a) is the average travel speed distribution map of the trajectory in the study area 1 (area outside the middle ring) with a time window of 7:00-8:00, (b) is the average travel speed distribution map of the trajectory in the study area 1 (area outside the middle ring) with a time window of 8:00-9:00, (c) is the average travel speed distribution map of the trajectory in the study area 2 (area inside the middle ring) with a time window of 7:00-8:00, (d) is the average travel speed distribution map of the trajectory in the study area 2 (area inside the middle ring) with a time window of 8:00-9:00;

[0080] from Figure 3 It can be seen that study area 1 is located outside the middle ring road. Under normal circumstances, the congestion level is not high during the morning rush hour. After the accident, the speed changes significantly due to the transfer of traffic flow. Area 2 is located inside the middle ring road. Under normal circumstances, the congestion level is relatively low during the morning rush hour. The speed changes before and after the accident are not obvious.

[0081] In this embodiment, when calculating the network resilience order parameter in step S4, the average travel speed of the trajectories during the morning peak hours of 19 working days from May 9 to June 7, 2016 and the 85% quantile change curves are plotted. Figure 4 and Figure 5 As shown in the figure, the mean of the average travel speed of the trajectory during the morning peak is in different ranges under normal and abnormal conditions, and as the accident is relieved, the distribution mean gradually returns to the normal range, indicating that the resilience order parameter indicator can be used to identify the start and end time of the impact of the accident on the traffic network. At the same time, the 85% quantile of the average travel speed of the trajectory during the morning peak is used to compare the extreme traffic conditions corresponding to the tail of the average travel speed distribution of the trajectory under normal and abnormal conditions. Compared with study area 2, the 85% quantile difference between the two states in study area 1 is greater, indicating that the accident has a greater extreme impact on travelers in study area 1.

[0082] Therefore, the above analysis shows that the road network resilience order parameter obtained by the method of this embodiment can effectively monitor the evolution trend of network-level resilience and measure the spatiotemporal impact of disturbance events.

[0083] The above is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any technician in the relevant technical field, without departing from the scope of the technical solution of the present invention, makes any form of equivalent replacement or modification to the technical solution and technical content disclosed in the present invention, which does not depart from the content of the technical solution of the present invention and still falls within the protection scope of the present invention.

Claims

1. A method for measuring the resilience order parameter of urban road network based on GPS trajectory data, characterized in that: Based on the floating car GPS trajectory data, the following steps are included: S1: Determine the scope of the study area and construct the road network topology of the study area; S2: Cleaning the original floating vehicle GPS trajectory data according to the road network topology; S3: Based on the road network topology, the cleaned data is used to perform trajectory map matching, so as to obtain the travel distance and travel time of each trajectory, and calculate the average travel speed of each trajectory; S4: Repeat steps S2 and S3 in each research time window to establish the empirical distribution of the average travel speed of the trajectory, and extract the mean and 85% quantile of the distribution as measurement indicators of the road network resilience order parameter.

2. The urban road network resilience order parameter measurement method based on GPS trajectory data according to claim 1 is characterized in that: Step S1 specifically includes: obtaining map data of the study area, taking intersections as nodes, road sections as edges, and road section lengths as weights, and constructing a directed network G=(N,A), where N represents a node set and A represents a road section set.

3. The urban road network resilience order parameter measurement method based on GPS trajectory data according to claim 1 is characterized in that: In step S2, the data cleaning includes time and space range limitation, trajectory denoising and trip segmentation.

4. The urban road network resilience order parameter measurement method based on GPS trajectory data according to claim 3 is characterized in that: The method of limiting the time and space range is: based on the road network topology and the research time window, excluding GPS track points that are more than 100 meters away from any road network element, and excluding GPS track points that are not within the research time window; The trajectory denoising method is as follows: the speed of the current GPS trajectory point is defined as the sum of the distances between the current GPS trajectory point and the adjacent GPS trajectory point divided by the time difference between the adjacent GPS trajectory points. If the value exceeds 120 km / h, or the angle between the current GPS trajectory point and the adjacent GPS trajectory point is less than 30°, the current GPS trajectory point is regarded as an abnormal point and removed; The segmentation rule of the itinerary is as follows: if the distance between more than 5 consecutive GPS track points is less than 20 meters, and the total time interval between these GPS track points exceeds 1 minute, then the point closest to the center among these GPS track points is selected as the stop point, and this is used as the basis for segmenting the itinerary.

5. The urban road network resilience order parameter measurement method based on GPS trajectory data according to claim 1 is characterized in that: In step S3, trajectory map matching is performed based on the hidden Markov algorithm to obtain the travel distance and travel time of each trajectory and calculate the average travel speed of each trajectory; specifically, it includes: S3.1: Based on each GPS track after cleaning Create GPS track points Corresponding candidate road segment set The observation probability is calculated by the following formula in, Represents at time t i (i=1,2,…,n) observed GPS track point coordinates; σ x is the GPS measurement standard deviation, For section a j Upper distance The nearest projection point, ||·|| Euclidean is the straight-line distance between two points; S3.2: From GPS track point Corresponding candidate road segment a j Move to GPS track point Corresponding candidate road segment a k The probability of is defined as the transition probability, and the transition probability is calculated by the following formula: Where β is the distribution parameter, ||·|| Route is the path distance between two points; and GPS track points and Corresponding to a j and a k The closest projection point on the S3.3: Based on the observation probability matrix and the transition probability matrix, the Viterbi algorithm is used to calculate and generate the most likely segment sequence ε = {a j ,a k ,…,a m } and the corresponding projection point sequence Calculate the trajectory travel time T X , track distance D X and the average travel speed of the trajectory Pace X .

6. The urban road network resilience order parameter measurement method based on GPS trajectory data according to claim 5 is characterized in that: In step S3.3, the trajectory travel time T X , track distance D X and the average travel speed of the trajectory Pace X The calculation formulas are: Trajectory travel time T X The calculation formula is: X =t n -t1; Track distance D X The calculation formula is: Average travel speed of track Pace X The calculation formula is: Among them, l(·) is the length of the segment, head(·) and tail(·) are the head node and tail node of the segment respectively.

7. The urban road network resilience order parameter measurement method based on GPS trajectory data according to claim 1 is characterized in that: Step S4 specifically includes: S4.1: Establish a research time window set TW = {tw1, tw2, ..., tw n }, for each time window tw i Execute steps S2 and S3 respectively to establish each time window tw in turn. i Empirical distribution of average travel speed of trajectories under Among them, X(tw i ) is the time window tw i All trajectories under |X(tw i )| is the number of trajectories, I(·) is the indicator function; S4.2: Define each time window tw i The average travel speed of the trajectory under and 85% quantile Pace(tw i ) 85％ , the calculation formula is as follows: in, To set the time window tw i The average travel speed of all tracks in Pace X Sort in ascending order value.

Images