Method and program product for identifying spatiotemporal location of traffic accidents using floating car data

Abstract

The application discloses a method and program product for identifying space-time position of traffic accidents by using floating vehicle data, and the method comprises the following steps: (1) obtaining floating vehicle data in a historical analysis period and floating vehicle data in a to-be-identified time period, calculating a historical average driving speed in the historical analysis period, and calculating an actual driving speed in the to-be-identified time period; (2) judging whether each link in each time interval is expected to be affected by an accident according to the historical average driving speed and the actual driving speed; (3) constructing an accident space-time position optimization model, wherein the model takes actual influence of each link in each time interval on each accident and space-time of each accident as a decision variable, and takes minimization of a difference between actual influence and expected influence on the accident as a target; (4) adding a constraint condition to the accident space-time position optimization model; and (5) solving the accident space-time position optimization model to obtain the space-time position of the accident. The application has high accuracy and high robustness.

Description

Technical Field

[0001] This invention relates to spatiotemporal location identification technology for traffic accidents, and more particularly to a method and program product for identifying the spatiotemporal location of traffic accidents using floating vehicle data. Background Technology

[0002] Accurate spatiotemporal localization of traffic accidents—that is, pinpointing the time and spatial location of their occurrence—is crucial for extensive accident correlation analysis and traffic management systems. However, this information is difficult to obtain directly from accident reports due to ambiguous textual descriptions and unreported accidents. To address this issue, some existing studies have utilized various data sources, such as social media data, and developed data-driven methods for accident identification. However, limited data availability and potentially low data quality, including missing and erroneous data, affect the accuracy of these methods in identifying the spatiotemporal location of accidents across a network. Furthermore, the spatiotemporal evolution of congestion caused by accidents in urban networks is far more complex than that of linear road segments, making it impossible for existing technologies to accurately identify spatiotemporal locations.

[0003] Patent document CN 116052418A discloses a data analysis method and apparatus for urban road networks, comprising: determining the spatiotemporal impact range of a sudden event based on acquired speed data; determining the spatiotemporal propagation law of the sudden event in the urban road network based on the spatiotemporal impact range; expressing the spatiotemporal propagation law with linear constraints; and estimating the spatiotemporal impact of the sudden event on the urban road network through an optimization model. This technology enables the prediction of the spatiotemporal impact range of an accident based on speed data, but it also fails to identify the spatiotemporal location of traffic accidents. Summary of the Invention

[0004] To address the problems existing in the prior art, the purpose of this invention is to provide a method and program product that can identify the spatiotemporal location of traffic accidents using floating vehicle data with high accuracy.

[0005] To achieve the above-mentioned objectives, the present invention provides the following technical solution:

[0006] A method for identifying the spatiotemporal location of traffic accidents using floating vehicle data includes the following steps:

[0007] (1) Obtain floating vehicle data for the historical analysis period and floating vehicle data for the time period to be identified, and calculate the historical average driving speed through each link in each time interval of the historical analysis period, and the actual driving speed through each link in each time interval of the time period to be identified.

[0008] (2) Determine whether each link is expected to be affected by the accident in each time interval based on the historical average driving speed and the actual driving speed;

[0009] (3) Construct an accident spatiotemporal location optimization model. The accident spatiotemporal location optimization model takes whether each link is actually affected by each accident in each time interval and the time interval and link where each accident occurs as decision variables, and aims to minimize the difference between the actual and expected impact of the accident on all links in all time intervals.

[0010] (4) Add constraints to the accident spatiotemporal location optimization model. The constraints include constraints on the source of the accident impact, constraints on the directional propagation of the traffic accident impact, constraints on the continuous propagation of the accident impact, additional constraints on the road network closed loop, minimum requirements for the spatiotemporal location range of the accident, and constraints on decision variables.

[0011] (5) Solve the spatiotemporal location optimization model of the accident to obtain the time interval and link of each accident in the time period to be identified, and use it as the spatiotemporal location output of the accident in the time period to be identified.

[0012] Furthermore, step (2) specifically includes:

[0013] (2.1) Based on the floating vehicle data and historical average driving speed during the historical analysis period, calculate the standard deviation of the historical speed through each link in each time interval of the historical analysis period;

[0014] (2.2) Based on the historical average driving speed, the historical speed standard deviation and the actual driving speed, calculate the expected accident impact indicator variable for each link in each time interval according to the following formula, wherein the expected accident impact indicator variable is used to indicate whether each link n in each time interval t is expected to be affected by an accident.

[0015]

[0016] In the formula, Indicates the expected impact of the accident as an indicator variable, when When t, it indicates that link n is expected to be affected by the incident within the time interval t; otherwise, it indicates that it is unaffected. n,t This represents the actual travel speed through link n within the time interval t. σ n,t α represents the historical average speed and historical speed standard deviation through link n within time interval t, respectively; N represents the number of links; and T represents the number of time intervals into which the time period to be identified is divided.

[0017] Furthermore, the objective function of the accident spatiotemporal location optimization model is:

[0018]

[0019] In the formula, This represents the decision variable indicating the actual impact of an accident, used to indicate whether link n is actually affected by accident k within the time interval t. The time interval t indicates the actual impact of the incident k on link n. This indicates that link n was not actually affected by incident k within the time interval t. The spatiotemporal location decision variable represents the time interval t and link n where accident k occurs. The time interval indicates that incident k occurred on link n within the time interval t. Time indicates that incident k did not occur on link n within the time interval t, and K represents the number of incidents.

[0020] Furthermore, the specific constraints on the sources of the accident's impact are as follows:

[0021]

[0022]

[0023]

[0024]

[0025]

[0026] In the formula, D n Denotes the set of downstream links of link n, |D n | represents the number of downstream links of link n, and m represents D. n The m-th link in Let be the actual accident impact decision variable used to represent whether link m is actually affected by accident k within the time interval t.

[0027] Furthermore, the constraint on the directional propagation of the impact of the traffic accident is specifically as follows:

[0028]

[0029] In the formula, D n Let m represent the set of downstream links of link n, and let m represent D. n The m-th link in Let be the actual accident impact decision variable used to represent whether link m is actually affected by accident k within the time interval t.

[0030] Furthermore, the constraint on the continued propagation of the accident's impact specifically includes:

[0031]

[0032]

[0033]

[0034] λ1+λ2+λ3≤2

[0035] λ1,λ2,λ3∈{0,1}

[0036] In the formula, n Denotes the set of downstream links of link n, |D n | represents the number of downstream links of link n, and m represents D. n The m-th link in Let λ1, λ2, and λ3 be the actual accident impact decision variables used to represent whether link m is actually affected by accident k within the time interval t, and let λ1, λ2, and λ3 be the propagation constraint parameters.

[0037] Furthermore, the additional constraints of the road network closed loop are as follows:

[0038]

[0039] In the formula, Δ represents the set of downstream links of link n, and |Δ| is the number of downstream links of link n.

[0040] Furthermore, the minimum requirement for the spatiotemporal location range of the accident is specifically as follows:

[0041]

[0042] In the formula, It is an indicator variable for the impact of the incident time; when at least one link is affected by incident k in time interval t, It equals 1, otherwise it equals 0, L t This indicates a preset time threshold. It is an indicator variable for the spatial impact of an accident, which is determined when link n is affected by accident k for at least one time interval. It equals 1, otherwise it equals 0, L n This indicates a preset space threshold.

[0043] Furthermore, the decision variable constraints are specifically as follows:

[0044]

[0045]

[0046]

[0047] A computer program product includes a computer program / instructions that, when executed by a processor, implement the above-described method.

[0048] Compared with existing technologies, the advantages of this invention are as follows: This invention constructs an accident spatiotemporal location optimization model following traffic wave propagation using floating car data (FCD). The model's input consists of historical and target time periods' speeds, which may contain measurement errors and missing data, but are generally available across the entire network. Spatiotemporal location identification is achieved by analyzing the origins related to the spatiotemporal progression of the accident's impact. To address missing and erroneous speed data, this invention uses accident shock wave propagation as a constraint to describe the impact of the accident on the speed evolution of adjacent links and time intervals, thereby compensating for limited information in damaged data. Furthermore, the impact of complex road network structures on accident shock wave propagation is considered. Numerical experiments validated the model, and the results show that the model achieves satisfactory results under different missing rates and noise intensities, verifying the model's accuracy and robustness. Attached Figure Description

[0049] Figure 1 This is a flowchart illustrating the method for identifying the spatiotemporal location of a traffic accident using floating vehicle data, as provided in an embodiment of the present invention.

[0050] Figure 2 This is a schematic diagram of a closed loop road network.

[0051] Figure 3 This is the time deviation of the present invention under different data missing rates;

[0052] Figure 4 This refers to the positional deviation of the present invention under different data missing rates. Detailed Implementation

[0053] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0054] Example 1

[0055] The technical solution of this invention involves several key concepts:

[0056] Floating Car Data (LCD) is a new type of traffic information detection technology. Its core is to use the location and time information collected by floating vehicles with GPS positioning capabilities (such as urban taxis and private cars) to calculate the speed of the vehicle's location and correlate this speed information with electronic maps to intuitively describe the traffic flow speed on the road.

[0057] Road network: Unlike highways or urban expressways, urban road networks have unique topologies, such as upstream and downstream relationships between links and some road loops.

[0058] Traffic wave propagation refers to the phenomenon of traffic flow propagating on roads, similar to wave motion in a fluid. By establishing a mathematical programming model, the spatiotemporal impact range of traffic accidents can be estimated, and the model's output can satisfy the propagation laws of traffic waves.

[0059] This invention provides a method for identifying the spatiotemporal location of traffic accidents using floating vehicle data, such as... Figure 1 As shown, it includes the following steps:

[0060] (1) Obtain floating vehicle data for the historical analysis period and floating vehicle data for the time period to be identified, and calculate the historical average driving speed through each link in each time interval of the historical analysis period, and the actual driving speed through each link in each time interval of the time period to be identified.

[0061] Specifically, the historical analysis period can be the floating vehicle data prior to the time period to be identified. The historical analysis period is divided into T equal time intervals. Based on the floating vehicle data, the travel speed of vehicles passing through link n within time interval t during the historical analysis period can be calculated, and then averaged to obtain the historical average travel speed. Similarly, the actual speed v of the vehicle passing through link n within the time interval t of the time period to be identified can be calculated. n,t , 1≤n≤N, 1≤t≤T, where N is the number of links.

[0062] (2) Determine whether each link is expected to be affected by the accident in each time interval based on the historical average driving speed and the actual driving speed.

[0063] This step specifically includes:

[0064] (2.1) Based on the floating vehicle data and historical average driving speed during the historical analysis period, calculate the historical speed standard deviation σ of each link within each time interval of the historical analysis period. n,t ;

[0065] (2.2) When an accident indexed by k occurs on a link, it disrupts the smooth flow of traffic, causing a decrease in speed. The impact of the accident propagates upstream over time, causing regional congestion in both spatial and temporal dimensions. Therefore, the speed caused by the accident is typically lower than the average speed. To measure this interruption and capture the potential impact of the accident on link n during time interval t, an indicator variable for the expected accident impact is defined by comparing the vehicle speed during the time period to be identified with the average vehicle speed during the historical analysis period. This indicates whether each link n is expected to be affected by an incident within each time interval t. The specific calculation formula is as follows:

[0066]

[0067] In the formula, when When t, it indicates that link n is expected to be affected by the incident within the time interval t; otherwise, it indicates that it is unaffected. n,t This represents the actual travel speed through link n within the time interval t. σ n,t Let represent the historical average speed and historical speed standard deviation through link n within the time interval t, respectively, and α represent the proportional parameter.

[0068] (3) Construct an accident spatiotemporal location optimization model. The accident spatiotemporal location optimization model takes whether each link is actually affected by each accident in each time interval and the time interval and link where each accident occurs as decision variables, and aims to minimize the difference between the actual and expected impact of the accident on all links in all time intervals.

[0069] Due to missing or erroneous data, the observed accident impact derived directly from velocity data may not conform to the propagation patterns of the accident shock wave. To reconstruct the actual accident impact, the objective of this invention is to minimize the difference between the actual and expected accident impact, specifically expressed by the following formula:

[0070]

[0071] In the formula, the decision variables of the accident spatiotemporal location optimization model include the actual accident impact decision variables. Decision variables related to the spatiotemporal location of the accident Used to indicate whether link n is actually affected by incident k within time interval t, when The time interval t indicates the actual impact of incident k on link n. This indicates that link n was not actually affected by incident k within the time interval t. Used to indicate the time interval t and link n where incident k occurred, when The time interval indicates that incident k occurred on link n within the time interval t. Time indicates that incident k did not occur on link n within the time interval t, and K represents the number of incidents.

[0072] (4) Add constraints to the accident spatiotemporal location optimization model. The constraints include constraints on the source of the accident impact, constraints on the directional propagation of the traffic accident impact, constraints on the continuous propagation of the accident impact, additional constraints on the road network closed loop, minimum requirements for the spatiotemporal location range of the accident, and constraints on decision variables.

[0073] Specifically, the constraints on the sources of the accident's impact are as follows:

[0074]

[0075]

[0076]

[0077]

[0078]

[0079] In the formula, D n Denotes the set of downstream links of link n, |D n | represents the number of downstream links of link n, and m represents D. n The m-th link in Let be the actual accident impact decision variable used to represent whether link m is actually affected by accident k within the time interval t.

[0080] There are five constraints related to the source of the incident's impact. The first constraint ensures that when incident k occurs on link n within time interval t, link n will be affected by incident k at time t. The second and third constraints ensure that when incident k occurs on link n within time interval t, neither link n at the previous time nor any downstream links of link n at time t will be affected by incident k. Where |D n | represents the number of downstream links of link n; the fourth constraint is used to ensure that when link n at the current time and all downstream links of link n at time t are not affected by incident k, incident k must have occurred on link n within the time interval t; the fifth constraint is used to ensure that when incident k occurs on link n within the time interval t, then both time t and link n are unique.

[0081] Specifically, the constraint on the directional propagation of the impact of the traffic accident is as follows:

[0082]

[0083] There is one constraint related to the directional propagation constraint of the impact of traffic accidents. This constraint ensures that when link n is affected by accident k in time interval t, and accident k does not occur on link n within time interval t, then this impact must propagate from the downstream link of link n at time t, or from the time before time t of link n, or both. It is worth noting that when This constraint always holds when the value of is 1.

[0084] Specifically, the constraint on the continued propagation of the accident's impact is as follows:

[0085]

[0086]

[0087]

[0088] λ1+λ2+λ3≤2

[0089] λ1,λ2,λ3∈{0,1}

[0090] In the formula, λ1, λ2, and λ3 are propagation constraint parameters.

[0091] There are a total of 5 constraints related to the continuous propagation of the impact of the accident. These 5 constraints together ensure the uninterrupted propagation of the impact of the accident. That is, for a certain link, if both the previous link and the downstream link were affected by the accident, then the link at this moment will also be affected by the accident.

[0092] Specifically, the road network closed-loop constraints are as follows:

[0093]

[0094] In the formula, Δ represents the set of downstream links of link n, and |Δ| is the number of downstream links of link n.

[0095] There is one additional constraint related to the road network closed loop. This constraint is used to prevent discontinuous propagation of accident shock waves caused by isolated closed loops. Unlike the direct sequential structure of highway segments, links in the road network can form complex closed loops. In such loops, one link can simultaneously serve as both an upstream and downstream link of another. This structure may create isolated areas of influence (see reference). Figure 2 An isolated closed loop always satisfies the above constraints, making it independent of other affected areas. Therefore, this independence leads to the discontinuous propagation of the accident shockwave, a result that contradicts real-world traffic dynamics and will further mislead the identification of the event's origin.

[0096] Specifically, the minimum requirement for the spatiotemporal location range of the accident is as follows:

[0097]

[0098] In the formula, It is an indicator variable for the impact of the incident time; when at least one link is affected by incident k in time interval t, It equals 1, otherwise it equals 0, L t This indicates a preset time threshold. It is an indicator variable for the spatial impact of an accident, which is determined when link n is affected by accident k for at least one time interval. It equals 1, otherwise it equals 0, Ln This indicates a preset space threshold.

[0099] There are a total of 8 constraints related to the minimum requirements for the spatiotemporal scope of accidents. These constraints can filter out accidents with minor impacts during the traffic accident identification process. Therefore, minimum event duration and minimum number of affected links are set, i.e., L. t and L n , as preset time and space thresholds.

[0100] Specifically, the decision variable constraints are as follows:

[0101]

[0102] There are a total of 5 constraints related to the decision variables. The last two constraints ensure that the values ​​of the two decision variables outside the boundary are both 0.

[0103] (5) Solve the spatiotemporal location optimization model of the accident to obtain the time interval and link of each accident in the time period to be identified, and use it as the spatiotemporal location output of the accident in the time period to be identified.

[0104] Based on the objective function and constraints of subsequent steps, the optimization model can utilize information provided by the input velocity data, and simultaneously, according to the spatiotemporal propagation of the accident shock wave, input unobserved accident shocks and correct for erroneous accident shocks. This optimization model can be effectively solved using the standard branch and bound algorithm.

[0105] Solving for decision variables at the spatiotemporal location of the accident Then, obtain the variable with a value of 1, and then extract the index numbers k, n, t to obtain the time interval t (time position) and link n (spatial position) where the accident k occurred.

[0106] The following numerical experiments demonstrate the embodiments of the present invention, mainly including the following steps:

[0107] Step 1: Experiment Setup

[0108] Numerical experiments were conducted on a road network. The simulation network used reproduced the area surrounding the Beijing National Stadium at a 1:1 scale, including 28 major intersections and 95 main roads. All boundary nodes served as both origin and destination points. Vehicles entered the network at four points: the top left and bottom right corners had a flow demand of 3500 pcu / h, and the bottom left and bottom right corners had a flow demand of 3250 pcu / h. The saturation flow configuration for each lane on the links was 1800 pcu / h / lane. Roads were divided into equal-length segments of 50 meters, with time discretization in equal 2-minute intervals. In the simulation, a loop detector was located at the midpoint of each segment (i.e., link) and recorded the speed of vehicles passing through that segment. Therefore, VISSIM could record the speed and position of all vehicles. The average speed of vehicles passing through each segment within a given time interval could then be calculated.

[0109] Step 2: Noise-free data experiment results

[0110] Based on the average performance of all 30 accidents, it can be found that the model of this invention can accurately identify the occurrence stage and direction of propagation of accidents. The average deviations of the time and location of these accidents were 0.6 minutes and 27.87 meters, respectively. Overall, these results indicate that the model has high accuracy in identifying the spatiotemporal location of traffic accidents when the input data is complete and clean.

[0111] Step 3: Results of the experiment with randomly missing data

[0112] Considering that network-wide speed data obtained from floating car data may be intermittently unavailable on certain links and time intervals, experiments were conducted to evaluate the performance of the model of this invention under data missing conditions. Specifically, the missing rate was set to [10%, 20%, 30%, 40%, 50%, 60%, 70%] by randomly ignoring observed speed data in the spatial and temporal dimensions. Figure 3 and Figure 4 As shown, experimental results indicate that the biases in distance and time increase with the increase in the missing rate, due to the decreasing amount of information related to the observed impact incidents. Notably, the model of this invention produces satisfactory results for all incidents when the missing rate is below 50%. Above this threshold, some incidents exhibit significant biases in either distance or time. Furthermore, the median biases in distance and time remain within acceptable ranges across all missing rates. This demonstrates the model's adaptability and reliable execution even with significant data loss. In summary, these findings demonstrate the robustness and reliability of the model proposed in this invention, even under conditions of substantial missing data.

[0113] Step 4: Noise Data Experiment Results

[0114] Another significant challenge in identifying network-wide incidents is measurement error and random velocity fluctuations, which can lead to erroneous data and hinder accurate observation of incident impacts. This further contributes to bias in determining the spatiotemporal location of incidents. The model of this invention addresses this challenge by minimizing the difference between the expected and observed spatiotemporal propagation of the incident impact. The performance of the model was evaluated at different percentages and magnitudes. Specifically, Gaussian white noise was introduced into the velocity data with magnitudes and percentages of [1, 2, 3]σ and [5%, 10%, 15%, 20%, 25%, 30%], where σ represents the standard deviation of the velocity data. Results show that distance and time biases increase with increasing error percentage. Similar results can be observed with increasing noise levels. Furthermore, for lower noise levels (e.g., 1σ and 2σ), the bias increases slowly with increasing noise percentage. However, at higher noise levels (e.g., 3σ), the bias increases more sharply, particularly when the noise percentage exceeds 20%. Overall, the performance of the model of this invention remains satisfactory even in the presence of erroneous data.

[0115] Example 2

[0116] This invention also provides a computer program product, such as an app on a mobile phone or tablet, or an installer on a computer. This program product includes a computer program / instructions that, when executed by a processor, implement the method described in Embodiment 1. The code for the computer-executable program used to perform the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0117] It should be understood that the embodiments and descriptions above are only the principles, main features and advantages of the present invention. Various changes and modifications can be made to the present invention without departing from the spirit and scope of the invention, and all such changes and modifications fall within the protection scope of the present invention.

Claims

1. A method for identifying the spatio-temporal location of a traffic accident using floating car data, characterized in that, The method comprises the following steps: (1) obtaining historical analysis period probe vehicle data and to-be-identified time period probe vehicle data, and calculating historical average travel speed of each link in each time interval in the historical analysis period and actual travel speed of each link in each time interval in the to-be-identified time period; (2) judging whether each link in each time interval is expected to be affected by an accident according to the historical average travel speed and the actual travel speed; (3) constructing an accident space-time position optimization model, wherein the accident space-time position optimization model takes actual influence of each link in each time interval by each accident and the time interval and the link where each accident occurs as decision variables, and minimizes the difference between actual and expected influence of accidents on all links in all time intervals as a target; (4) adding constraint conditions to the accident space-time position optimization model, wherein the constraint conditions comprise an accident influence source constraint, a traffic accident influence directional propagation constraint, an accident influence continuous propagation constraint, a road network closed loop additional constraint, a minimum requirement for the accident space-time position range, and a decision variable constraint; (5) solving the accident space-time position optimization model to obtain the time interval and the link where each accident in the to-be-identified time period occurs, as an accident space-time position output of the to-be-identified time period; the road network closed loop additional constraint is specifically: In the formula, denote actual accident influence decision variables, respectively used to indicate whether the link n is actually affected by the accident k in the time interval t, t-1, denote accident space-time position decision variables, used to indicate the time interval t and the link n where the accident k occurs, Δ denotes a set of downstream links of the link n, |Δ| denotes the number of downstream links of the link n, K denotes the number of accidents, N denotes the number of links, and T denotes the number of time intervals into which the time period to be identified is divided. the minimum requirement for the accident space-time position range is specifically: wherein is an incident time influence indicator variable, which equals 1 if at least one link is affected by incident k in time interval t, equals 1 otherwise, L t denotes a pre-set time threshold, is an incident space influence indicator variable, which equals 1 if link n is affected by incident k in at least one time interval, equals 1 otherwise, L n denotes a pre-set space threshold; the decision variable constraint is specifically:

2. The method for identifying the space-time position of a traffic accident using floating car data according to claim 1, characterized in that, step (2) specifically comprises: (2.1) calculating historical speed standard deviation of each link in each time interval in the historical analysis period according to the historical analysis period probe vehicle data and the historical average travel speed; (2.2) calculating an expected accident influence indicator variable of each link in each time interval according to the historical average travel speed, the historical speed standard deviation and the actual travel speed, wherein the expected accident influence indicator variable is used to indicate whether each link n in each time interval t is expected to be affected by an accident; wherein denotes an expected incident impact indicator, which indicates that link n is expected to be affected by an incident within time interval t, otherwise it indicates no impact, v denotes an expected incident impact indicator, which indicates that link n is expected to be affected by an incident within time interval t, otherwise it indicates no impact, v n,t denotes the actual travel speed through link n within time interval t, σ n,t denote the historical average travel speed through link n within time interval t, the historical speed standard deviation, respectively, a denotes a proportionality parameter, N denotes the number of links, and T denotes the number of time intervals into which the time period to be identified is divided.

3. The method for identifying the space-time position of a traffic accident using floating car data according to claim 2, characterized in that, the target function of the accident space-time position optimization model is: In the formula, represents the actual accident influence decision variable, which is used to indicate whether link n is actually affected by accident k within time interval t, when represents that link n is actually affected by accident k within time interval t, represents that link n is not actually affected by accident k within time interval t, represents the accident space-time position decision variable, which is used to indicate the time interval t and link n where accident k occurs, when represents that accident k occurs on link n within time interval t, represents that accident k does not occur on link n within time interval t, and K represents the number of accidents.

4. The method for identifying the space-time position of a traffic accident using floating car data according to claim 3, characterized in that, the accident influence source constraint is specifically: where D n represents the downstream link set of link n, |D n | represents the number of downstream links of link n, m represents the mth link in D n is the actual incident impact decision variable for indicating whether link m is actually affected by incident k within time interval t.​ 5. The method for identifying the space-time position of a traffic accident using floating car data according to claim 3, characterized in that, the traffic accident influence directional propagation constraint is specifically: where D n represents the set of downstream links of link n, m represents the mth link in D n , is the actual incident impact decision variable for indicating whether link m is actually affected by incident k within time interval t.

6. The method for identifying the space-time position of a traffic accident using floating car data according to claim 3, characterized in that, the accident influence continuous propagation constraint is specifically: λ1+λ2+λ3≤2 λ1, λ2, λ3∈{0, 1} where D n denotes the downstream link set of link n, |D n denotes the number of downstream links of link n, m denotes the mth link in D n is the actual incident impact decision variable for indicating whether link m is actually affected by incident k within time interval t or not, and λ1, λ2, λ3 are propagation constraint parameters.​ 7. A computer program product comprising a computer program, characterized in that, the computer program is executed by the processor to realize the method in any one of claims 1-6.

Images