Simulation-driven emergency control methods for highway accidents

Abstract

The emergency control method for highway accidents provided by this invention includes: S1, establishing a basic highway information database and a highway accident information database, and constructing and verifying a basic simulation model; S2, establishing a highway management strategy expert database and a simulation management plan; S3, when a highway accident occurs, performing online analysis of the temporal and spatial impact range of the accident; S4, dynamically matching traffic management strategies based on the highway management strategy expert database; S5, after the highway emergency control is completed, organizing the event information and emergency management strategy information, and storing them as new traffic management strategy samples in the highway management strategy expert database. This invention is comprehensive and widely applicable, and can reduce the impact of highway emergencies on transportation efficiency and improve traffic safety.

Description

Technical Field

[0001] This invention relates to the field of traffic control technology for highways, and more specifically, to a simulation-driven emergency control method for highway accidents. Background Technology

[0002] In recent years, my country's expressway mileage has increased year by year, ranking first in the world. However, expressways are prone to accidents, severe weather, natural disasters, vehicle mechanical failures, and hazardous chemical leaks, threatening people's lives and property. If appropriate control strategies are not implemented to ensure the operation of the transportation system when accidents occur on expressways, the incidents may trigger secondary traffic accidents and significantly reduce transportation efficiency.

[0003] Highways drive national socio-economic development, and their traffic safety and operational efficiency have received widespread attention. Emergency traffic control in the event of highway accidents is therefore crucial. The "Regulations on Emergency Management Procedures for Highway Traffic" classifies emergency responses into four levels based on the time and spatial extent of traffic disruption caused by accidents, and specifies management measures for each level. Traffic control strategies for highway accidents include ramp control, variable speed limits, variable lanes, toll station control, route guidance, and signal control on highway connecting sections. Quickly and accurately selecting appropriate traffic control strategies is a key challenge in highway emergency management. While numerous scholars both domestically and internationally have studied highway responses to accidents from the perspectives of risk warning, obstacle avoidance facilities, and emergency rescue, the following two issues remain largely unconsidered:

[0004] (1) There is little consideration for emergency control strategies under various accident events on highways. For example, existing research often limits accident events to traffic congestion, hazardous material leakage, severe weather and other modes for traffic control. In order to expand the applicability of the invention, it is necessary to study various accident events and ensure the accuracy and feasibility of emergency measures under different events.

[0005] (2) There is little consideration given to the situation where multiple control strategies are mixed on highways under a certain accident, such as adopting one or more control strategies to improve traffic efficiency and improve traffic safety based on the structural characteristics of different highways and the level of the accident. Summary of the Invention

[0006] The purpose of this invention is to provide a simulation-driven emergency control method for highway accidents, in order to overcome the shortcomings of the prior art.

[0007] To achieve the objective of this invention, the simulation-driven emergency control method for highway accidents provided by this invention includes the following steps:

[0008] S1. Establish a basic information database for highways and a database of information on highway accidents, and construct and verify a basic simulation model;

[0009] S2. Establish an expert database for highway management strategies and a simulation management plan;

[0010] S3. Online analysis of the temporal and spatial impact of highway accidents;

[0011] S4. Dynamically match traffic control strategies based on the expert database of highway control strategies;

[0012] S5. After the emergency control measures on the highway are completed, the incident information and emergency management strategy information are compiled and stored as a new traffic control strategy sample in the highway control strategy expert database.

[0013] Step S1 includes:

[0014] S101. Establish a national expressway basic information database and an expressway accident information database;

[0015] S102. Based on the variable cell transmission model, simulate the road conditions of the main line of the expressway, weaving area, ramps, toll stations and the road sections connecting to the ground.

[0016] S103. Select an actual highway, and by adjusting the variable attributes of the highway mainline, weaving areas, ramps, toll stations, and ground-connecting intersections or road sections, combine the simulated road condition map with the actual highway map, and run each basic simulation model to obtain average travel speed and traffic volume evaluation indicators.

[0017] S104. Based on the average travel speed and traffic volume data of the highway, test each basic simulation model. If the test requirements are met, proceed to step S2; otherwise, adjust the simulation model parameters, obtain the evaluation index again, and proceed to S103.

[0018] Verifying whether each basic simulation model meets the requirements includes:

[0019] In each basic simulation model, input the "average hourly traffic volume on highways" and output the "average travel speed" after 1 hour of simulation. "and "number of vehicles passing through Q" VCTM "Average travel speed on highways" in the highway basic information database "and "average hourly traffic volume on highways Q" h "Compare the different models separately, and if both of them satisfy the following formula, the simulation model is considered to meet the requirements;

[0020]

[0021]

[0022] Furthermore, step S2 includes:

[0023] S201. Establish an expert database for highway management and control strategies;

[0024] S202. Based on the basic simulation model of S1, set different traffic control strategies to form a new simulation model and store it with a number for matching and calling in step S4. At the same time, the simulation model sets the parameters of the simulation control scheme as hyperparameters to quickly simulate traffic control strategies under different parameters and obtain evaluation indicators.

[0025] Furthermore, the hyperparameters for different traffic control strategies include:

[0026] Ramp control hyperparameters include the cell number of the ramp, ramp flow input / output rate, ramp control start time, and ramp control duration;

[0027] The variable speed limit hyperparameters include the variable speed limit control start cell, the variable speed limit control end cell, the variable speed limit value, the speed limit control start time, and the speed limit control end time.

[0028] The hyperparameters of the variable lane control include the variable lane control start cell, the variable lane control end cell, the variable lane driving direction, the variable lane control start time, and the variable lane control end time.

[0029] The parameters for toll station control include the cell number of the toll station, the number of open ETC lanes at the toll station, the number of open manual lanes at the toll station, the number of open ETC / manual hybrid lanes at the toll station, the start time of toll station control, and the end time of toll station control.

[0030] The hyperparameters of the path guidance include the cell number of the starting position of the guidance, the cell number of the ending position of the guidance, the traffic flow of the highway after the implementation of the path guidance, the traffic flow of the new path after the implementation of the path guidance, the start time of the guidance control, and the duration of the guidance control.

[0031] The signal control hyperparameters at the off-ramp connection intersection include the cell number controlled by the signal, the signal control period, the number of phases, the phase sequence, and the effective green light time for each phase.

[0032] The parameters for calculating the traffic flow on the highway and the traffic flow on the new route after the implementation of the route guidance are as follows:

[0033] Based on the BPR impedance function and Wardrop's first principle of traffic flow assignment, the traffic flow q of the highway after the implementation of path guidance is... a Traffic flow q of the new route after the implementation of route guidance bThe following equations can be solved simultaneously:

[0034]

[0035] q a +q b =Q h +Q hnew (4)

[0036] In the formula: t a t represents the free-flow time from the start to the end of the route guidance process on the highway. b Let q be the free-flow time of the new path from the start to the end of the induced segment. a q b C represents the traffic flow on the highway and the new route after the implementation of route guidance. a C b Q represents the practical capacity of the highway and the new route, respectively, with α and β as parameters. The Federal Highway Administration recommends α = 0.15 and β = 4. hnew为 Traffic flow of the new route before route guidance is implemented.

[0037] Furthermore, the content of the highway management strategy expert database includes, but is not limited to: emergency management strategy number, emergency management strategy type, emergency management strategy content description, management strategy start time, management strategy end time, latitude and longitude of the management strategy implementation location, strategy implementation effect, accident number associated with the management strategy, and highway number associated with the management strategy.

[0038] Furthermore, online analysis of the temporal and spatial impact of unexpected events includes:

[0039] Based on the data from the highway accident information database, the cells of the highway are located, and the cell where the event is located is used as the center for the analysis of the scope of impact.

[0040] Plot the average travel speed data curve for each cell, and statistically analyze the data for three hours before and after the event, as well as the three-hour data for the same week and time period in the past 7 days, 14 days, and 21 days.

[0041] Set a threshold for the duration of outliers;

[0042] Outliers were analyzed using the IQR-based method.

[0043] Select all cells affected by the event and the degree of their impact;

[0044] A Bayesian-based event impact range prediction model is established. The input features are the daily traffic volume of the highway, the design speed, the average travel speed of the highway, the number of lanes on the road, the type of accident event, the start time of the event, the end time of the event, the latitude and longitude of the event location, the cell number where the event is located, the length of the distance between the cell under study and the cell where the event is located, the event level, the number of lanes blocked by the event, and the highway number associated with the event. The label is the degree of influence of the event on the cell.

[0045] Based on the attributes of real-time events, the predicted event duration is summed with the event start time to replace the event end time. Combined with the degree of influence of the event on the cells, after multiple predictions and sorting, the degree of influence of the unexpected event on all cells can be obtained, that is, the scope of the event's influence.

[0046] Step S3 includes:

[0047] S301. When an accident occurs on a highway, obtain detector data and internet statistics from the past month to the present for that highway.

[0048] S302. Analyze the duration of events based on survival analysis models;

[0049] S303. Analyze the scope of an event's impact using outlier identification based on the IQR method.

[0050] The survival analysis model analyzes the probability that the duration of an event, T, is longer than time t, using the following formula:

[0051]

[0052] Where f(x) and F(t) represent the density function and distribution function of the event duration T, respectively, T is a continuous random variable, and S(t) is the survival function, which is also the integral of the probability density function f(x).

[0053] Step S4 includes:

[0054] S401. An emergency strategy selection model is trained based on an expert database of highway control strategies.

[0055] S402. Select the primary emergency management strategy type for highway accidents using the emergency strategy selection model;

[0056] S403. Select real-time parameters for a secondary emergency management strategy for highway accidents using a basic simulation model;

[0057] S404. Output the parameters corresponding to the optimal evaluation index in the basic simulation model, as a detailed control plan for the highway accident.

[0058] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0059] (1) This invention is comprehensive and widely applicable. The emergency strategy control method of this invention is a hybrid strategy management method, which integrates key technologies such as event impact analysis, management strategy simulation, and automatic quantitative selection of schemes. It can form a complete highway emergency strategy management system. This invention can be adapted to highway networks under different scenarios, and therefore has strong practicality.

[0060] (2) The types of unexpected events and traffic emergency control strategies considered in this invention are diverse and comprehensive. The types of highways considered are also enriched as the sample database increases. Compared with the existing methods that rely on experience or adopt a single control strategy for highway traffic emergencies, the method of this invention is more scientific and reliable.

[0061] (3) The emergency strategy control method of the present invention determines the best control method for highways under different accident scenarios by using quantitative evaluation indicators, which can minimize the socio-economic losses caused by accidents, ensure the safety of traffic users and the operation of the traffic system. Attached Figure Description

[0062] Figure 1 A flowchart of the simulation-driven emergency control method for highway accidents provided by this invention;

[0063] Figure 2 An example of a highway simulation model based on VCTM in the emergency strategy control method provided by this invention;

[0064] Figure 3 Flowchart for analyzing the impact range of highway accidents in the emergency strategy control method provided by this invention;

[0065] Figure 4 This is an example of the average travel speed data curve of a certain cell on a highway in the emergency strategy control method provided by the present invention;

[0066] Figure 5 This invention provides an example of a hybrid strategy control method for highway traffic accident scenarios within the emergency strategy control method. Detailed Implementation

[0067] To more clearly illustrate the present invention, the following detailed description is provided in conjunction with embodiments and accompanying drawings. Those skilled in the art should understand that the specific descriptions below are illustrative rather than restrictive, and should not be construed as limiting the scope of protection of the present invention.

[0068] This invention is based on the situation where traffic is not completely interrupted but the capacity of a highway is affected by an accident. By evaluating the temporal and spatial impact of the event, adopting appropriate traffic control strategies, and determining the specific implementation plan of the strategies, this invention proposes a simulation-driven emergency control method for highway accidents.

[0069] This invention studies situations where highways are affected by unexpected events requiring traffic control, but traffic is not completely interrupted. First, a basic highway information database, a highway unexpected event information database, and an unexpected event emergency management strategy information database (i.e., a highway control strategy expert database) are established to integrate information data as samples. When an unexpected event occurs, the temporal and spatial impact range of the event is predicted by combining real-time internet data with the simulation model of the highway control strategy expert database. Then, an emergency strategy selection model and a simulation model based on Variable Cell Transmission Model (VCTM) are used to determine the traffic control strategy and its specific implementation plan. Finally, the unexpected event and its control strategy are recorded in the information database to form a new sample.

[0070] This invention considers unforeseen events such as traffic accidents, severe weather, natural disasters, vehicle mechanical failures, and hazardous chemical leaks. It also considers traffic emergency control strategies including ramp control, variable speed limits, variable lanes, toll station control, route guidance, and signal control. The invention addresses the diverse and complex types of highways and will be further refined as the database is expanded. The emergency strategy control method of this invention forms a complete highway unforeseen event response system with wide applicability, capable of reducing the impact of unforeseen events on transportation efficiency and minimizing socio-economic losses.

[0071] Please see Figure 1 and Figure 5 The present invention provides a simulation-driven emergency control method for highway accidents, comprising the following steps:

[0072] S1. Establish a basic information database for highways and a database of information on highway accidents, and construct and verify a basic simulation model.

[0073] In this embodiment, based on the highway basic information database and the highway accident information database, simulation model components for the main line, weaving areas, ramps, toll stations, and ground connection sections are first established based on the cellular transmission model. Then, the parameters of the simulation model components are modified and used multiple times to synthesize a complete simulation model of the highway that can simulate the traffic flow operation state. This is called the basic simulation model. Step S1 includes, but is not limited to:

[0074] S101. Establish a national basic information database for expressways and an information database for expressway accidents.

[0075] In one embodiment of the present invention, the contents of the national expressway basic information database include, but are not limited to: expressway number, name, expressway starting latitude and longitude, expressway ending latitude and longitude, expressway daily traffic volume, expressway average hourly traffic volume, expressway hourly traffic volume (24 sets from 0:00 to 23:00), design speed, expressway average travel speed, number of lanes, and expressway geometry.

[0076] The geometric alignment of the highway is stored in the form of an AutoCAD file, in which the latitude and longitude of the highway starting point are converted into planar coordinates and coincide with the origin coordinates of AutoCAD.

[0077] In one embodiment of the present invention, the contents of the highway accident information database include, but are not limited to: accident number, accident type, event start time, event end time, event location latitude and longitude, length of the road segment affected by the event, event level, number of lanes blocked by the event, and highway number associated with the event.

[0078] Once the "event location latitude and longitude" is converted into planar coordinates, it can be linked to the highway where the accident occurred and its specific location through the "highway number associated with the event".

[0079] S102. Based on the variable cell transmission model, five road conditions are simulated respectively: the main line of the expressway, the weaving area, the ramp, the toll station, and the road section connecting to the ground.

[0080] In one embodiment of the present invention, the Variable Cellular Transport Model (VCTM) refers to dividing the elevated highway into several segments of varying lengths, called cells, and discretizing time into uniform time periods, with the traffic flow status of the cells being updated once per time period.

[0081] Optionally, in the variable cell transmission model, the highway mainline and ramps have modifiable attributes such as number of lanes, lane width, road segment length, road start coordinates, road end coordinates, road segment speed limit, road surface type (e.g., cement, asphalt), presence of emergency parking lanes, altitude, gradient, and turning radius; the highway weaving area has modifiable attributes such as number of lanes, road segment length, road start coordinates, road end coordinates, road segment speed limit, road surface type (e.g., cement, asphalt), presence of emergency parking lanes, altitude, gradient, turning radius, width of each lane, and reversible direction of each lane. The attributes of road types connecting lanes (mainline-mainline, mainline-ramp, ramp-mainline, ramp-ramp) can be modified; the attributes of highway toll stations (number of toll lanes, road segment length, road start coordinates, road end coordinates, toll lane type (e.g., ETC toll, manual toll) and toll lane efficiency) can be modified; the attributes of highway-to-ground connecting road segments (number of lanes, lane width, road segment length, road start coordinates, road end coordinates, road segment speed limit, gradient, channelization scheme of the first intersection, and signal control scheme of the first intersection) can be modified.

[0082] The recording method for the channelization scheme and signal control scheme of the first intersection of the highway ground connection section is as follows: the entrance lane where the road section connecting with the highway is located is used as the initial number, and the other entrance lanes of the intersection are numbered in a clockwise direction; further, each lane of the entrance lane is numbered starting from the lane closest to the center line of the road, and the direction of each lane to the entrance lane is recorded; then, the start and end times of the green light, the start and end times of the yellow light, and the start and end times of the red light for each lane are recorded under the same signal timing scheme.

[0083] S103. Select an actual highway, and by adjusting the variable attributes of the highway mainline, weaving areas, ramps, toll stations, and ground-level junctions or road sections, combine the simulated road condition map with the actual highway map, and run each basic simulation model to obtain average travel speed and traffic volume evaluation indicators.

[0084] In this embodiment of the invention, a complete real-world highway is selected. By adjusting the variable attributes of the highway mainline, weaving areas, ramps, toll stations, and ground-connecting intersections or road sections, the simulation model (i.e., the national highway basic information database) is combined to fit the complete real-world highway as closely as possible. The simulation model is then run to obtain average travel speed and traffic volume evaluation indicators.

[0085] A complete highway can be viewed as consisting of several parts: mainline, weaving zones, ramps, toll stations, and ground-level connecting intersections or sections. Step S101 establishes simulation model components for the mainline, weaving zones, ramps, toll stations, and ground-level connecting sections based on a cellular transmission model. Then, in step S102, the parameters of these simulation model components are modified and reused to form a complete highway simulation model. For example, if a real-world complete highway consists of mainline 1, weaving zone 1, ramp 1, ramp 2, ground-level connecting section 1, and ground-level connecting section 2, then simulation models for the mainline, weaving zones, ramps, and ground-level connecting sections are required. Basic simulation model connections are established based on vehicle flow direction; for example, if vehicles need to flow from mainline 1 to weaving zone 1, then mainline 1 and weaving zone 1 are combined. The above explains which parts are selected for combination and how to combine them into a complete simulation model. After verification, different control strategies are set for the simulation model in step S203, which is then called by step S4. This invention improves efficiency by reusing basic simulation models and avoids remodeling each highway.

[0086] S104. The simulation model is tested based on the average travel speed and traffic volume data of the highway. If the test requirements are met, proceed to S2; otherwise, proceed to S103 to adjust the model parameters and obtain the evaluation indicators again.

[0087] In one embodiment of the present invention, the simulation model is input with "average hourly traffic volume on highways", and the simulation output after 1 hour is "average travel speed". "and "number of vehicles passing through Q" VCTM "Average travel speed on highways" in the highway basic information database "and "average hourly traffic volume on highways Q" h "Compare them separately. If both equation (1) and equation (2) are satisfied, then the basic simulation model is considered to meet the requirements."

[0088]

[0089]

[0090] S2. Establish an expert database for highway management strategies and a simulation management plan.

[0091] In this embodiment, step S2 includes:

[0092] S201. Establish an expert database for highway management and control strategies.

[0093] In one embodiment of the present invention, the content of the highway management strategy expert database includes, but is not limited to: emergency management strategy number, emergency management strategy type, emergency management strategy content description, management strategy start time, management strategy end time, latitude and longitude of the management strategy implementation location, strategy implementation effect, accident number associated with the management strategy, and highway number associated with the management strategy.

[0094] Once the "latitude and longitude of the location where the management strategy is implemented" is converted into planar coordinates, it can be linked to the highway where the management strategy is located and its specific location through the "highway number associated with the management strategy".

[0095] In one embodiment of the present invention, each highway may correspond to multiple different emergency management strategies.

[0096] Each highway may correspond to multiple emergency management strategies. These strategies include emergency management strategy types and real-time selectable parameters. The "emergency management strategy type" can be considered a primary strategy, which is divided into ramp control, variable speed limit, variable lane, toll station control, route guidance, and off-ramp connection signal control. The "real-time selectable parameters" can be considered a secondary strategy. In ramp control, the parameter is the traffic input / output rate of ramp control; in variable speed limit, the parameter is the variable speed limit value; in variable lane, the parameter is the direction of travel of the variable lane; in toll station control, the parameter is the number of open toll lanes of each type; in route guidance, the parameter is the guidance path and vehicle guidance ratio; and in off-ramp connection signal control, the parameter is the intersection phase and timing scheme design.

[0097] S202. Based on the simulation model of S1, different control strategies are set to form a new simulation model, which is then numbered and stored for use by S4. At the same time, the simulation model sets the parameters of the control scheme as hyperparameters to quickly simulate the control scheme under different parameters and obtain evaluation indicators.

[0098] In one embodiment of the present invention, the highway simulation model established based on VCTM is as follows: Figure 2 As shown in the figure, the embodiment simulates a complete highway, consisting of 26 cells. Cells 1 and 16 are starting cells, representing the starting points at both ends of the highway; cells 15 and 26 are ending cells, representing the ending points at both ends of the highway; cells 5 and 9 are ramp cells, representing the locations where ramp control can be implemented; cells 11-14 are speed limit cells, representing the sections between cells 11 and 14 where variable speed limit control can be implemented; cells 17 and 25 are toll station cells, representing the sections between cells 17 and 25 where toll station control strategies can be implemented, determining the number of lanes open or closed; cells 17-25 are variable lane cells, representing the sections between cells 17 and 25 where variable lane control can be implemented; cells 10, 28, and 29 are signal control cells; the remaining cells are ordinary cells.

[0099] In one embodiment of the present invention, the hyperparameters set for different control schemes are different: 1) Ramp control hyperparameters include the cell number of the ramp, the ramp flow input / output rate, the ramp control start time, and the ramp control duration. 2) Variable speed limit hyperparameters include the variable speed limit control start cell, the variable speed limit control end cell, the variable speed limit value, the speed limit control start time, and the speed limit control end time. 3) Variable lane hyperparameters include the variable lane control start cell, the variable lane control end cell, the variable lane driving direction, the variable lane control start time, and the variable lane control end time. 4) Toll station control hyperparameters include the toll station cell number, the number of open ETC lanes at the toll station, the number of open manual lanes at the toll station, the number of open ETC / manual hybrid lanes at the toll station, the toll station control start time, and the toll station control end time. 5) Route guidance hyperparameters include the cell number of the guidance start position, the cell number of the guidance end position, the traffic flow of the highway after the route guidance is implemented, the traffic flow of the new route after the route guidance is implemented, the guidance control start time, and the guidance control duration. 6) The signal control parameters at the off-ramp connection intersection include the cell number controlled by the signal, the signal control period, the number of phases, the phase sequence, and the effective green light time for each phase.

[0100] In one embodiment of the present invention, each control scheme can be set at multiple locations on the highway, and different control schemes can be combined; when the highway does not meet the implementation conditions of a certain control scheme, that control scheme is not selectable; wherein, the parameter solution method for the highway traffic flow and the traffic flow of the new route after the implementation of the route guidance is as follows:

[0101] Based on the BPR impedance function and Wardrop's first principle of traffic flow assignment (Wardrop's equilibrium principle, in which scholars proposed the first and second principles of traffic network equilibrium, laying the foundation for traffic flow assignment), the path guidance concept states that "the traffic flow q on the highway after the implementation of path guidance..." a "and "traffic flow q of the new route after the implementation of route guidance" b "By solving equations (3) and (4) simultaneously, we can obtain:

[0102]

[0103] q a +q b =Q h +Q hnew (4)

[0104] In the formula: t a t represents the free-flow time from the start to the end of the route guidance process on the highway.b Let q be the free-flow time of the new path from the start to the end of the induced segment. a q b C represents the traffic flow on the highway and the new route after the implementation of route guidance. a C b Q represents the practical capacity of the highway and the new route, respectively, with α and β as parameters. The Federal Highway Administration recommends α = 0.15 and β = 4. hnew Traffic flow for new routes before route guidance is implemented.

[0105] S3. Online analysis of the temporal and spatial impact range of highway accidents.

[0106] In this embodiment, step S3 includes:

[0107] S301. When an accident occurs on a highway, obtain detector data and internet statistics from the past month to the present for that highway.

[0108] In one embodiment of the present invention, when an accident occurs on the highway and is reported to the control center, the method is triggered to obtain the number of vehicles and average travel speed data of the highway from the past month to the present. The number of vehicles can be obtained by induction coil vehicle detectors or video vehicle detectors installed on the highway. The average travel speed data is crawled by map software at regular intervals. In this embodiment, the crawled data is set at 5-minute intervals according to the cells of the highway simulation model. Each cell is associated with each segment to obtain the average travel speed data of each cell of the highway, which serves as the basis for S302.

[0109] S302. Analyze the duration of events based on survival analysis models.

[0110] The survival analysis model is an analytical method introduced from the field of medical statistics (Press C. Statistical and Econometric Methods for Transportation Data Analysis, Second Edition [M]. CRC Press, 2010.). It is characterized by its targeted study of the duration of events and its strong capabilities in handling censored data and analyzing influencing variables.

[0111] In one embodiment of the present invention, survival analysis of the duration of highway accidents refers to a method of analyzing and inferring the duration of an event based on existing data, and studying the relationship between the duration of the event and numerous influencing factors and the degree of their influence. Censored data refers to data that has been cut off for various reasons, and is a key feature of survival analysis; the survival function (survival rate) represents the probability that the duration T of the event is longer than time t, denoted as S(t):

[0112]

[0113] Where f(x) and F(t) represent the probability density function and distribution function of the event duration T, respectively. T is a continuous random variable, and its survival function is also the integral of the probability density function f(x).

[0114] The hazard function (conditional death probability) represents the probability that an unexpected event will end within a subsequent time interval Δt after a duration of t, and is denoted as h(t):

[0115]

[0116] h(t) represents the probability that the event will end at a subsequent time interval Δt after a duration of t; T is the duration of the study; f(t) is the probability that the duration is t; S(t) is the survival rate. As can be seen from the expression, the larger the value of the hazard function as a conditional probability, the greater the probability that the event will end at the next moment.

[0117] In one embodiment of the present invention, lifetime plot analysis, Kaplan-Mayer analysis (Kaplan-Mayer is a univariate survival analysis), and Cox regression analysis are used as a set of survival analysis methods to study the duration of unexpected events. The lifetime plot is used to record and statistically analyze the state changes of the population under the analyzed event, and to statistically represent statistical measures such as the median. Kaplan-Mayer analysis is a nonparametric method used to estimate the survival function and the hazard function, denoted as T1 < T2…T… n For a sample representing the durations of n events, the Kaplan-Mayer analysis estimates the survival function S(t) as follows:

[0118]

[0119] In the formula, T to tal represents all events. Let i be the i-th event after sorting. Event i must be complete, uncensored data; censored data should be recorded separately. Kaplan-Mayer can quantitatively analyze the influence of a factor on the duration of an event and determine whether there is a significant difference in the influence of that factor on the outcome by combining it with the Log-rank test.

[0120] Cox regression analysis is a semi-parametric regression model that uses the ratio of the hazard function to the baseline hazard function as the dependent variable to reflect the influence of different independent variables. It is expressed as:

[0121] h(t,x)=h0(t)exp(β1x1+β2x2+…+β i x i (8)

[0122] In the formula, β1,β2…β i Here are the regression coefficients, x1, x2…x i Let h(t,x) be the independent variable, h(t,x) be the hazard function, and h0(t) be the basic hazard function, which represents the inherent hazard function of the event duration when no other factors are involved.

[0123] Parametric regression presupposes the form of the model and then uses data to estimate its coefficients. Nonparametric regression, on the other hand, does not assume a model form and directly fits the model to the data. Semiparametric regression involves a part of the model's structure being known and requiring parameter estimation, while another part of the structure is unknown. Cox regression is a type of semiparametric regression model.

[0124] In one embodiment of the present invention, this step obtains the regression coefficient of the hazard function for the duration of highway accidents through data analysis of the "sample information database." Based on the real-time event input variable values, the duration of the event can be predicted. The sample information database includes basic road segment information, accident information, and control strategy information. The "sample information database" refers to the "National Highway Basic Information Database, Highway Accident Information Database, and Highway Control Strategy Expert Database."

[0125] S303. Analyze the scope of an event's impact using outlier identification based on the IQR method.

[0126] In one embodiment of the present invention, based on the research approach of traffic wave models and using the "average travel speed data of each cell of the highway from one month in history to the present" obtained by S3, the event impact range analysis process is as follows: Figure 3 As shown, the overall approach is to first determine the scope of influence of the sample event, establish a predictive model for the scope of influence of the event, and then input the features of the real-time event to predict the scope of influence of the real-time event.

[0127] Step 1: Locate the cells of the highway based on data from the "Highway Accident Information Database," using the cell where the event is located as the center for the impact range analysis; Step 2: Plot the average travel speed data curve for each cell, specifically statistically analyzing data from three hours before and after the event, as well as three-hour data from the same week and time period over historical periods of 7, 14, and 21 days. The average travel speed data curve for a given cell plotted in this embodiment is shown below. Figure 4 As shown, based on historical data from three hours before and after the accident (13:00-16:00), and speed data from the same time period on three days in the same month (December 2, 2018, December 9, 2018, and December 16, 2018), after exponential smoothing, line graphs of historical speeds and accident times are obtained for each road segment. Figure 4 The horizontal axis represents the observation times for each data point over three hours (180 minutes) from 13:00 to 16:00, and the vertical axis represents the average travel speed within a cell. Step 3 sets a threshold for the duration of outliers based on professional experience; in this invention, it is set to 10 minutes. That is, if the average travel time of a cell exceeds the outlier threshold by 10 minutes compared to historical data, the cell is considered to be affected by the event. Step 4 is outlier analysis; this invention uses an outlier analysis method based on IQR (InterQuartile Range). Step 5 involves selecting all cells affected by the event and their degree of influence, expressed as the decrease in average travel speed. Step 6 establishes a Bayesian-based prediction model for the event's impact range, with input features including "daily average traffic volume on highways, design speed, and..." The data includes: average speed of vehicles traveling on highways, number of lanes on the road, type of accident, start time of the event, end time of the event, latitude and longitude of the event location, cell number of the event, distance of the studied cell from the cell of the event, event level, number of lanes blocked by the event, and highway number associated with the event. The label is "degree of influence of the event on the cell" (0 means no influence, positive value means positive influence and increased average speed of vehicles traveling on highways, negative value means negative influence and decreased average speed of vehicles traveling on highways). Step 7 is to replace the "end time of the event" with the sum of "predicted event duration" and "event start time" based on the attributes of the real-time event, and combine it with other features to output the result "degree of influence of the event on the cell". After multiple predictions and sorting, the degree of influence of the accident event on all cells can be obtained, that is, the scope of influence of the event.

[0128] S4. Dynamically match traffic control strategies based on the expert database of highway control strategies.

[0129] In this embodiment, step S4 specifically includes:

[0130] S401. An emergency strategy selection model was trained based on an expert database of highway control strategies.

[0131] S402. Select the primary emergency management strategy type for highway accidents using the emergency strategy selection model.

[0132] In one embodiment of the present invention, an emergency strategy selection model is trained using sample information database data of S2. The model can be based on machine learning methods such as Bayesian and random forest, or on neural networks. The input features are "highway number, average daily traffic volume of the highway, average hourly traffic volume of the highway, design speed, average travel speed of the highway, number of road lanes, accident type, event start time, event end time, event location latitude and longitude, length of the road segment affected by the event, event level, number of lanes blocked by the event, and highway number associated with the event". The output result is "emergency management strategy" (1 for ramp control, 2 for variable speed limit, 3 for variable lane, 4 for toll station control, 5 for route guidance, and 6 for off-ramp connection signal control). When selecting a control scheme for a real-time event, the input comes from known highway information, real-time accident attributes, and predicted event duration and impact range.

[0133] The emergency strategy selection model can be based on machine learning methods such as Bayesian and random forests, or on neural networks. The input features are: "highway number, average daily traffic volume of the highway, average hourly traffic volume of the highway, design speed, average travel speed of the highway, number of road lanes, type of accident, start time of the event, end time of the event, latitude and longitude of the event location, length of the affected road segment, event level, number of lanes blocked by the event, and highway number associated with the event." The output is "emergency management strategy" (1 for ramp control, 2 for variable speed limit, 3 for variable lane, 4 for toll station control, 5 for route guidance, and 6 for off-ramp connection signal control). When selecting a control scheme for a real-time event, the input comes from known highway information, real-time accident attributes, and predicted event duration and impact range.

[0134] S403. Select real-time parameters for a secondary emergency management strategy for highway accidents using a simulation model.

[0135] In one embodiment of the present invention, after selecting a Level 1 emergency management strategy type for an event, the simulation model corresponding to S2 is invoked, and the hyperparameters of the simulation model's control scheme are changed in a controlled manner. For each set of parameters, from the start of the simulation to the end of the event, evaluation indicators such as "average travel speed," "number of vehicles passing through," and "average vehicle delay" are output. Since some parameters are continuous variables, upper and lower limits are set for the variables, and after several sets of simulations, the evaluation indicators are supplemented by interpolation.

[0136] S404. Output the parameters corresponding to the optimal evaluation index in the basic simulation model, as a detailed control plan for the highway accident.

[0137] In one embodiment of the present invention, the total delay time of the vehicle from the start of the simulation to the end of the event is calculated for each set of parameters. The total delay time of the vehicle is the product of the number of vehicles passing through and the average delay of the vehicle. When the product is minimized, the evaluation index of the set of parameters is considered to be optimal.

[0138] S5. After the emergency control measures on the highway are completed, the incident information and emergency management strategy information are compiled and stored in the highway control strategy expert database as a new traffic control strategy sample.

[0139] In one embodiment of the present invention, the system reports a chain-reaction rear-end collision on a highway, causing abnormal congestion. The incident handling process is illustrated below: Step 1: The incident triggering system acquires the number of vehicles and average travel speed data for the highway from the past month to the present. Step 2: Based on the survival analysis model, the duration of the incident is analyzed, and based on the outlier identification method of the IQR method, the scope of the incident's impact is analyzed. Step 3: Combining the highway basic information database and the highway accident information database, using highway information, traffic accident attributes, and the predicted duration and scope of the incident as input, the emergency strategy selection model selects the primary emergency management strategy type for the incident as "1 ramp control, 5 path guidance, 6 off-ramp connection signal control". Step 4: The simulation model that has been trained offline and includes the "1 ramp control, 5 path guidance, 6 off-ramp connection signal control" management strategy is called. In the secondary strategy, the hyperparameters of ramp control are set as follows: the cell number of the ramp is set to [the cell number of the ramp at the incident point], the ramp flow input rate is set to [do not close lanes, close 1 lane], and the ramp control start time is set to [when emergency management personnel can reach the ramp]. The following parameters are set: [Fastest time for the road], [Range control duration] [10 minutes, 20 minutes, 30 minutes, 40 minutes], [Path guidance hyperparameters], [Induction start position cell number] [Number of the nearest intersection to the on-ramp from the event point], [Induction end position cell number] [Number of the nearest intersection to the next on-ramp from the event point], [Highway traffic flow after path guidance implementation] [50%, 40%, 30%, 20%, 10% of the original highway traffic flow], [Induction control start time] [Fastest time to implement guidance control], [Induction control duration] [10 minutes, 20 minutes, 30 minutes, 40 minutes], [Off-ramp connection intersection signal control hyperparameters], [Signal-controlled cell number] [Cell number contained in the off-ramp connection intersection], [Signal control cycle, number of phases, and phase sequence remain consistent with the original scheme], [Effective green light time of the phase of the off-ramp section], [30 seconds, 40 seconds, 50 seconds], [Control variables change the hyperparameters of the simulation model control scheme]. For each set of parameters, from the start of the simulation to the end of the event, the evaluation indicators "average travel speed", "number of vehicles passing", and "average vehicle delay" are output. Step 5: For each set of parameters, the total delay time of vehicles from the start of the simulation to the end of the event is calculated. The total delay time of vehicles is the product of "number of vehicles passing" and "average vehicle delay". The parameter set with the smallest product is considered to have the best evaluation indicators. Step 6: The event information and emergency management strategy information are organized and added to the information database as new samples.

[0140] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A simulation-driven emergency control method for highway accidents, characterized in that, Includes the following steps: S1. Establish a basic information database for highways and a database of highway accidental events, construct and verify a basic simulation model, and run the basic simulation model to obtain average travel speed and traffic volume evaluation indicators. S2. Establish an expert database for highway management strategies and a simulation management plan; S3. Online analysis of the temporal and spatial impact of highway accidents; S4. Dynamically match traffic control strategies based on the expert database of highway control strategies; S5. After the emergency control measures on the highway are completed, the incident information and emergency management strategy information are compiled and stored in the highway control strategy expert database as a new traffic control strategy sample. Step S2 includes: S201. Establish an expert database for highway management and control strategies; S202. Based on the basic simulation model of S1, set different traffic control strategies to form a new simulation model and store it with numbers for matching and calling in step S4. At the same time, the simulation model sets the parameters of the simulation control scheme as hyperparameters to quickly simulate traffic control strategies under different parameters and obtain evaluation indicators. The hyperparameters for different traffic control strategies include: Ramp control hyperparameters include the cell number of the ramp, ramp flow input / output rate, ramp control start time, and ramp control duration; The variable speed limit hyperparameters include the variable speed limit control start cell, the variable speed limit control end cell, the variable speed limit value, the speed limit control start time, and the speed limit control end time. The hyperparameters of the variable lane control include the variable lane control start cell, the variable lane control end cell, the variable lane driving direction, the variable lane control start time, and the variable lane control end time. The parameters for toll station control include the cell number of the toll station, the number of open ETC lanes at the toll station, the number of open manual lanes at the toll station, the number of open ETC / manual hybrid lanes at the toll station, the start time of toll station control, and the end time of toll station control. The hyperparameters of the path guidance include the cell number of the starting position of the guidance, the cell number of the ending position of the guidance, the traffic flow of the highway after the implementation of the path guidance, the traffic flow of the new path after the implementation of the path guidance, the start time of the guidance control, and the duration of the guidance control. The signal control hyperparameters at the off-ramp connection intersection include the cell number controlled by the signal, the signal control period, the number of phases, the phase sequence, and the effective green light time for each phase. The parameters for calculating the traffic flow on the highway and the traffic flow on the new route after the implementation of the route guidance are as follows: Based on the BPR impedance function and Wardrop's first principle of traffic flow assignment, the traffic flow on the highway after the implementation of path guidance is determined. Traffic flow on new routes after route guidance implementation The following equations can be solved simultaneously: （3） （4） In the formula: This refers to the free-flow time from the start to the end of the highway route guidance segment. The free-flow time of the new path from the start to the end of the induced segment. , These are the traffic flows on the highway and the new route after the implementation of route guidance. , These refer to the practical traffic capacity of highways and new routes, respectively. , For parameters, the U.S. Federal Highway Administration recommends , , Traffic flow for new routes before route guidance is implemented; The online analysis of the temporal and spatial impact of unexpected events includes: Based on the data from the highway accident information database, the cells of the highway are located, and the cell where the event is located is used as the center for the analysis of the scope of impact. Plot the average travel speed data curve for each cell, and statistically analyze the data for three hours before and after the event, as well as the three-hour data for the same week and time period in the past 7 days, 14 days, and 21 days. Set a threshold for the duration of outliers, where the threshold is 10 minutes. That is, if the average travel time of a cell exceeds the outlier threshold for 10 minutes compared to the historical time, the cell is considered to have been affected by the event. Outliers were analyzed using the IQR-based method. Select all cells affected by the event and the degree of their impact; A Bayesian-based event impact range prediction model is established. The input features are the daily traffic volume of the highway, the design speed, the average travel speed of the highway, the number of lanes on the road, the type of accident event, the start time of the event, the end time of the event, the latitude and longitude of the event location, the cell number where the event is located, the length of the distance between the cell under study and the cell where the event is located, the event level, the number of lanes blocked by the event, and the highway number associated with the event. The label is the degree of influence of the event on the cell. Based on the attributes of real-time events, the predicted event duration is summed with the event start time to replace the event end time. Combined with the degree of influence of the event on the cells, after multiple predictions and sorting, the degree of influence of the unexpected event on all cells can be obtained, that is, the scope of the event's influence.

2. The simulation-driven emergency control method for highway accidents according to claim 1, characterized in that, Step S1 includes: S101. Establish a national expressway basic information database and an expressway accident information database; S102. Based on the variable cell transmission model, simulate the road conditions of the main line of the expressway, weaving area, ramps, toll stations and the road sections connecting to the ground. S103. Select an actual highway, and by adjusting the variable attributes of the highway mainline, weaving areas, ramps, toll stations, and ground-connecting intersections or road sections, combine the simulated road condition map with the actual highway map, and run each basic simulation model to obtain average travel speed and traffic volume evaluation indicators. S104. Based on the average travel speed and traffic volume data of the highway, test each basic simulation model. If the test requirements are met, proceed to step S2; otherwise, adjust the simulation model parameters, obtain the evaluation index again, and proceed to S103.

3. The simulation-driven emergency control method for highway accidents according to claim 2, characterized in that, Verifying whether each basic simulation model meets the requirements includes: In each basic simulation model, input "average hourly traffic volume on highways" and output "average travel speed" after 1 hour of simulation. "and" number of vehicles passing through "Average travel speed on highways" in the highway basic information database "and average hourly traffic volume on highways" "Compare them separately, and if the following formulas are satisfied simultaneously, the simulation model is considered to meet the requirements; （1） （2）。 4. The simulation-driven emergency control method for highway accidents according to claim 2, characterized in that, The content of the highway control strategy expert database includes, but is not limited to: emergency management strategy number, emergency management strategy type, emergency management strategy content description, management strategy start time, management strategy end time, latitude and longitude of the management strategy implementation location, strategy implementation effect, accident number associated with the management strategy, and highway number associated with the management strategy.

5. The simulation-driven emergency control method for highway accidents according to claim 1, characterized in that, Step S3 includes: S301. When an accident occurs on a highway, obtain detector data and internet statistics from the past month to the present for that highway. S302. Analyze the duration of events based on survival analysis models; S303. Analyze the scope of an event's impact using outlier identification based on the IQR method.

6. The simulation-driven emergency control method for highway accidents according to claim 5, characterized in that, Survival analysis models analyze the duration of events, representing the event duration. T Longer than time t The probability of is obtained using the following formula: （5） in , These represent the duration of the event. T Density function, distribution function T For continuous random variables, The survival function is also the probability density function. The points.

7. The simulation-driven emergency control method for highway accidents according to claim 1, characterized in that, Step S4 includes: S401. An emergency strategy selection model is trained based on an expert database of highway control strategies. S402. Select the primary emergency management strategy type for highway accidents using the emergency strategy selection model; S403. Select real-time parameters for a secondary emergency management strategy for highway accidents using a basic simulation model; S404. Output the parameters corresponding to the optimal evaluation index in the basic simulation model, as a detailed control plan for the highway accident.

Images