A highway network macroscopic cascade failure prediction method under a large-scale emergency

Abstract

The application discloses a highway network macroscopic cascade failure prediction method under a large-scale emergency, and comprises the following steps: identifying whether each road section of a highway network is congested at different time periods according to detection data; judging the state of each road section at a current time node t, wherein the road section state comprises a free-flow road section, a recovery road section, a congested road section and a propagating road section; calculating the shock wave speed of different road sections and the propagation time of the shock wave on the road sections according to the conversion of the road section state; calculating the conversion rate of each road section state at the current time node t, generating a cascade failure evolution model of the highway network, and predicting the macroscopic evolution law of the cascade failure of the highway network. The application can explore the evolution law of the macroscopic cascade failure of the highway network, and predict the failure degree of the highway section at different time nodes.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of dynamic model and traffic big data, and particularly relates to a highway network macroscopic cascading failure prediction method under a large-scale emergency. BACKGROUND

[0002] In recent years, the intelligent construction of expressways provides convenience for researchers, and a large number of sensing devices such as GPS, vehicle electronic tags, and induction coils are installed in the expressway network, providing a large amount of accurate data for researchers, thereby providing many conveniences for the prediction of cascading failure.

[0003] The research on the space-time propagation of cascading failure in the expressway network mainly includes research based on dynamic traffic distribution, queuing models, and research based on shock wave theory, but there are still certain limitations in the related research. Deterministic queuing models and shock wave models need to master real-time traffic flow characteristics in detail, and the parameters of such models require a large amount of historical data, and the data of emergencies is difficult to obtain, most of the research is based on simulation, and it is difficult to realize in reality. In addition, the propagation law of cascading failure under the expressway emergency is still unclear, the model is difficult to establish, the parameter calibration in the existing research is difficult, and the universality is poor. The failure of the road section is not identified by using a suitable method, and the spread of failure in the expressway network is simulated. SUMMARY

[0004] The technical problem solved by the application is that the application discloses a highway network macroscopic cascading failure prediction method under a large-scale emergency, which can quickly and accurately predict the cascading failure of the expressway network from a macroscopic perspective, so that the highway network management personnel can quickly respond to the emergency and minimize the impact of the emergency on the expressway.

[0005] Technical scheme

[0006] A highway network macroscopic cascading failure prediction method under a large-scale emergency, the highway network macroscopic cascading failure prediction method comprises the following steps:

[0007] Step A, collecting detection data of detectors distributed on the highway network, and identifying whether each road section of the highway network is congested at different time periods according to the detection data;

[0008] Step B, combine the congestion recognition results of each road segment of the highway network at different time periods to determine the state of each road segment at the current time node t, the road segment state including a free flow road segment, a recovery road segment, a congestion road segment and a propagation road segment; wherein the free flow road segment refers to that the current road segment has not been congested in the time period [0, t]; the recovery road segment refers to that the current road segment has been congested in the time period [0, t] but is in a free flow state at the time node t; the congestion road segment refers to that the current road segment is in a congested state at the time node t and the congestion has propagated to the upstream road segment; the propagation road segment refers to that the current road segment is in a congested state at the time node t but the congestion has not propagated to the upstream road segment;

[0009] Step C, calculate the shock wave speed and the propagation time of the shock wave on the road segment according to the conversion of the road segment state;

[0010] Step D, calculate the conversion rate of each road segment state at the current time node t, generate a cascading failure evolution model of the highway network, and predict the macro evolution law of the cascading failure of the highway network.

[0011] Further, in step A, the detection data of the detector includes position information of the detector, detection time, and measured vehicle flow, vehicle density and vehicle speed.

[0012] Further, in step A, the process of identifying whether each road segment of the highway network is congested at different time periods according to the detection data includes the following steps:

[0013] According to the time node, divide the data, take the time node of the occurrence of the event as the starting time node, and identify whether the road segment is congested according to the speed data measured by the detector:

[0014]

[0015]

[0016] wherein j is the road segment number, v j (t) is the average speed of the j road segment at the time node t, in order to limit the speed of the road segment, ε j (t) is the average speed of the j road segment at the time node t, j (t) is the ratio of v , ρ is a given threshold, θ j (t) is a 0-1 variable, θ j (t) = 1 represents that the j road segment is in a congested state at the time node t, and θ j (t) = 0 represents that the j road segment is in a free flow state at the time node t.

[0017] Further, in step B, the following formula is used to calculate the shock wave speed of different road segments according to the conversion of the road segment state:

[0018]

[0019] In the formula, u w is the shock wave speed, and represents the shock wave speed generated when the traffic flow with flow rate Q1 and density K1 encounters the traffic flow with flow rate Q2 and density K2.

[0020] Further, in step D, the transition rate of each road section state at the current time node t is calculated, and a cascading failure evolution model of the highway network is generated:

[0021]

[0022] Wherein F(t), H(t), C(t) and R(t) are the number of road sections in the free flow state, the propagating state, the congestion state and the recovery state at the t time node respectively; P FH (t) represents the transition rate of the free flow road section to the propagating road section at the t time node, P HC (t) represents the transition rate of the propagating road section to the congestion road section at the t time node, P HR (t) represents the transition rate of the propagating road section to the recovery road section at the t time node, P CR (t) represents the transition rate of the congestion road section to the recovery road section at the t time node; λ is the mean value of the number of upstream road sections on the effective link of the road network; T FH (t) represents the time required for the transition of the free flow road section to the propagating road section at the t time node, T HC (t) represents the time required for the transition of the propagating road section to the congestion road section at the t time node, T HR (t) represents the time required for the transition of the propagating road section to the recovery road section at the t time node, T CR (t) represents the time required for the transition of the congestion road section to the recovery road section at the t time node.

[0023] Beneficial effects:

[0024] First, the highway network macroscopic cascading failure prediction method under large-scale emergencies of the application can be used for real-time monitoring of the cascading failure condition of the highway after the emergency, and the optimization strategy can be adjusted according to different failure scenarios.

[0025] Second, the highway network macroscopic cascading failure prediction method under large-scale emergencies of the application has portability and high universality, and can update the network state in real time according to the traffic parameter input, adapt to different highway networks, and does not need a complex parameter correction model.

[0026] Thirdly, the highway network macroscopic cascading failure prediction method under large-scale emergencies has high model precision, simple calculation, time effectiveness and accuracy, and can quickly and accurately predict the space-time characteristics of cascading failure. BRIEF DESCRIPTION OF DRAWINGS

[0027] Figure 1 A flow chart of the highway network macroscopic cascading failure prediction method under large-scale emergencies according to the embodiment of the present application;

[0028] Figure 2 A schematic diagram of detector distribution in Minnesota;

[0029] Figure 3 A histogram of the number of road segments in different states according to the embodiment of the present application;

[0030] Figure 4 A comparison diagram of cascading failure prediction according to the embodiment of the present application. DETAILED DESCRIPTION

[0031] The following embodiments enable a person skilled in the art to more comprehensively understand the present application, but do not limit the present application in any way.

[0032] Figure 1 A flow chart of the highway network macroscopic cascading failure prediction method under large-scale emergencies according to the embodiment of the present application. Referring to Figure 1 The highway network macroscopic cascading failure prediction method comprises the following steps:

[0033] Step A, collecting detection data of detectors distributed on the highway network, and identifying whether each road segment of the highway network is congested at different time periods according to the detection data;

[0034] Step B, combining the congestion identification results of each road segment of the highway network at different time periods, judging the state of each road segment at the current time node t, and the road segment state comprises a free flow road segment, a recovery road segment, a congested road segment and a propagating road segment; wherein the free flow road segment refers to a road segment that has not been congested in the time period [0, t]; the recovery road segment refers to a road segment that has been congested in the time period [0, t] but is in a free flow state at the time node t; the congested road segment refers to a road segment that is in a congested state at the time node t and the congestion has propagated to the upstream road segment; and the propagating road segment refers to a road segment that is in a congested state at the time node t but the congestion has not propagated to the upstream road segment;

[0035] Step C, calculating the shock wave speed and the propagation time of the shock wave on the road segment according to the conversion of the road segment state;

[0036] Step D, calculating the conversion rate of each road segment state at the current time node t, generating a cascading failure evolution model of the highway network, and predicting the macroscopic evolution law of the highway network cascading failure.

[0037] Specifically, the highway network macro-cascading failure prediction method comprises the following steps:

[0038] (1) Data arrangement and extraction: The detector data is arranged, and field data including detector position, detector measured flow, detector measured density and detector measured speed is extracted.

[0039] (2) Road section state identification:

[0040] (21) The data is divided according to time nodes, and the time node of the occurrence of the event is taken as the starting time node, and whether the road section is congested is identified according to the detector speed data extracted in step (1):

[0041]

[0042] Where j is the road section number, v j (t) is the average speed of the j road section at the t time node, is the speed limit of the road section, and ε j (t) is the ratio of v j (t) and v .

[0043]

[0044] In the formula, θ j (t) is a 0-1 variable, ρ is a threshold value given in advance, and θ j (t) = 1 represents that the j road section is in a congested state at the t time node.

[0045] (22) Further divide the road section state, if the j road section is in a free flow state at the t time node and has not been congested in the [0-t] time window, the j road section is a free flow road section at the t time node, otherwise it is a recovery road section; if the j road section is in a congested state at the t time node, and the congestion has propagated to the upstream node, the j road section is a congested road section at the t time node, otherwise it is a propagating road section.

[0046] (3) Shock wave speed calculation: on the basis of the road section classification in step (2), the shock wave speed is calculated according to the conversion of the road section state:

[0047]

[0048] In the formula, U w is the shock wave speed, when the traffic flow with a flow of Q1 and a density of K1 encounters the traffic flow with a flow of Q2 and a density of K2, the generated shock wave speed is u w .

[0049] (4) Cascade failure law summary: use the shock wave speed obtained in step (3), based on the FHCR model, to summarize the cascade failure evolution law at the road network level, the specific operation is as follows:

[0050] (41) Determine the state transition of each road segment at the current time node;

[0051] (42) According to different state transitions, calculate the shock wave speed of the road segment, and then calculate the propagation time of the shock wave on the road segment. FH (t) represents the time required for the free flow road segment at the t time node to transform into the propagation road segment, δ HC (t) represents the time required for the propagation road segment at the t time node to transform into the congestion road segment, δ HR (t) represents the time required for the propagation road segment at the t time node to transform into the recovery road segment, T CR (t) represents the time required for the congestion road segment at the t time node to transform into the recovery road segment;

[0052] (43) Record the conversion rate of each road segment state at the current time node, and predict the cascade failure evolution law of the highway network according to the available data:

[0053]

[0054] Where F(t), H(t), C(t), R(t) are the number of free flow, propagation, congestion and recovery state road segments at t time node; P FH (t) represents the conversion rate of the free flow road segment at the t time node to the propagation road segment; similarly, P HC (t), P HR (t) and P CR (t) are the conversion rate of the propagation road segment at the t time node to the congestion road segment, the conversion rate of the propagation road segment to the recovery road segment, and the conversion rate of the congestion road segment to the recovery road segment, respectively; λ is the average number of upstream road segments on the effective link in the road network.

[0055] The cascade failure evolution model of the highway network generated in step (4) uses a set of dynamic differential equations, and the data input is the data collected by each detector. The macroscopic evolution law of the highway network cascade failure is obtained by calculating the shock wave speed and the conversion rate and conversion time at each time node.

[0056] Example

[0057] This example uses the case of the collapse of the I-35W Bridge in Minnesota on August 1, 2007, and predicts the cascade failure of the highway network based on the FHCR model through data analysis. The method flow chart is shown in the attached Figure 1 , which mainly includes the following four stages:

[0058] (I) Data collection and extraction: The data extraction includes six fields of detector number, date, time node, detector flow, detector speed, and detector density of the study area. The resulting data is shown in Table 1. The time node data is accurate to the second, for example, 20070801180130 represents the data recording time of August 1, 2007, 18:01:30.

[0059] Table 1. FHCR data table

[0060] Detector number Date Time Flow Speed Density S1009 2007 / 8 / 1 20070801000030 720 86.3372 1.1905 S1010 2007 / 8 / 1 20070801180030 4320 46.7734 114.1248 S1011 2007 / 8 / 1 20070801201300 2400 70.6652 20.2631 S1012 2007 / 8 / 1 20070801080230 4600 57.3314 96.3036 ... ... ... ... ... ...

[0061] (II) Road section state identification: Two adjacent detectors form a road section. According to the detector data, the road section state at each time node after the bridge collapse is identified and the number of congested road sections is counted. As shown in the following table. Figure 3

[0062] (III) Shock wave speed calculation and cascade failure prediction: In order to verify the accuracy of the prediction results, this experiment compares the predicted data with the actual data after the bridge collapse, and explores the evolution law of the cascade failure of the highway network at the macro level. After the calculation of the shock wave speed, the following steps are mainly included:

[0063] (1) Determine the state transition of each road section at the current time node;

[0064] (2) According to different state transitions, calculate the shock wave speed of the road section, and then calculate the propagation time of the shock wave on the road section. FH (t) represents the time required for the free flow road section at time node t to convert to the propagating road section, T HC (t) represents the time required for the propagating road section at time node t to convert to the congested road section, T HR (t) represents the time required for the propagating road section at time node t to convert to the recovery road section, T CR (t) represents the time required for the congested road section at time node t to convert to the recovery road section;

[0065] (3) Record the conversion rate of each road section state at the current time node, and according to the available data, predict the cascade failure evolution law of the highway network:

[0066]

[0067] where F(t), H(t), C(t), and R(t) are the number of free flow, propagating, congested, and recovery state road sections at time node t, respectively. FH (t) represents the conversion rate of the free flow road section to the propagating road section at time node t, and similarly, P HC (t), P HR (t), and P​CR (t) is respectively the conversion rate of the link section in the propagation to the congestion link section, the conversion rate of the link section in the propagation to the recovery link section and the conversion rate of the congestion link section to the recovery link section; and λ is the mean value of the number of upstream link sections of the effective link in the road network.

[0068] The cascade failure prediction result is as shown in the following table 1 Figure 4

[0069] (Fourth) Comparison of prediction results: compare the number of predicted congestion link sections with the number of real congestion link sections, and use R 2 index as the measurement:

[0070]

[0071] wherein C(t) is the number of real congestion link sections at each time point, and the average value of the number of congestion link sections at each time point in the research time period.

[0072] The R 2 index of the two research areas is shown in table 2.

[0073] Table 2. R 2 index calculation table

[0074] Region [R 2 ]]> Los Angeles 0.9954 Minnesota 0.9941

[0075] As can be seen from table 2, the prediction accuracy of the present application is more than 90%, thus indicating that the macro cascade failure prediction method of highway network under large-scale emergency event proposed by the present application is feasible and has high prediction accuracy.

[0076] The macro cascade failure prediction method of highway network under large-scale emergency event of the present application, under the multi-source traffic big data environment constructed, comprehensively considers the network traffic flow theory and the technical method of the infectious disease model, combines the shock wave theory, analyzes the space-time distribution mechanism of traffic flow, considers the transition between network failure states and travel behavior adjustment, integrates the evolution law of traffic flow situation, forms a magnitude model of cascade failure evolution through data-driven reconstruction, performs prediction and optimization according to the model, dynamically updates the network cascade failure state, and reflects the evolution law of the whole process cascade failure with time from the road network level. On the basis of quickly and accurately predicting the highway network cascade failure from the macro perspective, the present application can quickly respond to the emergency event and minimize the influence of the emergency event on the highway.

[0077] ​The above are only preferred embodiments of the present application, and the protection scope of the present application is not limited to the above-mentioned embodiments. Any technical scheme falling within the concept of the present application shall fall within the protection scope of the present application. It should be noted that, for ordinary skilled in the art, some improvements and refinements without departing from the principles of the present application shall be considered as the protection scope of the present application.

Claims

1. A highway network macro-cascading failure prediction method under a large-scale emergency, characterized in that, The high-speed network macro cascade failure prediction method comprises the following steps: Step A, collecting detection data of detectors distributed on the high-speed network, and identifying whether each road section of the high-speed network is congested at different time periods according to the detection data; Step B, combining the congestion identification results of each road section of the high-speed network at different time periods, judging the state of each road section at the current time node t, the road section state including a free flow road section, a recovery road section, a congestion road section and a propagation road section; wherein the free flow road section refers to the current road section that has never been congested in the time period [0, t]; the recovery road section refers to the current road section that has been congested in the time period [0, t] but is in a free flow state at the time node t; the congestion road section refers to the road section that is in a congested state at the current time node t and the congestion has propagated to the upstream road section; the propagation road section refers to the road section that is in a congested state at the current time node t but the congestion has not propagated to the upstream road section; Step C, calculating the shock wave speed of different road sections and the propagation time of the shock wave on the road section according to the conversion of the road section state; Step D, calculating the conversion rate of each road section state at the current time node t, generating a cascade failure evolution model of the high-speed network, and predicting the macro evolution law of the high-speed network cascade failure; In step B, the shock wave speed of different road sections is calculated according to the conversion of the road section state by using the following formula: wherein u w is the shock wave speed, representing the shock wave speed generated when traffic flow with flow rate Q1 and density K1 encounters traffic flow with flow rate Q2 and density K2; In step D, the conversion rate of each road section state at the current time node t is calculated to generate a cascade failure evolution model of the high-speed network: Where F(t), H(t), C(t) and R(t) are the number of links in free flow, propagation, congestion and recovery state at time t, respectively; P FH (t) is the transition rate from free flow to propagation state at time t, P HC (t) is the transition rate from propagation to congestion state at time t, P RR (t) is the transition rate from propagation to recovery state at time t, P CR (t) is the transition rate from congestion to recovery state at time t; λ is the average number of upstream links of a link in the network; T FR (t) is the time required for a link in free flow to transit to propagation state at time t, T RC (t) is the time required for a link in propagation state to transit to congestion state at time t, T HR (t) is the time required for a link in propagation state to transit to recovery state at time t, T CR (t) is the time required for a link in congestion state to transit to recovery state at time t.

2. The method of claim 1, wherein, In step A, the detection data of the detector includes the position information of the detector, the detection time, and the measured vehicle flow, vehicle density and vehicle speed.

3. The method of claim 1, wherein, In step A, the process of identifying whether each road section of the high-speed network is congested at different time periods according to the detection data comprises the following steps: Divide the data by time node, take the time node of the occurrence of the event as the starting time node, and identify whether the road section is congested according to the speed data measured by the detector: where j is the link number, v j (t) is the average speed of link j at time node t, vj is the link speed limit of link j, ε j (t) is the average speed of link v j (t) is the ratio of v (t) to vj; ρ is a given threshold, θ j (t) is a 0-1 variable, θ j (t) = 1 means that link j is in congestion state at time node t, θ j (t) = 0 means that link j is in free flow state at time node t.

Images