Expressway passenger and freight mixed traffic flow modeling method based on METANET model optimization

Abstract

The invention discloses a highway passenger and freight mixed traffic flow modeling method based on METANET model optimization. The method comprises the following steps: constructing a dynamic critical density correction function based on a freight car proportion; establishing a differentiated lane changing behavior decision-making mechanism, respectively setting lane changing probabilities based on a Logit utility function and a suppression coefficient for cars and trucks, and enhancing the dynamic simulation capability of lane transverse flow distribution; and constructing a ramp interaction model coupled with the truck proportion, and analyzing a traffic capacity attenuation mechanism of the confluence area. According to the method, the macroscopic modeling advantage of the METANET model is kept, and the method has good structural compatibility and generalizability. According to the method, the prediction precision of the flow, the speed and the density in a multi-vehicle mixed traffic scene is superior to that of a traditional model, and the method is suitable for modeling, simulation and control of an intelligent high-speed system.

Description

TECHNICAL FIELD

[0001] The application relates to a highway mixed passenger and freight traffic flow modeling method, in particular to a highway mixed passenger and freight traffic flow modeling method based on METANET model optimization. BACKGROUND

[0002] Traditional highway traffic flow modeling is mostly based on the homogenization assumption, and typical representatives such as the METANET model are widely used in traffic state estimation and control strategy optimization. However, with the continuous rise of the freight proportion, the highway has shown obvious mixed passenger and freight characteristics. The truck has a significant impact on system stability and traffic capacity due to its large safety headway, low acceleration and limited lane changing ability. The existing model has the following problems: the critical density parameter is statically set, and the nonlinear compression effect of the phase transition threshold caused by the change of the truck proportion cannot be dynamically reflected; the lane changing rule does not distinguish between vehicle types, which easily causes the problem of truck retention in the fast lane; and the ramp interaction mechanism does not consider the heterogeneity of vehicle types, resulting in large deviation in the estimation of the capacity of the merging area.

[0003] To solve the above problems, the application constructs a new traffic flow modeling method which integrates dynamic critical density correction, differentiated lane changing decision and truck proportion coupled ramp mechanism. SUMMARY

[0004] The application aims to provide a highway mixed passenger and freight traffic flow modeling method based on METANET model optimization, which improves the lane use efficiency, improves the prediction accuracy of the formation time of highway congestion, maintains the macro modeling advantage of the METANET model, and has good structural compatibility and generalizability.

[0005] Technical scheme: The application provides a highway mixed passenger and freight traffic flow modeling method based on METANET model optimization, which comprises the following steps:

[0006] (1) constructing a dynamic critical density correction function based on the truck proportion;

[0007] (2) establishing a differentiated lane changing behavior decision mechanism;

[0008] (3) constructing a ramp interaction model coupled with the truck proportion to analyze the traffic capacity attenuation mechanism of the merging area.

[0009] Preferably, the step (1) comprises defining a saturated critical density, constructing a saturated exponential critical density function, and the saturated exponential critical density function formula is as follows:

[0010] ρ crit (p T )=ρ crit0 ·[1-β·(1-e-k p T )]

[0011] Where, ρ crit0 The critical density without trucks, p T β is the truck ratio, β is the compression amplitude coefficient, and k is the decay rate parameter.

[0012] Preferably, the differentiated lane-changing behavior decision mechanism specifically involves setting lane-changing probabilities based on the Logit utility function and the inhibition coefficient for cars and trucks respectively, thereby enhancing the dynamic simulation capability of lane lateral traffic distribution.

[0013] Preferably, in the differentiated lane-changing decision-making mechanism, a decay coefficient related to the truck ratio is superimposed on the Logit model, and the lane-changing probability formula is:

[0014] p truck =δ·p car ·(1-γ·p T )

[0015] Where γ is the lane change suppression coefficient and δ is the lane change directionality correction factor.

[0016] Preferably, the ramp interaction model with coupled truck ratios uses a nonlinear efficiency function to adjust the actual merging flow of the ramps.

[0017]

[0018] Where, p T 'This represents the proportion of trucks on the ramps.' η is the merging efficiency attenuation coefficient, η is the nonlinear merging efficiency coefficient, and Q is the merging efficiency coefficient. merge This represents the actual merging flow rate of the ramp.

[0019] Preferably, when the main line flow is low, the merging capacity is dominated by the effective flow of the ramps; when the main line is close to saturation, the merging capacity is limited by the main line capacity, and the ramp flow is suppressed.

[0020] Preferably, in step (2), in the traffic flow model, based on the density conservation equation, a lane-changing mechanism is introduced. The transfer of vehicles between different lanes will change the density of each lane. For the j-th lane, its density conservation equation is modified as follows:

[0021]

[0022] Where, ρ i,j (k) is the vehicle density of lane j at point i on road segment k at time k, Q j’→j Q is the traffic flow from lane j' to lane j. j→j’is the vehicle flow from the jth lane to the j'th lane, and ΔU is the utility difference between the target lane and the current lane.

[0023] Preferably, the merging area traffic capacity C merge is determined by the mainline traffic capacity C main and the ramp effective merging flow η(p T )·Q ramp together:

[0024] C merge = min(C main ,Q main +Q merge )

[0025] wherein Q main is the mainline current flow, and Q merge is the ramp actual merging flow.

[0026] A computer device comprising one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the programs are executed by the processors to implement the steps of the method for modeling mixed passenger and freight traffic flow on expressway based on METANET model optimization.

[0027] A computer readable storage medium having stored thereon a computer program, the computer program being executed by a processor to implement the steps of the method for modeling mixed passenger and freight traffic flow on expressway based on METANET model optimization.

[0028] Advantages: Compared with the prior art, the present application has the following significant advantages:

[0029] (1) A dynamic critical density function based on truck ratio driving is proposed, which describes the phase transition compression phenomenon caused by the poor acceleration performance of trucks and the large safety distance, significantly improving the prediction accuracy of congestion formation timing; (2) Lane differentiation lane changing rules are established to finely describe special behaviors such as truck retention and return, and to improve lane usage efficiency; (3) The ramp merging efficiency and vehicle type ratio are coupled to solve the problem of overestimation of bottleneck traffic capacity by traditional models under high truck ratio; (4) Through simulation experiments on the NGSIM US-101 data set, the improved model described in the present application is superior to the traditional METANET model by more than 15% in terms of RMSE of flow, speed, and density prediction, and has good adaptability and reliability in actual deployment. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 is the METANET road section division schematic diagram of the present application.

[0031] Figure 2 Model design framework for the present invention.

[0032] Figure 3 Three mechanism logical relationship diagram for the present invention.

[0033] Figure 4 Critical density-truck proportion curve.

[0034] Figure 5 Flow-density-truck proportion three-dimensional graph.

[0035] Figure 6 Ramp merging efficiency attenuation effect graph.

[0036] Figure 7 Traffic flow model verification graph based on NGSIM data set. DETAILED DESCRIPTION

[0037] The technical solutions of the present invention are further described below in combination with the drawings.

[0038] The present invention discloses a highway passenger and freight mixed traffic flow modeling method based on METANET model optimization, comprising:

[0039] (1) Construct a dynamic critical density correction function based on truck proportion;

[0040] The dynamic critical density correction function is established, the time-varying parameter β(pT) dependent on the truck proportion is used to quantify the dynamic compression effect of vehicle heterogeneity on the traffic phase transition point, and the modeling inaccuracy problem of left shift of critical density is solved.

[0041] Set the car proportion as 1-p T , and the truck proportion as p T , then the average headway h e =(1-p T )h c +p T h T , wherein h c is the car headway, and h T is the truck headway. According to the Greenshields model, the critical density is inversely proportional to the average headway: Substitute the average headway into the critical density formula:

[0042]

[0043] Since h T >h c , the increase of the truck proportion p T will cause the decrease of the critical density, indicating that the critical density is linearly negatively correlated with the truck proportion, and the linear relationship is obtained:

[0044]

[0045] where ρ crit0 is the critical density without trucks.

[0046] However, the relevant measured NGSIM data shows that the left-moving rate of critical density is exponentially attenuated with the proportion of trucks, i.e., it decreases rapidly at low proportion stage and slows down at high proportion stage. The traditional linear model cannot capture this marginal effect of decreasing characteristics.

[0047] Therefore, an exponential function is introduced for nonlinear correction. Based on Kerner's three-phase traffic flow theory, the denominator of the linear model is replaced by an exponential form The formula is p T When it approaches 1, the critical density approaches 0, which is inconsistent with the fact that the critical density should be 0 when the proportion of trucks is 100%. Therefore, the formula form is further adjusted to ensure physical rationality.

[0048] First, define the saturated critical density: when the proportion of trucks is 100%, the critical density is (0<γ<1), where is the saturated compression coefficient of trucks on critical density;

[0049] Construct a saturated exponential function: introduce a saturated exponential function, the formula is as follows:

[0050]

[0051] When p T = 0, ρ crit = ρ crit0 , when p T = 1, ρ crit = ρ crit0 ·(1-β), where 1-β = γ, β is the maximum compression amplitude of the proportion of trucks on critical density.

[0052] (2) Establish a differentiated lane-changing behavior decision mechanism;

[0053] This paper proposes a differentiated lane-changing decision that integrates the Logit decision model and the truck forced return mechanism. While maintaining the continuity of the lane-changing decision of ordinary vehicles, it restricts the truck fast lane retention behavior through the probability attenuation coefficient γ, making up for the homogenization defects of lane-changing rules.

[0054] Define the utility difference between the target lane and the current lane as:

[0055] ΔU = U target - U current = -α lane ·(ρ target - ρ current ​)

[0056] wherein, U target and U current is the lane utility value, a lane is the lane attractiveness coefficient, in China's traffic rules, the left lane is defined as the "overtaking lane" or "fast lane", which has a higher priority. Drivers usually consider the left lane to be more efficient, so it is given a higher attractiveness coefficient. According to traffic flow distribution theory, lane attractiveness is positively correlated with lane speed, density, safety, and other factors. The left lane has a higher average speed and lower density, so it has a stronger attractiveness. The recommended left lane capacity correction coefficient is 1.2, and the right lane is 0.8. p is the lane density. The negative correlation between the utility difference and the density difference indicates that drivers tend to choose lanes with lower density to improve traffic efficiency.

[0057] According to the Logit model, the lane changing probability is:

[0058]

[0059] wherein, θ is the lane changing sensitivity coefficient, and θ is preferably 0.15 to balance model accuracy and computational efficiency, reflecting the driver's sensitivity to lane density difference. The formula shows that ΔU is the utility difference between the target lane and the current lane. When ΔU>0, the target lane has higher utility, and the lane changing probability increases exponentially with the utility difference. For example, when Δρ=5veh / km, P 2→1 =0.68, i.e. there is a 68% probability that vehicles in a high-density lane will change lanes to a low-density lane.

[0060] The truck lane changing decision is based on the Logit model and superimposes a decay coefficient related to the proportion of trucks. The formula is

[0061] P truck = P logit ·(1-γp T )

[0062] wherein, P logit is the lane changing probability calculated by the model, γ is the decay coefficient, and p T is the current lane truck proportion. By dynamically adjusting the lane changing probability according to the proportion of trucks, it reflects that the higher the proportion of trucks, the stronger the system's constraint on their lane changing behavior. At the same time, trucks should be encouraged to return to the slow lane from the fast lane, and the lane changing direction needs to be differentiated. The final lane changing probability calculation formula is as follows:

[0063] p truck =δ·p car ·(1-γ·p T )

[0064] where δ is the lane-changing directionality correction factor, used to enhance the behavior tendency of trucks returning to the slow lane from the fast lane.

[0065] The Logit utility function considers the influence of lane density difference on vehicle lane-changing decision from a macroscopic perspective. Vehicles usually tend to change from a lane with high density to a lane with low density to improve driving efficiency, and the Logit utility function quantifies this tendency. In addition, the truck return rule stipulates that the lane-changing probability of trucks decreases in the case of slow driving speed in the fast lane, prompting trucks to return to the slow lane. The combination of the two lane-changing decisions simulates the influence of truck lane-changing behavior on traffic flow, maintains the stability of traffic flow, and avoids congestion in the fast lane due to slow truck driving.

[0066] Based on the traffic flow density conservation equation, the lane-changing mechanism is introduced, and the transfer of vehicles between different lanes will change the density of each lane. For the jthlane, its density conservation equation is modified as

[0067]

[0068] where ρ i,j (j, i, k) is the vehicle density of the jthlane at road section i and time k, q i,j (j, i, k) is the flow of the jthlane at road section i and time k, Q j’→j (j, i, k) is the vehicle flow from the j'th lane to the jthlane, and Q j→j’ (j, i, k) is the vehicle flow from the jthlane to the j'th lane. The lane-changing probability determines the size of the lane-changing flow, such as Q j→j’ (j, i, k) = p j →j’·Q j When the utility difference ΔU between the target lane and the current lane is large, the lane-changing probability p increases, which in turn affects the density change of each lane. At the same time, the traditional speed-density relationship needs to be modified after introducing the lane-changing mechanism. The lane-changing behavior changes the density distribution of the lane, thereby affecting the vehicle speed. After considering the lane-changing mechanism, the speed-density relationship can be expressed as:

[0069]

[0070] where v i,j (j, i, k) is the vehicle speed of the jthlane in road section i, v f,i,j (j, i, k) is the free flow speed of the lane, ρ i,j (j, i, k) is the original density of the lane, Δρ i,j (j, i, k) is the change in the density of the lane due to lane-changing behavior, and ρ j,i,j (j, i, k) is the jam density of the lane. For example, in road section i, the return of trucks from the fast lane to the slow lane increases the density of the slow lane, and the speed of the slow lane vehicles will decrease.

[0071] (3) Construct a ramp interaction model with coupled truck ratio and analyze the capacity decay mechanism of merging area.

[0072] A ramp interaction model coupled with the truck ratio is constructed to analyze the capacity attenuation mechanism in the merging zone. A nonlinear merging efficiency coefficient η is introduced to establish a quantitative mapping relationship between the truck ratio and the intensity of local bottlenecks. Merging efficiency, as a core indicator for measuring the smoothness of vehicle merging on ramps, is defined as the ratio of the actual effective merging flow to the maximum flow on the ramp, reflecting the interference intensity of the truck ratio on the merging process. The merging efficiency function is shown below:

[0073]

[0074] p' is the merging efficiency attenuation coefficient. T For the proportion of trucks on the ramp, when p' T =0 indicates that when there are no trucks, η=1, and the efficiency is the highest. T =1 indicates that the truck is fully loaded. Lowest efficiency.

[0075] Merging area capacity C merge This refers to the maximum number of vehicles that can safely pass through the merging zone per unit time, determined by the mainline capacity C. main and the effective inflow rate of the ramp η(p) T )·Q ramp The joint decision reflects the dynamic balance between the "mainline carrying capacity" and the "ramp merging demand".

[0076] C merge =min(C main Q main +Q merge )

[0077] Q main The current traffic for the main line, Q merge This represents the actual merging flow rate of the ramp;

[0078]

[0079] Among them, Q ramp,max This represents the maximum flow rate entering the ramp. When the mainline flow rate is low, the merging capacity is dominated by the effective flow rate of the ramp. When the mainline is close to saturation, the merging capacity is limited by the mainline capacity, and the ramp flow rate is suppressed.

[0080] The density conservation equation of the traditional METANET model only describes the longitudinal flow conservation along the mainline and does not include the lateral effects of ramp merging. By introducing a vehicle-type merging term based on merging efficiency modulation, it is modified as follows:

[0081]

[0082] Where, ρ merge,j (t) represents the change in lane j density caused by ramp merging. The following steps are used to achieve a refined model of the merging process.

[0083] The efficiency impact of merging flow and vehicle type breakdown:

[0084] Firstly, due to the merging efficiency, the actual total flow rate merging into the main line from the ramp is calculated as follows, based on the ramp truck ratio p. T The merged traffic flow is broken down into passenger car and truck components:

[0085]

[0086] Among them, the proportion of trucks p T The higher the actual inflow Q, the greater the actual inflow. merge The lower.

[0087] Density increment calculation by vehicle type:

[0088] Based on the flow-density-velocity relationship, the merging density of cars and trucks at the merging zone entrance is:

[0089]

[0090] Among them, v merge,car and v merge,truck These are the design speeds for cars and trucks in the merging zone, respectively.

[0091] In the traffic flow model, the CFL condition is used to ensure the numerical stability of the density conservation equation after discretization. Considering the CFL stability condition, the density update in the merging zone is as follows:

[0092]

[0093] Where Δx is the road segment length, Δt is the time step, and ρ max This represents the maximum permissible density.

[0094] In the specific implementation process, a merging ramp is located on a two-lane mainline of a highway. The mainline is divided into fast lanes and slow lanes, with the fast lanes only allowing cars and the slow lanes prioritizing trucks. The proportion of trucks (p) in the current lanes is statistically analyzed. T Dynamically update the critical density ρ of each segment crit And adjust the velocity-density relationship.

[0095] For lane changing between lanes, different strategies are adopted for cars and trucks. For cars, the lane changing probability is calculated based on the density difference Logit model, while for trucks, an additional attenuation weight γ·p is applied. T Control its lane-changing behavior.

[0096] In the merging zone, according to the ramp flow rate Q ramp and truck ratio p T 'Calculate the nonlinear merging efficiency η(p) T Then adjust the merging flow rate Q. merge The mainline density is updated. This method accurately predicts traffic flow, speed, and density in mixed traffic scenarios involving multiple vehicle types, and is suitable for modeling, simulation, and control of intelligent highway systems.

[0097] To examine the dynamic response characteristics and generalization ability of the METANET-DHR model in heterogeneous traffic flow scenarios, this invention selects the US-101 highway dataset from the NGSIM dataset for model effectiveness verification. The simulation results of the improved METANET-DHR model, the original METANET model, and the spatiotemporal evolution curves of macroscopic flow, speed, and density from the NGSIM US-101 measured data are compared side-by-side. Relative error, root mean square error (RMSE), and peak time synchronicity are used to quantify the model's prediction accuracy and spatiotemporal consistency. Figure 7 As shown, the differences in response of each model throughout the “formation-persistence-dissipation” process are demonstrated, highlighting the superiority of the improved model over the traditional METANET model in error control and phase transition capture, thus verifying the robustness and reliability of the METANET-DHR model.

Claims

1. A method for modeling mixed passenger and freight traffic flow on highways based on METANET model optimization, characterized in that, include: (1) Construct a dynamic critical density correction function based on the truck ratio; (2) Establish a differentiated decision-making mechanism for lane-changing behavior; (3) Construct a ramp interaction model with coupled truck ratio and analyze the capacity decay mechanism of merging area.

2. The method for modeling mixed passenger and freight traffic flow on highways based on METANET model optimization according to claim 1, characterized in that, Step (1) includes defining the saturated critical density and constructing the saturated exponential critical density function, the formula of which is as follows: Where, ρ crit0 The critical density without trucks, p T β is the truck ratio, β is the compression amplitude coefficient, and k is the decay rate parameter.

3. The method for modeling mixed passenger and freight traffic flow on highways based on METANET model optimization according to claim 1, characterized in that, The differentiated lane-changing behavior decision-making mechanism specifically involves setting lane-changing probabilities for cars and trucks based on the Logit utility function and the inhibition coefficient, respectively, to enhance the dynamic simulation capability of lane lateral traffic distribution.

4. The method for modeling mixed passenger and freight traffic flow on highways based on METANET model optimization according to claim 3, characterized in that, In the differentiated lane-changing decision-making mechanism, a decay coefficient related to the truck ratio is superimposed on the Logit model, and the lane-changing probability formula is: p truck =δ·p car ·(1-γ·p T ) Where γ is the lane change suppression coefficient and δ is the lane change directionality correction factor.

5. The method for modeling mixed passenger and freight traffic flow on highways based on METANET model optimization according to claim 1, characterized in that, The ramp interaction model with coupled truck ratios uses a nonlinear efficiency function to adjust the actual merging flow rate of the ramps. Where, p T 'This represents the proportion of trucks on the ramps.' η is the merging efficiency attenuation coefficient, η is the nonlinear merging efficiency coefficient, and Q is the merging efficiency coefficient. merge This represents the actual merging flow rate of the ramp.

6. The method for modeling mixed passenger and freight traffic flow on highways based on METANET model optimization according to claim 1, characterized in that, When the main line flow is low, the merging capacity is dominated by the effective flow of the ramps. When the main line is close to saturation, the merging capacity is limited by the main line capacity, and the ramp flow is suppressed.

7. The method for modeling mixed passenger and freight traffic flow on highways based on METANET model optimization according to claim 1, characterized in that, In step (2), the traffic flow model introduces a lane-changing mechanism based on the density conservation equation. The movement of vehicles between different lanes will change the density of each lane. For the j-th lane, its density conservation equation is modified as follows: Where, ρ i,j (k) is the vehicle density of lane j at point i on road segment k at time k, Q j’→j Q is the traffic flow from lane j' to lane j. j→j ' is the traffic flow from lane j to lane j', and ΔU is the difference in utility between the target lane and the current lane.

8. The method for modeling mixed passenger and freight traffic flow on highways based on METANET model optimization according to claim 1, characterized in that, The merging zone has a traffic capacity C merge Mainline passability C main and the effective inflow rate of the ramp η(p) T )·Q ramp Joint decision: C merge =min(C main ,Q main +Q merge ) Among them, Q main The current traffic for the main line, Q merge This represents the actual merging flow rate of the ramp.

9. A computer device, characterized in that, It includes one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the programs, when executed by the processors, implement the steps of a method for modeling mixed passenger and freight traffic flow on highways based on METANET model optimization as described in any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for modeling mixed passenger and freight traffic flow on highways based on METANET model optimization as described in any one of claims 1-8.

Images