• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Bifurcation analysis of visual angle model with anticipated time and stabilizing driving behavior

    2022-08-01 06:01:16XueyiGuan管學(xué)義RongjunCheng程榮軍andHongxiaGe葛紅霞
    Chinese Physics B 2022年7期
    關(guān)鍵詞:榮軍紅霞

    Xueyi Guan(管學(xué)義), Rongjun Cheng(程榮軍), and Hongxia Ge(葛紅霞)

    Faculty of Maritime and Transportation,Ningbo University,Ningbo 315211,China

    Keywords: visual angle,bifurcation analysis,anticipated time,stabilizing driving behavior,TDGL and mKdV equations

    1. Introduction

    In recent years, the demand for transportation has maintained a continuous growth, which has brought huge challenges to the smooth traffic and traffic safety. Facing the traffic problems,numerous of scholars proposed a series of traffic flow models to explain the complex traffic phenomena. These models[1–9]mainly model the traffic flow from three modules:macroscopic,microscopic and mesoscopic. For example,Sunet al.[1]proposed a car-following model, one of the microscopic models of vehicular traffic, which is concentrating on individual vehicles,and studying the characteristics of the following vehicles in the entire fleet by studying the influence of the changes in the movement state of the leading vehicle on the following vehicles, meanwhile the theoretical analysis and simulation of the improved model not only can provide deep exploring into the properties of the model but also help in better understanding the complex phenomena observed in real traffic fields. In addition,we can also subdivide these models into the car-following models (e.g., Sunet al.,[1]Geet al.[2]and Jianget al.[3]), the cellular automation models (Tianet al.[4]), the gas kinetic models (Nagataniet al.[5]), the hydrodynamic lattice models (Redhuet al.[6]and Changet al.[7])and the continuum models(Chenget al.[8]and Zhaiet al.[9])according to different modeling theories. Bifurcation, as an important research content of non-linear science,mainly studies the instability of the system solution caused by parameter changes and the changing behavior of the number of solutions.Studying the bifurcation in traffic flow will have important theoretical implications in analyzing traffic phenomena, traffic modeling,traffic control,etc.

    Facing numerous of traffic congestions, a number of mathematical models have been proposed to demonstrate the car-following behavior under various conditions with the development of urban transportation. Generally speaking,these models are mostly based on the stimulus-response framework,which was put forward at the General Motors research laboratories (Chandleret al.[10]and Gaziset al.[11]) at the first time. According to the framework,each driver responds when giving a stimulus. And the relationship between response and stimulus is as follows:

    In the framework, the stimulus could be mainly divided into four aspects: driver’s behavior factors, vehicle factors, road factors and other factors. However, when it comes to carfollowing model equation,the stimulus could be related to the location of a large number of cars or their speed, or maybe some other coefficient. The reaction which demonstrates the propulsion of follower has been transformed into vehicle’s speeds or accelerations. In different forms of car-following models, the reason for the different forms of vehicle motion equations lies in the different hypotheses of stimulus properties. The stimulus is probably composed of the speed, the speed difference between two consecutive vehicles,the headway, etc. Following are some classic car-following models that consider different stimulus.

    Gazis–Herman–Rothery (GHR)[11]car-following model is perhaps the most recognized model in the research of traffic flow, which is the foundation of the car-following model proposed by Hermanet al.[12]and Yanget al.,[13]displayed as

    whereλis the sensitivity of the relative velocity, Δvn(t-τ)represents the relative velocity of thenth car at timet-τ,τis reaction time of driver,in addition,andλhas some functional forms as follows:

    where Δxn(t)represents distance between two consecutive vehicles, Δxcritical(t) is the threshold value obtained from actual data analysis,vn(t)represents the velocity ofnth vehicle,whileA,A1andA2are constant. Moreover, Lee[16]believed that drivers make decisions relying on a period of stimulation rather than a transient stimulus,and for the reason he proposed an improved GM model with considering driver’s memory,which is the foundation of the extended car-following model considering driver’s memory and jerk proposed by Jinet al.[17]

    For whether Chandleret al.[10]or Gaziset al.[11]mentioned earlier, they all assumed that both cars are driving at the same speed and the headway is acceptable, but this is not consistent with the actual situation. To overcome this shortcoming,Newell[18]proposed a nonlinear car-following model by using the headway as a stimulus, and Helly[19]assumed that every driver will adjust the following distance to a desired distance according to his driving characteristics, and drivers strive to minimize the relative speed along with the distance between the actual headway and the required headway,which is defined as

    whereDn(t) is driver’s desired following distance. In view of the earlier work, Helly[19]calculated the final calibration results whereτ=0.5 s,A1=0.5 s-1,A2=0.125 s-2, andDn(t)=20+vn(t-0.5). And Helly’s improvement also laid the foundation for the innovation of Koshiet al.[20]and Xinget al.[21]For example, Koshiet al.[20]combined Helly’s model with GHR’s model and did a nonlinear extension, but it still has some shortcomings in horizontal curvature effect.

    Then in 1995, Bando[22]proposed the optimal velocity(OV) model, which describes some nonlinear phenomenon and car-following behavior in traffic, and the specific expression shows as following:

    whereαdonates the coefficient of sensitivity andVrepresents the optimal velocity that drivers prefer based on the condition of the headway.

    Optimal velocity model can reflect some actual traffic flow characteristics such as the disappearance of stability of continuous traffic flow, the queuing of the traffic and the entire process from the generation to the elimination of traffic congestion very well, but when facing high acceleration values, it has certain defects. Helbing and Tilch[23]compiled a large amount of empirical data to overcome this shortcoming and proposed a generalized force (GF) model. The same as GHR model, GF model considered the effect of relative speed, too, but the model ignored the situation where the velocity of the follower is equal to or less than the velocity of the leader. Then, Jianget al.[24]proposed the full velocity difference(FVD)model,which considers the positive relative velocity. Therefore, the FVD model can describe the actual traffic flow better compared to the OV and GF models during the car-following process. Its formula reads

    in whichV(θn(t))represents the relative velocity andλis its sensitivity coefficient. In 2019, Yanet al.[25]improved the FVD by taking the optimal velocity difference and the electronic throttle angle difference of the preceding vehicle into account so as to explain the current complex traffic situation.

    In the model, driver’s response is mainly coming from the stimulus of both speed difference and headway. However,since Michaels[26]introduced psychology-related theories into the car-following model for the first time,a series of follow-up studies (e.g., Winsum[27]) showed that psychological characteristics are indeed related to the stability of traffic flow. Its main performance is that visual angle can be used to replace headway together with angular velocity, which can be used to replace speed difference in some engineering car-following models. Therefore,Andersen and Sauer[28]proposed the driving by visual angle (DVA) model by taking visual angle into account based on Helly’s[19]model.

    Combined the FVD with DVA model,Jinet al.[29]developed the visual angle model(VAM)which takes both positive velocity difference and visual angle into account defined as

    whereV(θn(t)) represents the optimal velocity according to the visual angle of the driver, andθn(t) represents visual angle observed by the driver of carnat timet,as shown in Fig.1.

    As we all know, car-following behavior is a humanrelated process, and inserting visual angle into the FVD can access to 3D parameters information directly, which is closer to reality. However, in order to reduce gasoline consumption and stop-and-go while driving,many drivers always try to keep the vehicle as smooth as possible, which means that drivers will control the headway far enough to accelerate, but in order to avoid the repeated deceleration, the driver may drive smoothly at the original speed rather than accelerating. Therefore, in addition to the influence of visual angle and relative speed,driver’s stabilizing driving behavior can also affect the driving behavior, and for this reason, based on the VAM, in this paper we propose an improved car-following model by taking the effect of visual angle,anticipated time and driver’s stabilizing driving behavior into account.

    Bifurcation is defined as the process in which the nonlinear dynamic parameters of the system continuously change and reach a certain critical point,which causes the global nature of the system to change. In 1978, Huilgol[30]firstly introduced the bifurcation theory to the nonlinear dynamics of vehicle systems,which aroused the enthusiasm of traffic scholars around the world to study the stability of vehicle systems.With the continuous development of bifurcation analysis theory,Igarashiet al.[31]and Oroszet al.[32]selected different optimal velocity functions to study the bifurcation phenomenon in the OV model,and obtained the similar conclusions,which was later extended to the OVM with a single fixed delay by Yamamotoet al.[33]performed Hopf bifurcation analysis for the dissipative system with asymmetric interaction including the OVM. It is generally found that a bifurcation structure is related to the multiple traffic jams via the transitions of emerging clusters(i.e.,traveling waves). In order to further explore the relationship between traffic state and bifurcation,in 2004,Gasseret al.[34]conducted a series of car-following models with bifurcation analyses. They found that the instability of the neutral stable solution increased in optimal velocity models is normally owing to the happen of the Hopf bifurcation.Jinet al.[35]conducted a bifurcation analysis for the FVD model to study the effect of bifurcation behavior on traffic congestion and stability of the system. More recently, Zhanget al.[36]studied the traditional velocity difference model considering time delays. They found that bifurcation happened corresponding to the emergence of traffic congestion in continuous traffic flow. Nowadays,a great progress on traffic flow has been made. However, it is still essential to explore and grasp the nonlinear characteristic of traffic information so as to eliminate the traffic jams.

    Through the review of the development of traffic flow theory,we can find that few studies have comprehensively considered the driver’s visual angle,anticipated time and driver’s stable driving behavior to explore the impact of these characteristics on car-following behavior. While the article focuses on three human factors, comprehensively considers the driver’s visual angle, anticipated time and stable driving behavior to explore the impact of driver behavior factors on car-following behavior. Then the influence of various factors on the stability of traffic flow is obtained through the linear, nonlinear analysis and numerical simulation. Finally, the rationality of the improved model is verified by parameter calibration.

    The specific contribution of this paper are as follows:

    (1)Providing theoretical support for the establishment of intelligent transportation systems. In the process of developing self-driving vehicles,robots can be used to simulate human driving,and the optimal human psycho-physiological characteristics are set during the robot programming,so as to achieve the maximum safety and optimal efficiency of the traffic flow.

    (2)Providing references for traffic management systems.In terms of traffic safety, the optimal driver’s psychologicalphysical characteristics obtained from the study can be used as the reference standard to strictly review the driver’s driving qualification.

    The following is the framework of the article. First of all,we make an overview of the development of the traffic flow in Section 1.Then we take visual angle,anticipated time and stabilizing driving behavior into account and put forward a new visual angle model in Section 2. In Section 3,the neutral stability curve is carried out through linear stability analysis. In Section 4 we conduct bifurcation analysis for the improved model and study the influence of the three driver’s behavior on Hopf bifurcation. Through the numerical calculation, the time-dependent Ginzburg–Landau (TDGL) equation and the modified Korteweg–de Vries (mKdV) equation are given respectively in Sections 5 and 6. Then in Section 7, parameter calibration and numerical simulations are carried out and numerical results are discussed. Finally, we obtain the conclusions in Section 8.

    2. The extended car-following model

    All of the previous car-following models are developed based on the ideal state that the driver’s immediate response to traffic conditions is absolutely rational,meanwhile,drivers have a great ability to perceive the traffic information,such as headway, velocity and acceleration. However, there are few psychological studies can prove this. Although some basic factors affecting the driving behavior,such as time delays,inaccuracies,and anticipated time have been taken into account in the car-following model, the headway and the relative velocity always play the most direct role in driver’s perception.The improved model in this paper is proposed based on the theory,considering from a psychological perspective,that numerous of studies have showed that psychological effects can affect driving behavior, and the results of these studies can be applied to car-following models. The most essential research results are that visual information is strongly related to the driver’s driving.

    Considering both the physical and psychological characteristics of follower, we insert a visual angle of follower’s driver into the car-following model rather than the headway and the difference of speed,whose relationship is displayed in Fig.1,

    whereθn(t)is defined as the visual angle which is measured by the driver of thenth car at timet,Δxn(t)=xn+1(t)-xn(t)is the headway between the preceding (n+1)th car and the followingnth car,andln-1andwn-1represent the length and the width of the(n-1)th car,respectively.

    Fig.1. Car-following model considering the visual angle.

    In view of the above knowledge, we developed an improved car-following model considering the visual angle and anticipated time with stabilizing driving behavior.The dynamical equation is proposed as follows:

    whereαis known as the sensitivity coefficient of drivers to the difference between the current speed and desired speed.vn(t)denotes the speed of carnat timet,εandγare defined as the sensitivity parameter of time-delayed velocity difference and steady-state control effects respectively, andτis the anticipated time. Furthermore,Vrepresents the optimal velocity whose mathematical expression is defined as follows:

    whereve=maxV(x)>0 is called the desired speed.

    Equation (7) proved that the visual angle stimulus can really affect the response of driver to follower. Since carfollowing behavior is closely related to human decisionmaking and response,it needs to take human’s limits on judging headway and velocity stimulus into account. Therefore,considering driver’s visual angle factor can fit to the real traffic flow better.

    3. Linear stability analysis

    In this section, we conduct the prerequisites for the stability of the improved model by linear analysis. For a uniform traffic flow,all vehicles have the same characteristics,including everyone in the unified system is moving at the optimal velocityV(θ0)and the same headwayh,and the visual angleθn(t)remains the same. Obviously,we can obtain

    Integrating the two formulas, we can obtain the neutral stability conditions

    With small disturbances,the neutral stability condition of the uniform traffic flow is as follows:

    Whenαis satisfied with the condition, the traffic is stable with small disturbances.Otherwise,the traffic will become unstable.

    Fig.2. Neutral stability curve with different γ,w,ε and τ.

    Figure 2 depicts the neutral stability curve which describes the functional relationship between the headway and sensitivity coefficient space. As we can see,every neutral stability curve has a critical point (hc,αc) with differentγ,w,εandτ, respectively. The stability graph is separated into two parts by the critical curve, among which, one represents stability region,and the other represents instability region. Each point on the line represents a critical point for the stability to instability region. Whenα >αc, regardless of the headway,traffic flow will always remain stable, otherwise congestion may occur. In other words, the traffic flow will keep going above the neutral stability curve and congestion is impossible to happen. While under the curve,the traffic flow density increased along with the increased possibility of congestion.As shown in Figs. 2(a), 2(c) and 2(d), when fixing other parameters at a proper value,the vertices of the curve gradually decrease with the increases ofγ,εandτrespectively. However,they play different roles for traffic flow stability. Instead,F(xiàn)ig.2(b)describes the instability regions increase with the increasing ofwgradually when fixingε=0.2 s-1,γ=0.2 m-1andτ=0.5 s. From this, we conclude thatγ,εandτhave a positive effect in the traffic stability,which is opposite tow.

    4. Bifurcation analysis and simulation

    In this section,we use the theoretical analysis to explore the bifurcation behavior of the new model first of all so as to study the correspondence between the critical point of Hopf bifurcation and instability, and then conduct the numerical simulation to display the impact of various parameters on the bifurcation behavior intuitively. Equation(7)can be rewritten as

    Through analyzing Eq.(17),we obtain the linearized equation as follows:

    Use the Hopf theorem to find the Hopf bifurcation pointL*.It can be found that when a smooth family of eigenvaluesλ=ψ±iωappears, the Hopf bifurcation condition is satisfied,and the bifurcation occurs.

    Substitutingλ=ψ±iωinto Eq.(19),one obtains

    wherek ∈{1,2,...,N}.

    Separating the real and imaginary parts of Eq. (22), the result shows

    When Hopf bifurcation happens, there are only a pair of imaginary rootsλ1,2=±iω. In Fig. 3, all of other (N-2)eigenvalues are located in the left complex plane when a pair of pure imaginary roots appears at the first time(fork=1 andk=N-1).

    Fig.3. Eigenvalues of the system for N=5,α =1 s-1,ε =0.05 s-1,and τ =0.4 s.

    which is just contrary to the result shown in Eq.(15).

    Next, differentiation of the characteristic Eq.(27)brings about the essential condition for the Hopf bifurcation,i.e.,

    Obviously, from Eq. (27), the Hopf conditionψ(β1)<0 is satisfied absolutely. Therefore,the Hopf bifurcation occurs in our system. We can obtain the critical lengthL*from the following equation:

    The spatial variation of Hopf’s relation(28)is displayed in the three-dimensional diagram of Fig. 4. The yellow surface is graph of

    while the blue surface is the spatial variation curve of

    The intersection of these two surfaces corresponds to the critical point when Hopf bifurcation occurs.

    In order to display and compare the effect ofτ,εandγon Hopf bifurcation more intuitively, Fig. 5 shows the two-dimensional(L,N)-plane corresponding to Fig.4,respectively,which reveals the instability regions of the system,and in which the first pair of eigenvalues have positive real parts drawn in yellow. The Hopf curve consists of the set of all intersections(L,N)of the first pair of eigenvalues and the second pair of eigenvalues, that is, the boundary between the blue plane and the yellow plane in Fig. 4. It is uncovered that the sensitivity anticipated timeτ,εandγplay a different role on the occurrence of Hopf bifurcation. Figures 5(a) and 5(b) indicate the critical point (LC,NC) of Hopf bifurcation from(4.1,3.8)to(5.1,8.9)because of the increasing of anticipated timeτfrom 1 to 2 and fixing the sensitivity coefficientε=0.05 s-1andγ=0.8 m-1.In Figs.5(c)and(d),the critical point(LC,NC)of Hopf bifurcation changes from(4.1,3.8)to(5.4,10.8)with the increasing ofεwhile keeping anticipated timeτandγunchanged. It displays the instability area decreases because of the increasing ofε. In Figs.5(e)and 5(f),the critical point (LC,NC) of Hopf bifurcation changes from(5.8,11.6) to (3.9,1.8) with the decrease of the sensitivityγand fixing the anticipated timeτ=0.4 s andε=0.05 s-1.It shows the instability area increases with the decrease ofγfrom 0.8 m-1to 0.65 m-1. Therefore, parameterτ,εandγall can inhibit the occurrence of bifurcation,it means they all contribute to the system’s stability despite the different effect in the critical point region of Hopf bifurcation.

    Fig.5. Side view for α=1 s-1,ve=1 m/s with different values γ,ε and τ: (a)τ=1 s,ε=0.05 s-1,γ=0.8 m-1;(b)τ=2 s,ε=0.05 s-1,γ =0.8 m-1; (c) τ =1 s, ε =0.05 s-1, γ =0.8 m-1; (d) τ =1 s, ε =0.1 s-1, γ =0.8 m-1; (e) τ =0.4 s, ε =0.05 s-1, γ =0.8 m-1; (f)τ =0.4 s,ε =0.05 s-1,γ =0.65 m-1.

    5. The TDGL eqnarray

    In this section, we carry out the nonlinear analysis of Eq. (7) so as to obtain the TDGL equation that fits the improved model traffic behavior best near the critical point(hc,αc). The slow scales for space variablen, time variabletand constantbare introduced,and the slow variablesXandTare defined as follows:

    While Δxn(t)is defined as follows:

    wherehcstands for safety distance. By inserting Eqs.(29)and(30)to Eq.(31),along with the Taylor expanding to the fifthorder ofμ,the nonlinear partial differential equation is shown as follows:

    and substitute Eq. (35) into Eq. (34), the TDGL equation is transformed as

    where?Φ(S)/?Srepresents the function derivative. In addition to the simplest solutionS=0,another two stable solutions have also been shown,one of which is a uniform solution

    and the other is the kink solution

    wherex0is constant. Through analyzing Eq.(39),we can obtain a coexistence curve when following formula relations are met:

    Inserting Eq.(35)into Eq.(40),we can obtain the coexistence curve with the original parameters

    The spinodal line is based on the following conditions:

    By inserting Eq.(35)into Eq.(42),the relationship of the spinodal line is as follows:

    Based on the condition?φ/?S=0 and Eq.(43),we can calculate the critical point containing the original parameters

    6. The mKdV equation

    All ofgiare shown in Table 1.

    Table 1. The coefficients gi.

    In the table

    In order to export the regularized equation,we give the following definition:

    So,the standard mKdV equation withO(μ)correction term is obtained as

    Then,assuming thatR′(X,T′)=R′0(X,T′)+μR′1(X,T′)and considering theO(μ) correction term, in order to obtain the propagation velocitycfor the kink solution,it can be based on the solvability condition as

    where

    and the general velocitycobtained by calculation is as follows:

    whereV′′′<0, andτrepresents the reciprocal of the driver’s sensitivity coefficient.

    7. Simulation

    7.1. Parameter calibration

    In this part, 558 groups datasets of the 59th vehicle in the NGSIM database including velocity,acceleration and other factors are used to calibrate the optimal parameters. Considering that the modified dataset includes vehicle data in a noncar-following state, and existing the lane-changing behavior,therefore we make the following restrictions on car-following behavior.

    (1)Car-following behavior should occur in the same lane,in other words, the data bar of Lane-ID should be the same value.

    (2) Except for the first car, the selected leading car and following car,namely,preceding and following are not 0.

    (3) In order to ensure that the vehicle stays in the carfollowing state, the headway should not be too large. Otherwise,the behavior change of the previous car will hardly affect the behavior of the target car.

    Filtering by the above conditions,we can obtain the carfollowing data about the 59th car. Following is a partial data display.

    Table 2. Car-following partial data of the 59th car.

    According to the model in this article,we can determine the parameters to be calibrated in this article and the input variables of the car-following model, then using the least square method to perform nonlinear optimization of the calibration parameters. The specific objective function expression we adopt is proposed by Ossenet al.,[37]whose core idea is as follows:

    whereNis the data group number,jdonates the serial number of data,andrepresent actual acceleration and simulated acceleration respectively. Meanwhile, as an evaluation index describing the fitness of the parameter calibration, the smaller the PI, the closer the measured value is to the simulated value, that is, the better the applicability of the fitting result.

    Moreover, both models select parameter combinations optimized by genetic algorithm and within 5% of the error range as their analysis objects.Then normalize each parameter and systematically cluster the optimized processing results of each parameter,the comparison of each parameter distribution of the clustering results of the two methods in Table 3.

    As we can see in Table 3 that considering the anticipated time and stabilizing driving behavior can decrease the influence ofαwith the same data, which means anticipated time and stabilizing driving behavior can eliminate or weaken the adverse effects of optimal velocity in unified traffic flow effectively. In addition, the smallerhand the increase ofvmaxindicate that the minimum safety headway requirement is smaller in the stable traffic flow,and the maximum velocity that each vehicle can reach in the traffic flow has increased,which means the extended VAM is easier to steady than the VAM.

    Table 3. Calibrated parameters.

    Figure 6 displays the fitness between the simulations and observations. It is not difficult to see that the red and pink curve maintain roughly the same trend with the blue line,where the pink and red lines denote the fitted value of extended VAM and VAM respectively, and the blue line represents the actual observation, indicating that the initial calibration coefficient set of the VAM and the extended VAM both can effectively predict the movement trend of the actual traffic flow during the acceleration,uniform velocity and deceleration phases.However,comparing the red and pink lines,obviously,the fitted value set of the extended VAM is closer to the observation in most data despite larger deviations in a few inflection points. And synthesizing the error value of all the data, it is found that the PI value is reduced from 0.6624 to 0.6237 after considering the gain and time-delay feedback control, which means that the controlled visual angle model can better reflect the complex actual flow.

    7.2. Numerical simulation

    In this section, numerical simulation of the improved model has been conducted so as to prove the obtained conclusions by using discrete equations. The discrete process is as follows:

    In Fig. 7, the improved car-following model carries out a series of simulations while controlling other parameters unchanged. The patterns in Figs. 7(a)–7(d) correspond to the variablew(w= 1.6,1.8,2.0,2.2) respectively. Especially,whenw=1.6 m,the traffic flow has the highest stability that means the car width of the front car has a certain influence on the follower’s visual angle. As we can see that giving a small perturbation can induce initial stable traffic flow into an oscillating density wave. It can be clearly seen from the curvature change in Fig. 7 that the amplitude of the kink–antikink solution gradually increases with the increasing of parameterw.In other words,taking the vehicle width of the leader into account will interfere with the follower’s perspective,which will inevitably affect the stability of the system and increase the possibility of congestion at the same time. The wider the vehicle’s width,the more it can interfere with the driver’s perspective, and the greater the possibility of the system’s transition to instability region and traffic jams.

    Figure 8 displays that the headway profiles and density waves for the preceding vehicle widthwatt=6000 s corresponding to panels in Fig. 7 respectively. Obviously, as the width of the vehicle increases, the amplitude of the headway gradually increases,which means considering the width of the leader will restrain the stability of the traffic flow and prolong the time for the unstable traffic flow to stabilize.

    Fig.7. Space–time evolution of the headway from with the t=5000 s to 6000 s.

    Fig.8. Headway profiles of the density wave at t=6000 s correspond to Fig.6 respectively.

    Fig.9. Space–time evolution of the headway from with the t=4400 s to 4500 s.

    In Fig.9,we explore the effect of the anticipated time on the stability of traffic flow in the improved car following model through a series of simulations, where Figs. 9(a)–9(d) correspond toε(ε=0.1,0.3,0.5,0.7),respectively. It can be obviously seen from the change curvature in Fig.8 that considering the anticipated time contributes to stable movement of the system. Specially,whenε=0.7 s-1,the density wave change curvature is zero, at this time, the traffic flow steadily travels with a constant headway,and the entire traffic flow system has the highest stability. Therefore,we conclude that considering the driving characteristics of subject driver can effectively improve the stability of the traffic flow. The proposed new model considers the anticipated time and the subject driver uses the received original traffic information combines with his own prediction of the front traffic state to shorten the time responds to the front traffic situation, which provides a comprehensive improvement to the entire car-following behavior.

    Figure 10 displays the headway profiles and density waves for anticipated time att=4500 s corresponding to panels in Fig. 8 respectively. From Figs. 10(a)–10(d), considering the speed difference,the traffic congestion conditions can be effectively alleviated,and as the anticipated time increases,the amplitude of the density wave gradually decreases,which means the traffic flow tends to move with a stable headway.The improved model has improved to a greater extent in eliminating interference. From Fig.10,we obtain that parameterεcan promote the stability of traffic flow.

    Fig.10. Headway profiles of the density wave at t=4500 s correspond to Fig.8 respectively.

    Figure 11 mainly depicts the hysteresis loop relationships between the velocity and density, which are shown with deferentγ(γ=0.05, 0.1, 0.15, 0.2)respectively. As we can see the size of the loop gradually decrease with the value increasesγin Fig. 11. This also means that considering driver’s stabilizing driving behavior can effectively ease traffic congestion and promote the stability of the system at the same time.

    Fig. 11. The hysteresis loops relations of the velocities v1(t) and the density ρ1(t)=1/Δx1(t)respectively.

    8. Conclusion

    In this paper, we take the anticipated time and stabilizing driving behavior into the visual angle model, combining anticipated time with stabilizing driving behavior to explore driver’s sensitivity of change of angle information. The relationship of neutral stability is obtained by linear stability analysis. Meanwhile, bifurcation analysis for the improved carfollowing model considering the visual angle,anticipated time and stabilizing driving behavior is investigated. Hopf bifurcation analysis mainly consists of two modules, one is theoretical analysis and the other is numerical simulation. Through the Hopf bifurcation analysis we can obtain the critical points.Results show that bifurcation analysis results are highly coincide with the stability condition of linear stability analysis.Furthermore,from the three-dimensional bifurcation diagram,it is found that the critical points of Hopf bifurcation are located on the boundary between the stability region and the instability region. Hopf bifurcation leads to conversion of stable traffic flow to irregular traffic flow. In addition, the TDGL equation is obtained by nonlinear analysis,and then the coexistence curve,metastable curve and critical point are described according to the obtained thermodynamic potential.We obtain the mKdV equation of the model in the same way. Finally,the parameter calibration and numerical simulation results display that the anticipated time and stabilizing driving behavior both play a positive role on traffic flow, which is opposite to the visual angle. Based on the conclusions, the article will make contributions on providing theoretical support for the establishment of intelligent transportation systems and references for traffic management systems.

    Acknowledgements

    Project supported by the Natural Science Foundation of Zhejiang Province, China (Grant Nos. LY22G010001,LY20G010004), the Program of Humanities and Social Science of Education Ministry of China (Grant No. 20YJA630008), the National Key Research and Development Program of China-Traffic Modeling, Surveillance and Control with Connected & Automated Vehicles (Grant No. 2017YFE9134700), and the K.C. Wong Magna Fund in Ningbo University,China.

    猜你喜歡
    榮軍紅霞
    請(qǐng)你幫個(gè)忙
    Traffic flow prediction based on BILSTM model and data denoising scheme
    An extended smart driver model considering electronic throttle angle changes with memory
    Stabilization strategy of a car-following model with multiple time delays of the drivers?
    《烏鴉喝水》中的“想”
    A Study of Combination of English Language Teaching and Context
    大東方(2018年1期)2018-05-30 01:27:23
    讓動(dòng)作“活”起來(lái)
    “大愛(ài)有聲 義工行動(dòng) 關(guān)愛(ài)榮譽(yù)軍人”慰問(wèn)活動(dòng) 走進(jìn)廣東省第一榮軍醫(yī)院
    大社會(huì)(2016年6期)2016-05-04 03:42:20
    “光的直線傳播”“光的反射”練習(xí)
    接力生命花季無(wú)聲
    ——記王臺(tái)中學(xué)初一158班郭榮軍同學(xué)舍己救人的英雄事跡
    亚洲成人精品中文字幕电影| 国产精品香港三级国产av潘金莲| 国产熟女午夜一区二区三区| 国产精品免费视频内射| 午夜亚洲福利在线播放| 夜夜看夜夜爽夜夜摸| 久久精品人妻少妇| 国产乱人伦免费视频| 色老头精品视频在线观看| 国产人伦9x9x在线观看| 久久午夜亚洲精品久久| 亚洲精品国产精品久久久不卡| АⅤ资源中文在线天堂| 天堂av国产一区二区熟女人妻 | 亚洲欧美日韩无卡精品| 三级毛片av免费| 18禁观看日本| 精品一区二区三区四区五区乱码| 日本免费一区二区三区高清不卡| 女警被强在线播放| 波多野结衣高清无吗| 亚洲精品在线观看二区| 色噜噜av男人的天堂激情| 久久久久久久久久黄片| 亚洲无线在线观看| 久久精品成人免费网站| bbb黄色大片| 精品日产1卡2卡| 中文字幕久久专区| 脱女人内裤的视频| 欧美精品啪啪一区二区三区| 熟女电影av网| 搡老岳熟女国产| 久久精品影院6| 国产亚洲欧美在线一区二区| 日日干狠狠操夜夜爽| 一个人观看的视频www高清免费观看 | 精品熟女少妇八av免费久了| 久久久久久亚洲精品国产蜜桃av| 午夜成年电影在线免费观看| 色综合婷婷激情| 又黄又爽又免费观看的视频| 久久欧美精品欧美久久欧美| 午夜免费成人在线视频| 99精品欧美一区二区三区四区| а√天堂www在线а√下载| 久久久久久国产a免费观看| 日本 欧美在线| 97超级碰碰碰精品色视频在线观看| 国产片内射在线| 亚洲一区二区三区不卡视频| 免费在线观看影片大全网站| 一区福利在线观看| 国产精品九九99| 免费人成视频x8x8入口观看| 久久久久九九精品影院| 狂野欧美激情性xxxx| 中文字幕高清在线视频| 精品少妇一区二区三区视频日本电影| 老汉色∧v一级毛片| 亚洲成人精品中文字幕电影| 亚洲精品粉嫩美女一区| 母亲3免费完整高清在线观看| 久久九九热精品免费| 蜜桃久久精品国产亚洲av| netflix在线观看网站| 免费无遮挡裸体视频| 国产av在哪里看| 国产成年人精品一区二区| 国产成人精品久久二区二区91| 麻豆成人午夜福利视频| 国产单亲对白刺激| 啦啦啦韩国在线观看视频| 国产精品一区二区精品视频观看| 人人妻,人人澡人人爽秒播| 婷婷六月久久综合丁香| 久久中文字幕人妻熟女| 精品免费久久久久久久清纯| 1024手机看黄色片| 亚洲一区高清亚洲精品| 久久久久精品国产欧美久久久| 真人做人爱边吃奶动态| 天天一区二区日本电影三级| 亚洲电影在线观看av| 一级a爱片免费观看的视频| 精品乱码久久久久久99久播| 九九热线精品视视频播放| 制服人妻中文乱码| 亚洲av中文字字幕乱码综合| 国产精品国产高清国产av| 免费在线观看影片大全网站| 中文字幕人成人乱码亚洲影| 波多野结衣高清作品| 国产伦人伦偷精品视频| 亚洲五月婷婷丁香| 热99re8久久精品国产| 久久草成人影院| 窝窝影院91人妻| 欧美三级亚洲精品| 国产高清视频在线观看网站| 两性夫妻黄色片| 国内精品久久久久精免费| 国产一区二区三区视频了| www.999成人在线观看| 久9热在线精品视频| 日日爽夜夜爽网站| 村上凉子中文字幕在线| 最近最新免费中文字幕在线| 亚洲欧美日韩高清在线视频| av中文乱码字幕在线| 听说在线观看完整版免费高清| 亚洲成a人片在线一区二区| 特级一级黄色大片| 亚洲成人国产一区在线观看| 成人av在线播放网站| 国产成人精品久久二区二区免费| 日韩 欧美 亚洲 中文字幕| 国产精品 国内视频| www.熟女人妻精品国产| 18禁观看日本| 叶爱在线成人免费视频播放| 日本在线视频免费播放| 一进一出抽搐gif免费好疼| АⅤ资源中文在线天堂| 精品久久蜜臀av无| 夜夜看夜夜爽夜夜摸| www.999成人在线观看| 国产精品美女特级片免费视频播放器 | 日本三级黄在线观看| 又紧又爽又黄一区二区| 色综合站精品国产| 国产精品久久电影中文字幕| 两个人的视频大全免费| 最近视频中文字幕2019在线8| 欧美中文日本在线观看视频| 亚洲中文字幕日韩| 天天添夜夜摸| 人妻久久中文字幕网| 婷婷精品国产亚洲av| 丰满人妻一区二区三区视频av | 99热只有精品国产| 黄色片一级片一级黄色片| 国产亚洲精品av在线| 少妇人妻一区二区三区视频| av超薄肉色丝袜交足视频| 国产又黄又爽又无遮挡在线| 99国产极品粉嫩在线观看| 男女之事视频高清在线观看| 男女那种视频在线观看| 久久欧美精品欧美久久欧美| 色精品久久人妻99蜜桃| 久久 成人 亚洲| 日本五十路高清| 最近最新中文字幕大全免费视频| 精品第一国产精品| 久久久久久九九精品二区国产 | 精品不卡国产一区二区三区| www.999成人在线观看| 国产成人aa在线观看| 亚洲一区二区三区不卡视频| 宅男免费午夜| 老司机午夜十八禁免费视频| 国产精品免费视频内射| 91成年电影在线观看| 婷婷丁香在线五月| 999精品在线视频| 亚洲性夜色夜夜综合| 人妻久久中文字幕网| 一区二区三区激情视频| 叶爱在线成人免费视频播放| 精品久久久久久久久久免费视频| 国产成人系列免费观看| 国产欧美日韩一区二区精品| 欧美乱色亚洲激情| 岛国视频午夜一区免费看| 国产精品美女特级片免费视频播放器 | 午夜激情av网站| 精品乱码久久久久久99久播| 免费在线观看亚洲国产| 久久香蕉精品热| 亚洲av成人av| 亚洲欧美精品综合久久99| 国产亚洲av高清不卡| 久久香蕉国产精品| 欧美国产日韩亚洲一区| 日本五十路高清| 日本 欧美在线| 丁香六月欧美| 久久精品人妻少妇| 国产精品久久久久久人妻精品电影| 亚洲人成网站高清观看| 国产一区在线观看成人免费| 欧美成狂野欧美在线观看| 国产伦一二天堂av在线观看| 床上黄色一级片| 狂野欧美激情性xxxx| 久久精品91蜜桃| 极品教师在线免费播放| 午夜精品一区二区三区免费看| 亚洲美女黄片视频| 午夜精品久久久久久毛片777| 特级一级黄色大片| 国产亚洲欧美在线一区二区| 精品第一国产精品| 亚洲片人在线观看| 国产69精品久久久久777片 | 长腿黑丝高跟| 午夜成年电影在线免费观看| 久久人人精品亚洲av| 桃红色精品国产亚洲av| 亚洲国产精品合色在线| 天天添夜夜摸| 国产精品电影一区二区三区| 香蕉av资源在线| x7x7x7水蜜桃| 成人特级黄色片久久久久久久| 日韩欧美免费精品| 天堂av国产一区二区熟女人妻 | 国产精品亚洲av一区麻豆| 午夜两性在线视频| 成人午夜高清在线视频| 搡老熟女国产l中国老女人| 久久精品国产清高在天天线| 老司机福利观看| ponron亚洲| 日本 欧美在线| 国产成人精品久久二区二区免费| 亚洲熟妇中文字幕五十中出| 国产精品永久免费网站| 国产野战对白在线观看| av福利片在线| 在线国产一区二区在线| 国产免费av片在线观看野外av| 国内精品久久久久久久电影| 在线观看www视频免费| 国产亚洲欧美在线一区二区| 婷婷丁香在线五月| 无限看片的www在线观看| 日韩欧美一区二区三区在线观看| 1024手机看黄色片| 欧美精品亚洲一区二区| 天堂动漫精品| 女人被狂操c到高潮| 老汉色∧v一级毛片| videosex国产| 午夜成年电影在线免费观看| 国产在线观看jvid| 成年人黄色毛片网站| 可以在线观看毛片的网站| 在线观看一区二区三区| 1024手机看黄色片| 在线观看免费视频日本深夜| 中文字幕熟女人妻在线| 免费在线观看日本一区| 中文字幕人成人乱码亚洲影| 99久久国产精品久久久| 老熟妇乱子伦视频在线观看| 九色成人免费人妻av| www.精华液| 亚洲精品中文字幕在线视频| 波多野结衣高清无吗| 亚洲aⅴ乱码一区二区在线播放 | 国产精品,欧美在线| 久久午夜亚洲精品久久| 国产亚洲av高清不卡| 国产三级黄色录像| 九色国产91popny在线| 波多野结衣高清无吗| 十八禁网站免费在线| 精品久久久久久成人av| 国产精品久久久久久久电影 | 久久精品91蜜桃| 90打野战视频偷拍视频| 久久国产精品人妻蜜桃| 精品人妻1区二区| 真人一进一出gif抽搐免费| 亚洲一区二区三区不卡视频| 免费在线观看完整版高清| 成年人黄色毛片网站| www.精华液| 欧美日本亚洲视频在线播放| 禁无遮挡网站| 在线a可以看的网站| 床上黄色一级片| 级片在线观看| 亚洲乱码一区二区免费版| 久久久国产成人免费| 99国产精品一区二区三区| 欧美午夜高清在线| 久久这里只有精品19| 白带黄色成豆腐渣| 日韩三级视频一区二区三区| 久久久精品大字幕| av欧美777| 久99久视频精品免费| 欧美高清成人免费视频www| 亚洲精品美女久久久久99蜜臀| 夜夜夜夜夜久久久久| 亚洲精品一卡2卡三卡4卡5卡| 变态另类丝袜制服| 很黄的视频免费| 亚洲精品中文字幕在线视频| 免费观看人在逋| 国产av一区在线观看免费| 天天躁夜夜躁狠狠躁躁| 香蕉国产在线看| 精品高清国产在线一区| 国产精品av视频在线免费观看| 久久精品91无色码中文字幕| www日本在线高清视频| 免费电影在线观看免费观看| 少妇的丰满在线观看| 全区人妻精品视频| 啪啪无遮挡十八禁网站| 一区二区三区国产精品乱码| 亚洲国产看品久久| 两个人视频免费观看高清| 麻豆成人午夜福利视频| www.熟女人妻精品国产| 12—13女人毛片做爰片一| 亚洲精品久久成人aⅴ小说| 免费看十八禁软件| 欧美成人免费av一区二区三区| 91九色精品人成在线观看| 午夜福利高清视频| 99久久精品热视频| 亚洲五月天丁香| 精品久久久久久成人av| 国产免费男女视频| 男女午夜视频在线观看| 国产精品日韩av在线免费观看| 男人的好看免费观看在线视频 | ponron亚洲| 视频区欧美日本亚洲| 一进一出抽搐动态| 啦啦啦韩国在线观看视频| 999久久久国产精品视频| 男女做爰动态图高潮gif福利片| 午夜亚洲福利在线播放| 免费在线观看成人毛片| 无限看片的www在线观看| 天天躁狠狠躁夜夜躁狠狠躁| 欧美日韩乱码在线| 国产区一区二久久| 窝窝影院91人妻| 此物有八面人人有两片| 黄色视频,在线免费观看| 国产麻豆成人av免费视频| 91九色精品人成在线观看| 午夜影院日韩av| 中国美女看黄片| xxx96com| 国产高清视频在线播放一区| 国产精品爽爽va在线观看网站| 精品午夜福利视频在线观看一区| 曰老女人黄片| 色综合亚洲欧美另类图片| 欧美成人一区二区免费高清观看 | 校园春色视频在线观看| 五月伊人婷婷丁香| 1024手机看黄色片| 亚洲全国av大片| 桃红色精品国产亚洲av| 国产精品久久久久久久电影 | 国产欧美日韩一区二区三| aaaaa片日本免费| 好男人电影高清在线观看| 一本久久中文字幕| 熟妇人妻久久中文字幕3abv| av在线天堂中文字幕| 亚洲在线自拍视频| 国产成年人精品一区二区| 亚洲男人的天堂狠狠| 国产不卡一卡二| aaaaa片日本免费| av天堂在线播放| 国产熟女午夜一区二区三区| 一进一出好大好爽视频| 又爽又黄无遮挡网站| 精品电影一区二区在线| 国产成人一区二区三区免费视频网站| 欧美三级亚洲精品| 久久天躁狠狠躁夜夜2o2o| 婷婷精品国产亚洲av| 国产亚洲精品久久久久5区| netflix在线观看网站| 亚洲av成人av| 国产成人精品久久二区二区免费| 国产午夜精品久久久久久| 日日爽夜夜爽网站| 免费高清视频大片| 男男h啪啪无遮挡| 日韩欧美三级三区| 精品国产乱子伦一区二区三区| 在线观看一区二区三区| 免费在线观看成人毛片| 国产三级在线视频| 黄色片一级片一级黄色片| 国产视频一区二区在线看| 国产精品av久久久久免费| 国产熟女xx| 国产av不卡久久| 最新美女视频免费是黄的| 88av欧美| 国产精品永久免费网站| 美女午夜性视频免费| 久久久精品国产亚洲av高清涩受| 亚洲精品在线观看二区| 最近在线观看免费完整版| 久热爱精品视频在线9| 两个人视频免费观看高清| 成人精品一区二区免费| 久久精品影院6| 亚洲欧洲精品一区二区精品久久久| 午夜影院日韩av| 久9热在线精品视频| 亚洲熟女毛片儿| 成年人黄色毛片网站| av福利片在线观看| 色综合站精品国产| 香蕉久久夜色| 极品教师在线免费播放| 亚洲 欧美一区二区三区| 99精品欧美一区二区三区四区| www.精华液| 宅男免费午夜| 在线播放国产精品三级| 亚洲成人中文字幕在线播放| 国产久久久一区二区三区| 亚洲美女黄片视频| 国产麻豆成人av免费视频| 最好的美女福利视频网| 三级毛片av免费| 日本黄色视频三级网站网址| 99久久国产精品久久久| 老熟妇仑乱视频hdxx| 99riav亚洲国产免费| 白带黄色成豆腐渣| 99riav亚洲国产免费| www日本在线高清视频| 免费观看人在逋| 免费无遮挡裸体视频| 性色av乱码一区二区三区2| 国产亚洲精品综合一区在线观看 | 亚洲午夜理论影院| 老汉色av国产亚洲站长工具| 淫妇啪啪啪对白视频| videosex国产| 脱女人内裤的视频| 一二三四社区在线视频社区8| 精品高清国产在线一区| 国产欧美日韩精品亚洲av| 成熟少妇高潮喷水视频| 在线永久观看黄色视频| 欧美一级毛片孕妇| 美女免费视频网站| 亚洲国产中文字幕在线视频| 18禁国产床啪视频网站| 美女扒开内裤让男人捅视频| 亚洲,欧美精品.| 国产片内射在线| 久久亚洲精品不卡| 老司机午夜福利在线观看视频| 国产又色又爽无遮挡免费看| 精品午夜福利视频在线观看一区| 一级a爱片免费观看的视频| 国内久久婷婷六月综合欲色啪| 色噜噜av男人的天堂激情| 国产成人欧美在线观看| 99久久精品热视频| 欧美国产日韩亚洲一区| 国产精品香港三级国产av潘金莲| 国产成人av激情在线播放| 国产精品免费一区二区三区在线| 午夜福利成人在线免费观看| 黄色女人牲交| 午夜成年电影在线免费观看| 哪里可以看免费的av片| 欧美最黄视频在线播放免费| 黄色视频,在线免费观看| 香蕉久久夜色| 国产一区二区三区在线臀色熟女| 日本熟妇午夜| 国产黄色小视频在线观看| 黑人巨大精品欧美一区二区mp4| 久久久久国内视频| 伊人久久大香线蕉亚洲五| 无人区码免费观看不卡| 国产又色又爽无遮挡免费看| svipshipincom国产片| 亚洲精品美女久久av网站| 精品国产乱子伦一区二区三区| 国产成年人精品一区二区| 欧美高清成人免费视频www| 黄色视频,在线免费观看| 国产黄片美女视频| 欧美成人一区二区免费高清观看 | 97人妻精品一区二区三区麻豆| 狂野欧美激情性xxxx| 精品久久久久久,| 中出人妻视频一区二区| 午夜福利18| 精品久久久久久久久久久久久| 1024香蕉在线观看| 亚洲成av人片在线播放无| 久久中文字幕一级| 亚洲专区国产一区二区| 99riav亚洲国产免费| 亚洲av成人不卡在线观看播放网| 国产97色在线日韩免费| 亚洲成人国产一区在线观看| 日韩大尺度精品在线看网址| 老汉色av国产亚洲站长工具| 精品高清国产在线一区| 男女视频在线观看网站免费 | 免费在线观看黄色视频的| 91字幕亚洲| 亚洲国产中文字幕在线视频| 久久精品国产清高在天天线| 国产亚洲精品综合一区在线观看 | 国产乱人伦免费视频| 麻豆av在线久日| 久久香蕉精品热| 给我免费播放毛片高清在线观看| 精品福利观看| 成人午夜高清在线视频| 曰老女人黄片| 国产精品美女特级片免费视频播放器 | 757午夜福利合集在线观看| 精品久久久久久久末码| 色哟哟哟哟哟哟| 女人被狂操c到高潮| 色噜噜av男人的天堂激情| 亚洲 欧美一区二区三区| 国产黄a三级三级三级人| 亚洲va日本ⅴa欧美va伊人久久| 免费搜索国产男女视频| 国产精品综合久久久久久久免费| 国产日本99.免费观看| 一进一出好大好爽视频| 人人妻人人看人人澡| 久久中文字幕一级| 动漫黄色视频在线观看| 女警被强在线播放| 国产蜜桃级精品一区二区三区| 亚洲第一电影网av| 男女做爰动态图高潮gif福利片| 一区二区三区国产精品乱码| or卡值多少钱| 国产亚洲精品av在线| 久久久久九九精品影院| 麻豆久久精品国产亚洲av| 男插女下体视频免费在线播放| 三级国产精品欧美在线观看 | 欧洲精品卡2卡3卡4卡5卡区| 99久久精品热视频| 日日摸夜夜添夜夜添小说| 亚洲国产欧美人成| 欧美黄色淫秽网站| 欧美黑人欧美精品刺激| 欧美精品啪啪一区二区三区| 久久久精品欧美日韩精品| 久久国产精品影院| 成年人黄色毛片网站| 啦啦啦免费观看视频1| 国产精品 欧美亚洲| 亚洲狠狠婷婷综合久久图片| 好看av亚洲va欧美ⅴa在| 两个人视频免费观看高清| 欧美一区二区精品小视频在线| 成年免费大片在线观看| 欧美黑人欧美精品刺激| 啦啦啦韩国在线观看视频| 露出奶头的视频| 日韩中文字幕欧美一区二区| 久99久视频精品免费| 少妇被粗大的猛进出69影院| 脱女人内裤的视频| 亚洲五月天丁香| 老司机靠b影院| 欧洲精品卡2卡3卡4卡5卡区| 久久精品国产清高在天天线| 啦啦啦观看免费观看视频高清| 欧美性猛交黑人性爽| 国产熟女xx| 好男人在线观看高清免费视频| 此物有八面人人有两片| 国产人伦9x9x在线观看| 久久精品综合一区二区三区| 真人做人爱边吃奶动态| 99riav亚洲国产免费| 国产爱豆传媒在线观看 | 国产又色又爽无遮挡免费看| 丰满人妻熟妇乱又伦精品不卡| 久久久精品欧美日韩精品| 香蕉久久夜色| 欧美人与性动交α欧美精品济南到| 亚洲全国av大片| 精品欧美国产一区二区三| 亚洲国产看品久久| 最新美女视频免费是黄的| 国产高清有码在线观看视频 | 欧美黑人精品巨大| 欧美久久黑人一区二区| 亚洲aⅴ乱码一区二区在线播放 | 日本一区二区免费在线视频| 国产精品影院久久| 精品久久蜜臀av无| 久久久国产精品麻豆| 不卡av一区二区三区| 无人区码免费观看不卡| 丁香欧美五月| 中文字幕av在线有码专区| 男女之事视频高清在线观看| 久久精品人妻少妇|