道橋畢業(yè)設(shè)計(jì)外文文獻(xiàn)翻譯--雙車(chē)道公路上駕駛員超車(chē)行為的安全性評(píng)估

1、<p><b>　　附錄：</b></p><p>　　Safety Assessment of Driver Overtaking Behavior on Two-Lane Highways</p><p>　　Yaqin QIN1, Jian XIONG1, Xiujuan ZHU1, and Jianshi LI1</p><p&g

2、t;　　Faculty of Transportation Engineering，Kunming University of Science and</p><p>　　Technology (KUST), Kunming, Yunnan 650224, China; PH (86) 0871-3802298;</p><p>　　email: qyq_email@foxmail.com

3、</p><p><b>　　ABSTRACT</b></p><p>　　Safety study of driving behaviors to reduce traffic accidents is of great significance. The primary objectives of the study were to investigate the

4、 relation between driver overtaking behaviors and traffic safety. Twelve drivers of different proficiency were selected for an experiment on a driving simulation system platform. A simulation scene of overtaking on a two

5、-lane highway was set up. The subjects overtook in the virtual scenarios at different speeds. A total of twelve parameters - includin</p><p>　　INTRODUCTION</p><p>　　Two-lane rural roads make up

6、the majority of the road network in many countries. In 2009, China’s rural mileage reached 3.2 million km, and highway mileage of rural roads accounted for more than three-quarters of the total mileage (Li, 2009). Rural

7、roads are also dominant in traffic fatality statistics. Lamm et al. (2007) has estimated that more than 60% of all fatalities in traffic occur on two-lane rural roads.</p><p>　　Overtaking maneuvers on rural

8、two-lane highways is a common phenomenon. When drivers have potential to overtake and there is sufficient space to overtake on the road, overtaking demands will be created. In the process of overtaking, the driver determ

9、ines whether there is sufficient and adequate passing sight distance and time headway of the opposing lane and whether there is an adequate inserting gap of the same lane. Then drivers decide whether the overtaking shoul

10、d be implemented. Since overta</p><p>　　Greenshields et al. (1935) was the first to establish the minimum requirements for safe passing under average traffic conditions. Bar-Gera, H. and Shinar, D. (2005) ha

11、ve shown this maneuver is associated with an increase in crash risk, because it involves driving in the lane of the opposing traffic direction. At the same time, Harris (1988) found that most drivers are indeed aware tha

12、t overtaking is a risky maneuver by self-report ratings. At present, many existing studies (see reviews by Tang, </p><p>　　et al., 2007; Shao, et al., 2007) focus on overtaking modeling. A few studies assess

13、ing the safety of driving overtaking behaviors generally examine driver gender, age and other characteristics (David, et al., 1998), or by DBQ questionnaire to survey (Ozkan and Lajunen, 2005). It is difficult to obtain

14、drivers’ real-time direct performance of overtaking maneuvers due to the danger of the experiment on a real road. </p><p>　　To assess driving behaviors in overtaking, we employ a driving simulator. Various s

15、tudies have shown that driving simulators can provide reliable observations of drivers’ behaviors (Blana, 1996; Desmond, and Matthews, 1997; Van der Winsum and Brouwer, 1997; Ellingrod et al. 1997.). In this study, we fo

16、cus on individual differences in the safety of overtaking maneuvers with DBQ and real-time behaviors on a driving simulator. The results of the study are designed to distinguish unsafe drivers from</p><p>&l

17、t;b>　　METHOD</b></p><p>　　Participants</p><p>　　Twelve volunteers participated in the experiment. The sample comprised six men and six women between 24 and 55 years old (M= 28.08 years,

18、 S.D. = 5.6). There was an equal balance of males and females. All had a valid driver’s license. The mean number of years driving experience was 5.25 years and, on average, drivers drove 3.08 hours per week. All had norm

19、al or corrected-to-normal visual acuity, and did not take any kind of medicine.</p><p><b>　　Apparatus</b></p><p>　　A full-size advanced driving simulator (KMRTDS, developed by the si

20、mulation laboratory of Faculty of Transportation in Kunming University of Science and Technology, China) was used in this study. The simulated vehicle cab, an Axial, featured all normal displays and controls (steering, b

21、rakes, and accelerator) found in a vehicle. Different driving scenarios were projected onto a 1500 cycle screen, with sound effects of the vehicles in motion broadcasted by two-channel amplifiers. With the optimiz</p&

22、gt;<p>　　Experimental design</p><p>　　The purpose of this experiment is to evaluate the safety of driving behaviors during driver overtaking at different speeds on two-lane highways. In China, the lar

23、gest design speed of a two-lane highway is 80km/h. On the other hand, the sight distance of overtaking is different with the different speeds of experimental vehicles. In this experiment, we set the speed of experimental

24、 passed and oncoming vehicles to 30, 37.5, 45 and 60km/h respectively. The corresponding overtaking speeds were set at</p><p>　　Driving scenarios</p><p>　　The parameters of overtaking driving sc

25、enarios include static and dynamic parameters. The static parameters refer to the road alignments, traffic signs, markings and the surrounding natural environment and so on. The dynamic parameters indicate the parameters

26、 of dynamic driving vehicles, including the triggering movement regions of experimental vehicles in the scenarios, speed, driving route, distance between vehicles and so on</p><p>　　Static scenarios</p>

27、;<p>　　The design of the static overtaking scenario of this experiment (shown in Figure 2) is an approximately square scenario, composed of four straight sections of a rural secondary two-lane highway. Total dista

28、nce is 8.0 km, with 2.0 km long on each side, and the turning radius of the connection is 200 meters. The width of a single lane is 4.5 m (including the road shoulder), with no isolation facilities in the center. Both si

29、des of the road are grass and randomly distributed trees and villages, and</p><p>　　no fog, no rain, no snow weather and dry flat pavement conditions. The static scenario design of experiment is shown in Fig

30、ure 1</p><p>　　Dynamic scenarios help</p><p>　　In accordance with highway standards in China, specific section with overtaking sight distance in appropriate distance should be set on a two-lane

31、highway based on need and terrain. Pie and Wang (2004) have stated the length of the section should not less than 10%-30% of the total length of the route in normal circumstances.</p><p>　　In dynamic scenari

32、os, the experimental vehicles must drive according to pre-determined routes and speeds. When the leading vehicle which the driver maneuvered reached a specific place, the other experimental vehicles will be triggered to

33、start movement. These places are triggering points. The setting of dynamic parameters of experimental vehicles during overtaking process was explained as follows.</p><p>　　Figure 1. Design of static scenario

34、</p><p>　　Figure 2 is the dynamic scenario distribution of the overtaking process. Car0 is the leading car, and Car1 is the passed car, and Cars 2, 3, 4, 5 are oncoming cars. In order to increase driving rea

35、lism and prevent the driver from overtaking ahead of time, set front Car6 and 7 ahead of Car 2,3,4,5. The setting of dynamic parameters includes starting area, speed, driving track, and space of vehicles. Once Car0 reach

36、ed the trigger point, all other vehicles will begin to move at the set speeds and t</p><p>　　Since the time headway is a time interval when the bumpers of two traveling cars pass the same transect on a road

37、section, it mainly depends on the distance between the two cars and the speed of the following car (Olsten, 2005). Observed in the static condition, time headway (called H) can be directly obtained by measuring the momen

38、t when the two vehicles pass through the same transect. </p><p>　　H=TB-TF (1)</p><p>　　Where, TB——The time when the leading car passes through the detector(s). TF——The time when the foll

39、owing car passes through the detector(s). In a dynamic overtaking process, the time headway is the quotient of the headway between two cars and the speed of the following car, namely:</p><p><b>　　(2)&l

40、t;/b></p><p>　　Where, D——Headway of leading-and-following car (m)</p><p>　　Vf——the speed of the following car (km/h)</p><p>　　Liu (2007) found that, in a mountainous area two-lan

41、e secondary highway, when the headway time of the leading and following cars is below 3.1s and the speed difference between the two cars reached about 20 km/h, a demand to overtake is generated(). In this experiment the

42、headway time was set at 3s. In addition, in order to make front cars driving on the road seem more real, we let the Car1 trigger in Figure3 is the experimental dynamic scenario after loaded traffic flow.</p><p

43、>　　advance, that is taking S01 = 2D. Sn_n+1 value in different speed limit sections were shown in Table 1. </p><p>　　Figure 3. Dynamic overtaking scenarios in KMRTS</p><p>　　Measurements</

44、p><p>　　Overtaking Behavior Measurement</p><p>　　The subjects were required to drive in the KMRTS in order to determine the driver subjects’ overtaking behavior. Every subject finished four tests i

45、n the scenarios. There are 48 different overtaking behavior measurements. When they drove, they should increase their speed as close to the speed limit according to limit speed signs in the simulated scenarios (including

46、 40,50,60,0KM/H, a total of four speeds), and then complete an overtaking maneuver in each section according to their personal exper</p><p>　　During the experiment, the drivers’ operation behavior, driving s

47、peed, acceleration, overtaking time, distances traveled, distance between the overtake car and oncoming cars before and after overtaking were recorded. All these data were used to evaluate the drivers' driving behavi

48、or.</p><p>　　Self-report measurement</p><p>　　A DBQ was introduced as an self-report questionnaire survey, which was adapted by Sullman and others based on Reason’s Driving Behavior Questionnair

49、e (2002). The questionnaire is divided into three dimensions, namely, violations of traffic rules, error behaviors and offensive violations, containing twenty items. In this questionnaire, we used the Likert five-point s

50、coring method to record the drivers' scores. The scores are described by centesimal grades. After calculating the mean values of all</p><p>　　Determination of beginning and ending time of overtaking proc

51、esses</p><p>　　Timing the beginning and ending of the overtaking process includes: the time when the driver begins to overtake, begins to return to the previous lane and completes the overtaking. Determinati

52、on of these moments enables us to get the travel time and distance traveled at all stages in the process of overtaking</p><p>　　The determination of the moment when drivers begin to overtake can be divided i

53、nto two situations: (1) The time a driver turns on a left turning signal before overtaking and (2) if the driver did not turn on a left turning signal, the time determined by mathematical methods. This fixes the driving

54、trajectory by two straight lines. Time to overtake corresponds to the intersection of two lines from the moment the driver begins to overtake.</p><p>　　The end time of the overtaking process was the moment t

55、he vehicle returns to the original lane. We used a video playback with stopwatch to record the moments when drivers completed the overtaking maneuver completely and returned to the original lane. The moment was defined a

56、s the time a driver completed overtaking. Based on the start and end time of the driver's whole overtaking process, we can obtain the vehicle speed, acceleration, steering angle and other motion parameters in the per

57、iod of ov</p><p>　　RESULTS AND DISCUSSION</p><p>　　Overtaking performance</p><p>　　The results of twelve drivers’ operation behaviors and performances in four speeds during the over

58、taking are shown in Table 2.</p><p>　　On different speed limit roads, driver subjects all completed the overtaking process successfully except for 80km/h. Among them, on the speed limit 40km/h and 60km/h sec

59、tions all overtook Car2, but on the speed limit 50km/h section, driver subjects No.7, 8, and 12 followed the front car for too long a time and overtook too late, so they only overtook Car3. In the speed limit 80km/h sect

60、ion, due to the higher speed, driver subjects No.1, 6, 7, and 11 did not complete the overtaking process, and d</p><p>　　To further understand the drivers’ safety trends at different speeds, three parameters

61、 were selected for further analysis. That is, Tp (the completing time for the entire overtaking process), Sc (the distance between the main car and oncoming cars when returned to the original lane after completed the ove

62、rtaking process) and Vavg (average overtaking speed). Table 3 shows the overtaking behavior parameters results of driver subjects at speed limit 80KM/H and the average speed of four.</p><p>　　In Table 3, the

63、 negative values (-16.74 and -2.32) means there was a collision with oncoming cars before completing overtaking, while the value 0.00 indicated not overtaking any oncoming car in the whole overtaking process, namely the

64、overtaking process was not completed. Compared to other speed limits, it was relatively difficult to complete the overtaking process on the speed limit 80km/h section. The average time drivers completed the overtaking pr

65、ocess was 8.4s, shorter than the average time</p><p>　　In terms of the average Top, Vague, Sc of the twelve driver subjects under four speed limit conditions, we found that the cases of smaller Tp and Sc and

66、 larger Vavg reflected the condition of the driver’s overtaking strategy. The strategy is that drivers will improve their own safety by reducing the conflict opportunity with other vehicles when they estimate the overtak

67、ing conditions. But at this point, the drivers may be in an unsafe overtaking. For example, we can conclude the driver subject N</p><p>　　Self-report on usual driving behaviors</p><p>　　The DBQ

68、questionnaire results for twelve subjects were analyzed with the use of a Linker five-point scoring method. Each item has five selections which grades from 1 to 5 points. The 20 items score 100. The higher the score, the

69、 safer the driver's driving behavior. The scores of DBQ (S_DBQ) are between 77 and 95 points. The subject No.3 got the highest score, while subject No.4 the lowest</p><p>　　Further analysis from the scor

70、es of three dimensions shows that subjects had good judgments about error behaviors (S_EJ: M=29.5, D=5) and for offensive violations (S_AV: M=26.08, D= 6). Comparing the two dimensions, there are more violations of traff

71、ic rules among individual drivers (S_OR: M=28.42, D=9 S_OR). The results are shown in Figure 4</p><p>　　It can be stated that twelve subjects identified aggressive violations and misjudgments as unsafe drivi

72、ng behaviors, but some overlooked traffic rule violations. On the other hand, with the results of subjects overtaking performance and DBQ questionnaire, we found that although the DBQ reflected their subjective driving e

73、xperience, there are still great differences in real-world driving.</p><p>　　Safety assessment MODEL on overtaking</p><p>　　To evaluate the safety of subjects in the overtaking process, a multip

74、le linear regression analysis was used to judge their behaviors combined with driving behavior questionnaire score and their motion parameters in the overtaking process.</p><p>　　We selected DBQ questionnair

75、e score “y” as the dependent variable for the regression equation, and identified initially twelve motion parameters which probably influenced their</p><p>　　safety as independent variables, with x1, x2, ...

76、 x12 indicated in Table 4 .</p><p>　　Selection of influencing variables</p><p>　　We analyzed the correlation among twelve influencing independent variables in 45 of 48 observations ( three obser

77、vations were redundant). The correlation coefficient between variables is </p><p>　　Table 5. Correlation Coefficient Matrix</p><p>　　shown in Table 5. From the or relation coefficient matrix in

78、Table 5, there is a strong correlation between some variables (the data with *)</p><p>　　Modeling of multiple linear regression</p><p>　　Table 6. Results of Multiple Linear Regression Analysis&l

79、t;/p><p>　　Because there were strong correlations between some variables, a multivariate regression analysis using a stepwise method was performed After the excluded variables, the optimized model including var

80、iables x8, x9 and x10 are the influencing factors, as shown in Table 6</p><p>　　From Table 6, we can get predictors in the model, which are constant, x8, x9 and x10. There is no multi-nonlinearity among the

81、predictors because of VIF<5. The model is not so good in that the chi-square goodness of fit test is weak. It indicated that there were some other nonlinear correlations between them. The model of multi-linear regress

82、ion obtained is as follows：</p><p>　　Y?? 73.928?? 0.137?? X 8?? 0.046 * X 9?? 0.11* X 10 （3）</p><p>　　These three selected variables showed that the safety of overtaking behavior is attrib

83、uted to the speed and some distances during overtaking. The faster the speed is, the safer the behavior is。</p><p>　　CONCLUSION</p><p>　　Safety assessment of driving overtaking requires at least

84、 two things: first is a method providing the driver’s overtaking maneuvers and second a model assessing and controlling for the safety of overtaking behaviors.</p><p>　　The proposed driver’s overtaking safet

85、y assessment model assumes that average speed and some distances are relevant to the safety of overtaking. The distances include the distances from the passing vehicle to the passed and oncoming vehicle when the overtaki

86、ng ended. The model relies on a stepwise method of multiple linear regression analysis, using twelve parameters in a simulation experiment as independent variables and DBQ scores as the dependent variable.</p><

87、;p>　　In our results, the model was certainly somewhat limited by two factors. The first was the selection and quantity of subjects, and the second was the difference between driving simulators and on-road experiments.

88、 Concerning the first factor, we must be aware that DBQ has subjective results. Moreover, there were 48 observation values, this experiment had only twelve subjects. We introduced twelve influencing factors under inadequ

89、ate data; there was greater influence among variables. All these wil</p><p>　　Dealing with the comparability of the driving simulator and on-road experiments, previous studies have made sure there are parall

90、el observations. Based on these studies, we also can conclude that with the advent of more powerful graphics processors and renderers, simulators are increasingly appealing for studying and training drivers.</p>&

91、lt;p>　　This work was a first approach to the problem. In future work, we plan to improve the safety assessment prediction model by increasing the number of subjects and by using other methods to build the model</p&

92、gt;<p>　　雙車(chē)道公路上駕駛員超車(chē)行為的安全性評(píng)估</p><p>　　秦亞琴 雄健 朱秀娟 李建始</p><p>　　昆明理工大學(xué)交通運(yùn)輸工程學(xué)院，技術(shù)（KUST），云南昆明650224，中國(guó); PH值（86）0871-3802 298; 電子郵件：qyq_email@foxmail.com</p><p><b>　　摘要<

93、/b></p><p>　　安全駕駛行為的研究對(duì)減少交通意外具有偉大的意義。這項(xiàng)研究的主要目的是探討司機(jī)超車(chē)行為和交通安全的關(guān)系。十二個(gè)??不同能力的司機(jī)被選定為駕駛模擬系統(tǒng)實(shí)驗(yàn)平臺(tái)。成立一個(gè)雙車(chē)道的高速公路上超車(chē)的模擬場(chǎng)景。在虛擬場(chǎng)景以不同的速度進(jìn)行超越。在這個(gè)實(shí)驗(yàn)提取了十二個(gè)參數(shù)，包括速度，加速度，超車(chē)過(guò)程中的時(shí)間， 距離和其他參數(shù)的變化。這些數(shù)據(jù)連同一個(gè)DBQ（違例駕駛行為問(wèn)卷）的結(jié)果，進(jìn)行了分析和評(píng)

94、價(jià)的多元線(xiàn)性回歸方法。結(jié)果顯示司機(jī)的安全與三個(gè)運(yùn)動(dòng)參數(shù)有一個(gè)密切的關(guān)系。最后,研究提出了一種線(xiàn)性模型的司機(jī)超車(chē)行為在雙車(chē)道高速公路的安全評(píng)估。該模型可以幫助識(shí)別司機(jī)的安全和不安全意識(shí)來(lái)減少交通事故的數(shù)量。</p><p><b>　　引言</b></p><p>　　在許多國(guó)家雙車(chē)道的農(nóng)村公路占據(jù)了絕大多數(shù)的公路網(wǎng)絡(luò)。在2009年，中國(guó)農(nóng)村里程達(dá)到3.2萬(wàn)公里，高速公

95、路在農(nóng)村公路里程占到總里程超過(guò)四分之三（李，2009）。農(nóng)村道路交通死亡人數(shù)的統(tǒng)計(jì)也占主導(dǎo)地位。拉姆等人（2007年）估計(jì)，超過(guò)60％的交通死亡發(fā)生在兩車(chē)道的鄉(xiāng)村公路上。農(nóng)村雙車(chē)道公路上的超車(chē)是一種常見(jiàn)的現(xiàn)象。當(dāng)司機(jī)有潛在的超越意識(shí)和有足夠的超越空間,超車(chē)的情況將會(huì)發(fā)生。在超車(chē)的過(guò)程中，司機(jī)決定是否有足夠和充分的超車(chē)視距和車(chē)頭時(shí)距，以及是否有一個(gè)足夠的插入同一車(chē)道的差距。然后司機(jī)決定是否超車(chē)。由于超車(chē)條件和司機(jī)行為會(huì)有所不同，超車(chē)的過(guò)程

96、是非常復(fù)雜的。它受道路條件、視覺(jué)距離、車(chē)輛類(lèi)型,速度和司機(jī),其他事情的影響。</p><p>　　Greenshields et al。（1935年）是第一個(gè)建立的最低平均流量條件下安全通過(guò)的要求。Bar-Gera, H. 和Shinar, D. (2005)顯示這種操作是增加引起交通事故風(fēng)險(xiǎn)，因?yàn)樗婕暗今{駛在內(nèi)線(xiàn)的相反的方向的交通。同時(shí)，哈瑞斯（1988）發(fā)現(xiàn)，大多數(shù)司機(jī)確實(shí)知道超車(chē)是危險(xiǎn)動(dòng)作的自評(píng)報(bào)告。目前

97、,許多現(xiàn)有的研究(參見(jiàn)評(píng)審,2007;Geertje唐等人,2007;魏、等,2000;榮)等人,2007;Shao等人,2007)關(guān)注超車(chē)建模。一些研究評(píng)估安全駕駛的超車(chē)行為一般檢查司機(jī)的性別、年齡和其他特征(大衛(wèi),等人,1998),或者通過(guò)DBQ問(wèn)卷方式調(diào)查(主刀和拉尤寧就,2005)。很難獲得司機(jī)超車(chē)時(shí)的直接動(dòng)作由于危險(xiǎn)的實(shí)驗(yàn)在一個(gè)真正的道路。</p><p>　　我們雇傭駕駛模擬器來(lái)評(píng)估駕駛行為的超車(chē)。不

98、同的研究都表明,駕駛模擬器可以提供可靠的觀察駕駛員的行為(林松柏;1996年,德斯蒙德,和馬修斯,1997;Van der Winsum和Brouwer,1997;Ellingrod et al . 1997年)。在本研究中,我們關(guān)注個(gè)體差異的安全演習(xí)DBQ超車(chē)和實(shí)時(shí)行為在駕駛模擬器。這項(xiàng)研究的結(jié)果被設(shè)計(jì)用來(lái)區(qū)分不安全的司機(jī)和其他類(lèi)型的司機(jī)。</p><p><b>　　方法</b><