그동안 다양한 도로설계 안전성 평가 연구는 교통사고다발지역인 교차로에 집중되어 왔다. 또한 도로구간에서는 특정한 기하구조 요인이 교통사고율에 어떠한 영향을 미치는지에 대한 연구는 일부 수행되었으나, 도로구간에서의 안전성 평가를 위한 다양한 원인분석을 위한 연구는 여전히 미진한 실정이다. 따라서, 본 연구는 지방부 도로구간에서의 안전성 평가를 위해 도로구간을 지방부 2차로 도로와 다차로 도로구간으로 구분하여 통계적 기법을 적용하였다. 일반적으로 도로구간에서의 교통사고는 "0"의 빈도가 높게 나타나므로 통계적 분석 시에 이를 고려해야만 한다. 따라서 본 연구에서는 사고의 비선형적 요소를 설명하는 허들모형이 전통적인 포아송 및 음이항모형보다 도로구간에서의 발생하는 교통사고의 성격을 설명하는데 더욱 적절한 것으로 판단하여 분석을 시작하였다. 본 연구에서는 지방부 도로구간에서의 사고빈도 예측모형 개발 및 도로구간 안전성 평가를 위해서 지방부 2차로 도로구간과 다차로 도로구간으로 구분하였으며, 모델분석결과 교통사고를 유발시키는 변수가 각각의 유형에 따라 서로 다른 것으로 분석되었다.
그동안 다양한 도로설계 안전성 평가 연구는 교통사고다발지역인 교차로에 집중되어 왔다. 또한 도로구간에서는 특정한 기하구조 요인이 교통사고율에 어떠한 영향을 미치는지에 대한 연구는 일부 수행되었으나, 도로구간에서의 안전성 평가를 위한 다양한 원인분석을 위한 연구는 여전히 미진한 실정이다. 따라서, 본 연구는 지방부 도로구간에서의 안전성 평가를 위해 도로구간을 지방부 2차로 도로와 다차로 도로구간으로 구분하여 통계적 기법을 적용하였다. 일반적으로 도로구간에서의 교통사고는 "0"의 빈도가 높게 나타나므로 통계적 분석 시에 이를 고려해야만 한다. 따라서 본 연구에서는 사고의 비선형적 요소를 설명하는 허들모형이 전통적인 포아송 및 음이항모형보다 도로구간에서의 발생하는 교통사고의 성격을 설명하는데 더욱 적절한 것으로 판단하여 분석을 시작하였다. 본 연구에서는 지방부 도로구간에서의 사고빈도 예측모형 개발 및 도로구간 안전성 평가를 위해서 지방부 2차로 도로구간과 다차로 도로구간으로 구분하였으며, 모델분석결과 교통사고를 유발시키는 변수가 각각의 유형에 따라 서로 다른 것으로 분석되었다.
The past researches on roadway segment safety estimation focused on intersections, which are the primary traffic accident regions. The past researches on roadway segments, However, analyzed the effects of certain factors on the traffic accident occurrence rate by organizing the individual geometric ...
The past researches on roadway segment safety estimation focused on intersections, which are the primary traffic accident regions. The past researches on roadway segments, However, analyzed the effects of certain factors on the traffic accident occurrence rate by organizing the individual geometric structures of the roads, and there is still a dearth of researches on the development of a traffic accident estimation model for rural roadway segments. Therefore, this research focused on rural two-lane and multilane roadway segments and developed traffic accident estimation models through the application of statistical techniques. This is required to explain such high frequency of zero counts in the traffic accident data. In this research, it was found that the Hurdle model is more suitable than the Poisson or negative binomial-regression model for explaining the excess zeros case. In addition, main variables were chosen to estimate their effects on traffic accident occurrence at rural roadway segments, and the safety at such rural roadway segments was estimated. In this research, it was assumed that there are different factors that affect the safety at two-way lane and multilane roadway segments, and a traffic accident estimation model was developed by dividing the two-way lane and multilane roadway segments.
The past researches on roadway segment safety estimation focused on intersections, which are the primary traffic accident regions. The past researches on roadway segments, However, analyzed the effects of certain factors on the traffic accident occurrence rate by organizing the individual geometric structures of the roads, and there is still a dearth of researches on the development of a traffic accident estimation model for rural roadway segments. Therefore, this research focused on rural two-lane and multilane roadway segments and developed traffic accident estimation models through the application of statistical techniques. This is required to explain such high frequency of zero counts in the traffic accident data. In this research, it was found that the Hurdle model is more suitable than the Poisson or negative binomial-regression model for explaining the excess zeros case. In addition, main variables were chosen to estimate their effects on traffic accident occurrence at rural roadway segments, and the safety at such rural roadway segments was estimated. In this research, it was assumed that there are different factors that affect the safety at two-way lane and multilane roadway segments, and a traffic accident estimation model was developed by dividing the two-way lane and multilane roadway segments.
* AI 자동 식별 결과로 적합하지 않은 문장이 있을 수 있으니, 이용에 유의하시기 바랍니다.
문제 정의
It is hoped that the development of a traffic accident estimation model in this research will pave the way for the establishment of a suitable method of preventing traffic accidents at rural roadway segments. This research was based on the assumption that the characteristics of traffic accidents follow Poisson distribution and can be compared by developing suitable statistical models. The results of this research are as follows:
제안 방법
Based on the research results of the developed traffic accident estimation model for geometric structures, the elasticity was analyzed for each explanation variable according to the change in values of the variables. As shown in Table 6, in the case of multilane roadway segments, the higher the radius (1000/R) is, the higher the increase in the values of the variable, and the lower the radius is, the bigger the values of the variable, which means that sudden curves are formed.
Rural roadway segments outside the capital region, or outside metropolitan areas, and on plains and downhill areas, were chosen for this study. Field research was executed as follows: First, based on data obtained from past researches, the variables affecting the frequency of traffic accidents were arranged, and the research areas and roadway axes were chosen. Second, the final variable candidates were chosen by carrying out some sections of the past field researches, and by supplementing the field research plan.
Based on such results, it can be used to choose the priority design element for designing rural roadway segments. For the analysis of the effects of the various design elements on the frequency of traffic accidents, the analysis scope is focused on two-way lane and multilane roadway segments. Also, table 3 explains study sites statistics.
Such models’ results are presented in Table 4 and 5. For the use of the developed traffic accident estimation model for rural multilane roadway segments, the following variables were chosen: EXPO, radius, HC, VC, guardrail, crosswalk, terrain, land use, and bus stop. EXPO, grade, driveway, lighting, Terrain, bus stop, median, land use, and crosswalk are shown in the two-way lane roadway segment.
Therefore, to reduce the traffic accidents at roadway segments, the traffic, roadway, and traffic operation conditions should be analyzed when designing roads. In addition, this research focused on rural two-lane and multilane roadway segments and developed a traffic accident estimation model through the application of a statistical technique. Through these, a suitable traffic accident prevention method was developed for rural roadway segments.
Main variables were chosen to estimate their effects on traffic accident occurrence at rural roadway segments, and the safety at such rural roadway segments was assessed or estimated. In this research, it was assumed that there are different factors that affect the safety at two-way lane and multilane roadway segments, and a traffic accident estimation model was developed by dividing the two-way lane and multilane roadway segments. The results showed that the [+] variables that increase traffic accidents at rural multilane roadway segments are EXPO, radius, VC, guardrail, crosswalk, terrain, land use, and bus stop; those that increase [+] traffic accidents at two-way lane roadway segments are EXPO, driveway, median, land use, and crosswalk; and those that reduce (-) traffic accidents at two-way lane roadway segments are grade (0%) and lighting facilities.
(1994), Brinkman and Percbonok (1979) in their research results. In this research, to analyze the relationship between the frequency of traffic accidents and the various explanation variables that had been gathered, the changes in the elements (e.g., geometric structure) were segmented for each section at a continuous roadway segment. Moreover, the sections where traffic accidents have occurred and those where no traffic accident has occurred were simultaneously considered and should be applied in the construction of a traffic accident estimation model.
In this research, two alternative models for each type of roadway segment that was developed from the Hurdle model were developed. These models can used to determine significant variables affecting accidents.
To determine the effect of the various design elements on the frequency of traffic accidents, to avoid repeated effects between the explanation and subordinate variables, and to choose the variables for securing each variables independence, the design elements were analyzed under the condition where the reliable level of correlation among the variables was 90%. In this study, Poisson, negative binomial regression, and Hurdle models were developed and analyzed. The research results showed that the used roadway segment had too many zeros in its traffic accident frequency figure.
Field research was executed as follows: First, based on data obtained from past researches, the variables affecting the frequency of traffic accidents were arranged, and the research areas and roadway axes were chosen. Second, the final variable candidates were chosen by carrying out some sections of the past field researches, and by supplementing the field research plan. Third, through field researches, the geometric structure of roadways and traffic condition variables were collected.
The scope of the explanation variables in the results showed that data regarding the geometric structure of roadways, traffic conditions, and other variables can be collected. The geometric structure consisted of the road variables (curve radius, vertical curve, traveled width, roadway segment length, driveway lanes, shoulder shape and width) and the roadway design variables (number of lighting facilities, median shape and width). The traffic condition variables were limited speed, traffic volume, heavy vehicle volume, restricted driving facilities, and terrain, and the other variables were the relevant land-uses and the number of crosswalks, bus stops, and visibility improvement facilities.
The past researches on roadway segments, however, analyzed the effects of certain factors on the traffic accident occurrence rate by organizing the individual geometric structures of the roads, but there is a dearth of researches on the development of a traffic accident estimation model. Therefore, this research aimed to develop a traffic accident estimation model by applying a statistical technique on rural two-lane and multilane roadway segments. It is hoped that the development of a traffic accident estimation model in this research will pave the way for the establishment of a suitable method of preventing traffic accidents at rural roadway segments.
shows the research sections for the traffic accident analysis at rural roadway segments. This research combined all of the data collected from Region A and Region B, and divided the data into two-lane and multilane categories.
This research used ρ2(Likelihood ratio statistic) to determine the final models in order to conduct analyses significant factors affecting accidents.
This research was focused on rural roadway segments, and traffic accident models were developed for such roadway segments. A traffic accident model for urban roadway segments is also needed, though.
0. This traffic accident estimation model was to be used to prove the relation between the main explanation variables and the frequency of traffic accidents, as well as to assess the safety of the analysis section through traffic accident estimation. To determine the effect of the various design elements on the frequency of traffic accidents, to avoid repeated effects between the explanation and subordinate variables, and to choose the variables for securing each variables independence, the design elements were analyzed under the condition where the reliable level of correlation among the variables was 90%.
대상 데이터
Rural roadway segments outside the capital region, or outside metropolitan areas, and on plains and downhill areas, were chosen for this study. Field research was executed as follows: First, based on data obtained from past researches, the variables affecting the frequency of traffic accidents were arranged, and the research areas and roadway axes were chosen.
이론/모형
A traffic accident estimation model was developed using LIMDEP 8.0. This traffic accident estimation model was to be used to prove the relation between the main explanation variables and the frequency of traffic accidents, as well as to assess the safety of the analysis section through traffic accident estimation.
”Important data may be lost in the analysis when the fact that the data obtained through the use of the existing model and the traditional reasoning method includes too many zeros is not considered. To address this problem, the Hurdle model was applied in this research. The following is an explanation of such model.
성능/효과
1. The research results showed that although traffic accidents follow Poisson distribution, the used roadway segment had too many zeros in its traffic accident frequency figure. Therefore, it is required to explain such high frequency of zero counts in the traffic accident data.
2. Main variables were chosen to estimate their effects on traffic accident occurrence at rural roadway segments, and the safety at such rural roadway segments was assessed or estimated. In this research, it was assumed that there are different factors that affect the safety at two-way lane and multilane roadway segments, and a traffic accident estimation model was developed by dividing the two-way lane and multilane roadway segments.
Researchers have also examined the effect of vertical curves on safety but did not establish a clear relationship between vertical curves and accident frequency. In this study, it was found that traffic accident occurrence probability is higher in the case of uphill road sections at vertical curves (t=7.027). This finding agrees with the results of the research conducted by Harwood et al.
Many previous researches revealed that raised median have a trade-off safety effect: placing a curb in the median will largely reduce cross-median accidents but may increase sideswipe crashes and crashes that result from vehicles being deflected back into the traffic stream. In this study, it was found that traffic accidents increased in roads where a median had been installed. This finding was also shown in the results of the research that was conducted by Donnell et al.
In this study, Poisson, negative binomial regression, and Hurdle models were developed and analyzed. The research results showed that the used roadway segment had too many zeros in its traffic accident frequency figure. Therefore, it was found that the Hurdle model is more suitable than the Poisson or negative binomial regression model for explaining the data that were used in the research, as shown in Fig 4.
In the research conducted by Sheffer(1999), it was assumed that the intersection traffic accident rate is affected by the lead and lag left-turn phase methods, among the different signaling phase methods. The results of the analysis of the target intersection lead left-turn phase showed a 50% lower traffic accident rate, or higher than that of the lag left-turn phase. In addition, Bauer(1996) analyzed the correlation between the traffic volume and the geometric structure of the plane intersection, and developed a statistical traffic accident model.
(2000) found that as the slope of the roadway segment increases, the number of traffic accidents also tends to increase. The results of their analysis revealed that the traffic accidents increased by 3-14% when the slope was between 2 and 8%. Therefore, in the study conducted by Zegeer et al.
In addition, Bauer(1996) analyzed the correlation between the traffic volume and the geometric structure of the plane intersection, and developed a statistical traffic accident model. The results of their research also showed that the traffic volume has a greater effect on the frequency of traffic accidents than the geometric structure of the intersection does. Moreover, Bonneson(1993) analyzed the relation between the traffic volume at intersections and the frequency of traffic accidents using negative exponential distribution, refined the relation between major and minor roadway traffic volume and the frequency of traffic accidents, and developed a traffic accident estimation model.
(1986). The results of their researches showed that countermeasures are needed to prevent traffic accidents at vertical-curve interchange sections. The speed decrease of a big vehicle in an uphill section will affect another speeding vehicle and may cause traffic chaos.
Thus, it is the presence of additional conflict points at roadway segments that is associated with crashes. The results of this study also showed that the higher the driveway density is, the lower the rate of vehicle to vehicle traffic accidents at roadway segments. These findings support the notion that effective access management can improve safety at roadway segments.
David and Norman(1976) analyzed the frequency of traffic accidents according to the urban- and rural-area traffic volumes, using comparative analysis. The results showed that at unsignalized intersections in urban areas, if the number of traffic accidents is 20,000 veh/day or less, the 3-leg intersection and 4-leg intersection traffic accident frequency will be similar, but if the number of traffic accidents at an unsignalized intersection is 20,000 veh/day or more, the frequency of traffic accidents at four leg intersections will be more than twice the frequency of traffic accidents at three-leg intersections.
The results showed that the [+] variables that increase traffic accidents at rural multilane roadway segments are EXPO, radius, VC, guardrail, crosswalk, terrain, land use, and bus stop; those that increase [+] traffic accidents at two-way lane roadway segments are EXPO, driveway, median, land use, and crosswalk; and those that reduce (-) traffic accidents at two-way lane roadway segments are grade (0%) and lighting facilities.
This traffic accident estimation model was to be used to prove the relation between the main explanation variables and the frequency of traffic accidents, as well as to assess the safety of the analysis section through traffic accident estimation. To determine the effect of the various design elements on the frequency of traffic accidents, to avoid repeated effects between the explanation and subordinate variables, and to choose the variables for securing each variables independence, the design elements were analyzed under the condition where the reliable level of correlation among the variables was 90%. In this study, Poisson, negative binomial regression, and Hurdle models were developed and analyzed.
후속연구
Therefore, this research aimed to develop a traffic accident estimation model by applying a statistical technique on rural two-lane and multilane roadway segments. It is hoped that the development of a traffic accident estimation model in this research will pave the way for the establishment of a suitable method of preventing traffic accidents at rural roadway segments. This research was based on the assumption that the characteristics of traffic accidents follow Poisson distribution and can be compared by developing suitable statistical models.
참고문헌 (23)
Bauer, K.M. and Harwood, D.W.(1996). Statistical Models of At-grade Intersection Accidents, Final Technical Report, FHWA-RD-96-125, November
Bonneson, J.A., and McCoy, P.T.(1997). Capacity and Operational Effects of Midblock Left-Turn Lanes, NCHRP Report 395, TRB
Brickman, C. and Perchonok, K.(1979). Hazardous Effects of Highway Features and Roadside Objects Highlights, Public Road. Federal highway Administration
David, N.A., and Norman, J. R.(1976). Motor Vehicle Accidents in Relation to Geometric and Traffic Features of Highway Intersections. Vol 2-Research Report, Report No. FHWA-RD-76-129, Federal Highway Administration
Harwood, D.W., Council, F.M., Hauer, E., Hughes, W.E., and Vogt, A.(2000). Prediction of the Expected Safety Performance of Rural Two-Lane Highways. FHWA-RD-99-207, December.
Hauer, E.(1988). The Safety of Older Persons at Intersections, Transportation in an Aging Society: Improving Mobility and Safety for Older Persons, Vol 2, Special Report 218, TRB, pp. 194-252
Jackett, M. J.(1996). Accident savings from road lighting in NewZealand. Proceedings of roads '96 conference, Christchurch
Austin, R. and Carson, J.(2002). "An alternative accident prediction model for highway-rail interfaces", Accident Analysis & Prevention, Vol 34, pp. 31-42.
Bonneson, J. A. and McCoy, P.T.(1993) "Estimation of safety at two-way stop-controlled intersections on rural highways", TRR 1401, pp 26-37
Donnell, Eric T., John, M., and Mason, Jr.(2006). "Predicting the frequency of median barrier crashes on Pennsylvania interstate highways", accident Analysis & Prevention, Volume 38, Issue 3
Hauer, E., Council, F.M. and Mohammedshah, Y.(2007). "Safety Models for Urban Four-Lane Undivided Road Segments", Transportation Research Record 1897, TRB, pp96-105
Joshua, S.C.and Garber, N.J.(1990). "Estimating Truck Accident, Rate and Involvements Using Linear and Poisson Regression Models", Transportation Planning and Technology, Vol15, pp41-58
Jovanis, P. P. and Chang H. L.(1986). "Modelling the relationship of Accidents to Miles Travelled", Transportation Research Record 1068, pp. 42-51
Lau, M. Y. and May, A. D.(1988). "Accident Prediction Model Development: Signalized Intersections", Institute of Transportation Studies, UCB-ITS-RR-88-7, University of California-Berkeley, December.
Lyon, C., Oh, J., Persaud, B.N., Washington, S.P., and Bared, J.(2003). "Empirical Investigation of the IHSDM Accident Prediction Algorithm for Rural Intersections". Transportation Research Record 1840. pp. 78-86
McCoy, P.T., and Malone, M.S.(1989). "Safety Effects of Left-Turn Lanes on Urban Four-Lane Roadways", Transportation Research Record 1239, pp 24-36
Miaou, S. P. and Lum, H.(1993). "Modeling vehicle accidents and highway geometric design relationship", Accident Analysis and Prevention, 25(6), pp. 689-709
Milton, J. C. and Mannering, F. L.(1998). "Then relationship among highway geometrics, traffic-related elements and motor-vehicle accident frequencies" Transportation Research Record, 25, pp. 395-413
Mullahy, J.(1986). "Specification and testing of some modified count data models". Journal of Econometrics 3, pp.341-365
Poch, M. and Mannering, F.(1996). "Negative binomial analysis of intersection-accident frequencies", Journal of Transportation Engineering, 122(2), pp. 105-113
Shankar, V., Milton, J. C. and Mannering, F. L.(1997). "Modeling accident frequencies as zero-altered probability process : An empirical inquiry", Accident Analysis and Prevention, 29(6), pp.829-837
Sheffer, C. and Janson, B.N.(1999). "Accident and capacity comparisons of leading and lagging left-turn signal phasings", Transportation Research Board, National Research Council, January
Zegeer, C.V., Stewart, R., Council, F. and Neuman, T.R.(1994). "Accident relationships of roadway width on low-volume roads", Transportation Research Board. No.1445
※ AI-Helper는 부적절한 답변을 할 수 있습니다.