According to TAAS, about 91.66% of all accidents occur on straight roads and the occurrence of accidents on curved roads is very small at about 6.10%, but the death rate on curved roads is about 3.14 times more than that of straight roads. The selection of improvement points based on the number of accidents in the past seems insufficient to judge the dangerousness of roads. Therefore, through virtual driving simulation, the safety limit of road factors is determined by various combinations of road factors Ld, LTR, and β. Driving speed, wind speed, depth of rutting, friction, and longitudinal slope were considered as road factors in the beginning, however, considering its minimal effect on the result, longitudinal slope was excluded. A symmetrically combined straight ? clothoid - circular road with a radius of 280m and an intersection angle of 90 degrees were design through an OpenCRG, for a virtual driving simulation. By installing 6% of superelevation in accordance with KDS-44-20-10, and simplifying rutting shape into a trapezoid type, rutting defect was created in the virtual road under the 80 of designed speed. The most suitable vehicle for each type of road design criteria were selected to perform simulations according to the combination of 4 road factors and 5 levels. The interaction of each factors and approximate response values could be calculated through the response surface model. Through machine learning, the safety limits were explored by classifying them into safety, and risk?accident level.
A) Designing an expressway, main arterial road (based on the safety limit of a semi-trailer) - If the friction is 0.5 or less, and a driving speed of 40km/h or more is operated, a lane departure accident may occur at a driving speed of 85km/h or more.
- More than 2cm rutting may cause an accident depending on driving and wind speed. - The act of double lane change on a curved part may bring about a risk of an accident due to the loss of the vehicle''s steering ability, except in cases where the driving speed is 80km/h or less, and when there is no effect of wind speed and rutting.
B) Designing a secondary arterial road, collector (based on the safety limit of a tower truck) - In the case of a truck, there is highest risk of accidents due to overturning and the risk of overturning increases when wind speed is applied. - 2 to 3cm or more rutting may cause a lane departure and an overturn accident. - Drive at a driving speed of 90km/h or less, it is safe to keep the driving speed at 80km/h or when wind speed over 80km/h continues. - In all driving conditions, double lane change on a curve poses an accident due to loss of steering capabilities of the vehicle.
C) Designing a local road (based on the safety limit of a SUV and passenger car) - It is safe for wind speed under 40km/h with a driving speed of 90km/h. - Compared to passenger cars, SUVs are more affected by wind speed. - If the driving speed is 110km/h or higher, when changing to a double lane on a curved road, the probability of an accident is very high. Therefore a speed limit or no overtaking zone has to be specified. - 2 to 3cm or more rutting may cause a lane departure and an overturn accident.
If we make a proposal for road maintenance according to the simulation results, when the rutting is more than 2, there is a possibility of an overturn accident through the double lane change even if there is no effect of the crosswind. Therefore, it is necessary to rehabilitate immediately when rutting of more than 2cm occurs. Changing the lane for overtaking on the curved part in the rain may cause an accident due to lane departure or overturning, which results in a risk of secondary damage to adjacent vehicles. Therefore, an overtaking prohibited zone should be established. In addition, in the section where strong moments frequently occur when designing curved roads, a speed limit section should be established and road structures such as windbreaks should be installed.
제 1 장 서 론 11.1 연구배경 및 목적 11.2 연구범위 31.3 연구방법 및 절차 6제 2 장 주행 안전성 평가에 관한 이론적 고찰 92.1 기존연구 고찰 92.2 차량-도로-운전자 사이의 역학적 관계 모형 132.2.1 점-질량 모형 132.2.2 Multi-body 모형 142.2.3 차량-도로-운전자 모형 182.3 주행 안전성 평가기법 222.3.1 실험계획법 222.3.2 기계학습법 24제 3 장 주행 안전성 평가를 위한 시뮬레이션 환경설정 273.1 시뮬레이션 환경설정 273.1.1 시뮬레이션 S/W 검증 273.1.2 차량 설정 293.1.3 도로 설정 343.1.4 주행 환경 설정 393.2 시뮬레이션 변수 설정 403.2.1 변수 및 수준 설정을 위한 사전 시뮬레이션 413.2.2 주행 안전성 변수(반응변수) 설정 443.2.3 도로 환경 변수(요인변수) 설정 483.2.4 실험계획법에 의한 시뮬레이션 계획 533.2.5 시뮬레이션 범위의 한계 및 가정 조건 543.3 통합 시뮬레이션(co-simulation)에 의한 실행 553.4 안전성 평가 방법 57제 4 장 반응 표면 분석법에 의한 주행 안전성 평가 결과 604.1 승용자동차 604.2 소형자동차 754.3 대형자동차 874.4 세미트레일러 99제 5 장 기계학습법에 의한 주행 안전성의 한계치 분류 1115.1 승용자동차 1115.2 소형자동차 1185.3 대형자동차 1245.4 세미트레일러 130제 6 장 시뮬레이션 결과 고찰 1366.1 주행 방법에 따른 결과 고찰 1366.2 도로 설계 및 유지보수 판단 기준에 의한 고찰 138제 7 장 결론 및 향후 연구과제 1437.1 결론 1437.2 향후 연구과제 147<참고문헌> 148