كليدواژه :
تونل‌هاي شهري , فراواني تصادفات , مدل خطي تعميم‌يافته (GLM) , مدل جمعي تعميم‌يافته (GAM)
چكيده فارسي :
هدف از اين تحقيق، استفاده از مدل جمعي تعميم‌يافته (GAM) به‌عنوان رويكردي ناپارامتري جهت شناسايي عوامل ترافيكي مؤثر بر فراواني تصادفات در نواحي دسترسي و ورودي تونل‌هاي شهري و مقايسه‌ي نتايج حاصل از آن با مدل خطي تعميم‌يافته (GLM) به‌عنوان رويكردي پارامتري است. براي اين منظور اطلاعات تصادفات رخ‌داده در محدوده‌ي تونل رسالت به همراه داده‌هاي ترافيكي روزانه در طول سه سال متوالي (1389 تا 1391) از مركز كنترل و مديريت تونل‌هاي شهري تهران دريافت شد. پس از پردازش اطلاعات، تعداد 1047 تصادف در نواحي دسترسي و ورودي تونل به‌عنوان متغير وابسته و سه متغير لگاريتم حجم ترافيك، فراواني وسايل‌نقليه سنگين (درصد) و اختلاف سرعت از محدوديت سرعت بزرگراه به‌عنوان متغيرهاي مستقل نهايي جهت فرآيند مدل‌سازي انتخاب شدند. بر اساس نتايج مدل خطي تعميم‌يافته، اثر خطي متغيرهاي لگاريتم حجم روزانه (009 /0 p<)، فراواني وسايل‌نقليه سنگين (000 /0 p<) و اختلاف ميانگين سرعت روزانه وسايل‌نقليه عبوري از تونل نسبت به محدوديت سرعت بزرگراه (000 /0 p<) در فراواني تصادفات نواحي دسترسي و ورودي تونل معني‌دار گزارش شد.
چكيده لاتين :
The development of underground spaces and its benefits encompass all sectors of society. Because of its enclosed space, the safety of traffic passing through urban tunnels is very important. Recent years of statistical progress in accident modeling, including regression analysis with generalized linear models and generalized additive models, considered a new approach to identifying more complex relationships between independent and dependent variables. Despite the high frequency of tunnel crashes, lower than open roads, such as highways and freeways, research has shown that tunnel crashes are more severe. The purpose of this study was to use the Generalized Model as a non - parametric approach to identify traffic factors affecting accident frequency in urban tunnel access and input areas and a parametric approach is to compare results with the generalized linear model. For this purpose, the accident data was received from the Tehran City Tunnel Control and Management Center during the three consecutive years (2010 to 2012) along with traffic data. There are teleservice systems and video surveillance in tunnel traffic systems. The traffic system is used to collect information and traffic parameters from tunnel entry level, such as: velocity and occupancy rate, and video surveillance system, to manage traffic flow and record events, including accidents occurring inside the tunnel. Data in this paper including records of 1047 accidents in the tunnel's access and entry areas. Traffic volume, heavy vehicle percentage and speed deviation from the speed limit of the highway as the independent variables were selected. Based on the results of the generalized linear model, the linear effect of daily traffic volume, percentage of heavy vehicles and daily average speed deviation from speed limit was reported Meaningful in crash model frequency. However, the three - degree and first - order relationship for the heavy vehicle percent and daily average speed deviation was confirmed by the frequency of accidents using the generalized model, in addition to the significance of the daily volume logarithm. The result of comparing generalized and generalized linear model using a good fit criterion and Akaic criterion shows that the generalized additive model is better suited to estimating the frequency of accidents dependent variable. A generalized additive model of superiority of this model was shown by the higher fitting coefficient (0.99) and lower Akaic information criterion (3823) compared to the generalized linear model. Drivers at high speed approaching the tunnel are at higher risk of collisions. The velocity fluctuations generated when entering tunnels to adapt to tunnel environmental conditions, such as lighting conditions, may be one of the factors that adversely affect traffic safety when passing through the tunnel. One accident is often expected with the increase in traffic volume and the percentage of trucks, but this proposition is not true for all situations. With the increase in the percentage of heavy vehicle in free flow conditions, the frequency and repetition of lane changing and overtaking increase, which could be one of the reasons for the increase in accidents compared with almost dense traffic conditions.
