Journal of Transportation Research

Journal of Transportation Research

Investigating the Effect of Geometrical Factors on the Severity of Car Crashes on Suburban Roads Utilizing Artificial Intelligence and Deep Learning Algorithms

Document Type : Original Article

Authors
Sarah Ghaffari 1
Farhad Hosseinali 2
1 M.Sc., Grad., Department of Surveying Engineering, Faculty of Civil Engineering, Shahid Rajaee Teacher Training University, Tehran, Iran.
2 Associate Professor, Department of Surveying Engineering, Faculty of Civil Engineering, Shahid Rajaee Teacher Training University, Tehran, Iran.
10.22034/tri.2024.453155.3240
Abstract
Roads, as a fundamental component of land transportation, face the challenge of road accidents. These accidents have irreparable financial and human impacts on individuals and society. Consequently, the importance of examining road safety and the factors contributing to accidents is evident. This study focused on the two routes of Ardakan-Naeen and Naeen-Ardakan, analyzing them based on six geometric and environmental parameters: horizontal curvature, sight distance, distance from intersections, distance from bridges, shoulder width, and land-use. Each route was segmented into homogeneous segments according to these six parameters and was classified based on the severity of geometric and environmental factors contributing to accidents. Subsequently, by incorporating accident points into both routes, the severity of accidents for each segment was calculated using the EPDO index and categorized into five levels. Three deep Neural Network methods, namely RNN, CNN, and MLFNN, were then employed to predict the severity of accidents for each segment based on the geometric and environmental parameters. The RNN and CNN methods achieved an overall accuracy of approximately 20 percent for both routes, while the MLFNN method demonstrated significantly better results with an overall accuracy of around 90 percent for both routes. Finally, the potential for predicting accident severity on one route by training the model with data from the other route was examined, resulting in an overall accuracy of 88 percent when trained with the Ardakan-Naeen route data and overall accuracy of 78 percent in a reversed state. The last section of this research focuses on analyzing the extent to which improvements in geometric factors contribute to the reduction of accident severity. The results of this analysis indicated that on the Ardakan-Naeen road, the shoulder of the road had the most significant impact, while on the Naeen-Ardakan road, land-use was the primary factor in reducing accident severity.
Keywords
Road Accidents
Deep Artificial Neural Network
Segmentation
Accident-Prone Areas
Geometrics Parameters
Subjects
-Aguero-Valverde, J., & Jovanis, P. P. (2006). Spatial analysis of fatal and injury crashes in Pennsylvania. Accident Analysis & Prevention, 38(3), 618-625.
-Akin, D., & Akba, B. (2010). A neural network (NN) model to predict intersection crashes based upon driver, vehicle and roadway surface characteristics. Sci. Res. Essays, 5(19),2837-2847.
-Cafiso, S., Di Graziano, A., Di Silvestro, G., La Cava, G., & Persaud, B. (2010). Development of comprehensive accident models for two-lane rural highways using exposure, geometry, consistency and context variables. Accident Analysis & Prevention, 42(4), 1072-1079.
-Cafiso, S., D'Agostino, C., & Persaud, B. (2018). Investigating the influence of segmentation in estimating safety performance functions for roadway sections. Journal of Traffic and Transportation Engineering (English Edition), 5(2), 129-136.
-D’agostino, C. (2014). Investigating transferability and goodness of fit of two different approaches of segmentation and model form for estimating safety performance of Motorways. Procedia Engineering, 84,613-623.
-de Santos-Berbel, C., Essa, M., Sayed, T., & Castro, M. (2017). Reliability‐Based Analysis of Sight Distance Modelling for Traffic Safety. Journal of Advanced Transportation, 2017(1), 5612849.
-Dinu, R. R., & Veeraragavan, A. (2011). Random parameter models for accident prediction on two-lane undivided highways in India. Journal of Safety Research, 42(1), 39-42.
-Donahue, J., Anne Hendricks, L., Guadarrama, S., Rohrbach, M., Venugopalan, S., Saenko, K., & Darrell, T. (2015). Long-term recurrent convolutional networks for visual recognition and description. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2625-2634.
-Dong, C., Shao, C., Li, J., & Xiong, Z. (2018). An improved deep learning model for traffic crash prediction. Journal of Advanced Transportation, 2018(1), 3869106.
-Effati, M., Rajabi, M. A., Samadzadegan, F., & Blais, J. R. (2012). Developing a novel method for road hazardous segment identification based on fuzzy reasoning and GIS. Journal of Transportation Technologies, 2(01), 32-40.
-Effati, M., Rajabi, M. A., Hakimpour, F., & Shabani, S. (2014). Analysis of Spatial Factors Contributing on Concentration of Highway Corridors Crashes Using GIS and Data Mining. Journal of Geomatics Science and Technology, 4(2), 87-102 (in Persian).
-Elvik, R., Sagberg, F., & Langeland, P. A. (2019). An analysis of factors influencing accidents on road bridges in Norway. Accident Analysis & Prevention, 129, 1-6.
-Erdogan, S., Yilmaz, I., Baybura, T., & Gullu, M. (2008). Geographical information systems aided traffic accident analysis system case study: city of Afyonkarahisar. Accident Analysis & Prevention, 40(1), 174-181.
-Fink, K. L., & Krammes, R. A. (1995). Tangent length and sight distance effects on accident rates at horizontal curves on rural two-lane highways. Transportation Research Record, 1500, 162-168.
-Flahaut, B. (2004). Impact of infrastructure and local environment on road unsafety: Logistic modeling with spatial autocorrelation. Accident Analysis & Prevention, 36(6), 1055-1066.
-Gardner, M. W., & Dorling, S. R. (1998). Artificial neural networks (the multilayer perceptron)—a review of applications in the atmospheric sciences. Atmospheric Environment, 32(14-15), 2627-2636.
-Glennon, J. C. (1987). Effect of alignment on highway safety. State of the Art Report, 6,48-63.
-Gross, F., & Jovanis, P. P. (2007). Estimation of safety effectiveness of changes in shoulder width with case control and cohort methods. Transportation Research Record, 2019(1), 237-245.
-Hilton, M. H. (1973). Some case studies of highway bridges involved in accidents. Highway Research Record, 432,
41-51.
-Islam, M. R., Jenny, I. J., Nayon, M., Islam, M. R., Amiruzzaman, M., & Abdullah-Al-Wadud, M. (2021, August). Clustering algorithms to analyze the road traffic crashes. In 2021 International Conference on Science & Contemporary Technologies (ICSCT) (pp. 1-6). IEEE.
-Jima, D., & Sipos, T. (2022). The impact of road geometric formation on traffic crash and its severity level. Sustainability, 14(14), 8475.
-Khalilzadeh, M., Zoghi, H., & Kazemi, A. (2013). The Study of the Identification and Prioritization Methods of the Black Spots Correction (Full-text in Persian). RAHVAR Journal of Applied Research Bureau Traffic Police 9: 27-54.
-Khosravi, Y., Hosseinali, F., & Adresi, M. (2022). Assessing Road Accidents in Spatial Context via Statistical and Nonstatistical Approaches to Detect Road Accident Hotspot Using GIS. Geodetski vestnik, 66(3), 412-431.
-Kronprasert, N., Boontan, K., & Kanha, P. (2021). Crash prediction models for horizontal curve segments on two-lane rural roads in Thailand. Sustainability, 13(16), 9011.
-Lateef, A. A. A., Al-Janabi, S., & Al-Khateeb, B. (2019). Survey on intrusion detection systems based on deep learning. Periodicals of Engineering and Natural Sciences, 7(3),1074-1095.
-LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436-444.
-Macedo, M. R., Maia, M. L., Rabbani, E. R. K., Neto, O. C. L., & Andrade, M. (2022). Traffic accident prediction model for rural highways in Pernambuco. Case Studies on Transport Policy, 10(1), 278-286.
-Mia, M. M. A., Biswas, S. K., Urmi, M. C., & Siddique, A. (2015). An algorithm for training multilayer perceptron (MLP) for Image reconstruction using neural network without overfitting. International Journal of Scientific & Technology Research, 4(02), 271-275.
-Michie, J. D. (1980). Strategy for selection of bridges for safety improvement. Transportation Research Record, 757, 17-22.
-Mohammed, H. (2013). The influence of road geometric design elements on highway safety. International Journal of Civil Engineering and Technology, 4(4), 146-162.
-Moomen, M., Rezapour, M., Raja, M. N., & Ksaibati, K. (2020). Predicting injury severity and crash frequency: Insights into the impacts of geometric variables on downgrade crashes in Wyoming. Journal of Traffic and Transportation Engineering (English Edition), 7(3), 375-383.
-Nair, S. R., & Bhavathrathan, B. K. (2022). Hybrid segmentation approach to identify crash susceptible locations in large road networks. Safety Science, 145, 105515.
-Narkhede, S. (2018). Understanding confusion matrix. Towards Data Science, 180(1), 1-12.
-Transportation Officials. (2011). A Policy on Geometric Design of Highways and Streets, 2011. AASHTO.
-Rezapour, M., Nazneen, S., & Ksaibati, K. (2020). Application of deep learning techniques in predicting motorcycle crash severity. Engineering Reports, 2(7), e12175.
-Rosenblatt, F. (1958). The perceptron: a probabilistic model for information storage and organization in the brain. Psychological review, 65(6), 386.
-Sahaf, A., Mohammadi, M., & Abdoli, A. (2021). 3D Sight Distance Calculation and Estimation of its Effect on Road Accidents in GIS Environment. Shock and Vibration, 2021(1), 5298309.
-Sameen, M. I., Pradhan, B., Shafri, H. Z. M., & Hamid, H. B. (2019). Applications of deep learning in severity prediction of traffic accidents. In GCEC 2017: Proceedings of the 1st Global Civil Engineering Conference1,  Springer Singapore, 793-808.
-Schmidhuber, J., & Hochreiter, S. (1997). Long short-term memory. Neural Comput, 9(8), 1735-1780.
-Shariff, S. R., Maad, H. A., Halim, N. N. A., & Derasit, Z. (2018). Determining hotspots of road accidents using spatial analysis. Indones. J. Electr. Eng. Comput. Sci, 9(1), 146-151.
-Steenberghen, T., Dufays, T., Thomas, I., & Flahaut, B. (2004). Intra-urban location and clustering of road accidents using GIS: a Belgian example. International Journal of Geographical Information Science, 18(2),169-181.
-Tamburri, T. N., & Smith, R. N. (1970). The safety index: A method of evaluating and rating safety benefits. Highway Research Record, (332).
-Tharwat, A. (2021). Classification assessment methods. Applied computing and informatics, 17(1), 168-192.
-Vujović, Ž. (2021). Classification model evaluation metrics. International Journal of Advanced Computer Science and Applications, 12(6), 599-606.
-Wenqi, L., Dongyu, L., & Menghua, Y. (2017, September). A model of traffic accident prediction based on convolutional neural network. In 2017 2nd IEEE international conference on intelligent transportation engineering (ICITE), IEEE. 198-202.
-WHO (2018). World health statistics 2018: monitoring health for the SDGs, sustainable development goals: World Health Organization.
-Xu, M., & Zhang, J. (2024). MGL2Rank: Learning to rank the importance of nodes in road networks based on multi-graph fusion. Information Sciences, 667, 120472.
-Yu, S., Wu, Y., Li, W., Song, Z., & Zeng, W. (2017). A model for fine-grained vehicle classification based on deep learning. Neurocomputing, 257, 97-103.
-Zegeer, C. V., Stewart, R., Council, F., & Neuman, T. R. (1994). Accident relationships of roadway width on low-volume roads. Transportation Research Record,160-161.