Using Artificial Neural Network to Predict Collisions on Horizontal Tangents of 3D Two-Lane Highways

Authors: Omer F. Cansiz, Said M. Easa

Abstract:

The purpose of this study is mainly to predict collision frequency on the horizontal tangents combined with vertical curves using artificial neural network methods. The proposed ANN models are compared with existing regression models. First, the variables that affect collision frequency were investigated. It was found that only the annual average daily traffic, section length, access density, the rate of vertical curvature, smaller curve radius before and after the tangent were statistically significant according to related combinations. Second, three statistical models (negative binomial, zero inflated Poisson and zero inflated negative binomial) were developed using the significant variables for three alignment combinations. Third, ANN models are developed by applying the same variables for each combination. The results clearly show that the ANN models have the lowest mean square error value than those of the statistical models. Similarly, the AIC values of the ANN models are smaller to those of the regression models for all the combinations. Consequently, the ANN models have better statistical performances than statistical models for estimating collision frequency. The ANN models presented in this paper are recommended for evaluating the safety impacts 3D alignment elements on horizontal tangents.

Keywords: Collision frequency, horizontal tangent, 3D two-lane highway, negative binomial, zero inflated Poisson, artificial neural network.

Digital Object Identifier (DOI): doi.org/10.5281/zenodo.1057261

References:

[1] American Association of State Highway Transportation Officials. A Policy on Geometric Design of Highways and Streets. Washington, D.C., 2004.
[2] Transportation Association of Canada. Geometric Design Guide for Canadian Roads. TAC, Ottawa, Ontario, 2007.
[3] Canadian Council of Motor Transport Administrators. 2005 Annual Report - Road Safety Vision2010. 2005. www.tc.gc.ca/roadsafety/vision/menu.htm. Accessed July 20, 2007.
[4] Easa, S.M. and Q.C. You. Collision Prediction Models for Three- Dimensional Two-Lane Highway Alignments: I. Horizontal Curves. Transportation Research Record, Journal of Transportation Research Board, 2092, 2009, pp. 48−56.
[5] You, Q.C. and Easa, S.M. Collision Prediction Models for Three- Dimensional Two-Lane Highway Alignments: II. Horizontal Tangents. Presented at the 88th Transportation Research Board Annual Conference, Paper 09-3558, 2009.
[6] Lamm, R., B. Psarianos, and T. Mailaender. Highway Design and Traffic Safety Engineering Handbook. McGraw-Hill, New York, 1999.
[7] Fink, K. L. and R.A. Krammes. Tangent Length and Sight Distance Effects on Accident Rates at Horizontal Curves on Rural Two-Lane Roads. Transportation Research Record: Journal of the Transportation Research Board, No. 1500, 1995, pp. 162−167.
[8] Brenac, T. Safety at Curves and Road Geometry Standards in Some European countries. Transportation Research Record: Journal of the Transportation Research Board, No. 1523, 1996, pp. 99−106.
[9] Hauer, E. Road Grade and Safety. University of Toronto, Toronto, Ontario, 2001. www.roadsafetyresearch.com. Accessed July 8, 2007.
[10] Zador, P., H. Stein, J. Hall, and P. Wright. Relationship between Vertical and Horizontal Roadway Alignments and the Incidence of Fatal Rollover Crashes in New Mexico and Georgia. Transportation Research Record: Journal of the Transportation Research Board, No. 1111, 1987, pp. 27−41.
[11] Harwood, D., F. Council, E. Hauer, W. Hughes, and A. Vogt. Prediction of the Expected Safety Performance of Rural Two-Lane Highways. Publication FHWA-RD-99-207. FHWA, U.S. Department of Transportation. www.tfhrc.gov/safety/99207.htm. Accessed Dec. 18, 2006.
[12] Easa, S.M. and A. Mehmood. Optimizing Design of Highway Horizontal Alignments: New Substantive Safety Approach. International Journal of Computer-Aided Civil and Infrastructure Engineering, 23(7), 2008, pp. 560-573.
[13] SAS Institute Inc. SAS/STAT® 9.1 User-s Guide. SAS Institute Inc, Cary, North Carolina, 2004.
[14] Vuong, Q. Likelihood Ratio Tests for Model Selection and Non-Nested Hypotheses. Econometrica, 57(2), 1989, pp. 307−333.
[15] Shankar, V., J. Milton, and F.L. Mannering. Modeling Accident Frequency as Zero-Altered Probability Processes: an Empirical Inquiry. Accident and Analysis Prevention, 29(6), 1997, pp. 829-837.
[16] Cybenko, G., 1989. Approximation by Superpositions of Sigmoidal Function. Mathematics of Control, Signals, and Systems, 2, pp. 303−314.
[17] Hornik, K., Stinchcombe, M., White, H., 1989. Multilayer feedforward networks are universal approximators. Neural Networks, 2, pp. 359−366.
[18] MATLAB, Neural Network Toolbox. The MathWorks Inc., Natick, MA.
[19] Cansiz, O.F. Improvements in the Forecast Model of Vehicle Crashes Deaths Formed by Artificial Neural Network. Simulation: Transactions of the Society for Modeling and Simulation International, doi:10.1177/0037549710370842. First published on May 26, 2010.