The objective of the research presented in this paper was to develop relationships between expected level of safety and traffic exposure for the selected types of rural freeways. A safety performance function is presented which reflect the complex relationship between exposure usually measured in ADT and accident count for a unit of road section or a junction over a unit of time. Within the framework of the research functions were developed for rural freeways with four or six lanes; in the paper only mountainous freeways are presented in detail. Since the relationships explored are complex it should be established by employing statistical modelling techniques. Since multiple linear regression modelling is not suited and General Linear Models with Poisson or Negative-Binomial assumptions only provides poor to moderate fit to the data relationships in this study were developed using Neural Networks methodology. Results are presented as descriptions of road sections in terms of driver confidence, subjective senses of security together with operating speeds, propensity to make driving errors and the dynamics of the interaction between the roadway environment and the driver behaviour.
Abstract