The problem of pedestrian injury is a significant one throughout the world. In 2001, there were 4724 pedestrian fatalities in Europe and 4882 in the US. Significant advances have been made by automotive safety researchers and vehicle manufacturers to address this issue with respect to the design of vehicles, but the complex nature of pedestrian accident scenarios has resulted in great difficulty when using traditional statistical methods. Specifically, problems have been encountered when attempting to study the effects of individual parameters of vehicle front-end geometry on pedestrian head injury. This paper attempts to demonstrate the feasibility of applying the field of evolutionary computation to the problem of pedestrian safety by using a simple genetic algorithm to optimize the centre-line geometry of a car's front-end for the reduction of pedestrian head and thoracic injury. The fitness of each design is assessed by creating a multi-body mathematical model of the vehicle front and simulating impacts with models of different sized pedestrians, and ranking according to the combined injury scores.

A review of injuries in impact trauma reveals a plethora of ‘injury criteria’, many of which are enshrined in legislation. It is assumed here that injuries can be modelled as mechanical dissipative processes, and the formalism of continuum damage mechanics based on irreversible thermodynamics is applied to impact trauma. It is shown that peak virtual power (PVP) predicts the severity of injury, measured on the abbreviated injury scale, in around 90 per cent of cases for all types of injury to all body regions (brain, skull, thorax, spine, upper and lower extremities) for car occupants from the CCIS and NASS-CDC databases. Consideration of injury to body regions shows that PVP predicts the form of acceleration-based criteria, the head injury criterion and the viscous criterion. It is shown that in general the lower bound of severity of injury is proportional to δ V3 or (ETS)3, where ETS is equivalent test speed, for restrained vehicle occupants, and the upper bound proportional to δ V2 or (ETS)2 for unrestrained occupants. It is concluded that PVP is a suitable candidate for an objective universal injury criterion which can be correlated to real-world injury experience.

This document is the author's post-print version, incorporating any revisions agreed during the peer-review process. Some differences between the published version and this version may remain and you are advised to consult the published version if you wish to cite from it.

No abstract

No abstract