The enormous diversity of passenger car models on the market and the conquest of always new niches lead to a variety of most different vehicle front-end-structures. As a result, it becomes more and more difficult to categorize them clearly. This study develops a method to objectively classify vehicle models according to their front-end-structure. Therefore, all available models involved in frontal pedestrian accidents in the GIDAS (German In-Depth Accident study) database are measured and categorized by a statistical algorithm. It was the aim of the study to develop an objective and statistically proper method for the classification of vehicle front-end shapes. Therefore, vehicles involved in frontal pedestrian accidents in the GIDAS database were analyzed. After the geometrical measurement of 189 CAD models, they were classified into five groups using a cluster analysis. After that, the calculated clusters were statistically described and compared, leading to the result that the formation of logical and at the same time statistically comprehensible groups is possible. The classification can be repeated with more or less models at any time. The crucial advantage of this method is the fact that there are no visually based algorithms used for the classification and that there is no subjective evaluation involved as it has been in former classifications. The use of the cluster analysis would also be possible for further geometrically based analyses (visibility relations, vehicle classifications etc.). In addition, a three dimensional evaluation of front-end shapes would be possible.
Objective classification of front-end structures on the basis of impact-related geometric measures in frontal pedestrian accidents
2010
21 Seiten, 26 Bilder
Aufsatz (Konferenz)
Datenträger
Englisch
NTIS | 1980
|Impact phase in frontal vehicle-pedestrian collisions
Springer Verlag | 2016
|Impact phase in frontal vehicle-pedestrian collisions
Online Contents | 2016
|Impact phase in frontal vehicle-pedestrian collisions
British Library Online Contents | 2016
|