Precise Dummy Head Trajectories in Crash Tests based on Fusion of Optical and Electrical Data: Influence of Sensor Errors and Initial Values Graz University of Technology

Format:

Conference/Event

Author/Creator:

Sinz, Sinz, author.

Contributor:

Greimel, Robert

Klein, Christoph

Middendorff, Class

Moser, Jörg

Raguse, Karsten

Steiner, Christina

Conference Name:

SAE 2015 World Congress & Exhibition (2015-04-21 : Detroit, Michigan, United States)

Language:

English

Physical Description:

1 online resource

Place of Publication:

Warrendale, PA SAE International 2015

Summary:

AbstractPrecise three-dimensional dummy head trajectories during crash tests are very important for vehicle safety development.To determine precise trajectories with a standard deviation of approximately 5 millimeters, three-dimensional video analysis is an approved method. Therefore the tracked body is to be seen on at least two cameras during the whole crash term, which is often not given (e.g. head dips into the airbag).This non-continuity problem of video analysis is surmounted by numerical integration of differential un-interrupted electrical rotation and acceleration sensor signals mounted into the tracked body.Problems of this approach are unknown sensor calibration errors and unknown initial conditions, which result in trajectory deviations above 10 centimeters.To address these non-continuity and gauging problems a data fusion method (SimbaV) has been developed combining the advantages of video and electrical measurements to yield non-interrupted trajectories with standard deviations less than 1 centimeter. Numerical optimization is employed to transform video information into best-fit maximum-likelihood estimates of unknown electrical parameters. These parameters are then in turn used to predict the trajectory by integration of differential equations of motion in time regions where no video data is available.The principal methods of the simulation-based test analysis (SimbaV) are explained. Furthermore, the main influence parameters of the determination process are identified and their influences on the results are shown. It is possible to correlate sensitive sensor errors to car types and test configurations to reduce optimization parameters. This leads to more reliable results and the risk of a false optimization result is reduced significantly

Notes:

Vendor supplied data

Publisher Number:

2015-01-1442

Access Restriction:

Restricted for use by site license