| This research attempts to develop a method for simulation of the motion of an anthropomorphic test device (a crash dummy) using accelerometer data taken during an automotive vehicle crash test. This research is focused on improvement of the determination of the full dummy motions during an automobile crash test, for the purpose of clarifying the motions of parts which may be masked in the filmstrip taken during the test.;Typically, in previous studies, in automobile crash tests, accelerometers and amplifiers have not responded well at low frequency and have generated significant errors after impacts. When double integrating to get displacements from accelerations, errors from accelerometer signals have accumulated and led to poor final results in motion simulation.;The simulation using multiple accelerometers without simulating mechanical connections in certain parts of the dummy, such as the neck and lumbar spine, but depending on an overabundance of redundant accelerometer data to supply correct motion simulation on average. The simulation results show that the head, chest and pelvis separate markedly from each other during the crash event. This is explained by inaccurate accelerometer signals due to roll off of the amplifiers at low frequency.;Once the problems in the data have been recognized by the author, the research has focused on imposing simulated constraints between the parts of the Hybrid III dummy to overcome or, at least, minimize, the problem of the faulty data. For example, a universal joint and a spherical joint at the two ends of a fictitious rigid bar are used to maintain a constant distance between the centers of the head and chest without restricting rotational motions. A similar arrangement is used to constrain the distance between the chest and pelvis, in place of the lumbar spine. Through extensive trials to implement appropriate mechanical constraints for each part of dummy, the simulation results are improved, and all dummy parts now remain connected during the simulated crash event. |
