Search Results

The exploration of impact biomechanics via data mining is investigated in this paper. The issues that are particularly pertinent to the use of data miming technology on biomechanics databases are addressed. These issues include (a) relationship between the manikin tests and human tests; (b) extension from lower impact, non-injurious conditions to high impact, injurious conditions; (c) test data versus simulation data; (d) input-output categorization; (e) input-output abstraction and representation; (f) topics for new knowledge discovery; and (g) user scenarios. Technical treatments and considerations are made on the unique characteristics and requirements involved in the biodynamics data mining. They are (a) mixture of classification and numerical prediction; (b) isolated feature space; (c) multiple dependent variables; (d) high dimensionality; (e) algorithm and parameter selection; and (f) scalable data integration and knowledge discovery.

The Air Force Research Laboratory (AFRL) Biodynamics Data Bank (BDB) is a Microsoft Access database containing study information and test data on the response of human volunteers and test dummies to impact accelerations. The BDB currently includes general information describing the objective, test matrix, results, and other pertinent information on eighty-six test programs that were conducted on AFRL in-house impact test facilities. These facilities include a vertical deceleration tower, vertical impact device, horizontal impulse accelerator, and horizontal decelerator. Test data collected from over 6,000 impact tests on these facilities are stored in the BDB. The data include acceleration, force, and displacement time histories and peak values, and test environment parameters. Slow-motion videos of the tests are currently being digitized into AVI files, and will soon be included in the BDB. Also included are anthropometry measurements from over 200 human volunteer test subjects.