Comparison of Time-Frequency Classification Methods for Intelligent Automatic Jettisoning Device of Helmet- Mounted Display Systems

Helmet-Mounted Display Systems (HMDS) improve the situational awareness of an air force pilot in combat; however, they can increase the probability of neck injury to a pilot during a crash or ejection due to their added weight and center of gravity shift. Attempts with simple mechanical force/acceleration release systems to release the HMDS during an ejection event have been unsatisfactory since helmet accelerations during normal air combat maneuvering (ACM) can be near peak accelerations seen during a crash or ejection. The HMDS acceleration responses indicated being non-stationary signals, which resulted in use of time-frequency classification methods to differentiate between the crash/ejection events and those of normal aircraft maneuvering. Parametric and non-parametric approaches for the optimization of the time-frequency representations (TFR) with the goal of classifications between these two environments were compared. The non-parametric approach was determined to be superior.