Search Results

Scalograms based on shift-invariant orthonormal wavelet transforms can be used to analyze impulsive and transient sounds in the presence of more stationary sound backgrounds, such as wind noise or drivetrain noise. The visual threshold of detection for impulsive features on the scalogram (signal energy content vs. time and frequency,) is shown to be similar to the audible threshold of detection of the human auditory system for the corresponding impulsive sounds. Two examples of impulsive sounds in a realistic automotive sound background are presented: automotive interior rattle in a vehicle passenger compartment, and spark knock recorded in an engine compartment.

Recent applied mathematics research on the properties of the invertible shift-invariant discrete wavelet transform has produced new ways to visualize, separate, and synthesize impulsive sounds, such as thuds, slaps, taps, knocks, and rattles. These new methods can be used to examine the joint time-frequency characteristics of a sound, to select individual components based on their time-frequency localization, to quantify the components, and to synthesize new sounds from the selected components. The new tools will be presented in a non-mathematical way illustrated by two real-life sound quality problems, extracting the impulsive components of a windshield wiper sound, and analyzing a door closing-induced rattle.

One important design goal for spark-ignited engines is to minimize cyclic variability. A small amount of cyclic variability (slow burns) can produce undesirable engine vibrations. A larger amount of cyclic variability (incomplete burns) leads to increased hydrocarbon consumption/emissions. Recent studies have reported deterministic patterns in cyclic variability under extremely lean (misfiring) operating conditions. The present work is directed toward more realistic non-misfiring conditions. Production engine test results suggest that deterministic patterns in cyclic variability are the consequence of incomplete combustion, hence control algorithms based on the occurrence of these patterns are not expected to be of significant practical value.