Statistical Decision Theory Methods for Noise and Vibration Engineering

Statistical decision theory provides the noise and vibration engineer with the tools necessary to automate the decision process in practical applications such as automatic classification of acoustic events or fault-detection by vibro-acoustic methods. We review the basic principles of statistical decision theory in a tutorial fashion. We illustrate how these principles have been put to work on two practical systems that have been developed by our laboratory: an intelligent noise monitoring system able to distinguish between the acoustic signatures of different types of noise sources, and a vibro-acoustic non-destructive inspection system for refractory plates. The experimental results obtained prove the validity of the approach.