Real Time Implementation of a Military Impulse Classifier

A real time military impulse classifier has been developed to distinguish between impulsive events, such as artillery fire, and non-impulsive events, such as wind or aircraft noise. The classifier operates using an artificial neural network (ANN) with four scalar metrics as inputs. This classifier has been installed into two prototype noise monitoring systems, which are capable of establishing an accurate record of impulse events. This record can be used to assist in processing noise complaints and damage claims. The system continually monitors sound levels with a microphone array and activates when the sound level exceeds a given threshold. Once activated, the system processes the data to determine the classification, as well as the approximate bearing of the event