Detecting periodically varying, pulse-like noise strengths - forming correlation function from signals produced by frequency filtering and threshold value

摘要

The method detects pulse type, cyclic loudness alterations (chatter noise) e.g. from a helicopter, using a threshold value. The loudness levels are converted into loudness signals, after frequency filtering, from which a correlation function is formed in the following steps: Two average loudness signals (L1,L2) are formed with different transmission durations(tau 1,tau 2), where tau 2 is larger than tau 1. Standardised outness signals (L) are formed according to the equation: L =( L1 - L2)/L2. The auto-correlation function (AKF(L)) is formed from the standardised loudness signal (L). A comparison auto-correlation function (AKF(K)) is produced synthetically, on the basis of the knowledge of the max. of the auto-correlation function (AKF(L)) of the standardised loudness signal (L) and a specified periodicity. The difference between the auto-correlation function (AKF(L)) and the comparison auto-correlation function (AKF(K)) is formed, and the existence of the cyclic pulse type loudness alteration is indicated, if this difference is below the specified threshold value (P(H)). USE/ADVANTAGE - Allows detection of pulse type cyclic vol. of loudness in unambiguous way. Initiates alarm, e.g. siren, upon detection of helicopter or salvo of bullets; blocks operation of defensive weapons or sets defence or endangered personnel on alert.