Modern signal processing for temperature fluctuation based fire detectors

Abstract

We suggest use of the high frequency part of the temperature signal or the temperature fluctuations in addition to or as a replacement for the common fire detector which is based only on a lowpass filtered (smoothed) temperature signal. Furthermore, a signal processing method distinguishing fire from non-fire situations using fire detectors that measure temperature fluctuations is presented. This method is partly based on the interpretation of the sensor output signals as a linear filtered stochastic process which enabled the reconstruction of the input signals from known output signals. We show the advantage of this method through application to many fire and non-fire situations.