The use of an expert system to expand a multi sensor fire detector to a robust fire detector

Abstract

A new fire detector is presented which produces a lower rate of false alarms then a customary one, in rooms that are hard to supervise. The detector is based on the assumption that rooms have a specific use, so that often similar reasons lead to false alarms. Therefore, a customary detector is improved so that it is capable of saving the physical characteristics of non-fire events. When a critical situation arises in the supervised room, it is compared to past non-fire events. When a cause is recognized as a repetitive non-fire cause, the decision is taken in favour of the non-fire hypothesis, otherwise, in favour of the hypothesis.