A trend-duration and gradient detector for automatic fire detection

Fire signals from sensors have a trend-duration characteristic, but such disturbance trends last only a short time. Step-shaped signals from fire sensors do not have a large trend value, although they may indicate a rapidly growing fire. Conventional trend algorithms cannot recognize the signal gradient. In this paper, a trend-duration and trend-gradient detector is proposed for automatic fire detection. The detector can distinguish fire signals from noise using a trend-duration algorithm. The fire smoke density (if smoke sensors are used) can be determined with the gradient algorithm. In order to indicate the level of the smoke density, fuzzy logic is applied to the gradient determination.