Abstract:

Aiming at the problems existing in small space fire detection,such as single detection method, high false alarm rate and simple data collection and processing method, a multisensor detection device and method are proposed. The detection device can realize multi- angle or multi- directional collection of fire signals including fire smoke, CO gas, temperature, flame, etc. Using information fusion technology and algorithms, the law between characteristic information is found, to analyze and optimized group the data information. Using the method of artificial neural networks and fuzzy inference, the error of detection datais minimized byrepeated sample training, so as to determine the stage and type of the fire, andtake fire extinguishing measures promptly and effectively. The studies have proved that the multisensor fire detection devices and methods can meet the requirements for fire detection in small spaces, reducing the loss of public property, and have important significance for the further research of fire detection.

Key words: small space, fire detection, multivariate detection, artificial neural network, fuzzy inference, information fusion