Research on Data Fusion System of Fire Detection Based on Neural-network

There are some serious problems in traditional fire detection methods and technology, such as lower grade intelligence, high mis-warming rate, delay-warming etc. In this paper, multi-information parameter in the early fire was regarded as early information of fire. This information was trained by neural network technology. Result would be gain with data fusion technology. And the result of data fusion could be confirmed as bases of fire and apply in the fire detection. A 3-layers data fusion structure is used in the neural network. In this fire detecting system, the fire experiential characteristic and the fire data-fitting characteristics of fire signal data are used by the fuzzy inference system to get the last fire probability. Finally simulation experiments are done for typical flaming fire, typical smoldering fire and the fire under typical disturbance signals´ in environment. And satisfactory results are obtained.