The Application Of Multi-sensor Information Fusion Technology On Fire Detection

Fire is a combustion process that lost control, and it is a serious threat of human lifeand property security. Therefore, how to improve the performance of the fire detection system,to realize the early forecast fire, timely and accurate alarm has always been the focus ofresearch. The fire detection of traditional single sensor type is directed to one of the numerousfie parameter, adopt the simple threshold algorithm to determine the occurrence of fire.However, due to the randomness and uncertainly of the fire signal, the single parameterdetection is easy to cause the alarm omission and false. So, in order to achieve the rapidityand accuracy of the fire alarm, the multi-sensor fire detection technology based on artificialintelligence technology has became one of the hot research areas of fire detection.In this article, the information fusion of the fire detection is divided into three levels inthis detection algorithm: information layer, feature layer and decision layer. Information layeris mainly for the original data collection and the data preprocessing. First, the data oftemperature and smoke concentration and co concentration and infrared signal collected bythe detector are preprocessed, then though the local decision judgment, to extract the featurethe same group of fire parameter and sent to the feature layer if appearing to have abnormalsignal. Feature layer is mainly to have the extraction of the fire feature parameter for featurefusion. Using BP neural network feature fusion to recognize the three kinds of fire:smoldering fire and light fire and non-fire source interference, then drawing the conclusion oftheir respective probability of occurrence. Decision layer is mainly to make the final decisionoutput of the fire detection system. The article though the fire risk and fire damage todetermine the building fire protection level, and using them as the indirect criterion indecision layer, making the time of fire signal duration and the probability of light fire andsmoldering fire as the direct criterion, to proceed with the decision-making fuzzy reasoning,obtaining the fire alarm decision output divided into four levels. The fire feature parametersdetected by the fire detector as a direct criterion, and the fire protection as indirect criterion inthe algorithm, so the accuracy of the detection system is improved. Finally making the firecontrol and fire alarm level with associated, then selecting the scheme of the fire alarm andfire-fighting linkage control system.Through the simulation of the Chinese standard light fire SH4, smoldering fire SH1,and the kitchen environment of typical interference data, to verify that the fire detectionmethod based on the information fusion technology is feasible and effective.