Intelligent Building Fire Detection Based On Multi-sensor Information Fusion

In modern society, with the continuous development of science and technology as well asthe economy, the number of the concentrated areas, such as residential areas and school districtand so on, become larger and larger, which increased the fire damage degree. If fire can bediscovered in the early time and it can be extinguished，people will make the fire loss meet theminimum. Therefore, the role of fire detection technology in building fire system is veryimportant, the research is also very significant.The traditional fire detection system generally only used single detector, using its thresholdfor fire information judgment. However, now the building is getting more and more complex,such that the environment variables as well as the interference of signal is increasing, a singledetector can not reliably detect fire information. Therefore, in order to improve the accuracy ofthe fire detector to detect the information, many researchers are developing detector to replace thetraditional fire detectors to detect fire information.This article combines basic characteristics of fire detection technology to establish the modelof fire detection system based on multi-sensor information fusion,The whole system detectionalgorithm is divided into three layers, pixel layer, feature level and decision level. In the pixellayer, detector mainly collect and pre-process the data of CO gas, temperature, smoke and othercharacteristic parameters;In the feature layer, the information from pixel layer using artificialneural network and linear transformation can be merged, it achieve the target recognition ofsmoldering fires, fire, non-fire source identification as well as the fire probability which providedirect criterion and indirect criterion for decision-making layer; In the decision-making layer,according to information from the feature layer,the layer do some analysis and judgment, wheninformation provided by the feature layer that can determine the fire situation, it will be directlyoutput, conversely,it will use fuzzy logic reasoning judgement technology to make pixel layerdata fuse again to get the decision results. In the study of the text, Using MATLAB simulationsoftware to simulate three kinds of fire information respectively from kitchen interference signal,our country standard fire smoldering fires and standard open flame. In the characteristic layersimulation study, it shows error curve of different neural network implicit layers,whichestablished the number of hidden layer of BP neural network is7, and in the decision-makingsimulation study, the set of fuzzy rules draw the final output results. The simulation results showthat fire detection system based on the information fusion technology has the feasibility.