Study On The Fire Detection Of The Highway Tunnel

It's easy to save and control a tunnel fire in the initial stage. It could avoid the spread and reduce the loss of personnel and property, if we can detect the tunnel fire, give the alarm and do some measure as early as possible. So it's important to detect and alarm the tunnel fire in good time.The fire is a combustion process without control, the basic elements of fire are combustible, oxidizer and ignition source. The physics and chemistry phenomena in combustion process could be detected. They include outgrowth of burning, smoke, radiation and heat, so the photoelectric smoke soundex, temperature detection and flame detection could be used. Combined with the characteristics of the tunnel fire, dual wavelength flame detector, fiber heat detector and wiry heat detector of thermal alloy are selected to detect the tunnel fire. And the theory, installation, application and system structure are studied.Based on analyzing the characteristic of tunnel fire and the theory of data fusion, the space-time redundancy and complemented information of sensors are fused according some rules in order to get a consistency explanation and description of the object. A detecting algorithm of tunnel fire, which is based on data fusion technology, is proposed. The input analog signals are provided by smoke, temperature and light sensors. Using artificial neural network and fuzzy logic technology, some sensor signals are fused. And an auto-detection fusion system is designed for tunnel fire. It can shorten the time of alarm and reduce the rate of misinformation. The result shows that the system conquered the disadvantage of using fixed threshold in fire detecting algorithm.To achieve the auto-detection fusion system of tunnel fire, it should mostly be the achievements of algorithm for information layer, character layer and decision layer. The collected signals are pretreated in information layer. Then the pretreated information is fused by feature fusion of neural network in character layer. Thus the tunnel fire can be classed by open flame and smoldering fire. Using fuzzy-logic discursion the final result is obtained through the analysis and judgement at decision layer.