Research on the Extraction Method of Process Information of Early Fire Signature
The current fire detection technologies usually monitor the smoke and heat using smog-sensitive and thermo-sensitive detectors, and make a fire alarm by comparing the fire variables sampled instantaneously to a given absolute threshold. It is difficult for traditional methods to alarm fire fleetly and reliably because smoke and heat always generate in the late stages of a fire and transient sampled values of fire variables are often affected by some stochastic errors. In this paper, a new fire diagnosis technology was put forward based on process information of early fire signature. The concentration of carbon monoxide in detected environment was sampled continuously by using NEXUS Fourier transformation infrared spectrograph, the fire process eigenvector was made up of CO concentration as well as its first derivative and second derivative, the fire criterion was formulated by the offset distance from the locus of process eigenveetor to the grid origin on the phase space of process characteristic. This technology distinguishes real fire source from non-fire nuisance sources according to the criterion in a given period of time instead of transient fire variables sampled like traditional methods. The result of experiments has shown that the presented technology is feasible for early fire detection. On the condition that the critical offset is from 3 to I0, the presented method can alarm fire much early than any other traditional method and distinguish real fire from non-fires nuisance sources reliably.
early fire signature process information eigenvector early fire detection
ZHANG Rencheng DU Jianhua QIU Hualan
College of Mechanical Engineering and Automation,Huaqiao University,Quanzhou 362021,Fujian,China
国际会议
The 2008 International Symposium on Safety Science and Technology(2008年安全科学技术国际会议)
北京
英文
754-758
2008-09-24（万方平台首次上网日期，不代表论文的发表时间）