In this study, the effectiveness and feasibility of a flame detector using a 4-wavelength infrared sensor that can compensate for the non-fire alarm disadvantages of the existing infrared flame detector was analyzed. The proposed flame detector adds an additional infrared sensor to the existing flame detector that uses two or three infrared sensors. The information from the additional sensor was used to determine non-fire alarms. To determine the fire/non-fire, wavelength information of the flame emitted from the fire source is required. To obtain this wavelength information, various fire sources, such as halogen lamps, spectrum analysis, fan heaters, and arc welding, were analyzed. To analyze the wavelength characteristics of each fire source, spectrum analysis (fast fourier transform, FFT) of wavelengths was performed in the central processing unit (CPU) of the flame detector, and the intensity of each wavelength was measured. Additionally, two wavelengths (3.95 μm and 5.3 μm) were used to discriminate non-fire alarms, and the parameters for judging fire/non-fire more accurately were derived using the information on these two wavelengths. To confirm the operation characteristics of the 4-wavelength infrared sensor flame detector, a fire test was performed using n-heptane. The results of the experiment indicate that the fire signal was generated accurately in the fire source experiment using n-heptane, and the non-fire alarm could be accurately identified in the non-fire source experiment of halogen lamps, fan heaters, and arc welding.
Cornell Spectrum Imager: Open Source Spectrum Analysis with ImageJ
