Active gas/odor sensing system using automatically controlled gas blender and numerical optimization technique

摘要: Abstract As measurement of a vapor mixture composition is difficult technique, no method using sensing system has yet been established in spite great effort by many researchers. In this paper, the authors propose new gas/odor gas blender and nonlinear numerical optimization algorithm which concentration each component an unknown can be quantified. The vapors are internally blended ratio modified so that sensor array output pattern made equal to one. After several iterations, convergence obtained determined from vapor. Although conventional passive, considered as active one it performs exploratory behavior prior recognition. Here, gasoline measured under condition or two interference exist together. Gasoline adopted example odors passenger compartment car, since sometimes smells unpleasant. essential for designing car order keep comfortable passengers. sensors used here three semiconductor electrochemical sensors, chosen obtain high sensitivity gasoline. techniques simplex gradient descent these methods compared here. It found quantification error within ten ppm two- three-component vapors.