Нелинейно-регрессионный алгоритм обработки сигналов полупроводниковых химических сенсоров, обеспечивающий селективное детектирование примесей в искусственном воздухе

New method is developed for proceeding a conductance change response Δσ,μS of a temperature (T) modulated chemical sensor. The method provides reliable substance identification and measurement of its trace concentrations for such impurities in arti-ficial air as ammonia, acetone, hexane, propane, toluene, turpentine etc. Due to this method the response of the ΔσY for a substance Y in actual concentration C range is interpolated with a completely nonlinear regression via discriminant modelling func-tions Fi(z=103/T,Ai, bi,ci,..),i=1-4 or 5. For principal parameters the plots AiY(C) are built which compose the selectivity/gauge portrait of Y in the air. In case when the analogous parameters of unknown substance X fit this portrait the substance is iden-tified as Y. And the common abscissa of corresponding crossing points of AiX with the curves AiY(C) indicates the value of the X concentration in units been used for Y.