Gray-box modeling of nonideal sensors

Industrial process sensors must often operate in harsh environments that include both uncontrollable elements (e.g. variations in flow rates, temperatures, etc.) and aggressive conditions (e.g. corrosion, erosion, and fouling). This paper considers the problem of modeling these environmental effects on sensor performance. Starting with the first-order linear model often used to describe sensor dynamics, a gray-box philosophy is adopted that incorporates qualitative knowledge of how these phenomena influence steady-state, dynamic and noise characteristics. In addition, it has been observed experimentally that sensor fouling can lead to significant sensor nonlinearity, and this paper also considers the problem of modeling this nonlinear behavior