Nonlinear Modeling of Voltammetric Sensor Signals: Application to the E-Tongue Measurement

The phenomena associated with electrochemical processes contribute to considerable nonlinearity on the response of the voltammetric sensor system. Motivated by the lack of nonlinear models of voltammetric systems, this paper reports an improved model that incorporates the system nonlinearity. The identified nonlinear model based on Hammerstein-Wiener structure is found to produce a model fit up to 90% and shows a definite improvement over the existing models. Appreciable clustering pattern among the model responses for black tea samples further encouraged the development of regression models for estimation of an important taste-related compound in tea. The results establish that the proposed nonlinear model finds a successful application in the area of electronic tongue. Comparison of the prediction performance with those of the existing models and with the experimental response is also carried out.