Online Chemical Sensor Signal Processing Using Estimation Theory: Quantification of Binary Mixtures of Organic Compounds in the Presence of Linear Baseline Drift and Outliers

Abstract

Compact sensor systems for on-site monitoring of groundwater for trace organic compounds in the liquid phase are currently under development in our laboratories. Potential challenges include sensor baseline drift and the presence of outliers in the data, along with difficulties extracting the contribution of individual BTEX compound (benzene, toluene, ethylbenzene, and xylenes) from the sensor response to mixtures containing multiple chemically similar compounds. As a first step, the approach presented here permits online estimation of analyte concentrations in binary mixtures of BTEX compounds in the presence of linear baseline drift and outliers. This paper investigates a sensor signal-processing approach based on estimation theory, specifically, Kalman filter (KF), extended KF, and discrete low-pass filter. The approach permits online linear baseline drift correction, filtering of outlier points, and estimation of analyte concentration(s) in binary mixtures and single analyte samples, before the sensor response reaches steady state. Sensor signals from mixtures of BTEX compounds were analyzed because these compounds are good indicators of accidental releases of fuel and oil into groundwater. Models were first developed for the sensor response so that estimation theory can be used to obtain the sensor parameters. The baseline-drift correction technique uses KF to perform online linear extrapolation or interpolation. The presented combination of sensor signal-processing techniques was simultaneously tested using actual measured data. Unknown sensor parameters and identification of analytes in samples were obtained within a relatively short period of time (8 min or less for the present sensor system), well before the sensor response reaches equilibrium.