Drift correction for gas sensors using multivariate methods

Drift is one of the most serious impairments afflicting gas sensors. It can be seen as a gradual change in the sensor response over a long period of t ime when the external conditions an constant. This paper presents a new sim ple drift counteraction method based on PCA and PLS. The basic idea is to r emove the drift direction component from the measurements. The direction of the drift, p, is calculated from measurements for a reference gas. Project ing the sample gas measurements on this vector gives the score vector t. Th e drift component tp(T) can then he removed from the sample gas data, which we call component correction (CC). The method is tested on a data set base d on a reduced factorial design with four gases and a concentration gradien t of hydrogen. It is found that the method works efficiently for both cases . Copyright (C) 2000 John Wiley & Sons, Ltd.