Common Principal Component Analysis For Drift Compensation Of Gas Sensor Array Data

Abstract

Two Component Correction methods for drift compensation in chemical sensor arrays have been evaluated in terms of classification accuracy. A well established linear method based on computing Principal Components of the reference gas is applied to a dataset, where three different reference gases are examined. The proposed method in this contribution is based on computing a Common Component Principal Analysis, which discovers a variance direction followed by all gasses in feature space. This new method—employing no reference gas—has shown the same performance as the traditional approach with the best fitted reference gas.