A Probabilistic Approach for Drift Compensation of Gas Sensor Data

Long-term drift is widely recognized as a key challenge in gas sensing systems, undermining the reliability and accuracy of these systems over time, as the pattern recognition systems lose their capability of performing gas discrimination. In this work, we employ a probabilistic modeling framework to address the drift problem. This perspective yields a novel scheme for drift compensation, which is grounded on probabilistic modelling. The potential usefulness of the proposed approach is demonstrated through an experimental evaluation on the classification task, which reveals consistently higher accuracy and stability when compared to baseline methods. In addition to its robustness, the proposed method also presents several desired attributes such as, computational simplicity, continuous adaptability and suitability for online applications. The experiments were conducted using a dataset acquired in an interval of 7 months, comprising nearly 4000 measurements from three different gases at different concentration levels in an array containing 17 polymeric sensors.