Environmental Variation or Instrumental Drift? A Probabilistic Approach to Gas Sensor Drift Modeling and Evaluation

TL;DR

This paper presents a probabilistic model that effectively separates environmental variation from instrumental drift in gas sensors, improving the accuracy of sensor performance evaluation and calibration.

Contribution

It introduces a novel probabilistic approach using importance sampling to distinguish environmental effects from sensor drift in long-term gas sensor data.

Findings

Environmental variation significantly impacts drift evaluation

The probabilistic model improves drift assessment accuracy

Proper accounting of environmental factors enhances sensor calibration

Abstract

Drift is a significant issue that undermines the reliability of gas sensors. This paper introduces a probabilistic model to distinguish between environmental variation and instrumental drift, using low-cost non-dispersive infrared (NDIR) CO2 sensors as a case study. Data from a long-term field experiment is analyzed to evaluate both sensor performance and environmental changes over time. Our approach employs importance sampling to isolate instrumental drift from environmental variation, providing a more accurate assessment of sensor performance. The results show that failing to account for environmental variation can significantly affect the evaluation of sensor drift, leading to improper calibration processes.