Low-Complexity Polynomial Regression Framework for Environmental Drift Compensation in MOX Gas Sensing Systems

Abstract Metal-oxide semiconductor (MOX) gas sensors are critical components in low-cost environmental monitoring architectures; however, their cross-sensitivity to ambient temperature and relative humidity induces severe baseline drift, compromising measurement accuracy and long-term stability. Therefore, accurately modelling the nonlinear influence of ambient temperature and relative humidity on sensor resistance remains a significant challenge for environmental drift compensation. This study proposes a structured polynomial mathematical framework to optimize and implement drift compensation across three distinct commercial MOX sensors (MQ4, MQ136, and TGS2442). The proposed approach systematically evaluates and compares the performance of six higher-order polynomial regression formulations (M1-M6) incorporating varying degrees of interaction and non-linear terms. Model structures were comprehensively evaluated using 10-fold cross-validation to establish the optimal balance between mathematical complexity and prediction accuracy. Cross-validation results show that increasing model complexity yields significant error reduction initially, leading to the selection of Polynomial Model M4 as representative models that provide the best trade-off between prediction accuracy and mathematical complexity. Compared with the selected linear regression model, M4 reduced the cross-validation mean squared error (MSE) by 18.5% for the MQ4 sensor, 46.5% for the MQ136 sensor, and 25.0% for the pulse-heated TGS2442 sensor. External verification further demonstrated the effectiveness of the selected polynomial model on an independent dataset, where the standard deviation of the corrected sensor responses was reduced by 71.1% for MQ4, 57.2% for MQ136, and 15.5% for TGS2442, indicating improved measurement stability under previously unseen operating conditions. These findings demonstrate that the proposed polynomial regression framework provides a mathematically transparent, computationally efficient, and low-overhead drift compensation solution suitable for gas sensing applications.

Authors

Institutions

Publication Details

Journal
Sensing and Imaging
Published
2026-11-13
DOI
https://doi.org/10.1007/s11220-026-00917-1
Primary Topic
Gas Sensing Nanomaterials and Sensors
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Low-Complexity Polynomial Regression Framework for Environmental Drift Compensation in MOX Gas Sensing Systems

Md Ashequl Islam, Ammar Zakaria, Kamarulzaman Kamarudin, Abdulnasser Nabil Abdullah et al.
1 links
Sensing and Imaging
Gas Sensing Nanomaterials and Sensors
article

Low-Complexity Polynomial Regression Framework for Environmental Drift Compensation in MOX Gas Sensing Systems

Md Ashequl Islam, Ammar Zakaria, Kamarulzaman Kamarudin, Abdulnasser Nabil Abdullah, Syed Muhammad Mamduh, Ahmad Syamil
article en

Abstract

Abstract Metal-oxide semiconductor (MOX) gas sensors are critical components in low-cost environmental monitoring architectures; however, their cross-sensitivity to ambient temperature and relative humidity induces severe baseline drift, compromising measurement accuracy and long-term stability. Therefore, accurately modelling the nonlinear influence of ambient temperature and relative humidity on sensor resistance remains a significant challenge for environmental drift compensation. This study proposes a structured polynomial mathematical framework to optimize and implement drift compensation across three distinct commercial MOX sensors (MQ4, MQ136, and TGS2442). The proposed approach systematically evaluates and compares the performance of six higher-order polynomial regression formulations (M1-M6) incorporating varying degrees of interaction and non-linear terms. Model structures were comprehensively evaluated using 10-fold cross-validation to establish the optimal balance between mathematical complexity and prediction accuracy. Cross-validation results show that increasing model complexity yields significant error reduction initially, leading to the selection of Polynomial Model M4 as representative models that provide the best trade-off between prediction accuracy and mathematical complexity. Compared with the selected linear regression model, M4 reduced the cross-validation mean squared error (MSE) by 18.5% for the MQ4 sensor, 46.5% for the MQ136 sensor, and 25.0% for the pulse-heated TGS2442 sensor. External verification further demonstrated the effectiveness of the selected polynomial model on an independent dataset, where the standard deviation of the corrected sensor responses was reduced by 71.1% for MQ4, 57.2% for MQ136, and 15.5% for TGS2442, indicating improved measurement stability under previously unseen operating conditions. These findings demonstrate that the proposed polynomial regression framework provides a mathematically transparent, computationally efficient, and low-overhead drift compensation solution suitable for gas sensing applications.

Sensing and ImagingVol. 27(1)
Adelaide University (AU), The University of Adelaide (AU), Universiti Malaysia Perlis (MY)
Gas Sensing Nanomaterials and Sensors
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon