Methodology for estimating ethanol concentration with artificial intelligence in the presence of interfering gases and measurement delay

Abstract

Gas concentration detection is a vital aspect of environmental monitoring, industrial safety, and various other applications. Metal Oxide (MOX) sensors have gained considerable attention in this context due to their low cost and ability to detect several gases, although their selectivity is a potential drawback. In this paper, a data-driven approach for the detection and estimation of ethanol concentration using MOX sensors in the presence of interfering gases is presented, with a dual emphasis on the impact of delay between gas transmission and sensor detection and the incorporation of heater current and voltage, alongside sensor current readings. A delay in sensor response can lead to erroneous readings and potentially compromise safety and environmental assessments. To tackle this issue, an incremental method for the identification of the delay length is proposed, and its effectiveness in improving the estimations of gas concentration is demonstrated by implementing three distinct regression techniques: Linear Regression (LR), Support Vector Regression (SVR), and Random Forest (RF). The analysis is then extended to incorporate the utilization of heater current and voltage, alongside sensor current readings. The experimental results demonstrate the effectiveness of the proposed method in handling measurement delay processing and its robustness to the presence of interfering gases.