Artificial olfaction based on tafel curve for quantitative detection of acetone ethanol gas mixture

This work established an artificial olfaction based on a single mixed potential type gas sensor. By identifying the Tafel curve, artificial olfaction can determine the concentration of each component in the acetone and ethanol vapor mixture. To save a long testing circle, a finite element model was built to efficiently output the Tafel curves of the sensor for different concentrations of vapor mixtures. Based on the Tafel curve data obtained from the simulation, different classification algorithms were trained to determine the operating state of the sensor. Among them, eXtreme Gradient Boosting (XGB) had a 99% accuracy in determining whether the sensor is saturated or not. Then three different regression algorithms (XGB, Gradient Boosting Decision Tree (GBDT), and Random Forest (RF)) were trained to identify acetone and ethanol vapor concentrations. The results showed that XGB and GBDT were the most effective in identifying acetone and ethanol vapor. The mean absolute percentage errors for the identified acetone and ethanol in the gas mixture were 11.3% and 23%, respectively. The above results indicated that it was feasible and effective to use Tafel curves to detect VOC gases.

Automated summary

This study established an artificial olfaction system based on a single mixed potential type gas sensor. By analyzing the Tafel curve, the system accurately determined the concentration of acetone and ethanol in a vapor mixture. A finite element model was created to efficiently generate Tafel curves for different vapor concentrations, saving time in the testing process. Various classification and regression algorithms were trained using simulated Tafel curve data, with eXtreme Gradient Boosting (XGB) achieving a 99% accuracy in determining sensor saturation and identifying acetone and ethanol concentrations.