Classification of Gases and Concentration Levels Obtained from Sensor Array Detection as Electronic Nose

Abstract

The sensor array system implemented to detect carbon monoxide, ethanol, ethylene, and methane gases has been successfully classified. This gas detection system is also known as an electronic nose (e-nose). The combination of sensors used to detect these gases uses TGS2602, TGS2610, TGS2611, and TGS2612, where there are two types of each of these sensors combined. This study aims to investigate the classification of response profiles of gas sensor array to analyte molecules and their concentration levels using the classification method and compares these classification methods. There are five classification methods used, namely K-Nearest Neighbor (K- NN), Decision Tree (D-Tree), Random Forest, Artificial Neural Network Multi-Layer Perceptron (ANN-MLP), and Artificial Neural Network Single Layer Perceptron (ANN-SLP). The results obtained from this study indicate that the sensor array classification system's superior accuracy value can produce 100% on the K-NN method. Not only that, but the value of precision, recall, and f1-score for the classification method of the implemented K-NN algorithm also gets a value of 100%.