Improving the Chemical Selectivity of an Electronic Nose to TNT, DNT and RDX Using Machine Learning.

Anton Gradišek,Mario Trifkovič,Drago Strle,Vid Prezelj,Matija Koterle,Marion Van Midden,Ivan Kvasić,Bogdan Štefane,Igor Muševič,Helena Brodnik,Erik Zupanič

doi:10.3390/s19235207

Anton Gradišek, Mario Trifkovič + Show 9 more

Open Access

https://doi.org/10.3390/s19235207

Copy DOI

Abstract

We used a 16-channel e-nose demonstrator based on micro-capacitive sensors with functionalized surfaces to measure the response of 30 different sensors to the vapours from 11 different substances, including the explosives 1,3,5-trinitro-1,3,5-triazinane (RDX), 1-methyl-2,4-dinitrobenzene (DNT) and 2-methyl-1,3,5-trinitrobenzene (TNT). A classification model was developed using the Random Forest machine-learning algorithm and trained the models on a set of signals, where the concentration and flow of a selected single vapour were varied independently. It is demonstrated that our classification models are successful in recognizing the signal pattern of different sets of substances. An excellent accuracy of 96% was achieved for identifying the explosives from among the other substances. These experiments clearly demonstrate that the silane monolayers used in our sensors as receptor layers are particularly well suited to selecting and recognizing TNT and similar types of explosives from among other substances.

Highlights

The past couple of years have seen an increase in the amount of research on artificial noses for detecting targeted substances in the atmosphere
The goal of this study is to apply the methods of artificial intelligence to an existing e-nose demonstrator, which is described in our earlier publication [2]
In this paper we demonstrated that we can apply relatively simple machine-learning models to substances, where where we are assist in the classification of sensor-array responses to various chemical substances, independently varying both the concentration of the target substance in the nitrogen carrier gas and the gas flow

Summary

Introduction

The past couple of years have seen an increase in the amount of research on artificial noses for detecting targeted substances in the atmosphere. While the first generations of sensors were optimized to respond to a particular substance and were designed to detect it within a certain concentration range, sensor sensitivity is not a major problem anymore. Because these sensors can have the same electrical response to many different targeted substances, the necessary chemical selectivity can be hard to achieve. Research is shifting towards arrays of an increasing number of sensors that are able to distinguish between a series of different substances in various concentration ranges, as a dog’s nose would. The applications are in the food and beverage industry [3,4], for detecting fruit aromas or determining the ripening status [5,6], controlling the quality of vegetable oil [7,8], classifying different types of wines [9] or teas [10] and in detecting spoilage due to Sensors 2019, 19, 5207; doi:10.3390/s19235207 www.mdpi.com/journal/sensors

Objectives

Methods

Results

Discussion

Conclusion