We use a few cookies

Essential cookies keep the library working. With your permission, we'd also like to use analytics to see what's useful and improve the experience. Read our Privacy Policy or Cookie Policy.

Skip to main content
Home/e-Resources/Electrical Engineering/APPLICATION OF MACHINE LEARNING ALGORITHMS IN ELECTRONIC NOSE TECHNOLOGY
PeriodicalsElectrical Engineering

APPLICATION OF MACHINE LEARNING ALGORITHMS IN ELECTRONIC NOSE TECHNOLOGY

Checking your access to this record.

Abstract & Overview

Electronic nose (e-nose) technology is considered the main tool for gas identification and determination of its concentration. The artificial olfactory system includes three main components: a gas sensor or sensor matrix, receiving and pre-processing the signal device, and a pattern recognition algorithm. Gas identification is usually carried out using various methods of machine learning or deep learning. The simple and portable electronic nose systems that provide identification of the greatest possible number of gases with satisfactory accuracy and can be deployed on the low-power microcomputer and microcontroller bases are particularly promising for application in sensor electronics, IoT, robotics, etc. Therefore, the purpose of the work is to study the effectiveness of gas recognition by various machine learning methods that do not require significant computing resources and can be implemented using edge computing technologies. Particular attention was focused on improving the effectiveness of machine learning models for gas recognition. Models of multi-class classification of the sensor data obtained using the graphene field-effect transistor in the form of dependence of the reduced graphene oxide film resistance on the gate voltage have been implemented. The effectiveness of recognition of ethanol, ammonia, acetone, and toluene in the gaseous state by support vector machine (SVM), gradient boosting, k-nearest neighbors (KNN), and decision tree methods was studied using confusion matrices and classification reports. The multi-class classification models have been evaluated according to the accuracy, precision, recall, and F1-score metrics. The effectiveness of gas recognition by the proposed machine learning models in concentration ranges <5% and >5% was compared. It was found that the accuracy of identification of the analyzed gases significantly depends on their concentration. In the specified concentration ranges, the KNN method has the highest classification accuracy (0.72 and 1.00, respectively). The possibility of increasing the accuracy of gas recognition by optimizing the hyperparameters of machine learning models has been demonstrated. Key words : intelligent sensor, gas identification, machine learning, multi-class classification.

Cite this record

Igor B. Olenych, Oleh Osadchuk (2024). APPLICATION OF MACHINE LEARNING ALGORITHMS IN ELECTRONIC NOSE TECHNOLOGY. Electronics and Information Technologies, 25. https://doi.org/10.30970/eli.25.5

Copy in another style

Olenych, I. B., & Osadchuk, O. (2024). APPLICATION OF MACHINE LEARNING ALGORITHMS IN ELECTRONIC NOSE TECHNOLOGY. Electronics and Information Technologies, 25. https://doi.org/10.30970/eli.25.5
Author names are used as deposited, so check their order before you submit.

Metadata is harvested from open-access indexes and can be incomplete. Check a citation against the published version before relying on it — see our open access & attribution policy.