Machine Learning Optimization-Based Efficient Detection of Fuel Adulteration Using a Novel Circular Slotted Refractive Index Sensor

Document Type

Article

Publication Title

IEEE Sensors Journal

Abstract

The overall quality of fuel significantly impacts the durability and optimal performance associated with all fuel engines. Several irresponsible retailers adulterate lower-priced oily substances or components with fuel compounds to augment their earnings. The Symmetric Quarter-Arc Optical Refractive Index Sensor (SQAORIS) has been developed to determine adulteration in fuel to address this issue. This fuel adulteration includes petrol, kerosene, and diesel. The approach of neural network regression in machine learning (ML) has also been studied to showcase the actual value and predicted value of fuel adulteration. Its own unique quarter-arc design gives accurate and faster results while detecting the smallest variations caused by adulteration. It achieved optimum sensitivity values of 1623.52, 1616.82, and 1614.17 nm/RIU, and detection limit (DL) values of 0.000419, 0.000202, and 0.000386 for adulterated petrol, kerosene, and diesel, respectively. The optimum values for quality factors (QFs) are 1433.55, 848.71, 1479.24, and 866.90, and the optimum values for figure of merits (FOMs) are 1068.10, 594.69, 1016.86, and 586.97 for water, petrol, kerosene, and diesel adulteration, respectively. The optimum detection range (DR) of 1865.33 was achieved for kerosene adulteration. We have observed an optimum value of 0.9984 from ML prediction by the neural network regression method. In addition, due to its enhanced sensitivity and facilitating features, the sensor is going to serve a crucial role in practical uses in the near future.

First Page

40008

Last Page

40019

DOI

10.1109/JSEN.2025.3611702

Publication Date

1-1-2025

This document is currently not available here.

Share

COinS