Diesel Adulteration Detection with a Machine Learning-Enhanced Laser Sensor Approach

Abstract

This paper introduces a novel and cost-effective method for detecting adulterated diesel, specifically targeting contamination with kerosene, by leveraging machine learning and the refractive index values of mixed diesel samples. It proposes a laser-based sensor, employing COMSOL simulations for synthetic data generation to facilitate machine learning training. This innovative approach not only streamlines the detection process by eliminating the need for expensive equipment and specialized personnel but also enables on-site testing without extensive sample preparation. The sensor’s design, utilizing light refraction and reflection principles, allows for the accurate measurement of diesel adulteration levels. Validation results showcase the machine learning models’ high precision in predicting adulteration percentages, as evidenced by an R-squared value of 0.999 and a mean absolute error of 0.074. This research signifies a leap in sensor technology, offering a practical solution for rapid diesel adulteration detection, especially in developing countries, by minimizing reliance on advanced laboratory analyses. The sensor’s design aligns with the requirements for low-cost IoT technology, presenting a versatile tool for various applications.