Classification of Fuel Octane Ratings Using Gas Sensors and Partition Column

• Published in: 2024 International Seminar on Intelligent Technology and Its Applications (ISITIA) pp. 385 - 389
• Main Authors: Zulfahmi, Syarofan, Rivai, Muhammad, Mujiono, Totok
• Format: Conference Proceeding
• Language: English
• Published: IEEE 10.07.2024
• Subjects: Accuracy; Artificial neural networks; Classification algorithms; energy; Gas detectors; gas sensor; neural network; octane number; partition column; Partitioning algorithms; Petroleum; Seminars
• ISSN: 2769-5492
• DOI: 10.1109/ISITIA63062.2024.10667830

Gasoline is a product from petroleum that is included in a type of non-renewable energy, which is one of the fuels for internal combustion engines. Commercial gasoline prices are differentiated based on octane value. Octane value explains the ability of gasoline to withstand engine compression without knocking. The higher the octane number, the more resistant gasoline is to engine compression. Usually, octane number measurements are done using the American Society of Testing and Materials (ASTM) standards. However, this method requires a long testing time and is carried out by a certified operator. This study develops a system to classify the octane number of a gasoline fuel product. This system consists of a partition column and a gas sensor to form a specific pattern for each octane number. A combination of principal component analysis as feature extraction and an artificial neural network algorithm is used to classify the patterns produced by the gas sensor. The experimental results show that the gas sensor arrangement can provide a specific pattern for each type of gasoline with different research octane numbers, namely 90, 92, and 98. The artificial neural network algorithm can differentiate each type of gasoline with an accuracy of 83.33%.