Fuel Control System on CNG Fueled Vehicles using Machine Learning: A Case Study on the Downhill

Abstract

Compressed Natural Gas (CNG) is an affordable fuel with a higher octane number. However, older CNG kits without electronic controls have the potential to supply more fuel when driving downhill due to the vacuum in the intake manifold. Therefore, this article presents a development of a CNG control system that accommodates road inclination angles to improve fuel efficiency. Machine learning is involved in this work to process engine speed, throttle valve position, and road slope angle. The control system is designed to ensure reduced fuel consumption when the vehicle is operating downhill. The results showed that the control system increases fuel consumption by 25.7% when driving downhill which an inclination of 5ᵒ. The AFR increased from 17.5 to 22 and the CNG flow rate decreased from 17.7 liters/min to 13.8 liters/min which is promising for applying to CNG vehicles.