APPLICATION OF MACHINE LEARNING IN CALCULATION OF FUEL CONSUMPTION OF VEHICLES

Abstract

Improving the fuel consumption of ships is one of the measures that can bring efficiency and great economic profit in ship management because fuel cost is one of the biggest operating costs that shipping companies have to pay. However, estimating the fuel consumption of a ship is a difficult problem because the fuel consumption of a ship depends directly on many factors such as the condition of the main engine, the weight of the cargo, and the draft. Currently, statistical models have been established based on the actual data of the ship to estimate the fuel consumption of the ship as accurately as possible. In this study, the authors used the Multi-Layer Perceptron Regression model to estimate the fuel consumption for the M/V NSU JUSTICE ship under the Vietnam National University of Science and Technology. The accuracy of the model was determined by the K-fold evaluation method. Error measurements such as root squared error and absolute mean error are used to evaluate the accuracy of the estimated model. The results show that the error compared with the actual fuel consumption of the ship is low with an acceptable error.