Abstract
This paper presents the construction of an algorithm for finding optimal fuel routes for ships with sailing time constraints applied to cargo ships. Unlike previous studies, most of them used data collected from noon reports recording past operating data of ships, these data have a large sampling time and often do not have enough necessary parameters. This study proposes to build a simulation model using the hardware-in-the-loop simulation method to create a dataset with a variety of situations that ships will encounter when operating in reality at sea. In addition, an MLP neural network model combined with the A* algorithm will be used to build an algorithm for finding optimal fuel routes with sailing time constraints. Testing results show that the algorithm operates reliably with low errors.