Weather rerouting decision support model
A lightweight AI rerouting decision model that helps to determine whether running weather routing is beneficial. The goal of the model is to help minimize the high computational workload demanded by a commercial weather routing system that performs frequent route re-optimazations, e.g. every 6 hours based on the frequent arrival of new weather forecast data
A fully connected feedforward neural network model trained with data of a container ship assumed to be travelling in the North Pacific experimental area under different environmental conditions between September 1st 2022 to January, 31st 2023.
The goal of the weather rerouting decision model is to minimise the high computational resource demands associated with the periodic operation of commercial weather routing systems by helping to determine cases that do not need rerouting. The rerouting decision model is based on a fully connected network MLP applied to a binary classification problem (reroute or no reroute).