Monitoring ship performance via multi-way partial least-squares analysis of functional data

The multi-sensor systems installed on board of modern ships provide massive amounts of data that require opportune multivariate methods for continuous performance monitoring during voyages. In this paper, functional data are obtained from variables that describe operating conditions of a Ro-Pax cruise ship owned by the Grimaldi Group and are analysed via multi-way partial least-squares regression of the fuel consumption per hour. The proposed procedure is shown to well predict and monitor ship performance and to indicate if and when an anomaly may occur in ship operating conditions throughout each voyage.