摘要： Load estimation is crucial information for every type of war-ships because it is the necessary base for a series of studies and operations, such as economic operation of electric generators, load shedding and power management systems. In this paper a pattern recognition methodology is presented, which consists of four steps: (a) data selection, (b) data pre-processing, in order to modify or exclude obviously incorrect values, (c) the application of pattern recognition algorithm, (d) load estimation. The clustering algorithm used in this paper is a modified k-means algorithm, which is properly calibrated by using the ratio of within cluster sum of squares to between cluster variation (WCBCR). This methodology is indicatively applied to a Hellenic Navy MEKO type frigate for the estimation of its total load demand while the usefulness of the obtained results for the power system design and operation is proved as well.
