Load estimation of a small autonomous island power system using k-means method

The islands power systems, especially of the Aegean Sea, are distinguished by their peculiarities, since they are currently autonomous, with the prospect that in the near future most of the islands will be interconnected with the mainland power system, thus shutting down the operation of oil power plants. However, the interconnection is not fully assured in terms of reliability, doubting whether it is ultimately possible to discontinue conventional units. While an additional issue is the penetration of renewable energy sources, due to the high wind potential and high solar radiation and the formulation of hybrid systems, even with the use of energy storage devices. The literature review shows that there has been no systematic research on how to develop the power system of non-interconnected Greek islands with the mainland power transmission system. For this reason, the load growth of an autonomous island, namely Patmos, is firstly studied in this paper indicatively during the years 2013 - 2019 (until October 31, 2019), using pattern recognition methods. In particular, the chronological load curve is estimated by the k-means pattern recognition method, using the mean square error as an adequacy measure, in order to formulate the typical load curves of this island. The results of the present study can be used in next stage for the island's short-term and medium-term load forecasting, for the power plants construction planning, and for the feasibility study of demand management programs.