Segmentação dos postos de transformação e distribuição

Abstract

This work was developed at EDP Distribuição, in the field of data validation and availability, aiming to handle failures of electric power consumption records from smart meters data electric power transformer stations (PTD’s) in Portugal. For that purpose, the PTD’s were segmented using K-Means (a Clustering technique) using the power consumption records recorded over the period of a year. These values were previously normalized, in order to consider only the profile’s shape during the segmentation phase, and were grouped by hour and by typical week in each month, in order to reduce the number of variables to be processed by R. As result of Clustering, 3 typical profiles were obtained, under the shape of time series that corresponded to the period of a year, representative of all PTD’s. These profiles were used to estimate the power consumption records for the next year. In the end, the developed procedure was tested to make the predictions for next year’s (2016) November. Concerning this month’s predictions, it was obtained an error between predictions and real values of 27,3%, excluding possible outliers. This result was considered acceptable and thus, it was assumed this procedure could be used to fulfil the defined goals.

Date of Award	4 Jul 2017
Original language	Portuguese
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Vera Lúcia Miguéis Oliveira e Silva (Supervisor), Maria Silva (Co-Supervisor) & Susana Magalhães (Co-Supervisor)