Applying hierarchical fuzzy systems to predict unplanned feeder outages in the Yazd

Abstract

Weather conditions and various phenomena cause unplanned power outages. These factors should be identified in a proper time and therefore prevent the occurrence of power outages. The aim of this paper is to provide a system that predicts unplanned outages before they occur to take necessary actions for no outage, power restoration and reduce costs. Based on the primary experiments, weather data for two days ago and historical outage are used as input of the proposed system. First, pre-processing is performed on data and outages factors identified and then hierarchical fuzzy system is used. In the first level, four features obtained from data mining are considered as input of a Mamdani fuzzy system. The outputs of this fuzzy system with fifth feature obtained from data mining would be inputs of the second fuzzy system. The System's output would be next day's outage or no outage prediction. At the end, system is evaluated with Yazd and Meybod regional data and accuracy 94.23 is obtained.