Novel heuristic and SVM based optimization algorithm for improving distribution feeder performance

Abstract: Secondary distribution networks generally perform as well as its LV feeders are performing. The main problem that a feeder is experiencing would be the load unbalancing due to the stochastic nature of its individual single-phase loads: larger losses in certain phases accompanied by bad voltage regulation and voltage unbalance. In order to address this problem, it may be economical to install apparatus to automatically balance or partially balance the loads progressing from the end of the feeder towards the front using smart devices based on a three-ways switch selector and an artificial intelligence algorithm to minimize the neutral current. The main idea behind this paper is therefore to keep the three phases progressively balanced along the whole length of the line. A Support Vector Machines (SVM) implementation and a heuristic method are presented as the numerical algorithm