Fault Indicators Allocation To Maximize The Performance of A Fault Locator Based On Artificial Intelligence

Fault location is one of the main challenges in Advanced Distribution Automation of Active Distribution Networks . One of the commonly used strategies by utilities to deal with this challenge is the use of Fault Indicators, which indicate to the operator, the path taken by the fault current. In this context, this paper presents an artificial intelligence-based fault location strategy that determines the number and location of FI into ADN to maximize performance in fault locator. To achieve this objective, the ADN is divided into sections, and the FL problem is modeled as a classification problem to train an Artificial Neural Network (ANN). To determine the number of FIs to be installed and their location, the strategy uses the three-phase current magnitudes measured by the FI as features for an ANN model. Also, the strategy uses a multiverse optimization algorithm to identify the features that maximize the accuracy of the ANN model. The strategy was validated on the IEEE123-node test feeder. The results showed accuracy close to 99.4% with a reduction of 40% of the number of FIs when compared with other method. The strategy shows its simplicity and promising prospects to apply it in the utility’s investment planning.