Three-phase line overloading predictive monitoring utilizing artificial neural networks

Abstract

The aim of this study is to develop and evaluate an autonomous method to perform real time monitoring of power line overloading. To that end, an Artificial Neural Network (ANN) that is repeatedly trained every hour with the most recently acquired measurements is utilized for conducting automated monitoring. The ANN is trained by using the Levenberg-Marquardt algorithm synergistically with Bayesian regularization, which is used to avoid overfitting of the training data. Obtained results by applying the ANN to a set of simulated data taken with the Gridlab-d software exhibit the potentiality of the method in monitoring and predicting line overloading at each line of a three-phase line system in nearly real-time manner.