DISTRIBUTION CONGESTION PREDICTION USING ARTIFICIAL NEURAL NETWORKS FOR BIG DATA

摘要

The prosperity of modern civilization requires power to be generated and delivered to end users 24/7. A big issue regarding power systems is the stability of the power grid since each failure has a significant impact on participating parties. The system operator who is in charge of controlling and coordinating the grid must be all the time aware about the system state and stand ready to take actions in the event of an anomaly. The aim of this study is to develop and evaluate a congestion predictor acting in the big data environment, able to monitor a part of a distribution system and alert about upcoming overloading events giving that way enough time to responsible parties to react and prevent it. Artificial intelligence techniques such as artificial neural networks (ANN) are used in this work. The ANN is trained by utilizing the Levenberg-Marquardt algorithm. All the data concerning both the training and the evaluation of the network were obtained through the GridLAB-D simulation platform. The presented experimental results confirm the soundness of our work.