Forecasting Transmission System Outages Using Artificial Neural Networks

Estimating system outages is an important aspect of operating the electrical system efficiently. It helps system operators to prepare the necessary measures in case these outages take place and ensure the system continues to operate in a safe, secure, and reliable way. In recent years, artificial intelligence algorithms have been considered to be a major aid in forecasting, estimating, and diagnostic methods. This is mainly due to the amount of data available to train such algorithms as well as rapid developments in computers hardware capabilities. This paper proposes a method of predicting transmission system outages, specifically extra-high voltage AC transmission lines by analyzing historical records of outages due to different types of events. The estimation is done through Artificial Neural Networks (ANNs) represented in MATLAB and trained using backpropagation techniques. The optimization of the training algorithm is presented, and it plays a major role in shaping up the final feedforward part of the ANN. The proposed techniques show good results when evaluated using a multitude of metrics.