Using Super Resolution Perception for Anomaly Detection in Energy Load Profiles

Detection of anomalies in energy load profiles is crucial for the optimization of energy resources. Although there are many possible approaches to anomaly detection, in this paper it is proposed a novel approach based on the use of an artificial neural network already trained to solve the problem of Super Resolution Perception (SRP) as a method to detect anomalies in energy consumption. Furthermore, a comparison with a conventional autoencoder (AE) is provided to demonstrate that the solution to the SRP problem generates a competitive model which surpasses in several metrics to AE in the anomaly detection task without additional implementation costs.