CHARACTERIZATION OF DISTURBANCE IN ELECTRIC POWER SIG- NALS: A MACHINE LEARNING APPROACH

Resumo

Voltage variation in electrical networks is one of the problems that arise when it comes to electronic equip-
ment that is sensitive to voltage variations. As a way of classifying voltage variation phenomena, such as Voltage

Sag or Swell, this paper aims to use artificially created voltage signals in the time domain, which represent each
electrical fault, which will be analyzed in the frequency domain with techniques of time frequency analysis (TFA)

and entropy analysis, among other Features. Subsequently, with classic methods from the Machine Learning lit-
erature, classify the general electrical signals and identify the respective faults. As a working tool, the Python

language is used, as it is easy to implement and learn, in addition to being widely documented. Additionally, the
scikit-learn librarie is used, which are widely tested and documented in the literature.