A Novel Methodology for Classifying Electrical Disturbances Using Deep Neural Networks

Electrical power quality is one of the main elements in generation systems. At same time, it most significant challenges regarding stability and reliability. Due to different switching devices this type architecture, kinds generators as well non-linear loads are used for industrial processes. A result need classify analyze Power Quality Disturbance (PQD) prevent degradation system reliability affected by non-stationary oscillatory nature. This paper presents a novel Multitasking Deep Neural Network (MDL) classification analysis multiple electrical disturbances. The characteristics extracted using specialized adaptive methodology signals, namely, Empirical Mode Decomposition (EMD). methodology’s design, development, various performance tests carried out with 28 difficulties levels, such severity, disturbance duration noise 20 dB 60 signal range. MDL was developed diverse data set difficulty noise, quantity 4500 records samples has an average accuracy percentage 95% In addition, 90% analyzing important aspects studying crest factor, per unit voltage analysis, Short-term Flicker Perceptibility (Pst), Total Harmonic Distortion (THD), among others.