A two-level neural network approach for flicker source location

Identification of flicker sources is necessary to find who responsible for the measured and improve power quality. This paper puts forward a new method identifying with minimum measurement units. Contrary previous works where are considered single-frequency signal, autoregressive moving average (ARMA) used model active reactive variations. First, envelope network voltage at busbars derived by Hilbert transform. Then, appropriate indices extracted from spectral density (PSD) envelope. A novel two-level structure set ANNs proposed, which needs low number units locate sources. Using captured data different simulations various scenarios, Artificial Neural Networks (ANNs) trained categorize