A new hybrid complex ADALINE-LES filter for off-nominal frequency symmetrical components extraction

This paper presents a new digital filter to estimate the symmetrical components in power systems under unstable frequency conditions. The proposed hybrid filter comprises a complex adaptive linear neural network and a least error squares filter (CADALINE-LES). First, a new complex ADALINE structure is introduced that compresses the three-phase signals into a complex vector to produce a tuning vector. The estimation of the symmetrical components for off-nominal frequency operation is carried out using the produced tuning vector and the LES algorithm. The simulation results in PSCAD/EMTDC and MATLAB software confirm the accuracy and efficiency of the proposed method.