Direct neural method for harmonic currents estimation using adaptive linear element

Abstract

The use of nonlinear loads has increased in power systems consequently the harmonic currents have also increased causing detrimental effects to the supply system and user equipment. The aim of the present paper is to identify harmonics in order to obtain a perfect compensation by active power filter (APF). A study of harmonic currents identification by two different methods is conducted in this work. The instantaneous power (PQ) theory method requires two low-pass filters for the extraction of direct power components from total power components. However the direct neural method based on ADALINE neural network method estimates total harmonic current as well as harmonic components separately. Moreover, the identification of each component separately enables the selective compensation of harmonics by the active filter if the objective is to minimize the cost. The method is easy to implement in real time compared to PQ method.In present paper two algorithms based on conventional PQ method and direct neural method were developed in order to identify the harmonic currents. These developed algorithms are confirmed by experimental tests by implementing these techniques in a dSPACE controller in order to show their effectiveness. The obtained results are compared, discussed and analyzed.