An Ultra-Fast Master-Slave ADALINE for Hybrid Active Power Filter including Photovoltaic System

Abstract

This paper proposes a unique two-fold adaptive linear neural network (ADALINE) for extracting the sum of harmonics and reactive currents from the load currents in a three-phase hybrid power filter (HPF) network. The HPF linked with photovoltaic (PV) system and DC-DC boost converter to extract maximal power using maximum power point tracking (MPPT). The proposed detection algorithm for HPF is entitled as Master-Slave ADALINE (MS ADALINE), which is based on parallel adaptive filter theory. The Slave-ADALINE follows fixed and large step-size least mean square (LMS) algorithm for weight vector correction. During transients, this filter plays an important job. However, Master-ADALINE selects adaptable step-size LMS learning rule for weight vector adaptation. At last, the local averages of the squared errors of both the ADALINE's are worked out and fed to the decision controller circuit. This circuit equates the two magnitudes, and revises the Master-ADALINE weight vector and step-size parameter, accordingly. This recommended scheme boosts the convergence speed and improves tracking accurateness. The efficacy of MS ADALINE is proven by comprehensive simulation and experimental survey.