ROBUST SIGMOIDAL COST FRAMEWORK BASED LMS ALGORITHMS FOR SHUNT COMPENSATION IN GRID CONNECTED SYSTEM

Abstract

In recent times AC grid connected systems are experiencing numerous power quality problems mostly originating in distribution level. Use of power converter-based machines in industries, homes, shops, offices, and traction applications have seen an exponential increase. These equipment behave as nonlinear load, draw non-sinusoidal current and cause Power Quality deterioration of supply system. In this work power quality issues due to increased application of renewable energy resources and non-linear loads and in distribution system are addressed. There are a number of current-based power quality problems such as poor voltage regulation, poor power factor, harmonics distortion, unbalanced currents etc. DSTATCOM are becoming more applicable for improving the condition of power quality in supply systems due to cost benefits. Due to increase in computational capacities of modern computers adaptive filters are becoming popular in wide areas of noise/echo cancellation, adaptive control, image restoration and channel equalization. DSTATCOM with adaptive filtering technique are being used as a cost-effective Shunt Active Power Filter (SAPF). Most simple control strategies based on adaptive filter, for SAPF, are the Least Mean Square (LMS) algorithm and its versions, because of their simplicity and relatively robust performance. Various issues such as performance degradation in presence of outlier conditions like impulsive current and voltage variations which commonly happens during practical system operation. Several LMS-Adaline based Control Algorithms of DSTATCOMs have been implemented in various articles. In this project the aim is to implement a more robust LMS based algorithm for DSTATCOM control using sigmoid cost-framework, which reduces performance degradation due to impulsive interferences. Using sigmoid cost framework v sigmoidal least absolute difference (SLAD), sigmoidal least logarithmic absolute difference (SLLAD), sigmoidal least mean square (SLMS), sigmoidal least mean logarithmic square (SLMLS) and sigmoidal least mean fourth (SLMF) algorithms for DSTATCOM control are simulated in MATLAB/Simulink environment. A comparison between various algorithms is presented. A PV-DSTATCOM for PV-grid integration, using SLMLS algorithm and Maximum Power Point Tracking (MPPT) algorithm using incremental conductance method, is also simulated and results are analysed.