MULTILAYER ADAPTIVE NEURAL NETWORK FOR DC LINK VOLTAGE REGULATION IN GRID CONNECTED HYBRID SYSTEMS

Masood I. Nazir, Ikhlaq Hussain, and Aijaz Ahmad

Keywords

Permanent Magnet Synchronous Generator (PMSG), fuzzy logic control, artificial neural network, PV, Wind energy

Abstract

This paper proposes a hybrid learning algorithm-based improved neural network control of a hybrid wind/photovoltaic (PV) power system for grid-connected applications. It is a model-free approach dissuading the need of designing mathematical plant models besides being robust and easy to implement. The proposed algorithm reduces the surplus mean-square error (MSE) iteratively that results in a compact expression of step size. The proposed control combined with the hybrid re-weighted zero attracting least mean-square (RZALMS) algorithm improves the steady-state and dynamic performance of the system by mitigating power system problems such as voltage fluctuations and harmonic distortion. It also ensures efficient power flow among the grid, the hybrid source and the load. The efficacy of the system is verified in MATLAB/Simulink. The performance of the proposed system against conventional PI and fuzzy controllers in maintaining DC link voltage under steady-state and transient conditions has been added for relative assessment. The simulation results demonstrate the efficacy of the algorithm in handling grid disturbances, as well as insolation and windspeed changes. Improvements are also observed during dynamic conditions in terms of reduced fluctuations, steady-state error and peak overshoot.

Important Links:



Go Back