This paper presents a maximum power point tracking (MPPT) method founded on the integration of a model-based technique given by a growing neural gas (GNG) network and a perturb and observe (P&O) algorithm. The neural network is trained off line to estimate the solar irradiance and the maximum power point starting from a measurement of voltage and current on the photovoltaic source. A variable step size perturb & observe method is then utilized to track the true maximum power point. The method is set up for a DC/DC boost converter used in a multi-string PV architecture. The voltage control of the DC/DC converter is performed by a fuzzified PI, assuring the best dynamic performance and stability of the system in all working conditions.
A growing Neural Gas Network based MPPT Technique for Multi-String PV Plants
MC Di Piazza;M Pucci;A Ragusa;G Vitale
2010
Abstract
This paper presents a maximum power point tracking (MPPT) method founded on the integration of a model-based technique given by a growing neural gas (GNG) network and a perturb and observe (P&O) algorithm. The neural network is trained off line to estimate the solar irradiance and the maximum power point starting from a measurement of voltage and current on the photovoltaic source. A variable step size perturb & observe method is then utilized to track the true maximum power point. The method is set up for a DC/DC boost converter used in a multi-string PV architecture. The voltage control of the DC/DC converter is performed by a fuzzified PI, assuring the best dynamic performance and stability of the system in all working conditions.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.