Comparison of ANN models for estimating optimal points of crystalline silicon photovoltaic modules

Abstract

Various artificial neural network (ANN) structures have been utilized to determine the maximum power points of PV system. The most common methods are radial basis function neural network (RBF), adaptive neuro-fuzzy inference system neural network (ANFIS) and three layered feed-forward neural network (TFFN). These ANN methods are recognized with simple computational techniques and high pattern recognition capabilities to deal with non-linear characteristic and intermittent output of PV system. However, there still might be strong and weak points for these methods during the optimization process. Since the characteristic of crystalline Silicon PV modules technology is almost similar, it is possible to select a single prominent ANN structure for identification the optimum points of this type solar cell technology. The paper discusses the most suitable ANN structure for estimation the MPP crystalline Silicon PV modules through their optimum operating voltages. To reach this objective, the ANN models have been trained and verified for multi-crystalline Silicon based edge defined film-fed growth (EFG) and wafer solar cell technologies, mono-crystalline Silicon and thin-film Silicon solar cell technologies. Then, the performance of ANN models is compared with hill-climbing (HC) based MPPT technique in terms of tracking the MPP voltage and the energy index.

Access to Document

Other files and links

• Link to publication in Scopus

Fingerprint