CPV module electric characterisation by artificial neural networks.
Concentrating photovoltaic is a new technology with promising future expectations. However, it is in an early stage of development and it has much room for improvement. In order to gain knowledge about concentrating photovoltaic technology, real outdoor measurements are necessary to adjust models and to study the influence of the atmospheric conditions on the modules performance. The current-voltage curve of a module characterises its behaviour under specific meteorological conditions. In this work, multilayer perceptron models are applied to generate these characteristic curves using the influential atmospheric variables as inputs of the network. To train these networks an experimental campaign with real measures of the electric performance of concentrating photovoltaic modules as well as atmospheric conditions was carried out in Jaén from July 2011 to June 2012. In addition to a model based on I-V curves expressed as a list of points in Cartesian coordinates, we present an alternative model trained with curves defined by points in polar coordinates. A previous selection of the most representative samples from the initial dataset was performed to train the multilayer perceptron models using a Kohonen self-organizing map. This procedure improves the simulation of the curves under non frequent atmospheric conditions. Using the proposed models, it is possible to obtain the characteristic curve of a concentrating photovoltaic module with a high accuracy and fidelity. Since there is not any standard algebraic procedure to obtain I-V curves of this type of modules under different meteorological conditions, the proposed models are very interesting tools when estimating their electric performance.
Citation:Garcia-Domingo, B. et al. (2015) CPV module electric characterisation by artificial neural networks. Renewable Energy, 78. pp. 173-181