Artificial intelligence based-modeling for sizing of a Stand-Alone Photovoltaic Power System: Proposition for a New Model using Neuro-Fuzzy System (ANFIS)

This paper introduces a new approach for modeling of the optimal sizing parameters of stand-alone photovoltaic power (SAPVP) system especially in isolated sites where meteorological data are not available. In addition to traditional models, several artificial intelligence based technique are studied and compared. These include feed-forward, radial basis function network, recurrent network, modular network, and the adaptive wavelet-network. The proposed model consists to use an adaptive neuro-fuzzy inference scheme (ANFIS). The problem consists to predict the optimal parameters of SAPVP system in isolated sites, where the traditional models are not able to estimate these parameters in these sites. From these parameters we can determine the optimal configuration (PV-array size and battery size) of SAPVP system for a given load. A database of optimal sizing parameters has been developed by using numerical model for 200 locations in Algeria. The ANFIS model has been trained by using 200 known sizing parameters data, in this way the model was trained to accept and even handle a number of unusual cases. Known sizing parameters were subsequently used to investigate the accuracy of estimation, the unknown validation sizing parameters set produced very set accurate estimation with the correlation coefficient between the actual and the ANFIS model estimated data of 98.5% was obtained. Obtained results indicate that the proposed model can be successfully used for estimating the optimal sizing parameters of SAPVP system for any location in Algeria from only the geographical coordinates of the considered site. The methodology can be generalized using different locations over the world