Novel Technique of Sizing the Stand-Alone Photovoltaic Systems Using the Radial Basis Function Neural Networks: Application in Isolated Sites

摘要

The objective of this work is to investigate the Radial BasisFunction Neural Networks (RBFN) to identifying andmodeling the optimal sizing couples of stand-alonephotovoltaic (PV) system using a minimum of input data,These optimal couples allow to the users of stand-alone PVsystems to determine the number of solar panel modulesand storage batteries necessary to satisfy a givenconsumption. The advantage of this model is to estimate ofthe sizing PV system in any site in Algeria particularly inisolated sites, where the global solar radiation data is notalways available. A RBFN has been trained by using 200known sizing couples data corresponding to 200 locations.In this way, it was trained to accept and even handle anumber of unusual case, known sizing couples weresubsequently used to investigate the accuracy of predictionthe training of the RBFN model was performed withadequate accuracy. Subsequently, the unknown validationsizing couples set produced very set accurate predictionswith the correlation coefficient between the actual and theRBFN model identified data of 98% was obtained. Thisresult indicates that the proposed method can besuccessfully used for estimating of optimal sizing couplesof PV systems for any locations in Algeria, but themethodology can be generalized using different locations inthe world.