DEVISING A NOVEL METHOD TARGETED AT IMPROVING THE PREDICTION ACCURACY OF THE PRODUCED

IN THIS PAPER, WE HA VE DEVELOPED, USING AN ARTIFICIAL NEURAL NETWORK (ANN) APPROACH, A NOVEL ME THOD AIMING TO IMPRO VE THE FORECASTING A CCURACY OF TWO CRITICAL INDICATORS FOR PHOTOVOLTAIC POW ER PLANTS: THE AMOUN T OF PRODUCED AND CONSUMED ENERG Y. THE METHOD IS ESP ECIALLY USEFUL IN TH E CASE OF PHOTOVOLTA IC POWER PLANTS COMPRIS ING SOLAR PANELS FIE LDS LOCATED AT A CER TAIN DISTANCE. OF PARTICULAR INTEREST WAS TO DEVELOP AND I MPLEMENT CUSTOM ARTIFICIAL FITTING NEURAL NETWORK ARCHITECTURE S, IN ORDER T O ACHIEVE A SPATIAL INTERPOLATION WITH A HIGH DEGREE OF PRECISION, REGARDING THE INPUT METEOROLOGICAL PARAM ETERS CORRESPONDING TO THE RESPECTIVE FIELDS OF SOLAR PANELS. OUR ME THOD IS USEFUL FOR THE PHOTOVOLTAIC POW ER PLANTS OPERATORS THAT MUST PROVIDE DE TAILED REPORTS REGARDING THE FORECASTED QUANTITY OF ENE RGY PRODUCTION AND C ONSUMPTION, TO THE NATIONAL AUTHORITIES . THE DEVISED METHOD POSES ADVANTAGES IN ASSESSING WHETHER A CERTAIN AREA IS APPR OPRIATE FOR SUSTAINI NG A PHOTOVOLTAIC PO WER PLANT DEVELOPMENT, THUS IF FINANCIAL RE SOURCES ARE WORTH IN VESTING. ONCE IMPLEM ENTED AND COMPILED, THE METHOD BECOMES A SPECI ALIZED FRAMEWORK, THAT IS USEFUL FOR THE DEVELOPMENT OF A WID E RANGE OF CUSTOMIZE D APPLICATIONS FOR P REDICTING PERFORMANCE INDICATORS IN THE FIELD OF RENEWABLE ENERGY.

The body of this article is intentionally hidden on the public page. Please use the PDF reader or the PDF download for the complete text.