Systematic cross-validation of photovoltaic energy yield models for dynamic environmental conditions. (October 2017)

• Record Type: Journal Article
• Authors: Anagnostos, D.; Goverde, H.; Catthoor, F.; Soudris, D.; Poortmans, J.
• Is Part Of: Solar energy. Volume 155(2017)
• Page Start: 698
• Page End: 705
• Publication Date: 2017-10
• Subjects: Energy yield modeling -- Photovoltaics -- Dynamic modeling -- Neural networks; Solar energy -- Periodicals; Solar engines -- Periodicals; 621.47
• DOI: 10.1016/j.solener.2017.07.011
• Languages: English
• ISSNs: 0038-092X

Abstract

Highlights: Locations with varying environmental conditions benefit from dynamic energy yield PV models. Developed models showcase improved performance for a plethora of time resolutions and conditions. Optimization of neural network models is possible by inclusion of thermal information. Abstract: The continuous growth of photovoltaic installations globally carries hope for a sustainable future but also imposes challenges on all levels of energy production and distribution. Grid operators and designers have to cooperate more closely with monitoring service providers in order to sustain a flexible scheme of energy exchange. The basis of these calculations is accurate energy yield estimation models, which are able to capture all effects of the environment. Especially for locations with highly dynamic irradiance and environmental conditions, this remains a tough challenge. All photovoltaic energy yield models presented in this work aim at accommodating the inherent dynamism of these challenging locations at small time scales. A detailed, physics based electro-thermal energy yield model is validated along with other state-of-the-art models and performs 25% more accurately. Additionally, the results from the dynamic modeling are transferred to a neural network model, increasing the accuracy further up to six times better than any parametric solution.