Optimising Own PV Consumption with PV Energy Yield Predictions from Machine Learning Algorithms and Weather Data

Heck, HorstHorstHeckSchmidt, Armin JürgArmin JürgSchmidtSchüpbach, EvaEvaSchüpbachKuonen, FranziskaFranziskaKuonenBacha, SaniaSaniaBachaMuntwyler, UrsUrsMuntwyler2024-11-192024-11-192020-09-083-936338-73-610.24451/arbor.13355https://doi.org/10.24451/arbor.1335510.4229/EUPVSEC20202020https://arbor.bfh.ch/handle/arbor/41855Weather data are evaluated in view of their influence on high-quality PV energy yield predictions based on machine learning algorithms (MLAs). Optimisation experiments evidence that the prediction quality can be increased to over 30% by incorporating specific weather parameters in the ML-training. The results will feed into a planning tool for optimising the own consumption (including in wintertime) of PV plant owners. The outcome of this study also illustrates evolving best practice in using meteorological data to produce PV energy yield predictions with specific MLAs.enPredictionMachine LearningWeather DataPV Energy YieldQAQA76TKOptimising Own PV Consumption with PV Energy Yield Predictions from Machine Learning Algorithms and Weather Data-conference_item