A0930
Title: Modelling German renewable energy data
Authors: Miao Yu - Leibniz University Hannover (Germany) [presenting]
Abstract: The purpose is to model hourly German onshore wind energy and solar power production playing a crucial role in the transmission towards greener energy systems and CO2 avoidance. Wind energy closely co-moves with wind speed, while solar power aligns with daily sunshine hours. Both wind and solar data exhibit irregular trends and significant variability. Interpreting energy production as continuous, functional data analysis allows for explaining seemingly irregular data structures. When modelling functional data, differential equations are used to both describe changes in variables over time and predict future trends. Estimating a suitable linear differential operator during model building is essential. Analyzing wind and solar data, it finds that a second-order operator best fits the data by minimizing the sum of squared residuals, which indicates that wind and solar production exhibit nonlinear dynamics with an accelerating trend. Summer data fits well with a single-coefficient second-order operator, showing minimal fluctuations. Winter data exhibits stronger fluctuations, approximating a sinusoidal curve, requiring a complex operator. Solar energy, less volatile than wind, fits well with the estimated second-order operator throughout the year. Relying only on one renewable source leads to green energy shortages due to different estimated cycles of wind and solar energy, potentially increasing CO2 intermittency.
