A New Methodological Approach to Estimate the Long-Term Wind Power Production with the Best Accuracy Using Three Estimating Models
Year: 2025
Type: article
Abstract: Renewable energy sources are mainly used today to supply end-users cleaner and efficient electrical power. Wind is the most promised Renewable energy source, since is considered as the best cost-effective one. The wind park under study is consisted of by 31 wind turbines of 1.3 MW, having total nameplate power 41 MW, located in mountain Rodopi-1 site/Greece. The estimated amount of power for the same time/year, are compared with the actual one for a period of 10 years, 2011-2020. We compared the actual to estimated outcomes of each time/year to check which of 3 models is achieving the greater estimating accuracy. The prototype approach of our work lies on two facts, first, the given data are 100% actual, incorporating all losses of wind power produced/transmitted and second, we use 3 different models using the same data/time for estimating/predicting the wind power. The models used are Persistence, Multiple Regression Analysis and Adapted Neuro-Fuzzy Inference System. For the anytime estimation/prediction of power production of the Wind turbines we have used 120 recorded monthly wind speeds, air densities, gusts and power production, referred to the period 2011-2020. These data come from 2 sources, from 31 SCADA systems giving 100% actual information per 15 min and from nearby stations of National Meteo Service. The comparisons showed that the actual and estimated power were influenced by 34.8% of technical features of Wind turbines and by 66.2% by 3 variables, average wind speed (64.3%), wind density (12.8%) and Weibull Contribution of wind (22.9%). The most accurate estimation among 3 models was found the Adapted Neuro-Fuzzy Inference System one, having, simultaneously, the less power production uncertainty.
Authors: STATHAKIS Efthimios, BANDEKAS V. Dimitrios, FANTIDIS G. Jacob, BANTEKA D. Despoina
Institutions: Democritus University of Thrace, University of Nicosia
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7106477875
FWCI:
Cites:
Cited by:
Related to: