Abstract In recent years, renewable energy resources have drawn a lot of attention worldwide in developing a more sustainable society. Among various forms of renewable energies, wind power has been recognized as one of the most promising ones in many countries and regions including Northern Ireland and Ireland according to the National Renewable Energy Action Plans (NREAPs). However, due to the variability nature of wind power, the wind generation forecasting hours even days ahead proves to be imperative to enhance the flexibility of the operation and control of real-time power systems. In this paper, a variant Gaussian Process employing only nearby measured wind power data is proposed to make short term prediction of the overall wind power production for the whole island of Ireland. Multi Gaussian Process submodels are developed, and the model capability in reflecting the variability and uncertainty in the wind generation system is enhanced. In such method, local data could be utilized more efficiently and computation complexity is reduced at the same time. The forecasting results have been verified in comparison with standard Gaussian Process and persistence model, and improvements can be observed in terms of the model complexity and prediction accuracy. Moreover, a recently proposed teaching-learning based optimization algorithm (TLBO) is applied to build the Gaussian model, and simulations show its faster convergence speed and better global searching capability.
A Variant Gaussian Process for Short-Term Wind Power Forecasting Based on TLBO
01.01.2014
10 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
A Variant Gaussian Process for Short-Term Wind Power Forecasting Based on TLBO
British Library Conference Proceedings | 2014
|Gaussian Processes for Short-Term Traffic Volume Forecasting
Transportation Research Record | 2010
|Gaussian Processes for Short-Term Traffic Volume Forecasting
Online Contents | 2010
|Short-term Wind Power Forecasting Based on CEEMDAN-FE-KELM Method
British Library Online Contents | 2016
|