Hybrid forecasting system based on an optimal model selection strategy for different wind speed forecasting problems. (15th September 2019)

• Record Type: Journal Article
• Authors: Zhou, Qingguo; Wang, Chen; Zhang, Gaofeng
• Is Part Of: Applied energy. Volume 250(2019)
• Page Start: 1559
• Page End: 1580
• Publication Date: 2019-09-15
• Subjects: Hybrid wind speed forecasting system -- Artificial neural network -- Data analysis -- Model selection strategy -- Modified multi-objective optimization algorithm; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2019.05.016
• Languages: English
• ISSNs: 0306-2619

Abstract

Highlights: Data analysis eliminate the outliers and denoise of actual wind time series. The optimum forecasting model is selected by model selection strategy. Modified multi-objective algorithm optimizes optimum model's parameter. The multi test method indicate the accuracy of the proposed forecasting system. Abstract: Wind energy represents an important cornerstone of a sustainable and non-polluting electricity supply. With the aim of reducing greenhouse gases, wind speed forecasting has always been a crucial component of transmission-network operation and wind power station planning. In recent years, studies have shown that there is no single forecasting model that can be considered the best and applied in all cases for wind speed forecasting, as there are considerable differences among the wind speed time series. Therefore, a model selection strategy can prevent the worst model from being employed for wind speed forecasting. In this study, a novel wind speed forecasting system is developed, which includes four modules: data analysis, model selection strategy, forecasting processing combined with a modified multi-objective optimization algorithm, and model evaluation. This hybrid forecasting system retains the advantages of traditional forecasting models, and eliminates unsuitable forecasting models. The experimental results demonstrate that the proposed hybrid forecasting system not only effectively selects optimal forecasting models, but is also able to improve the wind speed forecasting performance. Thus, it can provide an effective tool for planning and dispatching in smart grids.