A combined forecasting model for time series: Application to short-term wind speed forecasting. (1st February 2020)

• Record Type: Journal Article
• Authors: Liu, Zhenkun; Jiang, Ping; Zhang, Lifang; Niu, Xinsong
• Is Part Of: Applied energy. Volume 259(2020)
• Publication Date: 2020-02-01
• Subjects: Short-term forecasting -- Combined model -- Forecasting accuracy -- Wind speed forecasting; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2019.114137
• Languages: English
• ISSNs: 0306-2619

Abstract

Graphical abstract: Highlights: An advanced combined model is developed for short-term wind speed forecasting. Latest data processing strategy is applied to seize the character of wind speed. Linear and nonlinear forecasting models are applied for wind speed forecasting. A modified multi-objective optimization is used to determine the optimal weight. The uncertainty analysis of each model is conducted by interval forecasting. Abstract: Wind speed forecasting has been growing in popularity, owing to the increased demand for wind power electricity generation and developments in wind energy competitiveness. Many forecasting methods have been broadly employed to forecast short-term wind speed for wind that is irregular, nonlinear, and non-stationary. However, they neglect the effectiveness of data preprocessing and model parameter optimization, thereby posing an enormous challenge for the precise and stable forecasting of wind speed and the safe operation of the wind power industry. To overcome these challenges and further enhance wind speed forecasting performance and stability, a forecasting system is developed based on a data pretreatment strategy, a modified multi-objective optimization algorithm, and several forecasting models. More specifically, a data pretreatment strategy is executed to determine the dominating trend of a wind speed series, and to control the interference of noise. The multi-objective optimization algorithm can help acquire more satisfactory forecasting precision and stability. The multiple forecasting models are integrated to construct a combined model for wind speed forecasting. To verify the properties of the developed forecasting system, wind speed data of 10 min from 4 adjacent wind farms in Shandong Peninsula, China are adopted as case studies. The results of the point forecasting and interval forecasting reveal that our forecasting system positively exceeds all contrastive models in respect to forecasting precision and stability. Thus, our developed system is extremely useful for enhancing prediction precision, and is a reasonable and valid tool for intelligent grid programming.