Renewable Energy 33 (2008) 35–41 Short term wind speed forecasting for wind turbine applications using linear prediction method G.H. Riahy à , M. Abedi Center of Excellence on Power Systems, Wind Energy Laboratory, Electrical Engineering Department, Amirkabir University of Technology, Tehran 15914, Iran Received 26 October 2006; accepted 23 January 2007 Available online 19 March 2007 Abstract In this paper a new method, based on linear prediction, is proposed for wind speed forecasting. The method utilizes the ‘linear prediction’ method in conjunction with ‘filtering’ of the wind speed waveform. The filtering eliminates the undesired parts of the frequency spectrum (i.e. smoothing) of the measured wind speed which is less effective in an application, for example, in a wind energy conversion system. The linear prediction method is intuitively explained with some easy to follow case studies to clarify the complex underlying mathematics. For verification purposes, the proposed method is compared with real wind speed data based on experimental results. The results show the effectiveness of the linear prediction method. r 2007 Elsevier Ltd. All rights reserved. Keywords: Wind energy; Wind speed prediction; Linear prediction 1. Introduction Variations of wind speed are in the range of seconds, minutes, hours, days, weeks, seasons and years. In long- term planning for wind turbine installations, the wind speed prediction should be for years ahead [1,2]. However, control system of a wind turbine, requires prediction times in the range of seconds ahead of our interest, because the major problem with the control of wind turbines is the delays associated with the wind turbine system. These delays affect the response of the system in respect to controller action [3–5]. In the literature [6–8], it is shown that the wind speed prediction is an important issue for wind energy conversion systems. Short term wind speed prediction can be used for dynamic control of a wind turbine, due to importance of short-term decisions. The short-term decisions could be classified as connection of a load, changing the pitch of the blades and/or any other control action which involves delays. 2. The linear prediction method Linear prediction method is a powerful technique for predicting time series in a time-varying environment. A time-varying process is a process where the underlying function of its measured parameters is time variant. This means a measured parameter of such a process cannot be represented by a unique mathematical function over a long period of time, but rather the equation of the function has to be updated over short periods of time. A wind speed signal typically belongs to a time-varying process. The linear prediction model, recursively represents time series of signal samples over a time interval [9], such as yðt þ T Þ¼ a 1 yðtÞþ a 2 yðt  T Þ þ ... þ a m yðt ðm  1ÞT Þ. ð1Þ In Eq. (1), a 1 , a 2 , y, a m are the linear prediction coefficients, ‘m’ is the model degree, ‘T’ the sampling time, y(t+T) the future observation and y(t), y(tT), y, y(tmT) are the present and past observations. In Eq. (1), the output is the linear combination of the present and past samples, hence it is called linear prediction. ARTICLE IN PRESS www.elsevier.com/locate/renene 0960-1481/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.renene.2007.01.014 à Corresponding author. Tel.: +98 912 133 7238; fax:+98 216 646 9961. E-mail addresses: gholam@aut.ac.ir (G.H. Riahy), abedi@aut.ac.ir (M. Abedi).