Forecasting and simulating wind speed in Corsica by using an autoregressive model

Wind power forecasting : state-of-the-art 2009.

“…For example, Contaxis et al [94] employed an autoregressive (AR) model (more precisely an AR [3]) to forecast the wind speed for time horizons ranging between 30 min. and 5 hr and used the values to control an isolated hybrid diesel/wind system and short-term operation scheduling; Kamal et al [95] used an ARIMA model to forecast the wind speed and estimate confidence intervals; Schlink et al [96] employed these models to forecast the wind speed for the next 10 minutes in an airport; Poggi et al [97] used an autoregressive model for each month in order to forecast the wind speed for the following 3 hr; Torres et al [98] used five Auto-Regressive Moving Average (ARMA) models to forecast the hourly average wind speed for a time horizon of 10 hr in five different locations with different topographic characteristics. With this model, over nine years it was possible to achieve a 20% error reduction as compared to persistence; Tantareanu [99] found that ARMA models can perform up to 30% better than persistence for 3 to 10 steps ahead in 4-s averages of 2.5-Hz sampled data.…”

Stochastic Modeling and Analysis for Environmentally Powered Wireless Sensor Nodes

“…For the stochastic energy model, no classical probability law could be suitably fitted to solar radiation empirical probabilities. On the other hand, the literature on climatology and renewable energy is already well established, using historical data, descriptive Markov chain models for various forms of environmental energy, such as solar radiation [9], [18], [23], wind speed [17] and ambient temperature, or autoregressive process models [22]. More exactly, [23] proposes a firstorder stationary discrete time Markov chain model for each month of the year, due to the big monthly variations, built from traces taken over a period of 20 years.…”

Economic performance indicators of wind energy based on wind speed stochastic modeling

“…Many scholars have proposed new models that can allow the prediction of wind speed minutes, hours or days ahead. Many of these models are based on neural networks [16], autoregressive models [17,18], Markov chains [19][20][21][22][23][24], hybrid models where the previous mentioned models are combined [25][26][27][28][29] and other models [30][31][32][33][34]. All of these models try to catch, from the past, information on the future.…”