GW Work

 

Building Time Series Model for Wind Power Forecast Open Access

Downloadable Content

Download PDF

With the depletion of unrenewable resources, new concepts of “sustainable development” “environment-friendly” are increasingly popular, which makes the utilization of renewable resources a global focal point. Comparing to the unrenewable resource, wind power, as an example of renewable resource, has several advantages like cost-effectiveness and free-pollution. According to World Wind Energy Association, the entire world wind power was up to 270 billion kilowatts in 2007, which shows the great potential of wind power industry. Naturally, a power prediction model is definitely necessary to better use wind power. Brown (1984) was the first one who applied forecast model in wind power industry. Since then, many mathematicians and statisticians spare no effort in promoting wind power forecast models. Many recent studies have focused on short-term forecasts of wind power and hybrid method However, the importance of time series character of wind power data is not highlighted by most of the previous researchs. We pay more attention to this character and apply time series specific model-ARIMA to forecast wind power. ARIMA (autoregressive integrated moving average model) is a generalization of ARMA (autoregressive moving average model), which is suitable for time series data either to better understand the data or to predict future points in the series (forecasting). ARIMA is usually applied in cases where data show evidence of non-stationarity, where an initial differencing step (corresponding to the "integrated" part of the model) can be applied one or more times to eliminate the non-stationarity. A data series is said to be stationary when its mean, variance, and auto-covariance are all constant, which can be tested by ADF test in R. Considering the intrinsic nonstationary character of wind power, we will apply ARIMA as our model to predict wind power. If wind power can be predicted more accurately, the usage of it will be more effective, which could greatly ease energy pressure faced by human. Our research includes six tasks, data gathering, data pre-processing, data importing and smoothing, model fitting, model examination, and model application. In this process, we build an ARIMA to forecast wind power. Then we apply this model to test whether it is useful to test out-sample data and refit the parameter to find the better model.

Author Language Keyword Date created Type of Work Rights statement GW Unit Persistent URL
License

Relationships

Items