Wind power forecasting based on time series model using deep machine learning algorithms. (2021)

• Record Type: Journal Article
• Authors: Chandran, V.; Patil, Chandrashekhar K.; Merline Manoharan, Anto; Ghosh, Aritra; Sumithra, M.G.; Karthick, Alagar; Rahim, Robbi; Arun, K
• Is Part Of: Materials today. Volume 47:Part 1(2021)
• Page Start: 115
• Page End: 126
• Publication Date: 2021
• Subjects: GRU -- RNN -- LSTM -- Wind power forecasting -- Deep Machine learning; Materials science -- Congresses -- Periodicals; 620.1
• DOI: 10.1016/j.matpr.2021.03.728
• Languages: English
• ISSNs: 2214-7853

Abstract

Abstract: Wind energy is a created due the uneven heating of the earth surface and Coriolis acceleration. Wind energy source is capable of continuously and sustainably producing energy from renewable sources (RES). However, energy generation from the wind power plant has number of issues, such as initial investment costs, wind power plant stationary properties and difficulty in identifying wind power zones. Three deep learning algorithms are utilized in the study for predict short-term wind power generation from wind speed data. We suggested a system that would effectively predict wind power values of wind power by utilizing machine learning algorithms. The machine algorithms adopted for this study is Long Short-Term Memory (LSTM), Gated Reference Unit (GRU) and Recurrent Neural Network (RNN). The models proposed are applied in six times to the projection of wind farm output. The error analysis to balance performance and other approaches is carried out. More concerns are also discussed on short-term wind energy forecasts accuracy improvement. Furthermore, the results have shown that machine learning models can be used in locations other than models. This research showed that, if the basic location model is reasonable, machine learning algorithms could be efficiently used before installation of the wind power plants in an unknown geographical area.