Short-Term and Very Short-Term Wind Power Forecasting Using a Hybrid ICA-NN Method

Abstract

Utilization ofwind power as one of renewable resources of energy has been growing quickly all over the world in the last decades. Wind power generation is significantly vacillating due to the wind speed alteration. Therefore, assessment of the output power of this type of generators is always associated with some uncertainties. A precise wind power prediction can efficiently uphold transmission and distribution system operators to improve the power network control and management. This paper presents a new Imperialistic Competitive Algorithm- Neural Network (ICA-NN) method to enhance the short wind power forecasting exactness at a wind farm utilizing data from measured information of online supervisory control and data acquisition (SCADA) as well as Numerical Weather Prediction (NWP). Moreover, a very short-term wind power prediction is accomplished based on the past values of wind speed and wind generation and then a comprehensive comparative literature review on the proposed methods in cases of short-term and very short-term is presented. In the proposed method, first, a prediction model of the wind speed is built based on Multilayer Perception (MLP) artificial neural network considering environmental factors (i.e. Humidity, wind speed, temperature, geographical conditions and other factors). Then, Imperialist Competitive Algorithm is used to update the neural network weights. The proposed method has ability of dealing with data jumping and is suitable for any wind power and wind speed foreseeing.