Title: Wind speed prediction and error distribution based on rational sample organisation for Elman recursion neural networks

Authors: Junfang Li; Buhan Zhang; Chengxiong Mao

Addresses: Electrical Power Security and High Efficiency Lab, Huazhong University of Science and Technology, Wuhan, 430074, China. ' Electrical Power Security and High Efficiency Lab, Huazhong University of Science and Technology, Wuhan, 430074, China. ' Electrical Power Security and High Efficiency Lab, Huazhong University of Science and Technology, Wuhan, 430074, China

Abstract: Wind speed prediction and the stochastic distribution characteristic of prediction error of wind speed are important for power system operation and planning. This paper presents a rational and efficient sample organisation method when using Elman recursion neural network (ERNN) for one step ahead average ten-minute wind speed prediction. After introducing the prediction model based on the ERNN, a rational sample organisation method is presented to make ERNN feasible. Prediction error of wind speed is simply assumed following Gaussian distribution in the previous articles, this paper exemplifies that the prediction error is more likely to follow Weibull distribution rather than Gaussian distribution. Compared with back-propagation (BP) network, the effectiveness of the ERNN is tested on a case about a Chinese wind farm with the historical data using MATLAB software. The case shows that the prediction model and method are effective for one step ahead average ten-minute wind speed prediction.

Keywords: wind speed prediction; Elman recursion neural networks; ERNNs; wind farms; power systems; Weibull distribution; wind power; wind energy; error distribution.

DOI: 10.1504/IJAMECHS.2011.045006

International Journal of Advanced Mechatronic Systems, 2011 Vol.3 No.5/6, pp.346 - 354

Published online: 18 Mar 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article