What is it about?

A nonlinear modeling method that can predict future data by learning the relationships of historical data.

Featured Image

Why is it important?

Due to the intermittence of wind power, it is very important to accurately predict it for the stability of power grid. Therefore, the prediction of its size through mathematical modeling can help the power system to reasonably allocate the proportion of wind power in the power grid.


In this paper, aiming at the prediction of an ultrashort-term wind power time series, a new O-KELMmethod with the evolutionary computation strategy is proposed on the basis of single-hidden layer feedforward neural networks. In the O-KELM framework, the input structure of network, kernel parameters, and regularization coefficient are optimized by using the BBO algorithm. Compared to the ELM method, the number of hidden layer nodes need not be given, the corresponding connection weights between hidden layers and input are also not necessary, and the unknown nonlinear feature mapping of the hidden layer is represented with the kernel function. In addition, the output weights of the networks can also be analytically determined using the regularization least squares algorithm, and hence, the KELM method provides better generalization performance at a much fasterlearning speed as well as much easier implementation in applications. In addition, the cosine migration model is also applied to the BBO optimization algorithm, and the optimized BBO1-KELM method further improves the network performance of the KELM than BBO2- KELM using the linear migration model.

meng li

Read the Original

This page is a summary of: Prediction of ultra-short-term wind power based on BBO-KELM method, Journal of Renewable and Sustainable Energy, September 2019, American Institute of Physics,
DOI: 10.1063/1.5113555.
You can read the full text:




The following have contributed to this page