Load forecasting based on kernel-based orthogonal projections to latent structures

Author

Lingcai Kong ; Yanpeng Ma

Author_Institution

Dept. of Math. & Phys., North China Electr. Power Univ., Baoding, China

Volume

9

fYear

2011

fDate

12-14 Aug. 2011

Firstpage

4451

Lastpage

4454

Abstract

The Kernel-based orthogonal projections to latent structures (K-OPLS) model is a recent novel data analysis method for both regression and classification. Compared with the classical orthogonal projections to latent structures (OPLS), it utilizes the kernel Gram matrix as a replacement of descriptor matrix to use the partial least squares (PLS) model. This enables it can effectively improve predictive performance, considerably in such situations where strong non-linear relationships between descriptor and response variables while retaining the OPLS model framework. In this paper, we first introduce the K-OPLS model. And then, a load forecasting model based on K-OPLS is proposed.

Keywords

least squares approximations; load forecasting; K-OPLS; Kernel-based orthogonal projections; PLS model; load forecasting; orthogonal projections to latent structures; partial least squares model; Autoregressive processes; Data models; Kernel; Load forecasting; Load modeling; Mathematical model; Predictive models; kernel PLS; load forecasting; orthogonal signal correction; partial least square;

fLanguage

English

Publisher

ieee

Conference_Titel

Electronic and Mechanical Engineering and Information Technology (EMEIT), 2011 International Conference on

Conference_Location

Harbin, Heilongjiang

Print_ISBN

978-1-61284-087-1

Type

conf

DOI

10.1109/EMEIT.2011.6023132

Filename

6023132