Kernel methods for short-term spatio-temporal wind prediction

Jethro Dowell; Stephan Weiss; David Infield

doi:10.1109/PESGM.2015.7285965

Kernel methods for short-term spatio-temporal wind prediction

Jethro Dowell, Stephan Weiss, David Infield

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution book

5 Citations (Scopus)

89 Downloads (Pure)

Abstract

Two nonlinear methods for producing short-term spatio-temporal wind speed forecast are presented. From the relatively new class of kernel methods, a kernel least mean squares algorithm and kernel recursive least squares algorithm are introduced and used to produce 1 to 6 hour-ahead predictions of wind speed at six locations in the Netherlands. The performance of the proposed methods are compared to their linear equivalents, as well as the autoregressive, vector autoregressive and persistence time series models. The kernel recursive least squares algorithm is shown to offer significant improvement over all benchmarks, particularly for longer forecast horizons. Both proposed algorithms exhibit desirable numerical properties and are ripe for further development.

Original language	English
Title of host publication	2015 IEEE Power and Energy Society General Meeting
Publisher	IEEE
Number of pages	5
ISBN (Print)	9781467380409
DOIs	https://doi.org/10.1109/PESGM.2015.7285965
Publication status	Published - 30 Sept 2015
Event	IEEE Power and Energy Society General Meeting, PESGM 2015 - Denver, United States Duration: 26 Jul 2015 → 30 Jul 2015

Conference

Conference	IEEE Power and Energy Society General Meeting, PESGM 2015
Country/Territory	United States
City	Denver
Period	26/07/15 → 30/07/15

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 7 Affordable and Clean Energy

Keywords

least squares approximation
wind power
kernel method
nonlinear methods

Access to Document

10.1109/PESGM.2015.7285965

Dowell-Weiss-Infield-IEEEPESGM2015-kernel-methods-for-short-term-spatio-temporal-windAccepted author manuscript, 247 KB

Cite this

@inproceedings{f058bc7458624a92aa7483a54b042e7e,

title = "Kernel methods for short-term spatio-temporal wind prediction",

abstract = "Two nonlinear methods for producing short-term spatio-temporal wind speed forecast are presented. From the relatively new class of kernel methods, a kernel least mean squares algorithm and kernel recursive least squares algorithm are introduced and used to produce 1 to 6 hour-ahead predictions of wind speed at six locations in the Netherlands. The performance of the proposed methods are compared to their linear equivalents, as well as the autoregressive, vector autoregressive and persistence time series models. The kernel recursive least squares algorithm is shown to offer significant improvement over all benchmarks, particularly for longer forecast horizons. Both proposed algorithms exhibit desirable numerical properties and are ripe for further development.",

keywords = "least squares approximation, wind power, kernel method, nonlinear methods",

author = "Jethro Dowell and Stephan Weiss and David Infield",

note = "(c) 2015 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.; IEEE Power and Energy Society General Meeting, PESGM 2015 ; Conference date: 26-07-2015 Through 30-07-2015",

year = "2015",

month = sep,

day = "30",

doi = "10.1109/PESGM.2015.7285965",

language = "English",

isbn = "9781467380409",

booktitle = "2015 IEEE Power and Energy Society General Meeting",

publisher = "IEEE",

}

TY - GEN

T1 - Kernel methods for short-term spatio-temporal wind prediction

AU - Dowell, Jethro

AU - Weiss, Stephan

AU - Infield, David

N1 - (c) 2015 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.

PY - 2015/9/30

Y1 - 2015/9/30

N2 - Two nonlinear methods for producing short-term spatio-temporal wind speed forecast are presented. From the relatively new class of kernel methods, a kernel least mean squares algorithm and kernel recursive least squares algorithm are introduced and used to produce 1 to 6 hour-ahead predictions of wind speed at six locations in the Netherlands. The performance of the proposed methods are compared to their linear equivalents, as well as the autoregressive, vector autoregressive and persistence time series models. The kernel recursive least squares algorithm is shown to offer significant improvement over all benchmarks, particularly for longer forecast horizons. Both proposed algorithms exhibit desirable numerical properties and are ripe for further development.

AB - Two nonlinear methods for producing short-term spatio-temporal wind speed forecast are presented. From the relatively new class of kernel methods, a kernel least mean squares algorithm and kernel recursive least squares algorithm are introduced and used to produce 1 to 6 hour-ahead predictions of wind speed at six locations in the Netherlands. The performance of the proposed methods are compared to their linear equivalents, as well as the autoregressive, vector autoregressive and persistence time series models. The kernel recursive least squares algorithm is shown to offer significant improvement over all benchmarks, particularly for longer forecast horizons. Both proposed algorithms exhibit desirable numerical properties and are ripe for further development.

KW - least squares approximation

KW - wind power

KW - kernel method

KW - nonlinear methods

UR - https://www.scopus.com/pages/publications/84956852712

U2 - 10.1109/PESGM.2015.7285965

DO - 10.1109/PESGM.2015.7285965

M3 - Conference contribution book

AN - SCOPUS:84956852712

SN - 9781467380409

BT - 2015 IEEE Power and Energy Society General Meeting

PB - IEEE

T2 - IEEE Power and Energy Society General Meeting, PESGM 2015

Y2 - 26 July 2015 through 30 July 2015

ER -

Kernel methods for short-term spatio-temporal wind prediction

Abstract

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Doctoral training centre in wind energy systems

Cite this

Kernel methods for short-term spatio-temporal wind prediction

Abstract

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Projects

Doctoral training centre in wind energy systems

Cite this