Simultaneous day-ahead forecasting of electricity price and load in smart grids

H. Shayeghi, A. Ghasemi, M. Moradzadeh, M. Nooshyar

Research output: Contribution to journalArticle

47 Citations (Scopus)

Abstract

In smart grids, customers are promoted to change their energy consumption patterns by electricity prices. In fact, in this environment, the electricity price and load consumption are highly corrected such that the market participants will have complex model in their decisions to maximize their profit. Although the available forecasting mythologies perform well in electricity market by way of little or no load and price interdependencies, but cannot capture load and price dynamics if they exist. To overcome this shortage, a Multi-Input Multi-Output (MIMO) model is presented which can consider the correlation between electricity price and load. The proposed model consists of three components known as a Wavelet Packet Transform (WPT) to make valuable subsets, Generalized Mutual Information (GMI) to select best input candidate and Least Squares Support Vector Machine (LSSVM) based on MIMO model, called LSSVM-MIMO, to make simultaneous load and price forecasts. Moreover, the LSSVM-MIMO parameters are optimized by a novel Quasi-Oppositional Artificial Bee Colony (QOABC) algorithm. Some forecasting indices based on error factor are considered to evaluate the forecasting accuracy. Simulations carried out on New York Independent System Operator, New South Wales (NSW) and PJM electricity markets data, and showing that the proposed hybrid algorithm has good potential for simultaneous forecasting of electricity price and load.

Original languageEnglish
Pages (from-to)371-384
Number of pages14
JournalEnergy Conversion and Management
Volume95
Early online date4 Mar 2015
DOIs
Publication statusPublished - 1 May 2015
Externally publishedYes

Fingerprint

Electricity
Support vector machines
Set theory
Profitability
Energy utilization
Mathematical transformations
Power markets

Cite this

Shayeghi, H. ; Ghasemi, A. ; Moradzadeh, M. ; Nooshyar, M. / Simultaneous day-ahead forecasting of electricity price and load in smart grids. In: Energy Conversion and Management. 2015 ; Vol. 95. pp. 371-384.
@article{675f8f63efef4bc183cf7ac0cb2da0b7,
title = "Simultaneous day-ahead forecasting of electricity price and load in smart grids",
abstract = "In smart grids, customers are promoted to change their energy consumption patterns by electricity prices. In fact, in this environment, the electricity price and load consumption are highly corrected such that the market participants will have complex model in their decisions to maximize their profit. Although the available forecasting mythologies perform well in electricity market by way of little or no load and price interdependencies, but cannot capture load and price dynamics if they exist. To overcome this shortage, a Multi-Input Multi-Output (MIMO) model is presented which can consider the correlation between electricity price and load. The proposed model consists of three components known as a Wavelet Packet Transform (WPT) to make valuable subsets, Generalized Mutual Information (GMI) to select best input candidate and Least Squares Support Vector Machine (LSSVM) based on MIMO model, called LSSVM-MIMO, to make simultaneous load and price forecasts. Moreover, the LSSVM-MIMO parameters are optimized by a novel Quasi-Oppositional Artificial Bee Colony (QOABC) algorithm. Some forecasting indices based on error factor are considered to evaluate the forecasting accuracy. Simulations carried out on New York Independent System Operator, New South Wales (NSW) and PJM electricity markets data, and showing that the proposed hybrid algorithm has good potential for simultaneous forecasting of electricity price and load.",
keywords = "Generalized mutual information, Load and price forecasting, MIMO predictor, QOABC algorithm, Smart grids, Wavelet packet transform",
author = "H. Shayeghi and A. Ghasemi and M. Moradzadeh and M. Nooshyar",
year = "2015",
month = "5",
day = "1",
doi = "10.1016/j.enconman.2015.02.023",
language = "English",
volume = "95",
pages = "371--384",
journal = "Energy Conversion and Management",
issn = "0196-8904",
publisher = "Elsevier Limited",

}

Simultaneous day-ahead forecasting of electricity price and load in smart grids. / Shayeghi, H.; Ghasemi, A.; Moradzadeh, M.; Nooshyar, M.

In: Energy Conversion and Management, Vol. 95, 01.05.2015, p. 371-384.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Simultaneous day-ahead forecasting of electricity price and load in smart grids

AU - Shayeghi, H.

AU - Ghasemi, A.

AU - Moradzadeh, M.

AU - Nooshyar, M.

PY - 2015/5/1

Y1 - 2015/5/1

N2 - In smart grids, customers are promoted to change their energy consumption patterns by electricity prices. In fact, in this environment, the electricity price and load consumption are highly corrected such that the market participants will have complex model in their decisions to maximize their profit. Although the available forecasting mythologies perform well in electricity market by way of little or no load and price interdependencies, but cannot capture load and price dynamics if they exist. To overcome this shortage, a Multi-Input Multi-Output (MIMO) model is presented which can consider the correlation between electricity price and load. The proposed model consists of three components known as a Wavelet Packet Transform (WPT) to make valuable subsets, Generalized Mutual Information (GMI) to select best input candidate and Least Squares Support Vector Machine (LSSVM) based on MIMO model, called LSSVM-MIMO, to make simultaneous load and price forecasts. Moreover, the LSSVM-MIMO parameters are optimized by a novel Quasi-Oppositional Artificial Bee Colony (QOABC) algorithm. Some forecasting indices based on error factor are considered to evaluate the forecasting accuracy. Simulations carried out on New York Independent System Operator, New South Wales (NSW) and PJM electricity markets data, and showing that the proposed hybrid algorithm has good potential for simultaneous forecasting of electricity price and load.

AB - In smart grids, customers are promoted to change their energy consumption patterns by electricity prices. In fact, in this environment, the electricity price and load consumption are highly corrected such that the market participants will have complex model in their decisions to maximize their profit. Although the available forecasting mythologies perform well in electricity market by way of little or no load and price interdependencies, but cannot capture load and price dynamics if they exist. To overcome this shortage, a Multi-Input Multi-Output (MIMO) model is presented which can consider the correlation between electricity price and load. The proposed model consists of three components known as a Wavelet Packet Transform (WPT) to make valuable subsets, Generalized Mutual Information (GMI) to select best input candidate and Least Squares Support Vector Machine (LSSVM) based on MIMO model, called LSSVM-MIMO, to make simultaneous load and price forecasts. Moreover, the LSSVM-MIMO parameters are optimized by a novel Quasi-Oppositional Artificial Bee Colony (QOABC) algorithm. Some forecasting indices based on error factor are considered to evaluate the forecasting accuracy. Simulations carried out on New York Independent System Operator, New South Wales (NSW) and PJM electricity markets data, and showing that the proposed hybrid algorithm has good potential for simultaneous forecasting of electricity price and load.

KW - Generalized mutual information

KW - Load and price forecasting

KW - MIMO predictor

KW - QOABC algorithm

KW - Smart grids

KW - Wavelet packet transform

UR - http://www.scopus.com/inward/record.url?scp=84924166400&partnerID=8YFLogxK

UR - https://www.journals.elsevier.com/energy-conversion-and-management

U2 - 10.1016/j.enconman.2015.02.023

DO - 10.1016/j.enconman.2015.02.023

M3 - Article

VL - 95

SP - 371

EP - 384

JO - Energy Conversion and Management

JF - Energy Conversion and Management

SN - 0196-8904

ER -