Doctoral Dissertations

Keywords and Phrases

Electricity load; Functional principal components; Splines; VARX

Abstract

"Forecasting electricity load is very important to the electric utilities as well as producers of power because accurate predictions can cut down costs by avoiding power shortages or surpluses. Of specific interest is the 24-hour daily electricity load profile, which provides insight into periods of high demand and periods where the use of electricity is at a minimum. Researchers have proposed many approaches to modeling electricity prices, real-time load, and day-ahead demand, with varying success. In this dissertation three new approaches to modeling and forecasting the 24-hour daily electricity load profiles are presented. The application of the proposed methods is illustrated using hourly electricity load data from the Atlantic City Electric (AE) zone, which is part of the Pennsylvania, New Jersey, and Maryland (PJM) electricity market. The first approach that is proposed can be used to make short-term forecasts of electricity load. This approach employs a hybrid technique utilizing autoregressive moving average method (ARMA) and cubic spline models. The second approach is suitable for obtaining long-term forecasts of the daily electricity load and employs cubic splines with time varying coefficients. These coefficients are modeled as a multivariate time series using a vector autoregressive model with exogenous variables to forecast the average daily electricity load profile for a future month. The last approach uses functional principal components to model the daily electricity load profile for each day as a linear combination of three eigenfunctions, with the coefficients of the day-specific linear combinations modeled as univariate time series using transfer functions. The fitted models from the three approaches were applied to data from a subsequent year and the results show that these models perform quite well"--Abstract, page iii.

Advisor(s)

Samaranayake, V. A.

Committee Member(s)

Paige, Robert
Olbricht, Gayla R.
Wen, Xuerong
Gelles, Gregory M

Department(s)

Mathematics and Statistics

Degree Name

Ph. D. in Mathematics

Publisher

Missouri University of Science and Technology

Publication Date

Spring 2016

Pagination

x, 106 pages

Note about bibliography

Includes bibliographic references (pages 102-105).

Rights

© 2016 Abdelmonaem Salem Jornaz, All rights reserved.

Document Type

Dissertation - Open Access

File Type

text

Language

English

Subject Headings

Electric power consumption -- Forecasting -- Econometric modelsElectric power consumption -- Forecasting -- Simulation methodsSplines

Thesis Number

T 10912

Electronic OCLC #

952594798

Download

Share

 
COinS