|Title:||Comparison of electric load forecasting algorithms|
|Subject:||Electric power-plants -- Load -- Forecasting|
Electric power-plants -- Load -- Data processing
Electric power-plants -- Load -- Mathematical models
Electric power consumption -- Forecasting
Hong Kong Polytechnic University -- Dissertations
Department of Electrical Engineering
|Pages:||vii, 80 leaves : ill. (some col.) ; 30 cm|
|Abstract:||An intelligent approaches to electric load forecasting which combines the techniques of Box-Jenkins (BJ) modeling with the evolutionary programming (EP) and neural network (NN) algorithms are presented. Box-Jenkins models have been shown to produce better forecasts than any other time series model especially when the data series contains both seasonal and nonseasonal patterns. However, many statisticians and analysts have been reluctant to use it because the modeling procedures are complicated which requires a great deal of experience and a thorough understanding on the theoretical features of the autocorrealation functions and the partial autocorrelation functions. In general, BJ modeling procedures consist of a four-stage cycle: model identification, parameter estimation, diagnostic checking, and validation forecasts. In this dissertation, the complex BJ model identification problem is addressed by using EP. EP is a new adaptive search technique which simulates the natural evolutionary process to reach the fittest individuals after repeated mutation, competition and selection procedures. It has the ability in determining the global minimum and thus leading to optimal model selection. By using this approach, it is attempted to simplify the BJ identification procedures, and at the same time, achieving the optimal load forecasting model. At the estimation stage, instead of using the traditional gradient search approach like nonlinear least square, the EP and NN algorithms are adopted separately for selecting the coefficients of the tentative model. NN is another well-known intelligent algorithm which has proved to be successful in pattern recognition. However, it has seldom been applied on parameter estimation, so the result of this dissertation can serve as a good reference for future research work. The performance of these approaches is investigated by using the monthly electricity consumption of Hong Kong at the validation forecasting stage. It is found that the forecast error are generally better than that of the traditional approach as obtained by using the SPSS commercial software package.|
|Rights:||All rights reserved|
Files in This Item:
|b14854260.pdf||For All Users (off-campus access for PolyU Staff & Students only)||6.31 MB||Adobe PDF||View/Open|
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item: