Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
ADAPTIVE LOAD PROJECTION
Abstract
Different modeling methodologies, identification, estimation and verification techniques are applied in developing a probabilistic model for the hourly load of an electric power system, for the purpose of short term load forecasting. New techniques have been developed to improve the adaptive capability of recursive algorithms, and suggestions are made for directions of future research. The model assumes the load is given by a discrete, nearly periodic, but non-stationary time series and residual phenomena. The latter is characterized by the output of a discrete time dynamic linear system driven by deterministic input, including weather variables and other phenomena. System identification techniques are used to determine the degree of differencing and the system order to obtain a best model from available data. Several estimation methods are tested with demonstrated results involving load forecast curves for variable periods. Based upon the investigation of the convergence properties of adaptive algorithms for use in modeling non-stationary systems, new algorithms are proposed to achieve better model tracking. Tests are carried out with real load and weather data collected by the Nebraska Public Power District to validate the proposed models' capabilities to forecast hourly load three days ahead and to adaptively track for periods in excess of one year. The conclusion is that the success of the techniques depends very much not only on the structure of the model, and the selection and modification of the independent variables, but also upon the adaptive mechanism. A new time varying algorithm provided the best results. The dissertation includes recommendations as to how to use the available techniques at all modeling stages and includes comments regarding the results obtained usng a wide variety of adaptive strategies. An earlier version of the program used in this dissertation served satisfactorily in a production environment for a period of about one year. The results of this research indicate that the newer model, just placed into production, should produce better results even from the beginning and continue to adapt for much longer periods of time, perhaps indefinitely.
Subject Area
Electrical engineering
Recommended Citation
BELIK, DOBROMIL DAVID, "ADAPTIVE LOAD PROJECTION" (1983). ETD collection for University of Nebraska-Lincoln. AAI8423760.
https://digitalcommons.unl.edu/dissertations/AAI8423760