Long-term load forecasting in electric power idustry /
The objective of this research study is to develop methods to improve long-term load forecasting practice in electric power industry. This objective is accomplished through the development of more accurate long-term load forecasts as well as more accurate assessment of the uncertainties involved. Sc...
| Main Author: | |
|---|---|
| Format: | Thesis Book |
| Language: | English |
| Published: |
[Place of publication not identified] :
[publisher not identified] ;
1997.
|
| Subjects: | |
| Online Access: | http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=731678061&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD |
| Summary: | The objective of this research study is to develop methods to improve long-term load forecasting practice in electric power industry. This objective is accomplished through the development of more accurate long-term load forecasts as well as more accurate assessment of the uncertainties involved. Scarcity of the historical observations imposes restrictions on effective solutions of the long-term load forecasting problem. This study shows that nonlinear estimators of controlled complexity are viable tools in long-term load forecasting. The estimator complexity level should be large enough to capture the general historical trends. However, it should be small enough to avoid memorization of all the historical events. A Monte Carlo filtering process is proposed to optimize different estimator structures and parameters, and to select a level of complexity. Forecasts developed using the proposed method outperformed the currently available forecasts in four Texas utilities in 85% of the developed models. Moreover, large errors in the forecasts of exogenous variables do not result in large errors in forecasts of energy sales or peak load demand, as it is observed in linear forecasting methods. An inherent characteristic of long-term load forecasting is the uncertainty in the identified model structure, model parameters, as well as future projection of different exogenous variables. An uncertainty assessment framework is proposed based on statistical bootstrapping methods. The bootstrapping method does not place any restriction on the probability distribution of the forecasting model residuals. The proposed uncertainty method is used to compute the variability of the energy sales or peak load demand. This becomes a measure of uncertainty in the forecasting model structure, parameters and future values of the exogenous variables. While the proposed forecasting and uncertainty assessment techniques generally result in lower forecast bias, no guarantee of narrower forecast variance can be given. That is, no well-defined relation between the forecast bias and forecast variance was observed. Using the developed uncertainty assessment framework, an assessment of uncertainty in energy sales in different classes of customers and peak load demand in the City of Bryan, Texas was performed. |
|---|---|
| Item Description: | Vita. "Major Subject: Electrical Engineering". |
| Physical Description: | xvii, 269 leaves : illustrations ; 28 cm. Issued also on microfiche from University Microfilm Inc. |
| Bibliography: | Includes bibliographical references (leaves 263-268). |