Knowledge-based and statistical load forecast model development and analysis
| dc.contributor.author | Moghram, Ibrahim Said | en |
| dc.contributor.committeechair | Rahman, Saifur | en |
| dc.contributor.committeemember | Besieris, Ioannis M. | en |
| dc.contributor.committeemember | Foutz, Robert V. | en |
| dc.contributor.committeemember | Mashburn, William H. | en |
| dc.contributor.committeemember | De La Ree, Jaime | en |
| dc.contributor.department | Electrical Engineering | en |
| dc.date.accessioned | 2015-07-09T20:43:24Z | en |
| dc.date.available | 2015-07-09T20:43:24Z | en |
| dc.date.issued | 1989 | en |
| dc.description.abstract | Most of the techniques that have been applied to the short-term load forecasting problem fall within the time series approaches. The exception to this has been a new approach based on the application of expert systems. Recently several techniques have been reported which apply the rule-based (or expert systems) approach to the short-term load forecasting problem. However, the maximum lead time used for these forecasts has not gone beyond 48 hours, even though there is a significant difference between these algorithms in terms of their data base requirements (few weeks to 10 years). The work reported in this dissertation deals with two aspects. The first one is the application of rule-based techniques to weekly load forecast. A rule-based technique is presented that is capable of issuing a 168-hour lead-time load forecast. The second aspect is the development of a comprehensive load forecasting system that utilizes both the statistical and rule-based approaches. This integration overcomes the deficiencies that exist in both of these modeling techniques. The load forecasting technique is developed using two parallel approaches. In the first approach expert information is used to identify weather variables, day types and diurnal effects that influence the electrical utility load. These parameters and hourly historical loads are then selectively used for various statistical techniques (e.g., univariate, transfer function and linear regression). A weighted average load forecast is then produced which judiciously combines the forecasts from these three techniques. The second approach, however, is free of any significant statistical computation, and is based totally on rules derived from electric utility experts. The data base requirement for any of these approaches do not extend more than four weeks ol hourly load, dry bulb and dew point temperatures. When the algorithms are applied to generate seven-day ahead load forecasts for summer (August) and winter (February) the average forecast errors for the month come under 3%. | en |
| dc.description.degree | Ph. D. | en |
| dc.format.extent | xvi, 273 leaves | en |
| dc.format.mimetype | application/pdf | en |
| dc.identifier.uri | http://hdl.handle.net/10919/54247 | en |
| dc.language.iso | en_US | en |
| dc.publisher | Virginia Polytechnic Institute and State University | en |
| dc.relation.isformatof | OCLC# 21329073 | en |
| dc.rights | In Copyright | en |
| dc.rights.uri | http://rightsstatements.org/vocab/InC/1.0/ | en |
| dc.subject.lcc | LD5655.V856 1989.M633 | en |
| dc.subject.lcsh | Electric power-plants -- Load | en |
| dc.subject.lcsh | Electric power systems -- Load dispatching | en |
| dc.title | Knowledge-based and statistical load forecast model development and analysis | en |
| dc.type | Dissertation | en |
| dc.type.dcmitype | Text | en |
| thesis.degree.discipline | Electrical Engineering | en |
| thesis.degree.grantor | Virginia Polytechnic Institute and State University | en |
| thesis.degree.level | doctoral | en |
| thesis.degree.name | Ph. D. | en |
Files
Original bundle
1 - 1 of 1