Artificial neural networks for short-term electric demand forecasting: Accuracy and economic value
- Johns Hopkins Univ., Baltimore, MD (United States)
- Case Western Reserve Univ., Cleveland, OH (United States)
- Electric Power Research Inst., Palo Alto, CA (United States)
Artificial neural networks (ANNs) are increasingly used by utilities to forecast short-term (1 to 7 day ahead) demands for electricity. Over the past decade, interest in improving the accuracy of short-term demand forecasting has sharply increased. This interest is stimulated in part by competition, which has motivated efforts to save costs, and in part by methodological advances. ANNs are exceptionally promising because they do not require specification of a particular functional relationship between inputs and outputs, and because of their ability to adapt as new data becomes available. In the first section of this paper, the authors report the results of a survey which quantifies the extent to which use of ANNs can improve upon current forecasting practice. Yet is the effort and expense required to develop and use these new systems justified by the benefits they bring? Improved accuracy has value only if it leads to better decisions. The authors attempt to answer that question in the remainder of the paper by, first, surveying electric utilities on their uses of forecasts and benefits they perceive from using ANNs and, second, simulating how improved accuracy lowers expected power generation costs. The results show that, on average, electric utilities can realize considerable savings by adopting more accurate ANN-based forecasting systems, but that the potential savings can vary greatly from utility to utility.
- OSTI ID:
- 319019
- Report Number(s):
- CONF-980426-; TRN: IM9909%%163
- Resource Relation:
- Conference: American power conference, Chicago, IL (United States), 14-16 Apr 1998; Other Information: PBD: 1998; Related Information: Is Part Of Proceedings of the American Power Conference: Volume 60-1; McBride, A.E. [ed.] [Illinois Inst. of Tech., Chicago, IL (United States)]; PB: 613 p.
- Country of Publication:
- United States
- Language:
- English
Similar Records
Quantification of Energy Savings and Demand Reduction for a Heat Pump Integrated with Thermal Energy Storage (Final Report)
Predictive Optimal Control of Active and Passive Building Thermal Storage Inventory