Artificial neural networks for short-term energy forecasting: Accuracy andeconomic value

Sixteen electric utilities surveyed state that use of ANNs significantly re duced errors in daily electric load forecasts, while only three found other wise. Data for five gas utilities reinforces this result: the mean absolute percentage error (MAPE) for ANN daily gas demand forecasts was 6.4%, a 1.9 % improvement over previous methods. Yet ANNs were not always best, implyin g opportunities for further improvement. The economic value of error reduct ion for electric utilities was assessed by examining operating decisions. F or 19 utilities surveyed, an average of $800000/year per utility is estimat ed to be saved from use of ANN-based forecasts. Most benefits resulted from improved generating unit scheduling; the utilities estimated such benefits to be up to $143 annually per peak MW of demand for each 1% improvement in MAPE. This estimated worth of accuracy improvement (roughly 0.1% of annual generation O&M costs) is confirmed by solving generation scheduling and di spatch models under various levels of forecast accuracy. (C) 1998 Elsevier Science B.V. All rights reserved.