Development of artificial neural network models to predict dailygas consumption

摘要

The development of feedforward artificial neural network (ANN)models to predict daily gas consumption is the subject of this paper. Amethodology based on network sensitivities and intuition is discussed.The methodology is applied to two regions in Wisconsin served by theWisconsin Gas Company (WGC). Training results show that ANN modelsreduce prediction root mean squared errors by more than half whencompared with linear regression models. The ANN predictions are comparedwith predictions made by WGC gas controllers for the first 97 days ofthe 1994-1995 heating season. The ANN prediction errors are 82.2% and69.7% of the WGC estimate errors for the two regions