Application of Improved GM(1,N) Models in Annual Electricity Demand Forecasting

This paper presents two improved models based on the first-order multi-variable grey model (GM(1,N)) for forecasting the electricity demand. The first model named IGM1(1,N) is developed through the optimization of background value by Lagrange mean value theorem (LMVT). Another model named IGM2(1,N) is established through the calculation of its boundary value using least square method (LSM). Despite of the uncertain external factors, the two models can ensure the prediction accuracy without requiring too much input data. Then grey correlation analysis method is used to choose the key external factors that have great influence on the electricity demand. Finally, the improved models are evaluated by forecasting the annual electricity sales of Guangzhou, China. The effectiveness of the improved models is validated by comparing with that of the general first-order one-variable grey model (GM(1,1)) and general GM(1,N), respectively.