Predicting electricity consumption based on optimized model of GM(1,1)

Optimized GM(1,1) model based on least absolute criteria is proposed in this paper. Since the initial condition of original GM(1,1) model is not very suitable, we use the modified latest data which generating from the accumulative generating operation as the new initial condition. And the least absolute criteria is applied instead of least square criteria to improve the stability and prediction accuracy of GM(1,1) model. Then the particle swarm optimization is adapted to the parameters optimization. At the end, the optimized GM(1,1) model is used to predict the whole social electricity consumption of China and the result shows its prediction accuracy is better than the original model and the GM(1,1) model with latest initial condition.