The Application of Variance Contribution Method in Mid-long Term Power

This paper presents a novel approach for long-term electric power load forecasting. A three-layer back propagation(BP) network is designed using the artificial neural network. The idea is to forecast medium and long term power load of Shanxi Province using the ability of ANN of nonlinear modeling. Seven factors are selected as Input Variables for the proposed ANN. The seven factors include GDP, heavy industry production, light industry production, agriculture production, primary industry, secondary industry, tertiary industry. Variance contribution method new defined is used for the optimization selection of correlative factors, and forecasting accuracy is discussed. Simulation results show that the optimization selection of input variables of neural network model is feasible and effective.