基于Shapley值的中国电力需求组合预测模型

Research on Combined Forecasting Model of Electric Power Demand of China Based on Shapley Value

针对中国1997-2013年的电力消费量，采用3次指数平滑模型和ARIMA模型分别进行了建模与拟合分析，通过合作对策理论中的Shapley值法求解了两模型所占权重，从而构建了所需组合模型，并用该组合模型预测了2015年和2016年的中国电力需求量．实例结果表明，组合模型有着更高的拟合精度，拟合相对误差平均值仅为1．52％．

By using the electric power consumption data from 1997 to 2013 of China, cubic exponential smoothing model and ARIMA model are adopted respectively for modeling and fitting analysis. Then the weights to each single forecasting model are allocated by using Shapley value method, and the combined forecasting model is constructed. The results show that the combined model has higher fitting precision, and the average value of relative error is only 1.52%. Finally, by using the combined model ,the electric power demand of China from 2015 to 2016 is predicted.