基于Shapley值的农机总动力组合预测方法  

Combinatorial Forecast of Agricultural Machinery Total Power Based on Shapley Value

作　　者：张淑娟[1] 赵飞[1] 

机构地区：[1]山西农业大学工程技术学院

出　　处：《农业机械学报》2008年第5期60-64,共5页Transactions of the Chinese Society for Agricultural Machinery

基　　金：山西省高校科技研究开发资助项目(项目编号:20051220)

摘　　要：应用合作博弈中的Shapley值方法,通过分配总误差来确定组合预测模型中各预测模型的权重,以此构建组合预测模型并对山西省农机总动力进行组合预测。结果表明,该组合预测模型的预测平均误差为1.81,低于选定的一元线性回归模型、三次指数平滑模型和BP神经网络模型的3.12、2.42和2.23;也低于基于方差倒数法以及基于离异系数法构建的组合预测模型的预测平均误差1.88和1.90。A combinatorial model was established to forecast the total power of agricultural machinery in Shanxi province. Applying the Shapley value method of cooperative game, the weight of each forecast model in the combinatorial forecast model, was determined by distributing the total combinatorial error. The results showed that the forecast average error of combinatorial forecast model is 1.81, which is lower than 3.12, 2.42 and 2.23 of one-variable liner regression model, cubic exponent smooth model and BP neural network model, and is also lower than 1.88 and 1.90 of the combinatorial forecast model based on the variance reciprocal method and the divergence coefficient method.

关 键 词：农机总动力 SHAPLEY值 组合预测 

分 类 号：S232.3[农业科学—农业机械化工程] C931.1[农业科学—农业工程]