Abstract: To solve the problem of the limited information and the large errors using the single prediction model, a weights determination method of the combinatorial forecast model is proposed based on combining the degree of grey incidence and Induced Ordered Weighted Harmonic Averaging（IOWHA） operator and the optimal weighted theory. Using the weights determination method, an optimal combination forecasting model is proposed based on the RBF neural network model and the GM model. The model can overcome two deficiencies of traditional combination forecasting method in the invariable weighting and using single error index as the principle. In the paper, combination forecasting of the logistics demand of China is given by using this model which is compared with the forecasting result of the RBF neural network model and the GM model. The result shows that this combination forecasting model is more accurate, and is an effective logistics demand forecasting model.

Key words: combination forecasting, Induced Ordered Weighted Harmonic Averaging（IOWHA） operator, degree of grey incidence, superior combination forecasting

摘要： 针对现有单项预测模型提供信息有限，预测误差大的问题，引用最优加权组合建模理论，将灰色关联度与IOWHA算子相结合，提出一种新的组合预测模型权重确定方法，并应用该权重确定方法构建了一种基于RBF神经网络预测模型和GM预测模型的最优组合预测模型。该模型能够克服传统组合预测方法的两个缺陷：加权平均系数不变和以单一误差指标为准则。利用该组合模型对全国物流需求进行组合预测，并与RBF神经网络模型、GM模型的预测结果进行了对比分析。结果表明，相对于单项预测模型，该组合预测模型的预测精度更高，是一种有效的物流需求预测模型。

关键词: 组合预测, 诱导有序加权调和平均（IOWHA）算子, 灰色关联度, 优性组合预测