【摘要】 阐述了组合预测方法的基本原理,并从理论上论证了组合预测方法相对于单一预测模型的优越性。针对以往的组合预测模型最优权重不能保证非负性的问题,提出了一种新的基于前向神经网络模型的最优权重确定方法,并引入了神经网络优选组合预测模型。通过分析和比较,此模型具有很强的自适应性和较高的预测精度。更多还原

【Abstract】 This paper elaborates on the principles of combination prediction,and it demonstrates the superiority of combination prediction to single prediction model.A new optimization weight method based on feed forward neural network is proposed here against the fact that former combination prediction model cant not guarantee the negative weight,and at the same time it introduces the optimization combination model for neural network.By analysis and comparison,this model has a strong self-adaptability and high prediction accuracy.更多还原