摘要
在现有组合预测模型中以线性或非线性规划方法求解优化问题遇到了较多的困难,而递归组合预测方法的过程又较为烦琐。针对这一问题,本文基于人工神经网络技术建立了一种新的非线性组合预测模型;并从避免不规则权重入手,提出了通过主分量分析提高参与组合预测信息有效性的方法,提高了预测精度。实践表明,效果是令人满意的。
This paper discusses the compound-forecasting method,which is the hotspot in current prediction research.The paper first sets up a nonlinear compound forecasting model based on ANN,then presents the idea of extracting effective information with individual forecasting method to participate in compound-forecasting.Satisfactory results have been achieved in practice.
出处
《西安邮电学院学报》
1999年第1期37-41,36,共6页
Journal of Xi'an Institute of Posts and Telecommunications
关键词
组合预测
人工神经网络
主分量分析
Compound-forecasting
ANN
Principal Component Analysis