计算机工程与应用 ›› 2015, Vol. 51 ›› Issue (16): 63-69.

• 理论研究、研发设计 • 上一篇    下一篇

基于证据理论优化的态势预测模型

汪永伟1,2,刘育楠1,2,赵荣彩1,常德显1,2,邱  卫1,2   

  1. 1.信息工程大学,郑州 450004
    2.河南省信息安全重点实验室,郑州 450004
  • 出版日期:2015-08-15 发布日期:2015-08-14

Situation forecast model based on evidence theory optimization

WANG Yongwei1,2, LIU Yunan1,2, ZHAO Rongcai1, CHANG Dexian1,2, QIU Wei1,2   

  1. 1.Information Engineering University, Zhengzhou 450004, China
    2.Henan Key Laboratory of Information Security, Zhengzhou 450004, China
  • Online:2015-08-15 Published:2015-08-14

摘要: 针对态势预测的多模型组合问题,提出了一种基于证据理论优化的态势组合预测模型。该模型首先对预测子模型进行训练,获得预测子模型的性能评价与指标权重分配;基于证据理论对多指标的权重分配结果进行融合,提高权重分配的精度;在预测完成后,基于指标可信度和证据理论对指标权重进行调整,优化多指标的评价能力。Matlab实验仿真结果表明,该模型能够依据态势曲线的变化动态优化组合权重,其预测精度优于典型预测模型。

关键词: 证据理论, 组合预测, 优化, 态势

Abstract: According to the problem of multi-model combination in situation forecast, a situation combination forecast model based on evidence theory optimization is proposed. The sub-model is trained to get the performance evaluation and the weight assignment. The weights assignment results of multi-index are fused and refined based on evidence theory. Multi-indexes weight is adjusted based on the confidence and evidence theory after situation forecast circle is finished, optimizing the evaluation ability of multi-indexes. Experiments results in Matlab show that the new model can dynamic adjust the weights in combination model, and it can get better forecast accuracy than typical forecast models.

Key words: evidence theory, combination forecast, optimization, situation