| 基于学习与证据理论的专家群体预测系统研究 |
| |
| 引用本文： | 朱卫东,杨善林,任明仑.基于学习与证据理论的专家群体预测系统研究[J].预测,2003,22(1):61-63,41. |
| |
| 作者姓名： | 朱卫东 杨善林 任明仑 |
| |
| 作者单位： | 合肥工业大学,管理学院,安徽,合肥,230009 |
| |
| 基金项目： | 国家自然科学基金资助项目 (70 1710 33)，安徽省自然科学基金资助项目 (0 0 0 436 0 7) |
| |
| 摘 要： | 基于学习与证据理论的专家群体预测系统方法，将学习专家预测的历史经验与个体专家预测结论的合成相结合，利用神经网络的学习功能与证据理论在证据合成中的系统完整性，形成了一种系统的专家群体预测方法。该方法将每个专家的预测结论看作为证据，面向具体应用问题通过对历史数据统计分析选择专家、建立学习样本，先考虑个体专家预测结论的独立性、重要性、可靠性与冲突，对个体专家预测结论的基本可信数进行修正，然后用Dempster合成规则合成。该方法用神经网络学习专家历史预测的经验，调整对基本可信数进行修正的修正系数，使证据合成的结果更符合解决问题的要求。实验结果表明本方法应用于解决实际问题时具有较好的效果。
|
| 关 键 词： | 神经网络 证据理论 专家预测 学习功能 |
| 文章编号： | 1003-5192(2003)01-0061-03 |
Experts Group Forecasting System Based on Learning and Theory of Evidence |
| |
| ZHU Wei-dong,YANG Shan-lin,REN Ming-lun.Experts Group Forecasting System Based on Learning and Theory of Evidence[J].Forecasting,2003,22(1):61-63,41. |
| |
| Authors: | ZHU Wei-dong YANG Shan-lin REN Ming-lun |
| |
| Abstract: | The learning and theory of evidence based experts group forecasting system takes advantage of both historic experiences of experts forecasting and the combination of individual forecast conclusion. It uses the learning power of ANN and the system integrity of theory of evidence in evidence combination, forming a systematic group forecasting method. This method uses the forecasting conclusion of individual expert as evidence, and selects experts and set up learning exemplars according to statistics of analysis of historic records. It first considers the independence, importance, reliability and conflicts, adjusts the basic probability numbers of individual expert, then combines these evidences using Dempster's rules. This method uses a neural network to learn experts forecasting experience, adjust the adjustment coefficients, make it more suitable to real environment. An empirical study shows its effect in solving application problems. |
| |
| Keywords: | learning artificial neural networks theory of evidence expert forecasting |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
