| 佟为明,李方,李中伟,王铁成.基于焦元信度折扣和概率排序的信息融合决策方法[J].高技术通讯(中文),2012,22(6):638~643 |
| 基于焦元信度折扣和概率排序的信息融合决策方法 |
| An information fusion and decision method based on belief discount and probability ranking of focal elements |
| 修订日期：2011-04-21 |
| DOI： |
| 中文关键词: 室内定位， 证据预处理， Pignistic概率， 信度折扣， 概率排序 |
| 英文关键词: indoor positioning, evidence pretreatment, Pignistic probability, belief discount, probability ranking |
| 基金项目:国家自然科学基金（51077015）和黑龙江省自然科学基金（E200914）资助项目 |
| 作者 | 单位 | | 佟为明 | 哈尔滨工业大学电气工程及自动化学院 | | 李方 | 哈尔滨工业大学电气工程及自动化学院 | | 李中伟 | 哈尔滨工业大学电气工程及自动化学院 | | 王铁成 | 哈尔滨工业大学电气工程及自动化学院 |
|
| 摘要点击次数: 2783 |
| 全文下载次数: 2036 |
| 中文摘要: |
| 针对基于贝叶斯推理的定位方法难于获取节点位置先验信息的问题，提出了一种应用于室内定位的Dempster Shafer信息融合与决策方法。采用Pignistic概率反映接入点证据在焦元信度分配上的冲突并根据冲突程度对信度打折，从而实现对冲突证据的预处理。通过合成预处理后的证据可消除定位过程中出现的一票否决和Zadeh悖论现象。根据焦元各自信任区间确定的分布函数，分别计算两焦元信度在整个信任区间上降序排列的概率，并将其作为在信任区间发生重叠的焦元间进行决策的依据。算例结果表明，上述方法对冲突证据的合成更加符合证据理论的本质，且收敛速度快，决策风险低。 |
| 英文摘要: |
| Considering that it is hard to obtain the prior information of node location using Bayesian inference based positioning methods, a Dempster Shafer based information fusion and decision method for indoor positioning is proposed in the paper. The Pignistic probability is adopted to reflect the conflict on belief assignment. The evidence pretreatment using belief discount of the focal elements is implemented according to the conflict degree. The method can eliminate one ballot veto and Zadeh’s paradox during positioning by fusing pretreated evidences. According to the distribution functions of two focal elements on their own belief intervals, the probabilities ranking in a descending order on the whole belief interval are calculated respectively and taken as the basis of decision making for the focal elements interlaping on the belief intervals. The results of the examples indicate that the proposed method is consistent with the essence of evidence theory in the fusion of conflict evidences, especially, it can accelerate the convergence and reduce the decision risk. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
