摘要
DSmT可以很好地解决高冲突证据的组合问题,但是它存在计算量大和主焦元的mass函数难以收敛的问题.针对这些不足,提出快速mass函数收敛算法,在融合过程中只计算辨识框架中单元素形成焦元所占的基本置信指派,并对其做归一化处理,重构mass函数,大大降低了计算量的同时也使mass函数迅速收敛向规定的阈值,以利于快速准确地进行归类判决.计算量的比较和仿真算例证明了该算法的有效性.
DSmT can solve the problems of high conflict evidence combination, but it has the disadvan- tage that the calculation is too large and the mass value of main element is difficult to converge. In order to solve this problem, a fast mass function convergence algorithm was proposed, which just calculates the basic belief of the single element in the frame of discernment to normalize them, and then rebuild the mass value, thus it can greatly reduce the calculation and make the mass function quickly converge to a certain threshold, which is convenient to quickly and exactly recognize and sort the targets . The comparison and simulation computation example have proved the validity of this algorithm.
出处
《湖南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2011年第1期89-92,共4页
Journal of Hunan University:Natural Sciences
基金
国家自然科学基金重点资助项目(60634030)
国家自然科学基金资助项目(60702066)
教育部新世纪优秀人才项目(NCET-06-0878)
航空科学基金资助项目(20090853013)
陕西省自然科学基础研究计划资助项目(2010JQ8032)
西北工业大学科技创新基金资助项目(2008KJ02025)
关键词
归类判决
DSMT
MASS函数
cluster computing
Dezert-Smaradache Thoery（DSmT）
mass function
