摘要
多传感器系统通常能够获取各种不同的量测数据,但是其信息的准确性和可靠性往往难以被保证,使用这些数据所做出的决策很有可能与事实相悖。鉴于此,在D-S证据理论框架内提出了一种基于改进Jousselme证据距离的多传感器决策融合方法。通过对相似性Jaccard系数矩阵分块化处理,以合理准确地描述传感器节点证据冲突,并借此计算各传感器节点的权值来修正证据源,最终通过D-S融合规则得到正确决策。数值实验结果显示提出方法的识别率最高可达92.52%,相比Muphy法高出了17.28%,而不确定度却降低了2个数量级,不但能够快速准确地识别传感器节点证据冲突,而且决策风险更小,因此适用范围更广。
Multi-sensor systems are able to obtain various measurement data,but it is difficult to guarantee their accuracy and reliability,the decision-makings using these data are likely to be contrary to the facts.In view of this,an approach to multi-sensor decision fusion based on improved Jousselme evidence distance was proposed in the framework of D-S evidence theory in this paper.By rationally dividing the similarity Jaccard coefficient matrix,the conflicted sensor node evidences were described accurately and the weight of each sensor node was calculated to modify the evidence source.Finally,correct decisions were made through D-S fusion rules.Numerical experimental results demonstrate that the highest recognition rate of the proposed method can be up to 92.52%,which is higher than that of Muphy's method by 17.28%.The uncertainty is reduced by two orders of magnitude,the proposed method not only recognizes the conflicted sensor node evidences quickly,but also has less risk of decision-makings within the wider applications.
作者
张雅媛
孙力帆
郑国强
ZHANG Ya-yuan;SUN Li-fan;ZHENG Guo-qiang(School of Information Engineering,Henan University of Science and Technology,Henan Key Laboratory of Robot and Intelligent Systems,Luoyang 471023,China)
出处
《仪表技术与传感器》
CSCD
北大核心
2019年第7期82-87,共6页
Instrument Technique and Sensor
基金
国家自然科学基金项目(U1504619,61573020,61473115)
河南省科技攻关计划项目(182102110397)
教育部产学合作协同育人项目(201602011006)
