基于TBM双层融合架构的航路属性异常检测

发布时间：2018-02-28 04:49

本文关键词： 信息融合航迹关联异常信源决策理论　出处：《电子学报》2017年03期 　论文类型：期刊论文

【摘要】：航路飞行目标的属性异常检测是确保及时发现飞行异常的关键问题.常用的概率框架需要受到先验信息的局限.可传递置信模型(Transferable Belief Model,TBM)不需要先验信息,能高效处理异质信息,但是传统的TBM无法处理时间上的不连续与不确定性,因此针对异常航路目标检测问题,将马尔可夫模型与TBM框架结合,建立了基于TBM的双层融合架构,实现了多特征融合航路属性异常检测.第一层是通过对多属性冲突信息的分析,实现对多特征的检测,并通过特征贡献度分析,对多特征信息进行打折后再融合;第二层是通过在时间序列上的指派融合,对比预测值和观测值差异,检测航路目标异常变化.仿真试验验证,在切换航路场景与偏离回归场景中,相较动态证据推理方法,本文方法具有更好的决策准确性与时间精确度.
[Abstract]:Attribute anomaly detection of flight target is the key problem to ensure the detection of flight anomaly in time. The commonly used probability framework needs to be limited by prior information. The transferable Belief Model TBM does not need prior information and can efficiently process heterogeneous information. But the traditional TBM can not deal with the discontinuity and uncertainty in time, so to solve the problem of abnormal path target detection, a two-layer fusion architecture based on TBM is established by combining Markov model with TBM framework. The first layer is to realize the detection of multi-feature by analyzing the multi-attribute conflict information, and through the analysis of feature contribution degree, the multi-feature information can be discounted and then fused again. The second layer detects abnormal changes of air route targets by assigning fusion on time series, contrasts the difference between predicted values and observed values. Simulation results show that dynamic evidence reasoning method is better than dynamic evidence reasoning method in switching route scene and deviation regression scenario. This method has better decision accuracy and time accuracy.
【作者单位】：西北工业大学自动化学院;光电控制技术重点实验室;
【基金】：国家自然科学基金(No.61135001,No.61374023,No.61374159) 航空科学基金(No.20125153) 光电控制技术重点实验室和航空科学基金联合资助(No.20125153027)
【分类号】：V328

【相似文献】