不可靠角度量测下的目标定位与追踪研究

发布时间：2018-06-27 11:31

本文选题：声阵列网络 + DOA　；参考：《浙江大学》2017年硕士论文

【摘要】：在目标定位与追踪领域,基于声信号的阵列传感网络系统具有能耗低、隐蔽性好的特点,在军事国防、野外监控等领域有重要应用。由于实际应用环境、设备等方面存在诸多不可控因素,难以保证各阵列节点均能给出完全可靠的量测信息。如何在部分量测不可靠的情况下获得高可靠的目标位置估计将是一项需要重点研究的工作。本文将针对以方位角为量测的声阵列传感网络系统,分别对单目标定位、单移动目标时空联合定位、多目标定位三个场景,在部分量测不可靠的情况下,给出具体的鲁棒定位方案。本文主要工作包含以下几个方面:首先,在单目标定位场景中,引入目标可行区域概念,并设计基于目标可行区域评价的异常量测检测方法和基于l1范数的单目标鲁棒定位方法。针对所设计的鲁棒定位方法求解困难的问题,提出一种基于区域拆分和剪枝的快速搜索求解方法。其次,在单移动目标时空联合定位场景中,基于非机动目标短时间近似匀速直线运动的前提假设,利用各节点一段时间的量测信息,从概率角度推导节点可靠性和量测可靠性的有效评估方法。基于现有定位方法,设计使用l1范数改进的时空联合单目标鲁棒定位方法,并基于反向评价机制设计有效的反欺骗鲁棒定位方法。最后,在多目标定位场景中,由于节点间量测关联关系未知,易因量测关联错误引入不属于任何目标的鬼点,针对该问题,本文将提出判别量测错误关联的若干充分条件。并对量测中存在异常的情况,扩展目标可行区域评价机制,设计适用于多目标场景的异常检测方法。最后将设计基于聚类思想的多目标定位方法,以进一步抑制对残留的少量错误关联和少量异常DOA的影响。
[Abstract]:In the field of target location and tracking, the array sensing network system based on acoustic signal has the characteristics of low energy consumption and good concealment. It has important applications in military defense, field monitoring and other fields. Because of the actual application environment, there are many uncontrollable factors in the equipment and other aspects, it is difficult to ensure that all the array nodes can give complete and reliable measurement information. How to obtain high reliable target location estimation under some unreliable measurement will be a key research work. This paper will focus on the single target location, single moving target spatio-temporal joint location and multi target location, which are unreliable in partial measurement, for the acoustic array sensing network system using azimuth as measurement. In this paper, the specific robust positioning scheme is given. The main work of this paper includes the following aspects: first, the concept of the target feasible region is introduced in the single target location scene, and the anomaly measurement detection method based on the target feasible region evaluation and the robust positioning method based on the L1 norm based monocular standard are designed. In order to solve the difficult problem, a fast search method based on regional separation and pruning is proposed. Secondly, based on the premise of a short time approximate straight line motion of a single moving target, the reliability of a single moving target is based on the assumption of a short time approximate straight line motion, and the node reliability is derived from the probability angle by using the measurement information of each node for a period of time. An effective method of evaluating the reliability of measurement. Based on the existing location method, a spatio-temporal combined single target robust positioning method improved by L1 norm is designed, and an effective anti deception robust positioning method is designed based on the reverse evaluation mechanism. Finally, in the multi target location scenario, the association relationship is unknown because the node interval is unknown. The error introduces the ghost point that does not belong to any target. In view of this problem, this paper will propose a number of sufficient conditions to determine the error Association of the discriminant quantity, and extend the feasible region evaluation mechanism of the target in the case of the anomaly in the measurement, and design the anomaly detection method suitable for the multi target scene. Finally, the multi target location based on the clustering idea is designed. Methods to further suppress the residual error and a small amount of abnormal DOA.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP212.9;TN929.5

【参考文献】