多雷达系统数据融合与航迹预测

为了便于准确定位传感器网络,分布式的单雷达系统首先进行各自的数据处理,包括:地理坐标换算至平面直角坐标;剔除孤立的异常点迹;采用模糊c-均值聚类方法和"动态分区",将单雷达数据中属于同目标的相似点迹归类集合;根据雷达观测和目标运动的特征,在每个点迹集合中设计门限滤波和相关矩阵检验,提取完整连续的目标运动的航迹;结合各航迹特征进行种类分析.接下来对属于不同雷达的航迹两两比较,找出有相交时间段的航迹,采用三次样条对两条航迹的进行内插和外推,再通过模糊综合函数对这两条航迹给出一个相似性度量,并取阈值为0.85.最后得出雷达间各航迹匹配关系.通过该雷达所观测到的航迹的稳定程度来近似估计其观察精度.首先对每一条航迹进行分段拟合得到其剩余方差,然后直接用每一条航迹的剩余方差来衡量雷达的观察精度,最后我们得出雷达的精度排序29107728,7724253720252539.对航迹融合,我们首先采用D-S证据理论并利用分析得到的雷达精度,对表示同一目标的航迹对进行融合.其次试图运用卡尔曼滤波对航迹进行融合:思路一是设法离线估计出噪声矩阵,得出系统噪声方差矩阵和观测噪声方差矩阵,从而用于标准卡尔曼滤波方程;思路二是探究较为实用的自适应滤波,兼顾Sage-Husa自适应滤波算法的高精度与强跟踪自适应滤波算法的可靠性,采用了一种混合算法给出收敛的估计.最终给出了雷达7728和2910的融合算例以及10秒钟的预测轨迹.最后,我们将导弹拦截飞机建模为三维的追逃问题,建立了运动学关系方程,最终归结为最小能量导引律问题.采用"模糊T-S线性模型"以及RH控制方法和伴随技术,在目标作对抗性机动条件下,获得了一个有效拦截的导引律.还对多雷达系统平均处理周期、数据融合系统的航迹处理周期进行了分析,对雷达网络实时性做了评价.

The Data Fusion and Tracks Prediction of Multiple Radar System

In order to get accurate positions of sensor networks,the distributed single radar system needs to conduct data processing,including:geographic coordinates conversion;removing isolated & bizarre target points;using"Fuzzy Clustering"&"dynamic partitioning"methods to classify the target points;With the target motion characteristics,setting threshold filtering and correlation matrix,to extract the continuous movement of the track; identifying various types of tracks with their specific features.Then we check every two tracks from different radars,and identify the intersection tracks in the same time period.Using cubic spline method to conduct the interpolation and extrapolation of the two tracks,we then give a similarity measure of these two traces with Fuzzy Function and threshold value is set 0.85.So we can conduct the conclusion of track matching relationships between the radars.Thirdly,we get residual variance by fitting each section of the track,and consider it as the accuracy of radar observations,so finally we come to the precision of radar sort 2910>7728,7724>2537>2025>2539.Fourthly,we use D-S proof theory,Kalman filter and the precision of radars above to realize the track fusion classified for the same target.Both precision of Sage-Husa adaptive filter algorithm and reliability of Strong Tracking algorithm are considered to achieve the convergence estimate.We provide the fusion example of radar No.7728 and No.2910 and the forecast track in 10 seconds.Finally,we focus on the three-dimensional model of missile intercepting aircraft,conducting equations based on kinematics relation;ultimately we can convert it to a minimum energy guidance law problem.With"TS fuzzy linear model",the RH control methods and associated technology,we access to an effective blocking guidance law under the target condition of adversarial maneuvering.The authors also research on the average processing period of multiple radar systems,the data processing cycle of track fusion system is analyzed,and real-time of radar network is also evaluated.