多传感器数据融合与航迹预测

多传感器数据融合技术是未来军事电子领域一个重要趋势.根据6个观测雷达的观测数据进行了数据融合算法的研究.在提取目标航迹对时,对每个雷达的数据依据一定的判定条件(时间变化,角度变化在一定范围内等),分别提取出不同的目标航迹对.在提取同一目标的航迹对时,先将目标航迹的一些异常点弃除,然后把时间重合的两段航迹提取出来,通过样条插值进行时间配准,共提取出多条相关的航迹组有3组.在使用雷达探测目标时,由于技术条件和方法等的限制,使雷达数据存在各种误差.利用卡尔曼滤波自适应算法估计出观测位置的噪声方差,对雷达偏差进行修正后,采用联合卡尔曼滤波算法对多条航迹进行融合,接着利用ARMA模型预测目标在未来10秒内的轨迹,最后,对目标在被锁定后的轨迹做出预测,结合导弹的爆炸范围求得导弹击中飞机的概率约为49.54%.

Date Combination and Track Prediction by Multiple Sensors

Using multiple sensors to solve the problems of data combination is an important trend in electronic area.In this passage,we do some date combination researches based on the data obtaining from 6 different radars.We extract different tracks by several conditional judgements(such as time variations and angle variations).After remove some abnormal points in every extracted tracks,we choose the cubic spline interpolation method to regist the intersected parts of two tracks,then we get 3 track groups,in which the tracks are related and stand for one target.Due to the restrictions of the equipment,the data collected by the radar may be not so accurate,so we use self-adaptive Kalman filter algorithm to estimate the noise variance and correct the deviation of the radar data.Next we applied the Kalman integrated filters to combine the tracks in all track groups.The track of the target in next 10s can be predicted by using ARMA Model.At last,with due considerations of the explosive range by a missile,we calculate the probability of a target is hit when the target is locked by the radar.