能量值和方位信息结合的粒子滤波算法

Energy and angle-based particle filter algorithm

针对水下无人航行器自主方位跟踪中出现的目标中断、虚警以及目标关联等多个问题,本文采用一种能量值和方位信息结合的粒子滤波算法,并应用在声呐目标检测跟踪中。在粒子滤波的基础上,引入波束形成给出的波束能量作为潜在信号源真实的概率。给出了多目标关联概率的粒子滤波表达,通过插值和自动阈值方法改善波束能量曲线,增强粒子滤波性能以及目标输出的平滑性。通过数值仿真及湖试的检验,该方法能够有效得到正确的多目标跟踪结果。

In view of the existing problems on underwater unmanned vehicle （ UUV） bearings for target tracking, including false alarm, temporary target disappearance, and target association, a particle filter algorithm based on beamforming energy and detection angle was studied and applied in sonar target detection and tracking. Beam ener-gy ,which is calculated by the beamforming method, was introduced to produce the probability that a potential tar-get source is real. A particle filter expression on multi-objective association probability was given. By interpolation and adaptive threshold setting,the beam energy curve was improved, the performances of particle filter were strengthened, and the target output became smooth. Both simulated and experimental data prove that the method could effectively realize correct multi-target tracking.