基于改进卡尔曼滤波技术的多波束避碰声纳应用研究

Applied Research on Multi-beam Collision Avoidance Sonar Based on Improved Kalman Filter Technique

避碰声纳可以为无人航行器提供障碍物的距离和方位信息,从而为无人航行器发现并避开障碍物提供准确的指导。传统的目标自动检测主要采取恒虚警算法,但由于背景的不断起伏变化,以及水下声场建模精度的限制,导致背景估计不准确,难以实现对水中目标的稳定跟踪。针对该问题,论文采用模拟退火算法结合卡尔曼滤波技术提高了系统目标探测能力。通过算法仿真及湖试试验,验证了该改进方法的有效性。

The collision avoidance sonar can provide distance and orientation information of the obstacle for unmanned naviga-tion device,thus providing accurate guidance for unmanned navigation device to discover and avoid obstacles. Traditionally,CFARalgorithm is mainly used for automatic target detection. However,due to the fluctuation of the background and the accuracy limita-tion of the underwater sound field modeling,the background estimation is not accurate and the stable tracking of the underwater tar-get is difficult to achieve. Aiming at this problem,we used simulated annealing algorithm combined with Kalman filter technology toimprove automatic target detection ability of the system. The effectiveness of the improved method is verified by the algorithm simula-tion and the lake test.