多模型自适应联邦卡尔曼滤波及其在鱼雷导航定位中的应用

Multi-model Adaptive Federated Kalman Filter and Its Application to Torpedo Navigation and Positioning

针对复杂环境所引起的鱼雷组合导航系统模型参数变化导致传统单一固定参数滤波器精度降低的问题,设计了一种适用于复杂情况的多模型自适应联邦卡尔曼滤波器。该滤波器采用多模自适应联邦滤波算法,利用多模型滤波器参数来逼近系统的动态性能,该滤波器同时采用了在信息融合时自适应调整信息分配系数的方法。实验室导航试验结果表明,文中设计的多模自适应联邦卡尔曼滤波器可以大大提高导航定位系统的估计精度、跟踪速度以及稳定性。

Aiming at the problem that traditional fixed parameter filter lose its precision due to parameter variation of torpedo integrated navigation system model in complicated environment, a multi-model adaptive federated Kalman filter is designed for a torpedo integrated navigation system. The Kalman filter adopts multi-model adaptive federated filtering algorithm, which can approximate the dynamic performance of the system by making use of multi-model filter parame- ters. Furthermore, an adaptive information distribution strategy for information fusion is employed in the federated filter. The results of navigation test in laboratory show that the multi-model adaptive federated Kalman filter can improve es- timation precision, tracking speed and stability of the navigation system greatly.