混合坐标下的卡尔曼滤波应用于水下被动目标跟踪

针对下水下被动目标跟踪数据率低、跟踪误差大的特点和难点，研究了混合坐标下的自适应推广卡尔曼滤波算法，并将其应用于水被动目标跟踪估计器设计，该算法充分应用了直角坐标系下动态方程的线性特性和极坐标系下测量方程的线性特性，针对两坐标系间协方差矩阵变换的近似，引入了虚拟噪声进行补偿。通过系统的MonteCarlo仿真结果表明：该算法在收敛速度和估计精度方面都优于单一坐标体系下的滤波算法。

Passive Underwater Target Tracking: A Hybrid Coordinate

Grossman combined the advantages of two coordinate systems——Cartesian and polar——for passive tracking of airplanes［4］. We apply this advantageous approach to passive tracking of underwater target and improve it in that we consider something (to be explained later) that Grossman overlooked and consequently we have to use adaptive extended Kalman filter (AEKF) instead of his non-adaptive one. The hybrid approach employs Cartesian system for state and state covariance extrapolation and polar system for state and state covariance updating. What Grossman overlooked is that the error of transformation is unavoidable when Cartesian system is transformed into polar system and vice versa. We introduce virtual noise in order to compensate for transformation error. Simulation results, given in Fig.2, show preliminaritly that our improved approach can achieve accurate state estimation.