基于机载单站无源定位中常用的扩展卡尔曼滤波具有近似线性方程的雅可比矩阵计算困难、不敏卡尔曼滤波(UKF)具有计算量较大的问题,提出了基于改进的变尺度最小斜度不敏卡尔曼滤波(SMSS-UKF)的机载单站无源定位算法,该方法采用最小斜度采样策略进行Sigma 点采样,减少Sigma点提高计算效率,利用变尺度不敏变换克服了采样点非局部效应问题。同时,针对大部分定位跟踪模型中状态方程为线性方程的特点,依据在线性状态方程情况下的贝叶斯理论,运用卡尔曼滤波状态预测的方法进行UKF的最优状态预测,使状态预测避免了不敏变换的数值近似误差和Sigma采样点计算的复杂性,提高了算法的运行效率和滤波精度。仿真实验的结果证明了SMSS-UKF滤波算法的有效性。

In order to solve the problems of difficulty for acquiring approximate linear equation of Jacobian matrix in extended Kalman filter and difficulty of complex calculation for unscented Kalman filter (UKF), a passive location algorithm for single airborne observer based on scaled minimal skew simplex unscented Kalman filter (SMSS-UKF) is presented. To decrease Sigma point and to improve the computational efficiency, minimal skew simplex sampling strategy is used in the algorithm to carry out Sigma point sampling. And scaled unscented transformation is used to overcome the problem of non-local effects of sampling. For state equation is linear equation in a majority of location & tracking model, according to Bayesian and linear equation, this paper derives that state prediction in UKF can use the state prediction of the Kalman filter to predict accurately. The numerical value approximate error in unscented transformation and the complexity of the Sigma point computing is avoided. Therefore the efficiency and filteringaccuracy of the algorithm is also improved. Validity of the SMSS-UKF algorithm is verified by the simulation examples.

TN958. 97

航空科学基金资助项目(20152096019)