捷联惯导初始对准大失准角系统误差模型中,当噪声具有不确定统计特性时,基于白噪声假设的无迹卡尔曼滤波算法鲁棒性较差.针对该问题,提出了一种基于H∞理论的鲁棒超球体无迹卡尔曼滤波算法.给出了计算量小的超球体采样策略,推导了H∞滤波的鲁棒机理,分离了鲁棒环节.将鲁棒环节引入超球体无迹卡尔曼滤波算法,得到鲁棒超球体无迹卡尔曼滤波算法,并分别在系统噪声和量测噪声为白噪声和有色噪声的条件下,对超球体无迹卡尔曼滤波和鲁棒超球体无迹卡尔曼滤波两种滤波方法进行了仿真实验.仿真结果表明,鲁棒超球体无迹卡尔曼滤波在白噪声情况下虽然精度有所降低,但是相对超球体无迹卡尔曼滤波具有了对有色噪声的鲁棒性,较超球体无迹卡尔曼滤波方法更适用于天向失准角为大角度并且噪声特性为有色噪声的情况.%The H∞ filtering was introduced to solve the problems of low filtering performance which derived from uncertainty noise caused by wind and wave and high frequency vibrancy in measurement equation of self-alignment of SINS for the carrier plane. First, the basic idea of self-alignment of SINS for the carrier plane on the sea environment was formulated. Second, the state space model for precision alignment of SINS in swing base with large amplitude was built. Finally, a filter of self-alignment based on H∞ filter for SINS of the carrier plane was designed. This method has effectively suppressed the external noise interference and ensured the robustness of the system. Simulation results show that the Kalman filter can not complete estimation of attitude error in 5 min, while H∞ filter can not only estimate attitude error, but also estimate the gyro constant drift error and the accelerometer bias error. The level alignment precision is within 2', while the azimuth alignment precision is about 12'. Meanwhile, it can obtain better rapidity and filtering accuracy. The method may after all be regarded as an effective self-alignment method on the sea.
展开▼
