考虑时间延迟的经纬仪测星评估惯导姿态误差方法

Method of using theodolite’s observation data of fixed star to evaluate SINS’s attitude accuracy considering time delay

利用经纬仪测星来解算惯导姿态误差是一种有效的惯导系统姿态精度动态评估方法。但在实际工程中,因机抖激光陀螺的抖动解调和传输延迟的影响,捷联惯导系统(SINS)提供给经纬仪的姿态信息带有不可忽略的时间延迟,造成惯导姿态误差动态估计精度较差。为此,提出了一种考虑时间延迟的经纬仪测星评估惯导姿态误差方法,推导了经纬仪观测量与惯导输出姿态误差和时间延迟的数学关系,建立了时间延迟估计和补偿模型。仿真分析表明,在观测星数较少且惯导数据中存在较大时间延迟时,现有方法在动态条件下的姿态估计误差可能超出惯导本身精度,从而不再可用。而所提方法能够有效估计和补偿时间延迟的影响,在时延20 ms且只有2颗星可观测条件下,也能得到高精度的估计结果,总姿态估计误差小于2″。船载动态试验表明,所提方法相比于现有方法估计精度平均提升约78%。

The method of calculating the attitude error with theodolite is an effective dynamic estimation method for measuring the attitude accuracy of ship inertial navigation system. In practical engineering, due to jitter demodulation and transmission delay of mechanically dithered ring laser gyroscopes, the attitude information provided to the theodolite by strapdown inertial navigation system(SINS) has a time delay that can not be ignored, resulting in poor dynamic estimation accuracy of inertial navigation attitude error. A method with time delay estimation and compensation is proposed to evaluate the attitude error of inertial navigation by theodolite’s observation data of fixed star. The mathematical relationship between theodolite’s observation data with inertial navigation attitude error and time delay is deduced, and the time delay estimation and compensation model is established. The simulation analysis shows that when fewer stars are observed and there is a large time delay in the inertial navigation data, the attitude estimation error of the existing methods under dynamic conditions may exceed the precision of the SINS and will no longer be available. However, the proposed method can effectively estimate and compensate the impact of time delay.Under the condition of 20 ms time delay and only 2 stars can be observed, high-precision estimation results can also be obtained, and the total attitude estimation error is less than 2 ″. The ship borne dynamic test shows that the average estimation accuracy of the proposed method is improved by 78% compared with the existing methods.