自适应格型算法对动调陀螺仪随机漂移的建模

针对动力调谐陀螺仪启动过程中随机漂移模型参数的缓慢变化及逐次启动漂移模型系数的不同，提出了一种利用递归最小平方格型（LSL）算法对动力调谐陀螺仪随机漂移模型系数进行自适应估计的方法。格型相同节级联的特殊结构决定了LSL算法在自回归（AR）模型阶数增加时不必改变算法结构。使得该算法可以在线进行，并且收敛非常迅速，可以针对动力调谐陀螺仪不同次启动快速建立该次启动的随机漂移模型。通过对动力调谐陀螺仪启动1min后的4组实测漂移数据处理及模型检验，证明了该方法的有效性。

Research on Dynamic Tuning Gyro Random Drift Model by Adaptive Lattice Algorithm

For the slow changes of random drift model's parameters during the spin-up of the Dynamic Tuning Gyro (DTG) and the different coefficients for different spin-up, an adaptive estimation to the coefficients of random drift model of DTG with Least Square Lattice (LSL) algorithm was introduced. The rank connection special structure of lattice determines that the LSL algorithm doesn't need to change the algorithmic structure when the stage number of AR model increases, which makes the algorithm be implemented on line. The convergent speed of LSL is rather fast, so it is convenient to establish the random drift model of DTG according to the different spin-up of DTG with high speed. The algorithm and the model are proved to be valid by the experimental drift data of DTG. The result shows the method is effective and feasible.