论文部分内容阅读
为补偿MEMS陀螺随机漂移,采用时间序列分析法对其进行自回归滑动平均(ARMA)模型辨识,提出一种滑动平均(MA)参数估计的新方法。先将陀螺随机漂移建模为带观测噪声的ARMA模型,在估计出自回归(AR)部分的参数后,针对AR滤波后的残差,推导出一种方差小的MA自协方差估计值,并将该估计值作为输入,利用Gevers-Wouters(GW)算法估计出MA部分的参数。仿真结果表明,MA参数估计精度得到提升的同时,参数估计可靠性也得到了增强。MEMS陀螺的随机漂移补偿实验进一步验证本文所提算法的补偿精度高于改进前。
In order to compensate the random gyro drift of MEMS gyroscope, the ARMA model is identified by time series analysis and a new method of sliding average (MA) parameter estimation is proposed. First, the gyro stochastic drift is modeled as an ARMA model with observed noise. After estimating the parameters derived from the regression (AR) part, a small variance MA covariance estimate is derived for the AR-filtered residuals Using this estimate as input, the parameters of the MA section are estimated using the Gevers-Wouters (GW) algorithm. The simulation results show that while the accuracy of the MA parameter estimation is improved, the reliability of the parameter estimation is also enhanced. MEMS gyro random drift compensation experiments further verify that the compensation accuracy of the proposed algorithm is higher than before.