基于TheoH方差的陀螺随机误差系数动态提取

针对运用动态Allan方差提取陀螺随机误差系数时,用截断窗截取原始信号造成方差估计置信度降低的问题,提出运用混合理论方差(Theo H方差)来代替Allan方差对截断窗内的数据进行分析,并提取出随时间变化的陀螺随机误差系数。Theo H方差改善了Allan方差计算时相关时间只能达到信号总时间的二分之一及长相关时间下方差估计置信度降低的问题,其计算的相关时间可以达到数据总时间的四分之三,有效改善了动态算法因数据截取造成误差系数估计置信度下降的缺陷。从对仿真信号和光学陀螺实测数据处理结果上来看,本文方法既能准确地对动态条件下陀螺量测信号的随机误差进行细化辨识,又能大幅提高中、长相关时间下方差估计的置信度。

Dynamic extraction of stochastic error coefficients for gyro based on TheoH variance

To solve the problem that the confidence on the estimate is poor when the dynamic Allan variance is used to analyze stochastic error of gyro, due to the reduced amount of data captured by the truncation windows, a new algorithm based on high confidence and hybrid theoretical variance (TheoH variance) is proposed. The Allan variance which is used to analyze the data in the truncation windows is replaced by TheoH variance in the proposed dynamic algorithm. The problems that the averaging time calculated with Allan variance is half of total data length and the confidence decreases at long-term averaging time are perfectly solved, the correlation time of the calculation can reach 3/4 of the total time of data, and the defect that the confidence level of the error coefficient estimates decreased is effectively improved, which is caused by the dynamic algorithm data interception. The measured data of dynamic test for optical gyroscope and simulation data are analyzed with the proposed algorithm and dynamic Allan variance. The results show that the proposed algorithm can describe the non-stationary of gyroscope more effectively and greatly improve the confidence of variance estimation at middle-term and long-term correlation time.