加速度计在分度头上的标定—加权最小二乘法

Calibration of an accelerometer on the indexing table:the weighted least squares method

关键词:  加速度计  加权最小二乘法  分度头  标定

Key words:  Accelerometer  Weighted least squares method  Indexing table  Calibration

摘要:

为了提升加速度计在重力场下的标定精度，提出了一种基于加权最小二乘法辨识加速度计误差模型系数的方法。该方法通过统计各个测试点位置加速度计输出方差，构造权重系数矩阵，降低了异方差的影响程度，有效地抑制了加速度计输出波动性带来的影响。通过加速度计在分度头上的15组24位置数据进行实验，比较了加权最小二乘法与工程上广泛应用的普通最小二乘法的辨识精度。实验结果表明，加权最小二乘法辨识结果的各个系数测试的标准差小于普通最小二乘法，基于加权最小二乘法辨识方案可有效提高在同等测试条件下误差模型系数的辨识精度。

Abstract:

In order to improve the calibration accuracy of accelerometers in the gravity field, a method based on the weighted least squares method is proposed to identify the error model coefficients of the accelerometer. This method constructs a weighted coefficient matrix by statistically analyzing the variances of the accelerometer output at each measurement point, reduces the influence of the heteroscedasticity, effectively suppresses the influence of the volatility of the accelerometer output. The identification accuracies of the weighted least squares method and the ordinary least squares method, which are widely used in engineering, are compared through experiments with 15 sets of 24-position data of the accelerometer on the indexing table. The results show that the standard deviation of the error model coefficients test with the weighted least squares method is smaller than that of the ordinary least squares method, and the identification scheme based on the weighted least squares method can effectively improve the identification accuracy of the error model coefficients under the same test conditions.