加速度计几种模型辨识方法的比较研究

Comparative study on model identification methods of accelerometers

加速度计离心试验中,为了更精确的得到加速度计的模型系数,比较研究了3种辨识方法:最小二乘方法(加权最小二乘)、总体最小二乘方法和EV模型方法。通过仿真得出在输出噪声和输入噪声为白噪声或者近似白噪声且离心机精度优于1×10-5的情况下,最小二乘与其他2种辨识方法辨识精度相当。最后通过试验对比了最小二乘方法与加权最小二乘方法,得出在此精度下,加速度计离心机试验的数据处理,应用加权最小二乘方法最为合适。

In order to acquire model coefficients of accelerometer on centrifuge trail,three identification methods are compared.They are least squares（LS）method,weighed least squares（WLS）method,total least squares method and error-in-variables（EV） method.Under the circumstances of white noise or closed to white noise in output and input,the centrifuge precision prior to 1×10-5,the identification precision of LS method is as high as the other two.WLS method is compared with LS method through the test.The test results demonstrate that WLS method is more suitable for data processing in accelerometer test on centrifuge.