摘要
针对高精度加速度计在重力场的标定试验,为了消除正交双表g2观测模型引起系统复共线性的影响和提高加速度计模型系数的辨识精度,文中提出了一种经验贝叶斯岭估计辨识方法并应用于正交双表误差模型的参数辨识中.从仿真分析和实验结果中看出,与传统的最小二乘法相比,经验贝叶斯岭估计能够消除系统的复共线性,可以分离出二次项系数K2,且辨识精度较高.
In order to eliminate the multicollinearity caused by a g2 model of dual orthogonal accelerometers and increase the accuracy of coefficient identification in the high precision accelerometer calibration test with a gravity field,this paper proposed empirical Bayes ridge estimation,which manipulated the model of dual orthogonal accelerometers.The simulation and experiment show that compared with conventional least squares estimation,empirical Bayes ridge estimation can not only eliminate the influence of the multicollinearity and separate K2,but also has higher identification accuracy.
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
北大核心
2011年第5期570-574,共5页
Journal of Harbin Engineering University
基金
国家"十一五"预研基金资助项目(51309050702)
关键词
加速度计
正交双表g2观测模型
复共线性
经验贝叶斯岭估计
accelerometer
g2 model of dual orthogonal accelerometers
multicollinearity
empirical Bayes ridge regression estimation