Application of Empirical Bayes Estimation in Error Model Identification of Two Orthometric Accelerometers

Abstract

The high accuracy accelerometer can be demarcated in multiposition tumbling experiment under 1g gravitational field. The g2 observation model of Two Orthometric Accelerometers can eliminate the corner error. there is a serious multicollinearity exit in this system because some model coefficient mix together. In view of above question, this article has given the arithmetic of Empirical Bayes Estimation(EB), and applied this method in model which is mentioned above. The result of the simulation and the experiment shows that compared with the conventional least squares method and the generalized diagonal ridge estimation, the Empirical Bayes Estimation can overcome the influence of the multicollinearity and can separate two coefficients which are The Second-Order Terms and the cross-coupling terms.