Optimized selection of observation model parameters based on QR decomposition

Based on QR decomposition, this paper proposes a method for optimized selection of observation model parameters. It includes several stages: first, by employing QR decomposition of observation matrix, the analysis of observation model is converted into the analysis of the upper triangular matrix R from QR decomposition; second, the deep analysis of the ill-condition of the model is carried out combining the observant structure; for the third stage, the reason for the production of ill-condition is further analyzed; the selection of parameters of the model is optimized in the final. The numerical experiments show that our method is feasible and can be implemented with ease.