Prediction of GAMIT baseline solution based on Bayesian classifier
Meiyun Hao, Xiankun Sun, Ling Yin, Qianyun Ding · 2018
When we do the GPS baseline vector solution with GAMIT/GLOBK, the NRMS (normalized root-mean-square) value of the solution results might be too large or the solution might be failing during the process. The usual way to deal with both situations is to look for the causes of the error and correct them manually. Then do the baseline solution again. In view of the large amount of GPS data and the long baseline, it might take a long time if the procedure of solution repeated. In order to solve this problem, we select five parameters (observation duration, sampling interval, epoch, starting point coordinate and baseline solution type) related to the solution results of GAMIT as the discriminant factors, and establish the GAMIT solution Bayes prediction classification model. The experimental results show that the Bayes discriminant model has good prediction performance and can efficiently predict the solution is normal or not.