System Error Compensation Algorithm in the Presence of Random Walk

The batch method and EM algorithm have been used to process the observational data of geostationary communication satellite.Using the inspection methods of skewness and kurtosis to test the normality of residuals processed by batch methos.In this case,the observation datas of Xinnuo statellite containing not only accidental errors,but also systematic errors.Batch method in dealing with systematic errors of observation data,the results are not reliable.Compared with the batch method,using random walk model to describe the system error,EM algorithm based on digital filtering method with systematic errors in the data processing is feasible,reached a certain extent,improved the precision of the data processing.