Abstract: To avoid the problem of modes mixing caused by the intermittent in the data,a method of the ensemble empirical mode decompostion(EEMD) is studied in this paper along with its error effects.In this method the white noise is used to improve the decomposion results.Then the free vibration response can by obtained by the random decrement technique(RDT) together with the improved decomposion results.This can provide the available data for Hilbert transformation to identify the modal parameters.Finally,this is applied on the GPS dynimic monitoring data of the Wuhan Baishazhou Bridge.The results show that the mode parameters can be identify by the EEMD together with RDT effectively.