In order to improve positioning accuracy of GNSS-INS train combination positioning system

a fine performance CKF nonlinear filtering algorithm is intended to be introduced to traditional framework of particle filter（PF） to form improved CPF algorithm.Combined with the sample datas provide by a train control system

to carry out simulation

compared with the traditional PF algorithm

the CPF algorithm has good filtering effect

small error

and can meet the requirements of the train operation nonlinear environment.The results show that CPF algorithm has good engineering application value in GNSS-INS train combination positioning system.