In order to identify the error parameters of inertial navigation platform in the centrifuge testing, a 6-position identification scheme was developed. The non-linear model of the inertial navigation platform for the identification was established, and the extended Kalman filter algorithm was applied to estimate the error parameters. The correlation coefficients among the error parameters calculated with the estimation error covariance matrix in the filter were used to analyze the observability of the system. The minimum sum of all correlation coefficients was considered as the optimization criterion, and then the genetic algorithm was utilized to obtain the optimal position combination. The simulation results show that the scheme is valid to identify all the error coefficients connected with the higherorder acceleration with high precision. 