In view of the problems that the state estimation of nonlinear error model has the shortcomings of long convergence time,low estimation accuracy,and large calculation cost,a twice rapid transfer alignment algorithm based on dual-model switching is presented.The commonly-used linear error model based on Euler angles and nonlinear error model based on quaternion were utilized. In the first phase,the nonlinear error model and extended Kalman filter were used.Then,when the attitude misalignment angle converges to a certain degree,the linear error model based on Euler angles and the conventional Kalman filter were adopted in the second alignment to get higher accuracy.The simulation and vehicle test results demonstrate that this method is feasible and can meet the requirements to accuracy and rapidity.