Design Methods For Self-calibration Test Of Inertial Navigation Platform System Based On Genetic Algorithm

The Inertial Navigation Platform(INP)system calibrates the tool error before performing the task.The pre-calibration is done in the laboratory with the rotation of the external turntable,but the error coefficient of the instrument changes as time and environment change.As a result,the inertial navigation system needs to be calibrated on a regular basis after deployment on the carrier.The Platform inertial navigation systems typically contain three rotating frames that can be self-calibrated without relying on external test equipment.The continuous rolling test method can use all the observation information in the rolling process to identify more error coefficients and have higher calibration accuracy,but the model equation is more complicated and there are more factors that affect the test accuracy.For the optimal experiment design of rotational trajectory of platform in continuous tumble test,a practical design method in D optimum sense is presented.In this method the state equation is established using psi-angle error model and the outputs of three accelerometers are used as observables to establish the observation equation.The determinant of information matrix is selected to be the performance index,the moment currents to gyroscope are control variables and then the design problem of optimal continuous rotational trajectory is transformed to an optimal control issue.The issue contains non-linear constraints,and the objective function is non-convex.The optimal control issue is solved using two optimization algorithms,SA and GA.Through the reasonable parameter configuration,the two algorithms are simulated and compared.Furthermore,the Multi-objective Optimization method is studied by using the recognition accuracy of each error coefficient as the optimization target.The main diagonal elements of the inverse matrix of the information matrix are chosen as the target parameters,and the non-inferior solution set is obtained by NSGA ? algorithm.Based on the continuous rotation path of the platform by single-objective and multi-objective optimization,the method of error identification for continuous roll self-calibration test is studied.The filter equation is linear time-varying,so use the Kalman filter to identify the errors.During the process of identification,process noise is introduced.The validity of the experimental design results is verified by simulation.The Kalman filter simulation results show that the optimal rotation path given by single-objective and multi-objective optimization obviously improves the recognition accuracy.