Research On Multi-objective Optimal Test Design Of Inertial Navigation Platform System

Due to the complicated structure of the inertial navigation system,the error parameter identification test process needs to be designed for good results.The single-objective test design can only optimize the test time or the accuracy.The multi-objective test design optimizes both the test time and the accuracy.at the same time and provides a set of test plans which is favorable for the execution of the test.In this paper,the inertial navigation platform system is studied,and the improved algorithm is proposed to solve the multi-objective experimental design problem of the error parameter identification test.First,the error model of the inertial navigation platform system is established.In order to simplify the problem,it is considered that the gyroscope error model has 21 error parameters,the accelerometer error model has 9 error parameters.Based on ? angle method,we establish the state and Observation equations of the inertial navigation platform system.System observability is the basis of parameter identification.Since the established error model is linear time-varying,the Piece-Wise Constinuous Systrm method is used to analyze the observability of the system.Under the premise that the estimation is unbiased and effective,the degree of observability is analyzed by Cramer-Rao lower bound,and the process of obtaining the information matrix is derived.In order to solve the multi-objective experimental design problem,this paper studies the multi-objective optimization algorithm and improves the existing multi-objective optimization algorithm MMOPSO.The improved algorithm optimizes the elite archive update method by avoiding the missing of pareto front.The improved global optimal solution selection method could promote the algorithm to move to the unrich region.The normalized decomposition method prevent the nonuniform pareto front caused by the difference of the objective values.Finally,the boundary reflection method is used to prevent the particles from flying out of the position boundary which leads to unnecessary function evaluation.The ZDT series test function,GD and SP evaluation criteria are adopted compare the improved algorithm(IMMOPSO),MMOPSO and the classical algorithm NSGAII in order to verify the better performance of the improved algorithm.Base on the optimal design theory,the relationship between various optimal criteria is analyzed.The expression of single objective optimization is obtained according to D optimal criterion.Considering the experiment time,the experimental design problem of the inertial platform error parameter identification can be considered as a constrained multi-objective optimization problem.A constraint processing mechanism should be added to solve the problem.We compare the effect of solving the multi-objective test design problem to NSGAII,which proves the effectiveness of the improved algorithm.Under the optimal trajectory,the error parameter identification method of inertial navigation platform is studied.The Kalman filter equation is obtained by discretizing the inertial navigation platform error model.The Kalman filter method is used to identify the error parameters,and the effectiveness of Kalman filter in identifying the inertial platform error parameters is verified.