Research On Modeling And Compensation Method Of Missile-borne Integrated Navigation Error In Complex Environmen

Missile-borne integrated navigation in complex environments is a hot topic of current research.In practical engineering,the missile-borne integrated navigation system faces harsh environments such as high overload,high dynamics,strong vibration and complex noise interference.Based on this background,this paper studies the impact of overload shock on the performance parameters of the Miniature Inertial Measurement Unit(MIMU).Then,a MIMU error calibration method is designed based on deep learning.Furthermore,we propose a state fault tolerance estimate algorithm under complex noise environments.The research contents of this paper mainly includes the following aspects:(1)Considering the high-overload in the missile-borne environment,the high overload impact on the performance parameters of the micro inertial measurement unit is studied.Different magnitudes of shock schemes and calibration test experiments are designed for MEA200high-overload MIMU.Use the drop hammer shock table to perform several times shocks on the MIMU and use the multi-function turntable to calibrate it before and after each shock.We compare the calibration test results before and after the shock and analyze the shock of different magnitudes of impact on the calibration and performance parameters of the micro inertial measurement unit.(2)Considering the complex noise characteristics of the micro inertial measurement unit,the error calibration model based on the deep learning method is studied.In order to enable the neural network to fully learn the error characteristics,a MIMU error excitation experiment method based on a three-axis thermostat position rate turntable is designed.Then,an experimental data set is established.According to the error characteristics,the input and output structure of the neural network is designed.Further,the appropriate hidden layer and hyperparameters of the neural network are determined through experiments.The comparison of semi-physical simulation experiments demonstrates the effectiveness of the proposed method.(3)Considering the asynchronous multi-rate,complex noise and sensor failures faced by multi-sensor fusion in the missile-borne integrated navigation system,a fault-tolerant estimation method of the optimal state under the multiplicative noise and cross-correlated noise is studied.The observation equation is reconstructed by using the asynchronous model based on timestamp,and the asynchronous data is synchronized to solve the problem of data asynchrony.Furthermore,the recursive form of the state estimation algorithm is derived by using the theories of orthogonal projection and Kalman filtering.A packet loss compensation mechanism based on generalized regression neural network and a fault diagnosis mechanism based on normalized filtering innovation are designed to reduce the influence of packet loss and sensor failure.The simulation results verify the effectiveness of the algorithm.