Research On Key Technologies Of Geomagnetic-aided Navigation Of Cruise Vehicle

Accrate positioning of cruise vehicle in complex battle filed is the key factor to ensure effective attack.At present,due to the constraints like technical level,geographical conditions and international environment,the guidance mode is severely restricted and has obvious limitations and promotion space.Geomagnetic navigation is a new advanced auxiliary navigation system.The earth magnetic is a vector field and is globally observable,in theory,any point in the near earth space has unique corresponding magnetic vector.At the same time,it is a passive navigation mode without climate and area limited.There is no accumulated error in positioning process,and abundant features could be applied to make up the shortcomings of existing approaches.In this paper,according to the geomagnetic characteristics and navigation demand of cuise vehicle,we make research for the key navigation technologies mainly in the following several aspects.In the light of the limitations of traditional approaches in geomagnetic reference map preparation,the gradually interpolation and correction preparation approach is proposed.Firstly,we make multifractal spectrum analysis on the geomagnetic anomaly field data.On base of this,for the points to estimate,Kriging method is firstly used to make initial interpolation,then in the neighborhood of corresponding point,we use multifractal theory to compute singularity coefficient,which is used to correct the interpolation result of Kriging.Through this approach,the problems of inadequate reference points and inaccurate estimation are efficiently avoided.To solve the boundary effect problem in regional geomagnetic reference map preparation,we put forward the nonparametric sampling compensation approach based on feature matching.Firstly,for the geomagnetic dataset,we improved the traditional bidimensional empirical mode decomposition(BEMD)to decompose the sample datasets into several components and determine the scale-feature of each component.The total Hilbert transform is used to compute instantaneous features of each point in small-scale components,and the sum of them could be compensated through self-sampling and feature matching at the boundary regions.The sum of large-scale components could be well modeled with trigonometric functions and compensated through prediction approach.The rectangular harmonic analysis approach in map preparation is used to verify the effectiveness and advantage of this method.For the aeromanetic interference related to the vehicle attitude produced in cuise process,and the problem that cruise vehicle is not applicable to traditional calibration flight,a real-time computation approach for interference model without calibration flight is proposed.A general interference model including permernant,induced and eddy current interence filed is constructed,according to the vehicle attitude that INS output and generation mechanism of interference filed,initialize the interference filed parameters.Kalman filtering approach and particle swarm algorithm are used to compute the model parameters;besides,we discussed the practibility of using off-line experimental approach to solve the interference model parameters.To solve the electronic magnetic interference problem,a separation method based on blind signal separation is researched.On base of the independence of electronic magnetic interference,earth magnetic and that related to vehicles attitude,according to Biot-Safar law,we build the magnetic blind separation model,and select the separation approach based on neural network to separate the electronic magnetic interference field,the effectiveness is verified through simulation approach.In view of the limitation of traditional filtering methods in geomagnetic-aided navigation system,to further improve the estimation accuracy of principle system,a projection filtering approach based on Hermite functions is put forward.The basic properties include orthogonality are used to deduce and simplify the computation process.The system state's probability density could be achieved by solving the partial differential equation.This approach could achieve the confinuous estimation of the system state,through setting appropriate truncation order number for the projection basis functions,we can ensure the estimation accuracy of the system's state.