A constructing method of reference maps for geomagnetic navigation using rectangular harmonic analysis and support vector machine

A constructing method of geomagnetic reference maps is proposed using rectangular harmonic analysis and support vector machine to conveniently correct both the magnetic field variation of vehicles and geomagnetic field variation in engineering application of geomagnetic navigation. Rectangular harmonic analysis method is adopted to predict reference data in heartland of the navigation region so that the accuracy of geomagnetic reference maps can be improved or conveniently corrected. The processes include calculation of geomagnetic field remnants, transformations of geographical coordinates and geomagnetic components, calculation of rectangular harmonic coefficients, and prediction of reference data in heartland region, etc. The support vector machine method is adopted to predict reference data in the margin of the navigation region to reduce the influence of edge effects, and the data include selection of kernel functions, optimization of parameters, model training, and prediction of reference data in marginal region, etc. Parameters of support vector machine are determined using particle swarm optimization algorithm to improve the computational efficiency of the method. Experimental results show that the accuracy of geomagnetic reference maps constructed by the proposed method is improved and the runtime of the proposed method is reduced markedly.