A method for high accuracy heading angle combined with ellipsoid calibration and BP neural network

Magnetometer is widely used to indicate the heading of vehicle by measuring the Earth's magnetic field. In this article, to solve the problem existing in 3D magnetometer, a method is presented for fitting a high precision heading angle. First of all, on the basis of analyzing the sensor working principle and error sources, the error model is established and the compensation algorithms is developed. Then using the ellipsoid fitting method to calibrate the magnetometer and the accelerometer simultaneously. After that, the maximum error of the heading angle of the magnetometer is about 3 degrees. This is due to the non-orthogonal coupling error between the axes of the magnetometer. The BP neural network method is used to fit its nonlinearity error. The results show that the heading angle error is controlled within 0.8 degrees.