Fuzzy calibration of a magnetic compass for vehicular applications

In a strapdown compass on a vehicle, three-axis magnetometers measure the Earth's magnetic field vector along the body axes of the vehicle to determine its heading angle. Owing to the local magnetic effects, the measurements frequently deviate from the geomagnetic field vector coordinated in the body frame. Therefore, online calibration of the compass should be considered to satisfy the requirements of the vehicle navigation system. In this paper, a new intelligent method is developed to implement online calibration of the compass system. First, a regression model is proposed to increase the convergence probability of the calibration process using the attitude angles in the measurement equations. Second, based on the knowledge of expert engineers, a Mamdani type fuzzy batch least-square (FBLS) algorithm is designed to estimate the calibration bias and scaling parameters. Generalized likelihood ratio (GLR) and the changes of estimated parameters are considered as the main information of the fuzzy system in which the length of data batch and the associated weighting factor are updated continuously. The results of simulations and experiments reveal the superiority of the proposed approach to the non-fuzzy methods. (C) 2010 Elsevier Ltd. All rights reserved.