Calibration of tri-axis magnetometers using an improved truncated singular value decomposition method

The performance of a three-axis magnetometer (TAM) is limited by various types of error; hence, the system must be calibrated prior to use. The convergence of calibration algorithms is quite sensitive to input data, and divergence yields incorrect parameter-estimation results when the maneuvers of TAM are constrained. Therefore, a calibration method, based primarily on improved truncated singular value decomposition, was proposed. The singular values were divided into groups; the small values were modified, whereas the large values were retained (without modification). Information with improved accuracy was obtained from these large values, while the less accurate information contained in the smaller values was modified, thereby avoiding divergent solutions. The experimental results revealed that the performance of TAM was significantly improved by the use of the proposed calibration method.