A Novel Cascaded Extended Kalman Filter for Error Compensation in Magnetic Angle Sensors

Abstract

Conventional magnetic angle sensors mostly use a bipolar magnet structure, which shows a relative low resolution and accuracy due to nonuniform magnetization, mechanical errors, and electrical errors. To address these issues, a novel magnetic angle sensor is designed by adopting structure with a central bipolar magnet (BPM) and an annular multipolar magnet (MPM). The central bipolar magnet is responsible for coarse positioning, while the annular multipolar magnet is used for fine angular measurement. To enhance the accuracy of the novel magnetic angle sensor, a cascaded extended Kalman filter (CEKF) is proposed to process and fuse the measurement data from both the central bipolar magnet and the annular multipolar magnet. This algorithm effectively fuses the magnetic field data from both types of magnets, thereby addressing the angle jump issue that can occur when the multipolar magnet transitions across adjacent poles and improving the overall measurement accuracy. Under various working conditions such as constant speed, constant acceleration, and static, the angle measurement error can be stably maintained within 0.1°, indicating that the proposed method has high robustness and accuracy.