基于高斯过程回归的三轴磁传感器校正方法研究  

Study on Calibration Method of Three-Axis Magnetic Sensor Based on Gaussian Process Regression

作　　者：唐小瑜 杨丽[1] 徐超群[2] 肖书婷 王斌[2] TANG Xiaoyu;YANG Li;XU Chaoqun;XIAO Shuting;WANG Bing(College of Electrical Information,Southwest Minzu University,Chengdu Sichuan 610093,China;Beijing Institute of Spacecraft Environment Engineering,Beijing 100094,China)

机构地区：[1]西南民族大学电子信息学院,四川成都610093 [2]北京卫星环境工程研究所,北京100094

出　　处：《传感技术学报》2021年第10期1340-1345,共6页Chinese Journal of Sensors and Actuators

摘　　要：针对三轴磁传感器的转向差校准,本文提出基于高斯过程回归的误差补偿方法,在均匀磁场环境下多次旋转得到磁测数据,通过高斯过程回归拟合传感器磁测数据与真实磁场之间的映射关系,完成三轴磁传感器的校正。仿真和实验结果均证明该算法对小样本数据具有良好的校正效果,同时分析对比了几种常见核函数对校正性能的影响,进一步验证高斯过程回归法对磁传感器校正的有效性,将磁测数据的最大峰峰值误差从1168.44 nT降低到2.75 nT,显著降低了三轴磁传感器的测量误差。Aiming at the measurement error calibration of the three-axis magnetometer,this paper proposes an error compensation method based on Gaussian process regression.The magnetic measurement data are obtained through multiple rotations in a uniform magnetic field.Gaussian process regression is used to fit the mapping relationship of the sensor’s magnetic measurement data and the real magnetic field,which can complete the calibration of the three-axis magnetic sensor.Simulation and experimental results both prove that the algorithm has a good correction impact on small sample data.At the same time,the influence of several common kernel functions on the correction performance is analyzed and compared,and the effectiveness of the Gaussian process regression method in the correction of magnetic sensors is further verified.The maximum peak-to-peak error of the data is reduced from 1168.44 nT to 2.75 nT,which significantly reduces the measurement error of the three-axis magnetometer.

关 键 词：三轴磁传感器 误差校正 高斯过程回归 核函数 

分 类 号：TM936.1[电气工程—电力电子与电力传动] TP181[自动化与计算机技术—控制理论与控制工程]