磁强计辅助MEMS惯性器件的新型数据融合算法

A New Data Fusion Algorithm Based on Magnetometer and MEMS Inertial Device

摘要:

为实现载体姿态精确测量,设计了一种新型数据融合算法,用以替代传统的卡尔曼滤波器;首先,研究了陀螺仪、加速度、磁强计的姿态测量方法,通过试验分析测量误差分量,然后根据误差频域的差异性提出一种互补滤波器,并在二阶低通滤波器模型下推导出新型数据融合算法模型,最后,以FPGA为处理单元、MEMS惯性元件和磁罗盘为传感器单元开发一套小型姿态检测系统,进行可行性分析和精度估算;从载体的三维转动试验结果可以看出,新型融合算法测量值和卡尔曼滤波器输出值之差保持在±0.07°以内,并且不需要对传感器噪声特性精确标定；在运算时间方面,互补滤波器仅需360 μs,卡尔曼滤波器是它的3倍多,因此新型数据融合算法效率更高。

Abstract:

In order to realize precision determination of attitude, a novel data fusion algorithm, replacing the traditional Kalman filter, is designed in this paper. Firstly, methods of detecting attitude, respectively based on gyroscope, accelerometer and magnetometer, are discussed. The deviation derived from measurement results is analyzed via trials. Considering the different spectrum of deviations due to different sensors, the idea of complementary filter is then proposed. A novel model of data fusion algorithm is further inferred behind the second order low-pass filter. Finally, taking FPGA as processor, MEMS unit and a magnetic compass constructed as sensor module, a small attitude determination system is developed to evaluate whether the module is practicable and whether to reach required accuracy. Experiments upon the tri-dimension rotated platform show that the deviation of result is no more than ±0.07° compared to Kalman filter. Besides, this model is not required known the accuracy noise feature. With respect to the consumed time of algorithms working on the hardware, time required on complementary filter is 360μs, and Kalman filter is 3 times more than it. Thus, the presented algorithm has high efficiency.