EKF与互补融合滤波在姿态解算中的研究

Research on Attitude Algorithm of EKF and Complementary Filtering Fusion

传统的信号滤波方法不能有效的融合多传感器测量数据,或者融合中失去过去信号状态信息。针对这一问题,提出了扩展卡尔曼滤波(Extended kalman filtering,EKF)与互补滤波融合的信号处理策略。借助STM32微处理器采集MPU9250惯性测量传感器的原始数据,运用多传感器信息融合的处理算法,比较了互补滤波姿态解算结果和对互补滤波过程中所得的四元数运用EKF矫正后进行姿态解算的结果,以及互补滤波解算的欧拉角运用EKF矫正后的姿态数据。通过实验中3种解算结果与理论值的对比,得出结论:采用互补滤波会存在一定超调量,且结果波动较大,存在较大的噪声,对互补滤波过程中的四元数进行EKF滤波虽能降低解算结果的噪声,但仍存在超调量。而应用EKF矫正互补滤波解算出的欧拉角能同时解决超调量和降低噪声误差,抑制了随机波动,起到了更好的解算效果。

The traditional signal filtering methods cannot fuse the measurement data from multiple sensors effectively,or lose the past signal state information during the fusion. In order to solve this problem,the signal processing strategy by combing the extended Kalman filtering fusion with complementary filtering was proposed. The raw data of MPU9250 inertial measurement sensor by using STM32 microprocessor were collected. The algorithm theory of multi-sensor information fusion processing was applied. Compared among the result of attitude data of original complementary filtering,the result of attitude data calculated by complementary filtering quaternion,which was adjusted by EKF,and the result of attitude data of euler angles adjusted by EKF,which was calculated by complementary filtering. Compared the three calculated results with the theoretical value,concluded that,the result had a certain amount of overshoot,a comparatively large fluctuation and large noise was existed when applied complementary filtering separately. The noise of the result reduced when applied EKF to the quaternion in complementary filtering process,overshoot was still existed. Applied EKF to adjust euler angles calculated by complementary filter could solve the problem of overshoot and reduce noise error,and the random fluctuation suppressed,got a better measurement result.