微机电陀螺信号盲均衡迭代反卷积算法

摘要

The mathematical model of a mobile mini-robot running in unknown environment cannot be effectively built due to gyroscope’s noise interference, and needs to remove the noise only from the observing signal and estimate the original signal. In this paper, an iterative deconvolution algorithm of blind equalization for MEMS gyroscope signal is presented. The transversal filter for the gyroscope signal to implement deconvolution calculation is employed, and the signal is estimated by Bayesian methods. The error function is established and combined with LMS algorithm to achieve the automatic adjustment of equalization parameter. The verification of the algorithm is carried out on the mini-mobile robot. The experiment results show that the angular velocity and noise signals can be effectively extracted from the mixed signals, and the amplitude of noise signal is decreased to about 1/10 of the original. After the mobile robot has run 275.41 s, the error of its final yaw angle for mobile robot is reduced from 13° to 1.46°.%小型移动机器人在未知环境下运行，陀螺所受噪声干扰无法建立有效的数学模型，需要仅从观测信号中把噪声去除，并估计出原始信号，根据该特点提出一种微机电陀螺信号盲均衡迭代反卷积算法。该算法利用横向滤波器对陀螺信号进行反卷积运算，使用贝叶斯方法对信号进行估计，建立了误差函数并与 LMS 算法组合，实现了均衡器参数的自动调整，在小型移动机器人上进行了算法实验验证。实验结果表明，该算法可以有效分离角速度信号与噪声信号，其噪声信号幅值减小约10倍，移动机器人运行275.41 s抵达终点的偏航角误差从13°下降到1.46°。