针对MEMS加速度计在复杂应用环境下输出信号中存在野值,且其噪声统计先验知识不足的问题,运用了一种抗野值自适应Kalman滤波算法,来提高其测量精度。该算法用一个修正函数加权于自适应Kalman滤波方程的新息上,根据新息的方差和均值变化自适应调整修正权值,从而保持序列原有性质。通过对输出信号的算法验证,表明该算法能够有效在线去除野值,防止滤波发散,在提高MEMS加速度计测量精度上有一定的可行性。

The output signal of MEMS accelerometer may have outliers when used in complex environment,and the prior distribution to its noise characteristics is known insufficiently. An adaptive Kalman filtering algorithm tolerant to outliers was used to improve the measurement accuracy of the MEMS accelerometer. By adding a weighting function to the adaptive Kalman filters new information,the method can adjust the weighting factor adaptively according to the changing of the variance and mean value of new information,so it can keep the initial properties. The method was used for output signal. The result showed that the method can eliminate the outliers on line effectively and prevent filtering divergence. Therefore it is feasible for improving the measurement accuracy of the accelerometer.