基于改进卡尔曼滤波的陀螺仪误差补偿算法

针对基于卡尔曼滤波的MEMS陀螺仪误差补偿算法中量测噪声方差选取不准确的问题,提出一种基于改进卡尔曼滤波的陀螺仪误差补偿算法.卡尔曼滤波通常采用统计特性估计得到固定的量测噪声方差,无法自适应地估计不同环境下陀螺仪噪声特性.该算法将卡尔曼滤波与神经网络相融合,使用卡尔曼滤波新息矩阵作为神经网络输入,通过神经网络得到新息协方差矩阵,以此自适应地估计卡尔曼滤波量测噪声方差.将该算法应用到陀螺仪信号误差补偿中,使用Allan方差分析法对原始信号以及误差补偿后的陀螺仪信号进行分析,实验结果表明该算法能够有效地抑制陀螺仪随机误差,提高MEMS陀螺仪的精度.

A Compensation Algorithm of Gyroscope Error Based on Modified Kalman Filter

Aiming at the problem that the measurement noise covariance of the compensation algorithm of MEMS gyroscope error based on Kalman filter is determined inaccurately,a compensation algorithm of gyroscope error based on modified Kalman filter is proposed.The measurement noise of Kalman filter is usually estimated by statistic charac-teristics,so that the noise characteristics of gyroscope in different environment can't be estimated adaptively. The Kalman filter and neural network are combined in this algorithm. The innovation matrix of the Kalman filter is employed as the input of neural network and the innovation covariance matrix can be determined through the neural network.The measurement noise covariance can be calculated by the innovation covariance matrix adaptively and this algorithm is applied to the compensation of gyroscope error. Allan analysis method is used to analyze the raw signal and the result of the compensation algorithm. The experiment result shows that the random error can be controlled effectively by the algorithm proposed in this paper and the accuracy of MEMS gyroscope is improved.