Bias Accuracy Maintenance Under Unknown Disturbances by Multiple Homogeneous MEMS Gyroscopes Fusion

Bias accuracy of micromachined gyroscope is deteriorated severely when suffering from unknown environmental disturbances (UD), such as temperature, vibration, and shock. In this article, gyros array using a two-stages UD-decoupled fusion strategy is for the first time proposed to maintain high-accuracy bias under unknown disturbances. In detail, the evolvement model of the individual gyroscope among four-gyros array is first deduced to decouple UD. Then, local estimator is devised to simultaneously derive the estimates of the state, the bias, and the UD of each gyro among the array in the first stage, respectively. Further, due to the nonoptimal local state estimates caused by the inherent response difference between each gyro, the global state fusion estimator weighted by dynamic coefficients is devised to achieve the uniformly optimal state estimate of gyros array in the second stage. The experimental test shows that RMSE of the estimate based on the proposed method is evaluated as 0.13°/s, which is reduced by 92.4%, 87%, 82.4%, and 51.9% compared with kalman filter (KF), internal model approach (IMA), distributed optimal linear fusion estimator (DOLFE), and minimal learning parameter-neural network (MLP-NN), respectively. The tested scale factor of the array is calibrated as 5.9 mv/°/s and the corresponding bias stability is calculated as 0.019°/s, which is improved by 68.8%, 54.8%, 34.5%, and 17.4% compared with KF, IMA, DOLFE, and MLP-NN, respectively. Not only is bias accuracy under UD not deteriorated, on the contrary, the accuracy is improved significantly by 3.7 times compared with that without UD. The proposed array-based UD-decoupled method paves an effective path to apply microgyro in actual environment engineering.