Gyroscope Array Analysis Based on EMKF Algorithm

The utilization of a gyroscope array that comprises multiple MEMS gyros that are homogenous and low-cost is shown to be an effective approach that can be employed to decrease measurement errors and bolster navigation performance of inertial sensors by taking advantage of redundant information. This research proposes an improved Kalman filtering algorithm that is founded on the EM algorithm. The algorithm takes into account the influence of gyroscope noise correlation, and models the gyro array more precisely. A maximum mathematical expectation principle is utilized to estimate the gyro noise covariance matrix. This enables the real-time estimation of Q and R matrix, and circumvents the need to pre-treat the gyroscopes with Allan variance analysis. Experimental evaluation was conducted using a gyro-integrated array that is made up of four MEMS IMUs to assess the navigation performance of the proposed filter algorithm. The results show that by applying the EMKF process, the bias instability is reduced by 73.45%, the angle random walk is reduced by 83.18%, and the rate random walk is reduced by 72.32%, resulting in an accuracy that is more than twice the traditional Kalman filter.