针对小体积、低成本微惯性组件,提出了复合标定方法。该方法通过人工神经元估计耦合系数消除不同敏感轴的耦合影响。联合智能控制和扩展卡尔曼滤波方法,提出了位姿检测的自适应智能滤波方法。该算法中引入外部速度信息作为反馈,避免误差模型中依赖理论值进行计算受到的较大影响。在噪声统计特性未知或非先验条件下,根据量测噪声大小,在线调整系统噪声和观测协方差参数,改善滤波精度。搭建了导航实验系统,采用所提方法进行了实验,结果验证了算法的有效性。

For low cost, small size micro Inertial measurement unit (IMU),a compounded calibration based on artificial neural network (ANN) is proposed. It is adopted to identify the coupling coefficients to eliminate the adverse coupling effects between different rotation axes. Combining intelligent control with extended Kalman filter (EKF) an effective intelligent filter method for attitude and position determination is proposed. Using the external velocity information, the INS error can be extracted. Weak robustness would be solved when internal information of alone INS is considered as feedbacks. Process and measurement noise covariance matrix will be estimated to improve the accuracy of the filter in the condition of uncertain noise characteristic. The testing portable platform is set up. Navigation experimental tests are implemented to evaluate the proposed approach.