改进的自适应Kalman滤波方法及其在AUV组合导航中的应用

An improved self-adaptive Kalman filter algorithm and its application in integrated navigation systems for AUV

针对自主水下机器人(AUV)的工作特点与执行水下作业任务时对导航的需求,构建了基于航位推算的AUV组合导航系统体系结构,建立了水下机器人运动方程与观测方程,采用自适应卡尔曼滤波对水下机器人传感器信息进行数据处理。针对自适应卡尔曼滤波方法的缺点,采取渐消记忆指数加权方法引入了遗忘因子,并采用预报残差的方法求解最佳遗忘因子,同时采取措施保证了系统噪声估计方差阵和测量噪声估计方差阵的半正定性和正定性,避免了滤波发散现象。海试实验结果表明,改进的自适应卡尔曼滤波具有良好的滤波效果,可以满足水下机器人执行各种作业任务的水下导航定位精度。

According to the working characteristics of autonomous underwater vehicles （AUVs） and their navigation re- quirements when performing underwater tasks, the architecture of integrated navigation systems for AUVs based on dead-reckoning was designed. The motion equation and the observation equation of underwater vehicles were con- structed, and a self-adaptive Kalman filter was adopted for processing the data from underwater vehicles＇ sensors. To overcome the disadvantages of the self-adaptive Kalman filter, the forgetting factor was introduced based on the fading exponent method, and the residual prediction algorithm was used for computing the optimal forgetting factor And some measures were taken to ensure the half positive of the matrix of system noisy estimation and the positive of the matrix of measure noisy estimation, which can avoid divergence. The sea experimental results show that the im- proved self-adaptive Kalman filter method is effective, and can meet the AUVs＇ demand in navigation and positio- ning when they carry out underwater missions.