基于交互式模型的多AUV协同导航鲁棒性滤波算法

摘要

To deal with the problems of abnormal measurement of multiple autonomous underwater vehicle (AUV) in cooperative navigation process, a new cooperative navigation algorithm based on the interacting model is proposed to meet the requirement of navigation.Firstly, on the basis of the multiple AUV cooperative navigation model, the basic cooperative navigation filtering process is presented.Then, a generalized maximum likelihood estimation filtering algorithm is carried out to deal with the measured noise.Further, the generalized maximum likelihood estimation in the form of Schweppe is used to solve the anomaly of measurement.Finally, the interactive multiple model algorithm is used to solve the problem of the decrease of the filtering accuracy caused by the measurement noise.Simulation results show that the proposed algorithm is adaptive robust.%针对多自主式水下潜器(autonomous underwater vehicle,AUV)在协同导航过程中量测异常等问题,提出一种基于交互式模型的多AUV协同导航滤波算法.首先以建立多AUV协同导航基本模型为基础,给出基本的协同导航滤波过程;通过广义最大似然估计的滤波算法对受污染的量测噪声进行处理;进一步地,利用Schweppe形式下的广义最大似然估计解决量测出现的异常情况;运用交互式多模型算法解决由量测噪声时变而造成的滤波精度下降问题.最后仿真结果表明该协同导航滤波算法具有良好的自适应鲁棒性.