电子测量与仪器学报

李伟,郝顺义,黄国荣,马仕杰.改进自适应 ADMCC-HCKF 算法及在 SINS / CNS / GNSS 中的应用[J].电子测量与仪器学报,2021,35(8):79-85

改进自适应 ADMCC-HCKF 算法及在 SINS / CNS / GNSS 中的应用

Improved adaptive ADMCC-HCKF algorithm and application inSINS / CNS / GNSS integrated navigation

DOI：

英文关键词:maximum correlation entropy kernel function non-Gaussian noise cubature Kalman filter integrated navigation

基金项目:

作者	单位
李伟	1.空军工程大学航空工程学院
郝顺义	1.空军工程大学航空工程学院
黄国荣	1.空军工程大学航空工程学院
马仕杰	1.空军工程大学航空工程学院

Author	Institution
Li Wei	1.College of Aeronautics Engineering, Air Force Engineering University
Hao Shunyi	1.College of Aeronautics Engineering, Air Force Engineering University
Huang Guorong	1.College of Aeronautics Engineering, Air Force Engineering University
Ma Shijie	1.College of Aeronautics Engineering, Air Force Engineering University

摘要点击次数: 457

全文下载次数: 667

中文摘要:

针对传统容积卡尔曼滤波器(CKF)在非高斯噪声下滤波精度下降以及传统最大相关熵(MCC)算法收敛速度较慢的问题,提出了一种改进的自适应相关熵高阶容积卡尔曼滤波(ADMCC-HCKF)算法。该方法依据 MCC 迭代过程的误差变化自适应调整核宽大小,核宽能够改变核参数对输入数据的敏感性,从而提高算法收敛速度及对非高斯噪声的处理能力。基于非高斯噪声环境,搭建 SINS / CNS / GNSS 组合导航实验,研究结果表明,改进的 ADMCC-HCKF 算法相比传统 HCKF 和基于常规 MCC 的 HCKF 算法具有更强的鲁棒性,在降噪性能及对非高斯噪声的适应性角度均有所提升的同时,滤波精度较 HCKF 算法提高了 9. 63%。

英文摘要:

Aiming at the problems that the decline of accuracy appears in traditional CKF under non-Gaussian noise and the slower convergence speed of the traditional MCC algorithm, an improved adaptive correlation entropy high-degree cubature Kalman filter algorithm (ADMCC-HCKF) is proposed. This method adaptively adjusts the kernel width according to the error changes of the MCC iteration process, kernel width can influence the sensitivity of the kernel parameters to the input data, thereby improve the convergence speed of the algorithm and the processing ability of non-Gaussian noise. Under the non-Gaussian noise environment, we build a SINS / CNS / GNSS integrated navigation experiment, the results show that under non-Gaussian noise conditions, the improved adaptive ADMCCHCKF algorithm shows stronger robustness than traditional HCKF and conventional MCC-HCKF, at the same time, it has batter noise reduction performance and resistance to non-Gaussian noise. In terms of filtering accuracy, compared with the HCKF algorithm, an average improvement of 9. 63%.

查看全文查看/发表评论下载PDF阅读器