| 引用本文: | 许 红， 刘欣蕊， 邢逸舟， 全英汇. 基于参数解耦的变分贝叶斯自适应卡尔曼滤波[J]. 雷达科学与技术, 2024, 22(3): 291-299.[点击复制] |
| XU Hong, LIU Xinrui, XING Yizhou, QUAN Yinghui. Variational Bayesian Adaptive Kalman Filtering Algorithm Based on Parameter Decoupling Method[J]. Radar Science and Technology, 2024, 22(3): 291-299.[点击复制] |
|
| 摘要: |
| 针对噪声协方差矩阵失配情况下的状态估计问题，本文基于变分贝叶斯框架，提出了一种适用于过程噪声协方差矩阵和测量噪声协方差矩阵均未知条件下的参数解耦的变分贝叶斯自适应卡尔曼滤波算法。所提算法选取预测误差协方差矩阵作为变分优化变量，并引入了其马尔可夫演化模型，构造了参数解耦的变分推断模型。同时，采用固定点迭代优化实现状态、预测误差协方差矩阵和测量噪声协方差矩阵的联合后验概率分布求解，并设计了算法的收敛性判断准则。仿真结果验证了算法的有效性。 |
| 关键词: 自适应状态估计 卡尔曼滤波 变分贝叶斯 噪声协方差矩阵 参数解耦 |
| DOI：DOI:10.3969/j.issn.1672-2337.2024.03.007 |
| 分类号:TN953；TN957.5 |
| 基金项目:国家自然科学基金（No.62301408）；博士后科学基金（No.2022M722503） |
|
| Variational Bayesian Adaptive Kalman Filtering Algorithm Based on Parameter Decoupling Method |
|
XU Hong, LIU Xinrui, XING Yizhou, QUAN Yinghui
|
|
1. Hangzhou Institute of Technology, Xidian University, Hangzhou 311231, China;2. School of Electronic Engineering, Xidian University, Xi’an 710071, China
|
| Abstract: |
| In the context of state estimation problems under mismatched noise covariance matrices, a parameter?decoupled variational Bayesian adaptive Kalman filter （PD?VB?AKF） algorithm is proposed in the paper within the framework of variational Bayesian （VB） method. The filter can be applicable when both the process noise covariance matrix （PNCM） and the measurement noise covariance matrix （MNCM） are unknown. The proposed algorithm chooses the predicted error covariance matrix （PECM） as the variable to optimize through variational techniques and introduces a Markov evolution model to construct the parameter?decoupled variational inference model. Furthermore, it utilizes the fixed?point iteration optimization to solve the joint posterior probability distribution of the state, PECM and MNCM, and outlines the convergence criteria of the algorithm. The simulation results validate the effectiveness of proposed algorithm. |
| Key words: adaptive state estimation Kalman filtering variational Bayesian noise covariance matrices parameter decoupling |
