基于支持向量回归机的扩展卡尔曼滤波  

Extended Kalman Filter Based on Support Vector Regression

作　　者：申文斌[1] 裴海龙[1] 

机构地区：[1]华南理工大学自动化科学与工程学院,广东广州510640

出　　处：《计算机仿真》2011年第4期156-159,175,共5页Computer Simulation

基　　金：国家自然科学基金项目(60574004);国家自然科学基金资助重点项目(60736024);教育部科技创新工程重大项目培育资金项目(708069)

摘　　要：在信号滤波算法优化问题的研究时,扩展卡尔曼滤波算法的精度依赖于系统模型精确性。采用一种改进的扩展卡尔曼滤波算法研究了状态量和观测量相同的系统。用滤波后的状态量和当前观测量以得到实时噪声,求出过程噪声方差阵,在此基础上利用支持向量回归机算法对过程噪声方差阵进行训练,从而得到新的过程噪声方差阵,用此阵作为下一次扩展卡尔曼滤波的过程方差阵,以后继续上述迭代过程。结果证明方法极大的提高了滤波精度。仿真说明方法的有效性。Concerning the defect that the accuracy of the extended Kalman filter（EKF） depends on the accuracy of system model,in this paper,we use an improved EKF algorithm to research a system whose state variables are equal to observing variables.Firstly,we use the filtered state variables and current observing variables so as to obtain the real time noises,from which we then find the process noise variance matrix.On this basis,we train the process noise variances by using support vector regression（SVR） to acquire new ones,which we consider as the process variance matrix in the EKF next time.Next,we continue this iterative process.With this method,the accuracy of filter is improved greatly.At the end of this paper,we present some concrete simulative examples to show the validity of this method.

关 键 词：扩展卡尔曼 噪声方差阵 支持向量回归机 滤波精度 

分 类 号：TP2[自动化与计算机技术—检测技术与自动化装置]