多变量传递函数矩阵辨识的二级QR分解快速递推算法
A Fast Recursive Algorithm With Two-Grade QR Edcomposition for Multivariable Transfer Frnction Matrix Identification
-
摘要: 提出一种多变量传递函数矩阵辨识的二级QR分解算法,避免了子系统辨识方法 对可测数据的重复处理;在不扩大维数的条件下,获得使总体损失函数最小的估计值.该算法 与HOUSEHOLDER变换的快速递推算法结合,不仅大大地减少了辨识所需的运算量,而且 可减少LS算法中增益矩阵计算的误差积累和传递,提高辨识精度.Abstract: In this paper, a fast recursive algorithm with two-grade QR decomposition for multivarible transfer function matrix identification is presented. By using the algorithm, repeatedly processing of the measured data of subsystems identification algorithms can be avoided, and without expanding of parameter vector dimension, a parameter estimation that minimizes the overall cost function can be obtained by using this algorithm. Combined with the fast HOUSEHOLDER transformation, this algorithm has better numerical stability and can enhance the estimation accuracy.
计量
- 文章访问数: 2321
- HTML全文浏览量: 51
- PDF下载量: 1054
- 被引次数: 0