Box-Jenkins模型阶次与参数同时估计的递推算法

Recursive algorithm for simultaneous identification of Box-Jenkins model order and parameters

研究了Box-Jenkins模型阶次与参数的同时估计问题。基于信息压缩阵的UD分解技术和广义增广最小二乘原理,提出Box-Jenkins模型阶次与参数同时估计的一种速推算法,减少了辨识计算量,改善数值稳定性,提高了辨识精度。仿真结果表明该算法的有效性。

The problem of simultaneous identification of Box-Jenkins model order and parameter are investigated. By using the UD-factorization of the augmented information matrix (AIM) and the generalized extend least squares (GELS), an recursive algorithm for simultaneous identification of Box-Jenkins model order and parameters are presented. The proposed results can reduce computation labor, improve numerical stability and precision of identification. Simulation shows the effectiveness of the algorithm.