论文部分内容阅读
提出了两种新的有效的最小二乘算法——改进的双对角化最小二乘算法ＭＢＬＳ－Ⅰ与ＭＢＬＳ－Ⅱ．在存在舍入误差的条件下，证明了算法的收敛性．该算法具有几乎不受舍人误差影响的优点，优于一般常用的最小二乘算法，包括数值性态极佳的ＳＶＤ算法．同时，基于该算法及ＳＶＤ算法，构造出了一种新的ＮＡＲＭＡＸ模型结构与参数辨识的一体化算法．仿真结果证明了此新算法的优越性．
Two new efficient least-squares algorithms are proposed-MBLS-Ⅰ and MBLS-Ⅱ, which are improved doubly-diagonalized least squares algorithms. Under the existence of rounding error, the convergence of the algorithm is proved. This algorithm has the advantage of being less affected by the error of tricking person, which is better than the commonly used least squares algorithm, including the SVD algorithm with excellent numerical performance. At the same time, based on this algorithm and SVD algorithm, a new integrated algorithm of structure and parameter identification of NARMAX model is constructed. Simulation results show the superiority of this new algorithm.