论文部分内容阅读
本文根据预报误差(RPE)算法的思想,对多输入-多输出 ARMAX 模型 A(z)y_n=B(z)u_n+C(z)ε_n 发展了一种新的辨识算法.由于在算法中加了矩阵多项式 C(z)的稳定性检验,这一种算法比推广最小二乘算法具有更快的收敛速度,这个结果在计算机仿真中得到了证实.
In this paper, a new identification algorithm is developed for the multi-input multi-output ARMAX model A (z) y_n = B (z) u_n + C (z) ε_n according to the idea of prediction error (RPE) The stability test of the matrix polynomial C (z), which is faster than the generalized least squares algorithm, has been proved. The result is verified by computer simulation.