符号算法在φ-混和输入情况下的收敛性

CONVERGENCE ANALYSIS OF THE SIGN ALGORITHM WITH φ-MIXING INPUT

本文在E.Eweda 2000年文章结果的基础上,用符号算法(SA)考察了一类自适应辨识系统。证明了在输入和噪声信号是φ-混合情况下,对充分小的步长和任意初始权向量,权数偏离的绝对均值具有时间平均渐进收敛性。

Based on result of E. Eweda(2000), this paper has considered the problem of adaptive plant identification using sign algorithm (SA). Asymptotic time-averaged convergence for the mean absolute weight misalignment was proved for sufficient small step size and any initial weight vector, with the input and noise sequences being mixing.